Scientific Foundations of Ophthalmology

FRS

SECTION I

STRUCTURE AND ULTRASTRUCTURE

1

THE AQUEOUS OUTFLOW PATHWAY IN VERTEBRATE EYES

RAMESH C. TRIPATHI

Publisher Summary

Despite wide variations in the mode of life and in the habitat of vertebrates, the close similarity in the ontogeny and in the basic features of their eyes is most remarkable. In all species, the cavity of the anterior chamber acts as a reservoir for a clear watery fluid—the aqueous humor—which through its hydrodynamics is responsible for maintaining the normal intraocular pressure. This, in conjunction with the fibrous tunics of the eyes, provides the stability of ocular dimensions for the performance of visual function. There is considerable evidence that in most vertebrates, the aqueous humor is continuously formed in the posterior chamber, largely as a secretory product from the cells of ciliary epithelium. The comparative morphology and physiology of the pathways responsible for continuous drainage of the aqueous humor in vertebrate eyes may be discussed under two main headings: the conventional drainage pathway and the accessory drainage routes. Available evidence suggests that the accessory drainage routes do not contribute significantly to the bulk outflow of aqueous. This chapter focuses on the conventional drainage pathway. Except for minor variations, there is a structural basis for the bulk drainage of aqueous humor throughout the vertebrate phylum, in the same way as there is for a bulk formation.

Introduction

Despite wide variations in the mode of life and the habitat of vertebrates (whether terrestrial, arboreal, amphibious or aquatic), the close similarity in the ontogeny and basic features of their eyes is most remarkable. In all species, the cavity of the anterior chamber acts as a reservoir for a clear watery fluid, the aqueous humour, which through its hydrodynamics is responsible for maintaining the normal intraocular pressure, and this, in conjunction with the fibrous tunics of the eyes, provides the stability of ocular dimensions for the performance of visual function. There is considerable evidence that in most vertebrates the aqueous humour is continuously formed in the posterior chamber, largely as a secretory product from the cells of the ciliary epithelium.

The comparative morphology and physiology of the pathways responsible for continuous drainage of the aqueous humour in vertebrate eyes may be discussed under two main headings:

The ‘conventional’ drainage pathway is responsible for the bulk drainage of the aqueous humour from the angle of the anterior chamber into the channels located in the limbal region of the eye. An obstruction in this pathway leads to raised intraocular pressure and the clinical condition of glaucoma.

The accessory drainage routes include uveoscleral drainage, diffusion along the iris vessels, posterior drainage (through the vitreous into the retina and optic nerve) and transcorneal flux. Available evidence suggests that these routes do not contribute significantly to the bulk outflow of aqueous and this chapter will be concerned only with the conventional drainage pathway.

As most information available relates to the eyes of mammals, and in particular those of primates, it is appropriate to discuss it in descending order of evolution.

Primates

In primate eyes, the conventional drainage pathway, entirely confined to the limbal region, consists of the trabecular meshwork, the canal of Schlemm, and the intra- and episcleral system of collector channels, including the aqueous veins. The angle of the anterior chamber is clearly defined, probably as a consequence of evolution and the massive development of the ciliary body, which forms a compact triangular structure (Fig. 1).

FIG. 1 Photomicrograph of ciliary and limbal regions of human eye in meridional section. Note clearly defined angle of the anterior chamber (AC), compact ciliary muscle (CM), ciliary epithelium, (CE). The aqueous drainage pathway consists of trabecular meshwork (TM), Schlemm’s canal (SC) and intrascleral collector channels (IS). (× 54.)

The trabecular meshwork, located in the inner limbus, spans the angle as a wedge-shaped band with the apex inserted into the peripheral termination of Descemet’s membrane and the deeper corneal lamellae, and the base connected to the scleral spur, the anterior face of the ciliary body and the iris root; hence the distinction of corneoscleral trabeculae, uveal trabeculae and iris processes (pectinate fibres) respectively.

The corneoscleral and outer uveal trabeculae are flattened, perforated sheets, orientated circumferentially, parallel to the surface of the limbus. The inner one or two layers of uveal sheets are, however, cord-like, and are orientated predominantly in a radial net-like fashion, enclosing large oval, circular or rhomboidal spaces. The thicker, radially orientated iris processes, reminiscent of the pectinate ligaments seen in lower mammals, are widely spaced. Although there are variations in the thickness of the trabeculae and in the proportion of the connective tissue components, histologically they are made up of collagen fibres together with elastic tissue, and a zone of basal-lamina material covered by a layer of flattened endothelial cells.

The trabecular meshwork forms a labyrinth of extracellular intercommunicating spaces of variable size and shape. In the normal eye, the ease of flow of aqueous humour through these spaces is indicated by the fact that dyes, colloidal suspensions and particulate matter of certain sizes can be readily traced when introduced into the anterior chamber. From the trabecular meshwork, the aqueous humour passes into the canal of Schlemm, a circumferential vessel, elliptical in shape and located between the compact corneosclera and the trabecular meshwork. Its trabecular wall is composed of a zone of cell-rich supporting tissue or endothelial meshwork of variable thickness which intervenes between the outermost corneoscleral trabecular sheet and the endothelial lining of the canal (Fig. 2). The extracellular spaces in this region appear to be narrower than in the remainder of the trabecular meshwork, and it is thought that this region may account for some resistance to aqueous outflow.

FIG. 2 Survey electron micrograph of the trabecular wall of Schlemm’s canal (SC) of normal human eye in meridional section. The endothelial lining (N, cell nuclei) is characterized by a single layered membrane, many cells of which contain giant vacuoles (V). OS, open extracellular spaces of the supporting tissue zone or endothelial meshwork. (× 4375.

The structure and function of the canal of Schlemm has been the subject of much debate amongst morphologists, physiologists and clinicians, especially as to whether or not the outflow of aqueous humour across the trabecular wall of Schlemm’s canal takes place via pores (the ‘open’ and ‘closed’ systems respectively). These apparently opposing concepts probably originated a century ago. The supporters of the ‘open’ system concept believed that there were direct communications across the endothelial lining of Schlemm’s canal of a size discernible by light microscopy. Consequently, the possibility of the openings being tubules lined by endothelium, intercellular gaps or some undefined pores was considered. Histological studies with and without injection experiments, however, failed to clarify the size and the nature of the openings. The advocates of the closed system concept, on the other hand, have insisted that the endothelial lining of Schlemm’s canal is a continuous membrane and that the drainage of aqueous humour must therefore depend upon seepage, some form of passive filtration or even an active transport mechanism. To resolve these apparent contradictions and with the advent of electron microscopy, a number of ultrastructural studies have been undertaken to clarify the functional morphology of the exit pathway of the aqueous.

Ultrastructural studies of the endothelial lining of the canal of Schlemm have now clearly shown it to be characterized by a single layer of cells resting on a tenuous and interrupted basal lamina (Fig. 2). Adjacent cells are joined by poorly defined ‘tight’ junctions which occupy only a small area of the cell surface. Schlemm’s canal shows many tortuous blind channels or diverticula lined by endothelium and extending for a variable distance into the meshwork.

Scanning electron microscopy of the endothelial lining of the trabecular wall of Schlemm’s canal clearly reveals the individual cells to be generally spindle-shaped with a central bulge and tapering rounded ends (Fig. 3). The long axis of each cell is usually orientated parallel to the circumference of the canal. The cells measure some 40–100 μm in length and 5–12 μm in width, centrally. Ultrathin sections of the cells examined by transmission electron microscopy show them to contain the usual intracellular organelles: a centrally located oblong nucleus responsible for an apical bulge in the cell, a moderate number of mitochondria, smooth- and rough-surface endoplasmic reticulum, Golgi apparatus, membrane-bound dense bodies, multivesicular bodies, centrioles, glycogen and free ribonucleic acid granules. The cells are rich in fine cytoplasmic filaments, orientated along their long axis.

FIG. 3 Scanning electron micrograph of the endothelial lining of the trabecular wall of Schlemm’s canal viewed from the luminal aspect. Note the spindle-shaped appearance of the cells, their long axes being parallel to the canal circumference. The central bulges correspond to the location of cell nuclei and giant vacuoles. The asterisk denotes a collapsed vacuole. (× 2000.)

A specialized feature of many lining endothelial cells is the presence of giant vacuoles (Fig. 2), which over the past decade has aroused wide interest. The possibility that they were artefacts has also been considered by many workers, but it is now established that the vacuoles are a real morphological entity, and not a product of pathological or post-mortem change, since they are found in fresh ante-mortem tissue and also following in situ fixation by a variety of reliable techniques. Electron microscopy of normal eyes clearly reveals that the vacuoles are several micrometres in size, are membrane-bound, and essentially electron-optically empty; and when centrally located within the cell tend to indent the nucleus (Fig. 2).

Serial section analyses reveal that at a given time the majority of the vacuoles show a basal opening towards the trabecular aspect ranging in size from 0·1 to 3·5 μm. It is therefore envisaged that, in life, the vacuoles are in direct communication with the aqueous humour in the trabecular meshwork. A small proportion of vacuoles, however, show in addition apical (luminal) openings ranging in size from 0·1 to 2·5 μm and are thus interpreted as constituting a vacuolar transcellular channel (Tripathi, 1971a,b, 1974).

Using a variety of tracers (colloidal suspensions, graded microspheres, whole blood, etc.) introduced into the anterior chamber at a physiological pressure, it can be shown that the vacuoles readily become filled with tracer material (Tripathi, 1971a,b; 1974). Figure 4 shows the passage of an erythrocyte through a vacuolar structure. The filling of the vacuoles with colloidal ferritin serum solution, and its passage through a vacuolar transcellular channel into the lumen of the canal, is seen in Fig. 5. In our experimental studies, however, no leakage through the intercellular route was seen. Although consideration must be given to the possible role of micropinocytotic vesicles in the transfer of fluid, their contribution to the bulk drainage of aqueous humour would seem to be small, since during the experiment (approximately thirty minutes) tracer particles of colloidal dimensions were only rarely seen in these caveolae. In any case, this mode of transport cannot account for the rapidity and size of the tracer elements which leave the anterior chamber and can be recovered from the exit channels. In the absence of any other direct openings, the vacuolar structures are considered as having a significant role in the bulk outflow of aqueous humour across the endothelial barrier of Schlemm’s canal (Tripathi, 1971a,b; 1974).

FIG. 4 Electron micrograph of the endothelial lining of Schlemm’s canal (SC) showing filling of giant vacuoles (V) following intracameral injection of ferritin serum solution. Note the passage of the tracer into the canal through a vacuolar transcellular channel (arrows denoting basal (bottom) and apical (luminal) openings). The intracellular junctions (Z) are intact, (× 16 250.)

FIG. 5 Selected electron micrographs of serial sections of a vacuolar transcellular channel with an erythrocyte in transit, (a) Part of the erythrocyte (E) projects into the vacuole through its basal opening (arrow); (b) protrusion of a membraneous structure (probably a ghost erythrocyte) is seen through the apical opening of the vacuole (arrow). SC, Schlemm’s canal. (× 10 000.)

The passage of fluid and particulate matter from Schlemm’s canal into the intra- and episcleral system of collector channels, including the aqueous veins, is by laminar flow.

Lower Mammals

In comparison to that of primates, the angle of the anterior chamber in the lower mammals is not clearly defined, because the ciliary muscle is poorly developed and is divided into two leaves, one of which lies against the inner aspect of the sclera while the other is continuous with the fibrous baseplate of the ciliary body. The resultant cleft is criss-crossed by fibrocellular strands (the trabeculae), the most prominent of which extend from the root of the iris to the inner aspect of the peripheral cornea and are known as pectinate ligaments. The channels responsible for the drainage of aqueous humour (the angular aqueous plexus)* from the intercommunicating spaces of the angular meshwork* are analogous to the canal of Schlemm of primates, and are located between the angular meshwork and the compact tissue of the corneosclera. The comparative morphology of the angle of a few representative lower mammals (placentals) is illustrated in Figs. 6–8. A comparable anatomical arrangement is also seen in the angle in non-placental mammals (the marsupials (Fig. 9) and the monotremes).

FIG. 6 Photomicrograph of meridional section of the eye of a lower mammal showing morphological organization of angular region: dog (× 46) AC, anterior chamber; PL, pectinate ligament; CC, ciliary cleft; CM, ciliary muscle; AP, angular aqueous plexus; CE, ciliary epithelium. (See Figs. 7a, 7b, 8, and 9.)

FIG. 7 (a) Electron micrograph of an obliquely cut endothelial cell lining the angular aqueous plexus (× 5000). N, nucleus; V, giant vacuole with a basal opening; AP, angular aqueous plexus, (b) Photomicrograph of meridional section of the eye of a lower mammal showing morphological organization of angular region: horse (× 56). AC, anterior chamber; PL, pectinate ligament; CC, ciliary cleft; CM, ciliary muscle; AP, angular aqueous plexus; CE, ciliary epithelium. (See Figs. 6, 7a, 8, and 9.)

FIG. 8 Photomicrograph of meridional section of the eye of a lower mammal showing morphological organization of angular region: rabbit (× 189)—inset: electron micrograph (× 9100) of the endothelial lining of the angular aqueous plexus showing filling of a giant vacuole following intracameral injection of thorotrast. Z, intact cell junctions; AC, anterior chamber; PL, pectinate ligament; CC, ciliary cleft; AP, angular aqueous plexus; CE, ciliary epithelium. (See Figs. 6, 7a, 7b, and 9.) At this level a basal opening is not seen.

FIG. 9 Photomicrograph of meridional section of the eye of a lower mammal showing morphological organization of angular region: kangaroo (× 75)—inset: high power photomicrograph (× 1200) showing the presence of giant vacuoles (V) in the endothelial lining of the angular aqueous plexus. AC, anterior chamber; PL, pectinate ligament; CC, ciliary cleft; CM, ciliary muscle; AP, angular aqueous plexus; CE, ciliary epithelium.

In all these species two main features are noteworthy:

(1) The angular aqueous drainage channels form a plexus rather than a single vessel.

(2) Part of this plexus is often completely surrounded by the tissue of the angular meshwork, and recalls the appearance of the channels or diverticula described earlier in the canal of Schlemm of primate eyes.

It is to be noted further that the angular aqueous plexus in all these creatures seems to have a common microscopic appearance, its most prominent feature being the continuous lining and the presence of giant vacuoles in the lining endothelial cells, exactly comparable to those found in Schlemm’s canal in primates (Fig. 7a and insets in Figs. 8 and 9). The comparability of these vacuolar structures and their unit membrane-bound profiles are unequivocally demonstrated by electron microscopy. Serial section analyses of giant vacuoles reveal that at a given time a proportion of vacuoles constitute vacuolar transcellular channels by virtue of having basal and luminal openings, a situation exactly comparable to that seen in primates.

Similarly, the tracer studies show that following intracameral injection at a physiological pressure, the vacuoles readily become filled with the tracer substance, and its leakage into the lumen of the angular aqueous plexus through the vacuolar transcellular channels is seen. Functionally the intercellular junctional complexes of the lining endothelial cells offer an effective barrier to the leakage of colloidal suspensions, of particle size approximately 10 nm.

As in the canal of Schlemm in primates, the angular aqueous plexus is directly connected anteriorly with the intra- and episcleral venous channels, and in some animals aqueous veins have also been identified. In certain animals, especially carnivores, however, one of the intrascleral venous channels is of enormous calibre and is known as the circle of Hovius: its significance is not clearly understood.

Birds

In common with the lower mammals, the angle of the anterior chamber in birds shows the presence of a ciliary cleft criss-crossed by trabeculae which are remarkably rich in elastic tissue (Fig. 10). The angular aqueous plexus is a large sinus-like structure, which may be divided into two or three parts by an intervening artery and its branches. The endothelial lining of the sinus shows features closely comparable to the angular aqueous plexus in mammals, namely, the presence of a continuous membrane and giant vacuoles in many lining endothelial cells of the trabecular wall (Fig. 10 inset). Figure 11 (a and b) shows the incidental finding of a macrophage passing through a vacuolar structure which forms a transcellular channel. Tracer studies in avian eyes have shown results closely comparable to those in mammalian eyes, that is, the passage of electron-dense tracers through vacuolar structures and no leakage across the intercellular junctions.

FIG. 10 Photomicrograph of meridional section of angular region in pigeon eye (× 52). AC, anterior chamber; PL, pectinate ligament; CC, ciliary cleft; AP, angular aqueous plexus intervened by an artery (A); CE, ciliary epithelium; SO, scleral ossicle. Inset: electron micrograph (× 5400) of the inner wall of the aqueous plexus. V, giant vacuole with basal opening (arrow) in a lining cell; N, bulging nucleus of adjacent lining cell.

FIG. 11 Selected electron micrographs of serial sections cut meridionally through a vacuolar transcellular channel (V) with a macrophage (M) in transit. AP, angular aqueous plexus; Z, cell junction. (× 7910.)

Lower Vertebrates (reptiles, amphibians and fishes)

In common with birds and lower mammals, reptiles and most amphibians show a ciliary cleft in the angular region. It is bridged similarly by fibrocellular strands or the trabeculae of the angular meshwork (Figs. 12 and 13). In some fishes (e.g. the goldfish), however, the greater part of the iridocorneal angle may be entirely occupied by organized mesothelial cells (the ‘annular ligament’—an inappropriate term). Regional variations are also frequently seen and these seem largely related to the existence of a defined angular aqueous plexus in the corresponding region (Tripathi, 1974).

FIG. 12 Meridional section of the angular region of terrapin (× 67)—inset: high power photomicrograph (× 840) of the angular aqueous plexus, showing the presence of giant vacuoles (V). AC, anterior chamber; CC, ciliary cleft; CM, ciliary muscle (tensor choroidae); AP, angular aqueous plexus; SO, scleral ossicles.

FIG. 13 Meridional section of the angular region of a toad (× 124). Note also the presence of protractor lentis (PtL) muscle at this level of section. AC, anterior chamber; CC, ciliary cleft; CM, ciliary muscle (tensor choroidae); AP, angular aqueous plexus; CE, ciliary epithelium.

In some reptiles the angular aqueous plexus forms an incomplete ring. It is present only in the dorsal and ventral segments of the angle in amphibians, and only in a localized ventral segment in many fishes (Figs. 14 and 15). The endothelial lining of the angular aqueous plexus in all these species has features in common with that of higher vertebrates, especially with regard to the presence of vacuolar structures in the endothelial lining cells (Figs. 12 inset, 14b and 15 inset). Morphological and tracer studies have shown comparable results to those described previously, namely the passage of tracers across the endothelial barrier through vacuolar structures constituting transcellular channels (Tripathi, 1974).

FIG. 14 (a) Photomicrograph of a vertical section through the anterior segment of the eye of a trout (× 11). Note that the angular aqueous plexus is confined to the ventral segment of the angle (left) corresponding to the location of companula (Cp). The angular drainage system is partly accommodated by a localized corneal thinning. (b) Photomicrograph of meridional section passing through the ventral segment of the angular region of trout eye (× 287) AC, anterior chamber; AM, angular meshwork; AP, angular aqueous plexus; V, giant vacuoles in the endothelial lining.

FIG. 15 Photomicrograph of meridional section passing through the ventral segment of the angular region of the eye in the dogfish. Inset: electron micrograph (× 7250) of the inner wall of the angular plexus (AP) in dogfish. Arrow denotes the basal opening of the giant vacuole. AM, angular meshwork; AC, anterior chamber.

Conclusion

From the ultrastructural and physiological studies of a wide variety of vertebrates, it now seems clear that despite certain variations in the morphological organization of the iridocorneal angle, normally the aqueous humour flows freely through the angular meshwork. The presence of a mucinous substance in this region, for which there is more evidence in lower vertebrates, is said to influence the rate of flow (see Tripathi, 1974). The main structural barrier is, however, created by the intact endothelial lining of the angular aqueous plexus. The pathway for the bulk outflow of aqueous humour is through transcellular channels formed by vacuolar structures (Tripathi 1971a,b; 1974). A possible sequence of events is as follows (Fig. 16):

FIG. 16 Diagrammatic representation of the endothelial vacuolation cycle illustrating the formation of a dynamic system of pores and the mechanism of the bulk outflow of the aqueous humour across the endothelial barrier of the angular aqueous plexus. The earliest vacuole probably forms as an infolding of the plasma membrane on the basal aspect of the lining endothelial cells (stage 2). A progressive enlargement of this infolding leads to the formation of a giant vacuole (stages 3 and 4) which, by opening on the apical aspect of the cell surface (stage 5), creates a vacuolar transcellular channel (continuous arrow). The transcellular channels in fact provide the pathway for the bulk outflow of the aqueous humour down a pressure gradient (i.e. from the angular meshwork to the lumen of the aqueous plexus). After a certain time interval, the basal opening is resealed and the cell returns to its non-vacuola ted state (stage 1) (After Tripathi, 1971).

The vacuole probably forms initially as a membranous depression or indentation on the basal surface of the endothelial cell. This gradually enlarges and eventually opens on the apical aspect of the cell, thereby creating a transient transcellular channel. Finally, the cell cytoplasm moves towards the basal aspect of the cell to re-seal the opening and the cell returns to its non-vacuolated stage. It is thus conceived that the vacuoles with basal openings are only stages in the formation of temporary transcellular channels. The bulk outflow of aqueous humour from the angular meshwork across the endothelial barrier of the angular aqueous plexus takes place down a pressure gradient through the giant vacuoles forming transcellular channels, that is, having both basal and apical openings. Our impressions are, however, that at a given time only the requisite number of vacuolar transcellular channels are formed, and this probably accounts for the observed resistance to aqueous outflow across the endothelial barrier of the angular aqueous plexus. Hence this is an important factor in the maintenance of intraocular pressure. This concept would seem to bridge the gap between those who believed in pores and those who did not.

From the morphological configuration of transcellular channels (that is, a variable tortuosity of their course; their relatively large basal openings compared with the apical openings, the latter being located at the convex surface of the vacuole, and not usually opposite the basal opening; and the presumed modulating influence of the endothelial cells as indicated by the abundance of cytoplasmic filaments) and our experimental studies, it would appear that the vacuolar transcellular channels probably act as oneway valves, predominantly allowing fluid to pass through them from base to apex far more easily than in the reverse direction (Tripathi, 1974).

The many factors possibly responsible for the initiation and maintenance of the endothelial vacuolation cycle, such as the role of a hydrostatic pressure gradient across the endothelial barrier of the angular aqueous plexus, the rate of flow, the presence of mucinous substances, the composition and physical properties of the aqueous reaching the endothelial lining, the special functional property of the lining cells, neuronal control, and ocular haemodynamics, have been discussed in previous publications (Tripathi, 1971a,b; 1974).

In summary, it may be stated that except for minor variations, there is a structural basis for the bulk drainage of aqueous humour throughout the vertebrate phylum, in the same way as there is for a bulk formation. The normal presence of giant vacuoles in the endothelial lining of the aqueous plexus in species as diverse as dogfish, pigeon and man would suggest that the bulk outflow of aqueous humour by the dynamic system of vacuolar transcellular pores is a fundamental biological mechanism of enormous antiquity, and that this process probably holds the secret of the maintenance of aqueous dynamics in anatomically closed cavities.*

REFERENCES

Duke-Elder, Sir, Stewart, The eye in evolutionSystem of ophthalmology, Vol. 1. London: Kimpton, 1958.

Rochon-Duvigneaud, A.Les yeux et la vision des vertébrés. Paris: Masson, 1943.

Tripathi, R.C. Mechanism of the aqueous outflow across the trabecular wall of Schlemm’s canal. Exp. Eye Res. 1971; 11:116.

Tripathi, R.C. The aqueous outflow in normal and glaucomatous eyes. Brit. J. Ophthal. 1971; 56:157.

Tripathi, R.C., Comparative physiology and anatomy of the aqueous outflow pathwayDavson, H., eds. The Eye, Vol. 5. London: Academic Press, 1974.

Tripathi, R.C., Tripathi, B.J. Vacuolar transcellular channels as the outflow pathways of cerebrospinal fluid. J. Physiol. (Lond.). 1974; 239:195.

*In introducing the terms angular aqueous plexus and angular meshwork, I have aimed at an etymological nomenclature that may be widely applicable to vertebrate eyes (see Tripathi, 1974b).

*A close similarity in the morphology and physiology of the cellular lining of the angular aqueous plexus and that of the arachnoid villi of the brain, especially with regard to the normal presence and functional role of the vacuolar structures, has recently been shown for the first time (Tripathi, 1973; 1974; Tripathi and Tripathi, 1974).

2

THE RETINAL RECEPTORS AND THE PIGMENT EPITHELIUM

J. MARSHALL

Publisher Summary

This chapter discusses the morphology of the retinal receptors and the pigment epithelium and some of the relationships that exist between them in the human retina. One of the most striking aspects of retinal morphology is the uniformity of stratification. Throughout the vertebrates, the retinal matrix is composed of three major classes of the neuronal cell: the receptor cells, the intermediate neurones, and the ganglion cells. It is the anatomical juxtaposition of adjacent regions of similar cells that results in a complex layered structure. The extent of individual layers varies between species; however, in all retinas, eleven distinct layers may be recognized. Receptor cells are divided into two groups: rods and cones. The microanatomy of rods and cones conforms to a basic pattern, and both receptors may be subdivided into four analogous regions. Thus, the archetype photoreceptor has a transductive region, a region for the maintenance of cellular homeostasis, a nuclear region, and a transmissive region. The detailed morphology of analogous regions in rods and cones may be more or less similar, depending on the species. However, in the human retina, two cell populations are clearly distinguishable both functionally and morphologically.

Introduction

It is the purpose of this chapter to describe the morphology of the retinal receptors and the pigment epithelium, and some of the relationships which exist between them in the human retina. Although some comparison will be made with other mammals there will be no discussion of species differences or nomenclature within the wider limits of the vertebrate subphylum. Readers requiring information on this broader topic are referred to the excellent review by Crescitelli (1972), and those requiring detailed topographic anatomy are referred to Polyack (1957).

One of the most striking aspects of retinal morphology is the uniformity of stratification. Throughout the vertebrates the retinal matrix is composed of three major classes of neuronal cell, the receptor cells, the intermediate neurones (bipolar, amacrine and horizontal cells), and the ganglion cells. It is the anatomical juxtaposition of adjacent regions of similar cells which results in the complex layered structure. The extent of individual layers varies between species, but in all retinas eleven distinct layers may be recognized. The retinal receptor cells contribute to four of these layers (Fig. 1).

FIG. 1 Photomicrograph of a transverse section of an extramacular region of monkey retina. The layers are: (1) inner limiting membrane (2) nerve-fibre layer (3) ganglion-cell layer (4) inner plexiform layer (5) inner nuclear layer (6) outer plexiform layer (7) outer nuclear layer (8) outer limiting membrane (9) inner segments (10) outer segments (11) pigment epithelium. The receptor cells contribute to layers 10, 9, 7 and 6, and thus 10 to 6 is the extent of a single cell. (× 800.)

Receptor Morphology

Classically receptor cells are divided into two groups, rods and cones. These structures were named by the early microscopists from the geometric appearance of the cells; they are now used by physiologists to discriminate between two populations of cells with different functions. Rods are thought to be responsible for scotopic vision, and cones to be the photopic or chromatic receptors. This assignment of functional capacities to cells on the basis of crude morphological data has led to some confusion in the literature, because in many species convergent evolution has resulted in receptor cells with demonstrably different functions but similar morphology (Pedler, 1965).

The micro-anatomy of rods and cones is seen to conform to a basic pattern, and both receptors may be subdivided into four analogous regions (Fig. 1). Thus the archetype photoreceptor has a transductive region (outer segment), a region for the maintenance of cellular homeostasis (inner segment), a nuclear region (outer nuclear layer), and a transmissive region (the outer plexiform or synaptic layer). The detailed morphology of analogous regions in rods and cones may be more or less similar, depending on the species. However, in the human retina two cell populations are clearly distinguishable both functionally and morphologically.

In freshly fixed human retina, rod cells measure approximately 120 μm from the top of the outer segment to the pedicle surface. Cones appear to measure only 75 μm in the periphery, but this may be an artefact (p. 10). The exact dimensions of receptor cells are difficult to determine because their inner and outer segments are usually observed to be at an angle between the outer limiting membrane and the surface of the pigment epithelium. The mechanism underlying the angular displacement is still under discussion. While some authors claim that the observed displacement is a preparative artefact induced by histological processes, other authors suggest that the receptor outer segments are orientated towards the pupil or the posterior nodal point of the lens and that this enhances light capture by the photoreceptors. Measurement of photoreceptors is further complicated at their distal end by the fovea and the fibre layer of Henle (p. 13), which cause an angular displacement in the inner connecting fibre of both rods and cones some distance from the macular region.

Outer Segments

Rods.

Rod outer segments are approximately cylindrical, in width 2–5 µm and length 30 μm. In preparations examined under the light microscope rod outer segments appear uniformly densely stained; in occasional preparations pale transverse streaks or bands may be observed. The latter phenomenon occurs when a retina has experienced osmotic shock or has been damaged in some way, and is an artefact resulting from the separation of rod discs.

When rod outer segments are sectioned parallel to their long axis and examined under an electron microscope, they are seen to consist of a boundary or cell membrane which encloses a stack of membrane systems (Fig. 2). By sectioning the outer segments at right angles to their long axis or by using special ultrasonic techniques to rupture the boundary membrane, the enclosed stack of membrane systems is seen to originate from the cut edges of structures like hollow coins, the rod discs. Each disc consists of a boundary membrane and an enclosed space, the intradiscal space. Towards each edge of the flattened disc the membranes are further apart, increasing the intradiscal space (see Fig. 2).

FIG. 2 (a) Electron micrograph of a transverse section of a human rod cell, showing the junction between the inner and outer segments. The inner segment contains mitochondria (M) and is connected via the cilium (C) to the outer segment (S). (× 16 000.) (b) Electron micrograph of a transverse section through the outer segment of a monkey rod cell. The disc membranes appear as closely appositioned pairs with a smaller intradiscal space (1) than the interdiscal space (2). The discs are all isolated from the cell membrane (3). (× 25 000.) (c) A single rod disc isolated from a rat rod outer segment and photographed looking down the long axis of a receptor cell. The incisure (arrowed) is the region of the disc which is in association with the cilium. In man and monkey the rod discs are crenated or lobed. (× 29 000.) (Courtesy of Dr C. M. H. Pedler.)

The discs are isolated from the boundary membrane and from each other. The space between each disc is slightly more electron-dense than the intradiscal space and is called the interdiscal space. Each rod contains between 600 and 1000 discs, and in man the edges of the discs may be crenated or lobed. The discs are the structures with which the photopigment rhodopsin is associated.

The exact way in which the rhodopsin molecules are orientated within this system is in debate (p. 182), as also are the ways in which the light-induced changes in the rhodopsin molecule are transmitted to the boundary membrane. Even the dimensions of the discs are in dispute, and electron microscopists (Hogan, Alvarado and Weddell, 1971), and protagonists of X-ray diffraction (p. 182), each produce different figures (Table 1). It is well known that the discs behave rather like osmometers, and that the relative dimensions of the inter- and intradiscal spaces vary with the osmotic strength of the surrounding medium. In view of this, measurements based on material processed for electron microscopy are unreliable because of the ionic imbalance which this causes. Recent experiments using glutaraldehyde, polymerized by a condensation reaction with urea as an embedding medium, have resulted in disc measurements during electron microscopy identical to those determined by X-ray diffraction. These measurements must be considered more reliable than those following conventional tissue processing, as the glutaraldehyde-urea system embeds material without loss of water or of water- and lipid-soluble components.

TABLE 1

COMPARISON OF THE DIMENSIONS OBTAINED BY X-RAY DIFFRACTION AND ELECTRON-MICROSCOPE STUDIES ON THE DISC MEMBRANES OF ROD OUTER SEGMENTS

It has been stated that the rod discs are isolated from the boundary membrane; however, towards the end of the outer segment adjacent to the inner segment small disclike structures are often observed with membranes continuous with the plasma membrane of the cell. Such systems are found in human rods at all ages and suggest that discs are formed by the convolution and ingrowth of the boundary membrane, forming a sac which subsequently buds from the parent membrane and fuses to form the hollow disc structure. Thus the intradiscal space originates in the extracellular space. It is dangerous to deduce temporal events from the static spatial analysis provided by electron microscopy, but Young (1969) has demonstrated that the discs in rods of many species, including the monkey, are continuously renewed throughout life. He has shown that new discs are formed in the region of the inner segment and that they are progressively displaced towards the pigment epithelium. The time taken for a newly formed rod disc to traverse the outer segment varies between species, but in the monkey it is approximately nine to thirteen days. Young also found evidence that material of receptor origin was being taken up into the pigment epithelium. He concluded that rod discs have a limited life and that spent discs are lost to the pigment epithelium (p. 17). An equilibrium therefore exists in this system and there is no change in the length of the receptor outer segment during adult life.

Cones.

Cone outer segments vary in size and shape with retinal location, from thin cylindrical rod-like structures at the fovea to squat conical forms in the far periphery.

Extrafoveal Cones.

The squat conical outer segment described by some early histologists as too short to have any association with the pigment epithelium is an artefact. Observations on 1 μm sections cut from resin-embedded material have shown that many of the outer segments which in conventional wax histology would have been classified as rod cells are in fact cones. Examples of such an observation are shown in Fig. 3, where the diameter of the conical outer segment becomes reduced in thickness until, about half-way along its length, it assumes the dimensions of a rod outer segment. At this point the cone outer segment often becomes very constricted and therefore, due to the thinness of histological sections, the proximal and distal portions of the outer segments appear unrelated.

FIG. 3 Photomicrograph of the junction between the receptor cells and the pigment epithelium in an extrafoveal region of a monkey retina. Most of the cones (C) show a point of inflection (arrowed) close to the ciliary region of the surrounding rods (R). This together with an undulation in the thickness of cone cells gives rise to the apparently conical outer segment. (× 1470.)

In preparations examined by electron microscopy cone outer segments are seen to contain membrane systems similar to those of rods, but to differ in their dimensions. Cone disc membranes usually have wider intra- and interdiscal spaces than those of rods, and in sections at right angles to the long axis of the outer segment do not show the crenation of their boundary membranes. In the recent literature there has been much discussion as to the nature of the intradiscal space in cones. In a series of experiments in which retinas were immersed in solutions which could subsequently be induced to give electron-dense salts, Cohen (1970) was able to demonstrate that in cones, but not rods, reaction product was found in the intradiscal space. Cohen concluded that in the species he examined cone discs were not isolated structures but merely extensive infoldings of the boundary membrane of the cell. If this is so, then clearly the interdiscal space is continuous with the extracellular space.

In some species these observations have been verified by electron microscopy; for example, in the pigeon and frog the boundary membrane of many cones can be seen to be a continuous corrugated system incorporating the cone discs. Also, electrophysiological experiments on many of the lower animals suggest that such a system is necessary to account for the observed recordings and projected ionic fluxes (p. 188).

Further circumstantial evidence for this basic difference between rods and cones comes from the experiments of Young (1971). In his autoradiographic studies Young found that in rods a discrete band of radioactive protein material was incorporated into the cilium region of outer segments, whilst in cones very much less labelled material was incorporated and was diffused throughout their outer segments. This diffuse distribution remains static. Two conclusions have been drawn from these experiments, first that the diffuse distribution represents uptake into a single membrane system, and secondly that the cone membrane system is static. Young (1971a) had expanded on this second conclusion and states that the conical shape of the cone outer segment arises from the limited development of the cone membrane. The small discs at the tip of the cone outer segment are those formed during early development and progressively larger discs are added throughout the growth period. When the cone stops growing during the early postnatal months the total membrane system is preserved and static.

Young has also found similar differences in the ways in which rods and cones incorporate labelled proteins in the retina of the monkey. However, both human and monkey cones always appear to have a boundary membrane and to contain isolated disc structures, and continuity of these membranes is rarely observed. In all retinas from these species, cone disc membranes are much less uniform than those of rods. Most cones show membrane discontinuities at both ends of their outer segments (Fig. 4). At the ciliary end, disc membranes may be observed in sections as a disorganized array of vesicular or tubular structures, whilst towards the pigment epithelium disc membranes often become disorientated and misaligned and again degenerate into vesicular configurations. The extent and distribution of disc membrane loss appears to be a phenomenon of ageing, and in retinas from aged individuals ‘finger-print’ whorl-like disturbances of disc membranes and vesicular degenerations may be observed throughout the cone outer segment. These degenerations are particularly marked in the outer segments of foveal cones.

FIG. 4 Electron micrograph of a human extrafoveal cone showing the junction between the inner (M) and outer (S) segments. The disorganized membranes of the outer segment are typical of cones, as also are the tubular or vesicular formations (arrowed). The cilium is not seen in the plane of this preparation. (× 21 350.)

Foveal Cones.

The outer segments of cones in the human fovea are rod-like in shape. Typically these receptors are long thin cylindrical structures approximately 2 µm by 45 µm. In most preparations they do not show any angular displacement in the foveola, although a common artefact observed in many preparations is a corrugation of these organelles resulting from handling and shrinkage during preparation.

When examined in the electron microscope the outer segments of foveal cones are seen to have an ultrastructure similar to that of extrafoveal cones, except that the diameters of the discs do not vary throughout their length. Degenerate or misaligned membrane systems are extremely common in foveal cones and their number increases with age.

The Cilium.

At the junction of the inner and outer segment the cell body of both rods and cones constricts to about 0·3 µm. In light-microscopic preparations the two portions of the cell appear to be separated; however, using the electron microscope a connecting cilium may be distinguished. This structure is always eccentric and in the majority of receptor cells provides the only link between the inner and outer segments (Fig. 2). The cilium contains nine pairs of tubules arranged in a ring, and these run along its axis. There is no central pair of tubules in the receptor cilium as there is in mobile cilia. This structure appears to be anchored in the inner segment by ciliary rootlets (or root filaments) which can be observed as a condensation of filaments traversed by striations with a 70nm periodicity. In the outer segment the paired tubules rapidly converge to form a single tubule, and these penetrate the outer segment to varying distances.

The function of the cilium is uncertain although, again, evidence from Young’s autoradiographic studies suggests that it may represent a transport channel between the manufacturing processes in the inner segment and the energy-requiring systems in the outer limb. This region of the cell is further complicated because in some receptor cells the inner and outer segments are connected by a broad and well-developed cytoplasmic bridge as well as a cilium. If material is still transported through the cilium in such receptors then it must be a preferred channel.

Inner Segments

The inner segments of both rods and cones are divided into an outer ellipsoid which terminates in the cilium, and an inner myoid which runs into the outer nuclear layer. Biochemical and autoradiographic evidence suggests that most of the homeostatic mechanisms necessary for the maintenance of the receptor cell, and in particular of the outer segment, occur in the inner segment.

Rods.

The inner segments of rods are short cylinders, slightly wider than the outer segments. In the light microscope they tend to stain less intensely than the outer segments and in badly fixed preparations show small vacuolated areas. They may be clearly distinguished from cone inner segments both by their shape and their staining responses to techniques such as Mallory’s triple stain, with which rod inner segments stain blue whilst those of cones stain red.

In the electron microscope, rod ellipsoids are seen to contain many large mitochondria, and these are almost always orientated along the long axis of the cell. The cytoplasm also contains ribosomes, neurotubules, some smooth-surfaced endoplasmic reticulum, and glycogen. In some pathological conditions the glycogen deposits increase and large rosette forms are sometimes found. This phenomenon is particularly noticeable in the retina of animals subjected to prolonged exposure to light.

The exact point of transition between the ellipsoid and the myoid is clearly difficult to determine. However, in most receptors the myoid contains few if any mitochondria and is usually the region in which the receptor cell’s Golgi bodies are situated. Round the circumference of the cell in this region the cell membrane sends out many small projections or villi. These processes enter into contact with processes from adjacent rod receptor cells and there is now some evidence from electrophysiology that in rods these contacts are functional and may account for some adaptation in the peripheral retina. In other words, electrical information may be exchanged between receptor cells at a level other than that of their synapses.

Cones.

The ellipsoids of extrafoveal cone cells are much larger than those of rods, and are conical, tapering towards the outer segment. They are packed full of mitochondria and contain many more than those of rods. Within the fovea the cone ellipsoids are rod-like and cylindrical. The myoids of human cones are similar to those of rods but tend not to taper as abruptly because cone nuclei tend to be larger and situated towards the outer limiting membrane (Fig. 4).

The Outer Limiting Membrane

Insulating each receptor cell at the level of the myoid is the outermost portion of the glial Müller cell. The Müller cells enter into special relationships with both types of receptor cell and form the so called zonulae adherentes (Fig. 5). These appear as regions of membrane specialization round the circumference of the myoid and a corresponding condensation within the adjacent cytoplasm in both receptor myoids and Müller cells. The juxtapositioning of these dense areas and their uniformity throughout the retina led the early histologists to believe that this was a membrane system, and they called this region the outer limiting membrane of the retina. Some electrophysiologists claim that the network of zonulae adherentes represents a high resistance system limiting the flow of ions. However, electron-dense tracer studies show that free diffusion may occur, because even when relatively large molecules are injected to one side or the other of this system a diffuse distribution is rapidly achieved.

FIG. 5 (a) Electron micrograph of the outer limiting membrane (8) of a monkey retina. The densification of cell cytoplasm in association with the zonula adherens can be seen within the Müller’s fibres (F) and both rods (R) and cones (C). (× 3 350.) (b) Electron micrograph of the membrane associations contributing to a zonula adherens within the outer limiting membrane of a human retina. (1) and (2) are receptor cells and (3) is a Müller’s fibre process. The zonula adherens is a circumferential attachment characterized by the presence of a widened intercellular space (arrowed) and conspicuous bands of dense filaments in the adjacent cytoplasm. (× 53 350.)

The Müller cells send villus-like processes past the outer limiting membrane, and these project into the extracellular space round the receptors but in most cases do not extend past the myoid. The function of this so-called glial or Müller fibre basket is not known; however, some workers have suggested that the extensive surface area created by the villi must be important in transporting metabolites from the choriocapillaris via the pigment epithelium into the outer retina.

The Outer Nuclear Layer

In the human retina the outer nuclear layer is composed of receptor-cell bodies and their nuclei and inner connecting fibres, and Müller cell processes. The only nuclei present in this layer are those of rod or cone cells. The layer varies in thickness with topography but is thickest at the fovea and thinnest at the ora serrata.

Receptor Nuclei.

The slender geometry of the outer portions of the receptor cells together with their close packing and vast number has led to the multiple layering of receptor nuclei. Rod nuclei are roughly spherical, with a diameter of approximately 5 μm; clearly the acuity of the human eye could not be achieved if the receptor nuclei were confined to a single layer. Rod nuclei stain intensely with most histological stains but often show irregularities in chromatin distribution. Cone nuclei in the extrafoveal retina are situated close to the outer limiting membrane, where they form a single discontinuous layer (Fig. 3). They are larger than those of rods, ovoid, and stain less intensely. The contrast between the pale background stain and the chromatin deposits has led many workers to describe cone nuclei as tigroid nuclei.

In the foveal region the cone nuclei become smaller and multilayered like those of rods, and at the foveola this layer represents the innermost retinal layer. In most histological preparations of this region many cone nuclei have extremely dense or pyknotic staining characteristics. Such staining seems normal, although the number of pyknosed cells seems to increase with age. Rarely, cone nuclei are seen to be situated on the outer side of the outer limiting membrane; this occurrence is far more common in retinas that have been exposed to trauma.

The Inner Connecting Fibres.

The inner connecting fibres of the receptor cells run from their nuclei to the synaptic regions in the outer plexiform layer. The internal rod fibres are extremely thin and because of their tortuous route through the outer nuclear layer they are rarely observed in the light microscope. In contrast the thick internal cone fibres are often seen over their entire length in thick sections. The fibres of both cells may be described as axons, and like axons they have a well-developed system of neurotubules (microtubules) running along their length. In this region these cells contain few other organelles except for ribosomes and occasional mitochondria.

In the fovea the lateral displacement of the inner retinal layers results in extensive development of the internal connecting fibres of the foveal cones. The dense foveal cone population and the consequent complexity of the inner retinal organization cause some cone nuclei to be separated from their synaptic pedicles by lateral displacements of 500 μm. This lateral displacement is gradually reduced with distance from the macula, but some displacement is still observed in the mid-periphery in both rods and cones. The length and size of these cone fibres is clearly demonstrated after laser irradiation of the fovea, when damaged receptors show an intensive staining reaction over their entire length (Fig. 6). When the fibres leave the outer nuclear layer they run almost horizontally through the retina until they terminate in the receptor synapse. The displacement that these fibres induce between the receptor nuclei and the synaptic layer is quite large and is called the fibre layer of Henle. The ultrastructure of the cone fibres within the fibre layer of Henle is similar to that of more peripheral cones, but possibly because of their increased length they seem to have several nodular swellings at regular intervals, and contain relatively more mitochondria. The fibre layer of Henle is thought by many workers to contain the yellow macular pigment.

FIG. 6 (a) Photomicrograph of damage produced by argon laser irradiation in the macula of a human retina. The degenerating inner connecting fibres originating from receptor cells that have been irradiated (arrowed) can be seen traversing the fibre layer of Henle (H) before terminating in the outer plexiform layer (6). The lateral displacement of the receptor pedicles results in foveal cones being up to 500 μm in length. (× 428.) (b) Photomicrograph of a tangential section of a monkey foveola showing the cone nuclei (N) and the radial array of inner connecting fibres (arrowed) which form the fibre layer of Henle. (× 334.)

The Outer Plexiform Layer

The outer plexiform layer represents the inner limit of the receptor cells, and is composed of the synaptic complexes of these cells together with components from both horizontal and bipolar cells. It is beyond the scope of this chapter to discuss the detailed microstructure of the connections between the receptor cells and the intermediary neurones, but these and the ramifications in the inner retina have been well described by Dowling (1970).

The Receptor Synapses.

The synapse or spherule of rods is less than a quarter of the size of those of cones and for the most part are external to the cone synapses in the outer plexiform layer. They stain more densely than the cone pedicles and contain many synaptic vesicles. The processes from secondary neurones penetrate deep into the rod spherule within a single invagination of its surface membrane. Within the rod spherule and opposite these invaginated processes a synaptic ribbon is found together with a concentration of synaptic vesicles. At the present time this synapse is thought to be mediated by chemical transmitters; however, there is no evidence as to their identity.

Cone synapses or pedicles have both superficial and multiple invaginated contacts with the neurones of the inner retina. By labelling cells, using techniques such as the Golgi silver impregnation method, those responsible for specific types of receptor contact have been identified (Dowling, 1970). Cone pedicles themselves often send out long processes called basal filaments, which make superficial contacts with adjacent cone cells. The function of these inter-receptor contacts is not known.

The Pigment Epithelium

The pigment epithelium of the human retina consists of a single layer of cells which appear hexagonal when viewed from the surface of the retina, or when sections are cut at right angles to the long axis of the receptor cells; and approximately rectangular when cut transversely. The cells vary in shape and size with position in the fundus (Friedman and Tso, 1968). They tend to be larger and less regular in the peripheral retina, while at the macula their lateral dimensions are reduced but they increase in height. They are frequently bi- or multinucleate and their cytoplasm is full of smooth-surfaced endoplasmic