sign in sign up
<<
Home , Cone cell, Fovea centralis, Foveola, Ganglion cell layer, Inner plexiform layer, Parafovea

Chapter 1

Organization of the Adult Fovea 1 Anita Hendrickson

Contents macular region which is outlined by the superi- 1.1 Anatomy of the Human Fovea 1 or and inferior temporal branches of the cen- 1.1.1 General Anatomy 1 tral and vein (Fig. 1.1). The fovea 1.1.2 Photoreceptor Distribution, Types, lies on the visual axis of the such that a and Numbers in the Human 3 beam of passing perpendicularly through 1.1.3 Inner Retinal Associated the center of the will fall on the fovea. The with Cones in Central Primate Retina 9 fovea has long been recognized as the site of 1.1.4 Vascular and Glial Specializations maximal (Helmholz 1924). Al- of the Fovea 12 though the fovea only occupies 0.02% of the to- 1.1.5 Pigment Epithelium Numerical Relationships with Foveal Photoreceptors 13 tal retinal area and contains 0.3% of the total cones, it contains 25% of the ganglion cells 1.2 Anatomy of the Old (Curcio and Allen 1990; Curcio et al. 1990), il- and New World Monkey Fovea: What Are the Differences lustrating its importance in primate vision. A with Human Foveas? 14 large portion of the central visual centers is de- voted to processing foveal information. For in- 1.3 What Are the Anatomical Requirements to Create a Fovea? 17 stance, 40% of the primary pro- 1.3.1 Midget Ganglion Cells 17 cesses the central 5° of the which is 1.3.2 High Cone Density and Types approximately the retinal area of the foveal pit of Foveal Cones 18 (Tootell et al. 1988; Curcio et al. 1990). A major 1.3.3 Absence of Rods 18 function of the , pretectum, 1.3.4 Striate Cortex Expansion 19 cranial nuclei of III, IV,and VI and asso- 1.3.5 Vascular Specializations 19 ciated centers is to integrate cortical informa- References 20 tion and retinal input so that eye movements keep a point in space focused on the fovea of both to facilitate binocular high-acuity vi- sion (Goldberg 2000). The macular region contains a yellow pig- 1.1 ment which is particularly intense around the Anatomy of the Human Fovea fovea (Delori et al. 2001). The fovea also is iden- tified by the foveal pit which indents the vitreal 1.1.1 surface of the retina (Fig. 1.2). This pit and the General Anatomy concentration of yellow pigment around it (“yellow spot” or “macula lutea”) were recog- The is a characteristic feature of nized in human and monkey retina by anato- all primate except the nocturnal New mists such as Buzzi, Soemmerring, and Frago- World owl monkey Aotes. It lies on the temporal nard in the late eighteenth century (cited in side of the within the area centralis or Polyak 1941). Although Ramon y Cajal made 2 Anita Hendrickson

Fig. 1.1. The central human retina or macu- la, with an idealized view of its blood vessels. The optic disc (OD) is to the left. The macula is subdivi- ded into the central-most , which is surrounded in turn by the fovea, and wider perifo- vea. Each is indicated by concentric circles. Blood vessels form a ring outlining the foveal avascular zone, which marks the inner limits of the foveal pit

Fig. 1.2. Modification of a drawing of the macaque mon- on the right indicates the forming the wider key foveola and foveal slope from Polyak (1941). A cone inner portion of the foveal avascular zone and the arrow with an outer and inner segment in the foveola is filled on the left the narrower outer ring of capillaries of the in to show the length of the cone and displacement foveal avascular zone. The markings on the lower scale of the synaptic pedicle (P) from the cell body.The arrow each indicate 100 µm

significant advances in our understanding of the intricate relationships of primate retinal general retinal cellular anatomy and connectiv- neurons, including those forming the fovea. ity using the Golgi impregnation method (Cajal The high-resolution anatomical relationship 1892), it was the work of Stefan Polyak (1941) between neurons and within the human and Brian Boycott and John Dowling (1969), al- and monkey fovea also has been described us- so using Golgi impregnation, which revealed ing electron microscopy (Dowling 1965; Missot- Chapter 1 Organization of the Adult Primate Fovea 3 ten 1965, 1974; Yamada 1969;Borwein et al. 1980; by the elongated cone called the fibers of Schein 1988; Krebs and Krebs 1991). Henle, but it also contains the synaptic pedicles In the adult (Fig. 1.1), the fovea of the foveolar cones (Figs. 1.2, 1.3C,F). Rods centralis lies 4 mm temporal and 0.8 mm inferi- first appear in the fovea and also have elongat- or to the center of the optic disc (Hogan et al. ed axons (Fig. 1.3D,E). These long photorecep- 1971). The human foveal pit generally is round, tor axons are diagrammatically shown for a sin- approximately 1000 µm at its widest, and gle foveolar cone in Fig. 1.2 and indicated by ar- 200–240 µm deep (Fig. 1.2). This is about half rows in Fig.1.3E.They are formed during pit de- the thickness of adjacent retina, as can be seen velopment to allow foveal photoreceptors to re- in Fig. 1.3A. All of these dimensions can vary tain their synaptic connections to bipolar and slightly between individuals (Polyak 1941) and horizontal cells during the peripheral-ward dis- also are subject to the methods used to preserve placement of inner retinal neurons and the cen- and examine the fovea.A study of many species tral-ward displacement of photoreceptors of New World monkeys finds that, despite a (Hendrickson 1992; Provis et al. 1998). Capillar- fivefold variation in eye size and retinal area, ies are present in the inner retina up to the fo- the dimensions of the fovea remain constant veal slope, where they form a foveal avascular (Franco et al. 2000). This strongly suggests that zone (FAZ, see Fig. 1.4A) about 400–500 µm the same basic mechanism(s) must operate in wide (Figs. 1.1, 1.2, 1.4). Two more peripheral all to create this stable structure. zones are identified around the fovea but these The central retina or macula is divided into do not have sharp boundaries. The region next four concentric zones (Fig. 1.1), with the foveola to the fovea is the parafovea, a zone about in the center surrounded in turn by the fovea, 500 µm wide (Figs. 1.1, 1.3D). It also has a thick parafovea, and . The centralmost re- and rods are more nu- gion of the pit is called the foveola (Figs. 1.2, merous. The is still eight 1.3A,B,G). Here the center of the pit is slightly cells thick near the fovea but decreases to four flattened and is overlain by the highest cone cells thick at the peripheral edge of the parafo- density in the retina. The foveola is 250–350 µm vea. The remainder of the central retina is in diameter and represents the central 1° 20b of called the perifovea, which is 1.5 mm wide with the visual field.At its thinnest point, the cellular the peripheral perifoveal edge close to the nasal portion of the foveola is slightly more than side of the optic disc (Figs. 1.1, 1.3E). The perifo- 100 µm thick and is formed only by vea ganglion cell layer decreases to one cell bodies surrounded by Müller glial cell process- thick at its peripheral edge. The perifovea con- es (Yamada 1969; Burris et al. 2002). However tains many more rods than cones in its outer the foveola has the longest outer and inner seg- nuclear layer. ments in the retina and these indent the cellular layers, forming an inward curve called the fovea externa (Fig. 1.3A,G). The inner retinal layers of 1.1.2 the foveola, including the outer plexiform layer, Photoreceptor Distribution, Types, are moved laterally onto the foveal slope, al- and Numbers in the Human Retina though occasional neurons are found on the fo- veal floor (Röhrenbeck et al. 1989; Curcio and The fovea is the center of photoreceptor topog- Allen 1990). The fovea (Figs. 1.1, 1.2, 1.3C,F) in- raphy in the primate retina. The first quantita- cludes the adjacent 750 µm around the foveola, tive analysis of the human photoreceptor layer making it 1.85 mm wide or 5.5° of the central was done by Osterberg (1935), using sections visual field. The fovea contains all of the layers from a single celloidin-embedded 16-year-old of the retina, including the thickest part, which eye. The advantages of immunocytochemical is called the foveal slope. This region is charac- labeling of specific photoreceptor types com- terized histologically by having a layer of gan- bined with recent advances in optical imaging glion cells up to eight deep and a very thick out- and computerized counting of retinal whole- er plexiform layer. Most of this layer is made up mounts have been used by Curcio and col- 4 Anita Hendrickson

leagues (Curcio et al.1990,1991) to considerably by Osterberg (1935). Cones can be differentiated extend that pioneering effort. from rods on the basis of size, allowing the The mean number of photoreceptors per eye creation of -coded computer-generated based on a sample of seven human retinas maps for cone (Fig. 1.5A) and rod (Fig. 1.5B) dis- between 27 and 44 years old (Curcio et al. 1990) tribution in human retina. Different types of is 4.6 million cones and 92 million rods, which photoreceptors have been identified using im- is slightly lower than the 6 million cones found munocytochemical labels. Each photoreceptor Chapter 1 Organization of the Adult Primate Fovea 5

Fig. 1.3A–G. Photomicrographs of human central retina: formed mainly of long fibers of Henle and foveolar cone A, F, G are from a 72-year-old retina and B–E, a 13-year- P are first found at this location. The first R cell bodies old retina, as seen in semithin plastic sections of glutar- also are seen at the bottom of the ONL. The increase in aldehyde-fixed retina. (PE pigment epithelium; OS outer thickness of cone IS is obvious. D The parafovea has the segments; IS inner segments; C cone; R rod; ONL outer thickest GCL. R cell bodies have increased sharply in nuclear layer; P synaptic pedicle; OPL outer plexiform number in the ONL, but C cell bodies still form a single layer; INL , IPL , uniform layer at the outer edge of the ONL. E The peri- GCL ganglion cell layer.) A A low-power view of the fo- fovea is thinner overall, due to shorter OS and a much vea and foveal slope. The foveola (fov) is the thinnest thinner GCL. R are now the predominant cell in the part of this region, but has the longest OS. The outer fo- ONL. The fiber of Henle cone axon (arrows) can be tra- veola is indented in this region, forming the fovea exter- ced from the cone cell body into the OPL. F, G The fovea na. Layers are marked on the left side of D for B–D. (F) and foveola (G) of this 72-year-old retina have lon- B This 13-year-old foveola contains tightly-packed cone ger and thinner cone IS and OS than the young retina cell bodies, long and very thin IS and OS, as well as Mül- shown in B, but the GCL is thinner throughout. Note ler glial processes. C The inner foveal slope has all reti- that the foveola (G) contains only long thin OS and IS, nal layers, although the GCL is still thinner than its ma- some cone cell bodies (CB) surrounded by pale Müller ximum on the peak of the slope (not shown). The OPL is cell processes. Scale bar in G for B–G

Fig. 1.4A, B. Confocal micrographs of the foveal avascu- plexiform layer and B the inner wider FAZ at the level of lar zone (FAZ) in a retinal wholemount from a young the ganglion cells. Apparent blunt endings of some ves- adult macaque monkey (modified from Provis et al. sels actually are anastomoses between the two layers of 2000). The endothelial cells are immunocytochemically capillaries which form a cone around the inner walls of labeled with antibodies to CD31 and Von Wildebrand’s the foveal slope. Scale bar 200 µm factor. A shows the outer FAZ at the level of the outer type has a characteristic which, (L) selective cones and rods. S, M, and L cones combined with 11-cis-retinal forms a unique also are called blue, , and red cones, re- phototransduction molecule in the outer seg- spectively, for the color which they ment. The amino acid sequence of each opsin support. The amino acid sequence in S opsin determines the basis of wavelength selectivity and rod opsin are sufficiently different from of the three cone types and the single rod type each other and from the amino acid sequences (Nathans et al. 1986; Jacobs 1998; Sharpe et al. of M and L so that both riboprobes and 1999). The human retina contains a mixture of antibodies have been generated which specifi- short- (S), medium- (M), and long-wavelength cally label rods,S cones,and M/L cones in histo- 6 Anita Hendrickson Chapter 1 Organization of the Adult Primate Fovea 7 logical preparations such as retinal whole- photoreceptor outside the fovea (Fig. 1.5B), out- mounts. A map of S cones determined by these numbering cones 20:1 on average (Curcio and methods is shown in Fig. 1.5E. However, M and Hendrickson 1991). Rod density is relatively L opsins differ by only a few amino acids so uniform across most of the retina,but it is high- thus far it is not possible to generate riboprobes est in a ring at approximately 20° (Osterberg or antibodies which can distinguish between M 1935; Curcio et al. 1990). The highest rod density and L cones (reviewed in Bumsted et al. 1997; is found superior to the optic disc in most reti- Xiao and Hendrickson 2000). Recently in vivo nas, where this rod “hot spot” contains 176,000 studies in humans using adaptive (Roor- rods/mm2 (Fig. 1.5B). da and Williams 1999; Roorda et al. 2001) or In the macula, cone density rises (Figs. 1.5C, electroretinographic flicker photometry (Car- 1.6A) and rod density drops (Figs. 1.5D, 1.6A), roll et al. 2002) has made it possible to directly until, within the central 500–600 µm of the fo- observe living cones to determine M and L ra- vea, photoreceptor topography changes radi- tios. It also is possible to determine the ratio of cally.At 500 µm,the ratio of cones to rods is 1 : 1, L to M cones in vivo (Carroll et al. 2002) or in with an equal density of 40,000/mm2 (Fig. vitro (Hagstrom et al. 1998) using quantitative 1.6A,E). The last significant number of rods is Rt-PCR molecular techniques. found at 300 µm (Figs. 1.5D, 1.6A), with a few as Cone density in the human periphery is close to the foveolar center as 100 µm. S cones 2000–4000/mm2 (Fig. 1.5A). M and L cones continue in significant numbers up to 100 µm, form the majority, with S cones only being but are absent from the central foveola (Figs. 6–12% of total cones (Curcio et al. 1991). Cone 1.5E, 1.9C,D). This S cone-free region often is density remains higher along the nasal hori- not centered on the spot of highest cone den- zontal meridian so that the same eccentricity in sity, and its shape, size, and the number of S the nasal retina will have 2–3 times as many cones adjacent to the foveola varies between in- cones as the eccentricity in temporal retina. In- dividuals (Curcio et al. 1991; Bumsted and Hen- dividual S cones are typically separated from drickson 1999). Peak S cone density is 5100/ one another by intervening photoreceptors, mm2 near the foveola, where S cones constitute forming a mosaic in the peripheral retina. only ~3% of all cones. Because S cone density Analysis of this mosaic indicates that it is ran- drops more slowly than M and L cone density domly arranged (Curcio et al. 1991; Martin et al. in the fovea, S cone percentage actually rises to 2000; Roorda et al. 2001). Although direct visu- 6–7% in the parafovea. The central 100 µm of alization has not been done for peripheral M the foveola (Figs. 1.5C, 1.6A–C) is centered over and L cones (see above), their arrangement in the deepest part of the foveal pit and contains more central retina suggests that M and L cones only M and L cones. Direct visualization of nor- each form small clusters which also are ran- mal adult foveal and perifoveal human cone domly arranged (Roorda et al. 2001). Quantita- populations using has shown a tive PCR finds that the majority of human reti- remarkable range of M-to-L ratios in the fovea. nas show a decrease in M cones with increasing The average is 2L:1M, but individuals with nor- eccentricity, with many human retinas having mal can range between 8M:1L and very few M cones in the far periphery (Hag- 8L:1M (Roorda and Williams 1999; Carroll et al. strom et al. 1998). Rods are the predominant 2002). The central foveola contains the highest

Fig. 1.5A–F. Computer-generated color-coded maps of photoreceptor densities. Lines of isoeccentricity are photoreceptor density (A–E) and ganglion cell density spaced at 0.4 mm. Color-coded scales indicating cell (F) in the human retina. These maps display the left eye density _1000/mm2 are located below each map with the with the nasal retina to the left. A, B, F The entire retina highest density in each scale to the right.(A–D from from fovea in the center to the at the edge, Curcio et al. 1990; E from Curcio et al. 1991; F from Cur- with the black oval representing the optic disc. Lines of cio and Allen 1990) isoeccentricity are spaced at 5.94 mm. C–E only Foveal 8 Anita Hendrickson Chapter 1 Organization of the Adult Primate Fovea 9 cone density in the retina (Figs. 1.5C, 1.6A–C) outside the foveola center, inner segments are which ranges in normal young adults between 6 µm wide and 24 µm long, while outer-seg- 99,000–324,000/mm2, with a mean of 199,000/ ment width is the same but length has de- mm2 (Curcio et al. 1991) or 208,000/mm2 (Cur- creased to 27 µm (Yuodelis and Hendrickson cio and Allen 1990). However, the actual num- 1986). This change in size also is reflected in the ber of cones forming this high density is a small drop in cone density from 200,000mm2 at the fraction of the total, with only 7,000–10,000 in foveola (Fig. 1.6B,C) to 100,000/mm2 just the centralmost 300 µm and 90,000 in the cen- 150 µm away (Fig. 1.6D) and 15,000/mm2 at tral 2 mm of human retina (Curcio et al. 1990; 1.5 mm (Fig. 1.6F). Wässle et al. 1990). Therefore our highest visual These studies show that human foveal photo- acuity depends on the functional integrity of receptor topography is a series of narrow con- less than 100,000 cones out of almost 5 million, centric rings. Only M and L cones form the emphasizing how critical optimal health of the highest cone-packing density in the center. The fovea is for good vision throughout life. next ring starts at 100 µm, where S cones are Foveal cone diameter changes dramatically added and finally rods are added at 300 µm. over the central 5 mm, which is well illustrated Throughout these rings, M, L, and S cone den- in Fig. 1.6B–G. Foveolar cones are the longest sity drops and rod density increases so that a and thinnest in the retina, as well as being the rod-to-cone ratio of 1:1 is reached at 500 µm. most tightly packed (Figs. 1.2, 1.3B,G, 1.6B,C). During foveal development, a similar arrange- Their outer segments lie in the foveola, but ment of concentric rings can be identified at fe- their cell bodies are displaced into more lateral tal week 15–16, when all photoreceptor opsins fovea. Foveolar cones have a very long axon or are expressed (Bumsted and Hendrickson 1999; fiber of Henle which extends to the synaptic Xiao and Hendrickson 2000).Almost nothing is pedicle, well into the fovea. This axon is formed known about the molecular controls underlying during pit development to retain synaptic con- the development of this complex topography. tact with inner retinal neurons as these are dis- placed more peripherally (Hendrickson 1992; Provis et al. 1998). The overall total length of fo- 1.1.3 veal cones is 500–700 µm. Adult foveal cones Inner Retinal Neurons Associated measured in plastic sections have outer seg- with Cones in Central Primate Retina ments 1.2 µm wide and 40–50 µm long, while the slightly tapered inner segments are 3 µm at There are almost no inner retinal neurons in the outer limiting membrane, 2 µm at their tip, the foveola, because they have been moved pe- and 30–35 µm long (Yuodelis and Hendrickson ripherally during development (Fig. 1.2) to 1986). Cone length decreases and cone diame- form the foveal pit.Given the high density of fo- ter increases rapidly with increasing distance veal cones, there must be a similar high density from the foveal center. For instance, 400 µm of bipolar and ganglion cells to transmit infor-

Fig. 1.6A–G. Changes in the rod-to-cone ratios across crease in cone diameter and drop in packing density the central human retina. A This graph shows the rapid just 125 µm from the foveolar center. Although a few decrease in cone density (open squares) and rapid in- rods are present at this eccentricity, none are shown. crease in rod density (solid squares) across the foveola E From 660 µm eccentricity where the rod (R)-to cone- and fovea. A 1 : 1 ratio is reached at 500 µm from the fo- (C) ratio is slightly more than 1 : 1, and cone diameter veolar center. Below the graph are DIC images of human has increased markedly. F At 1.4 mm or the outer edge of photoreceptors in a retinal wholemount. This type of the parafovea where the rod-to-cone ratio is 6 : 1. Note image was used to calculate the color-coded computer that cones are now well separated by intervening rods. graphs in Fig. 1.5, and the densities in the graph above. G From 5 mm, where the peak density of rods is found. B, C The difference between normal young individuals (Modified from Curcio et al. 1990) in peak cone-packing in the center of the foveola. D In- 10 Anita Hendrickson

mation received by each individual cone. The fovea which have very tiny dendritic fields, and combination of high inner- density and he infers that they receive from a sin- pit formation creates the characteristic foveal gle cone or bipolar. These cells are termed the slope which has the thickest ganglion cell layer midget system (Fig. 1.7A), and form the basis of in the retina and also contains the synaptic subsequent work that shows that individual fo- pedicles of the foveolar cones (Figs. 1.2, 1.3C,F). veal cones have a “private line” synaptic circuit Each photoreceptor type has a specific subset (Kolb 1970). Foveal M and L cones have a large of inner retinal neurons associated with it synaptic pedicle 7–9 µm in diameter, and each which forms the M or L cone, S cone, or rod cir- pedicle contains 21 basal membrane infoldings cuits. Although a detailed description of these marked by a synaptic ribbon structure (Chun circuits is beyond the scope of this chapter (see et al. 1996; Haverkamp et al. 2000). Each infold- Boycott and Dowling 1969; Schein 1988; Wässle ing is presynaptic to two main cell types, a cen- and Boycott 1991; Dacey 1999), a brief discus- tral dendritic process from an invaginating sion will help understand why the fovea sup- midget bipolar cell and two lateral processes ports our highest visual acuity. from horizontal cells (Fig. 1.7A, midget bipolar, Polyak (1941) has pointed out that there are MBp, on right). On the base of the pedicle, more classes of bipolar and ganglion cells around the conventional synapses are made onto the flat-

Fig. 1.7A–D. The L and M cone and S cone circuits found ve a blue ON bipolar (BBp) which provides invaginating in primate central retina. A Individual L and M cones dendrites to two to four S cones, so several S cones sha- are synaptically connected in the outer plexiform layer re a single blue bipolar. The BBp synapses on a small, bi- (OPL) to an invaginating ON midget bipolar (MBp) on stratified ganglion cell in the inner part of the IPL. This the right and a flat OFF MBp on the left. Each MBp only same ganglion cell receives mixed L and M cone input to receives input from a single cone. In the inner plexiform its dendrites in the outer IPL. C, D Midget human gang- layer (IPL), each MBp in turn provides all of the bipolar lion cells at the same scale near the fovea and in the pe- synaptic input to a midget OFF (left) and midget ON ripheral retina. Note that the peripheral midget gangli- (right) ganglion cell. Note that this input is stratified, on cell has a much larger dendritic tree and receives with the OFF bipolar/ganglion cell synapses in the outer synaptic input from four midget bipolar cells portion, and the ON in the inner portion. B S cones ha- Chapter 1 Organization of the Adult Primate Fovea 11 tened dendritic tips of a flat midget bipolar cell “blue” bipolar is a specialized type that is post- (Fig. 1.7A, MBp on left). Each midget BP type synaptic only to S cones (Fig. 1.7B, BBp). Even provides dendrites to a single cone and each near the fovea these blue bipolars contact more cone contacts one of each midget type (Kolb than a single S cone, which reduces visual acu- 1970). The axon of the invaginating midget bi- ity in the blue system. The blue bipolars syn- polar terminates deep in the inner plexiform apse in the inner part of the inner plexiform layer, while the flat midget bipolar terminates layer with the inner dendrites of a specialized in the outer part of the inner plexiform layer. blue bistratified ganglion cell. Its dendrites in The invaginating bipolar provides almost all of the outer part of the inner plexiform layer re- the bipolar synaptic input to the tiny dendritic ceive mixed M and L input from other types of tree of a single midget ganglion cell which lies cone bipolars. in the inner part of the inner plexiform layer. The is the output cell of The flat midget bipolar has the same relation- the retina.Young human retinas contain a mean ship, but terminates on a tiny dendritic tree in of 1.07 million ganglion cells with a range of the outer part of the inner plexiform layer. The 710,000–1.54 million (Curcio and Allen 1990). pioneering work of Kolb and colleagues (re- This range appears to be due to individual vari- viewed in Kolb 1994) has shown that this strat- ation in absolute ganglion cell number and ret- ification reflects processing of circuits stimu- inal area. The distribution of ganglion cells lated when light goes on in a point in space (in- across human retina is shown in Fig. 1.5F,where ner part of inner plexiform layer) and when it can be seen that it fairly closely matches cone light goes off (outer part of inner plexiform distribution (Fig. 1.5A). The thickness of the layer). Thus a more modern view of the foveal ganglion cell layer helps define the zones in midget system is that it conveys both on and off central retina (Polyak 1941; Hogan et al. 1971). information from a single cone through two The thickest ganglion cell layer begins on the different midget bipolar cells, each of which peak of the foveal slope where it is up to eight onto two different midget ganglion cells deep, and this thick region continues into cells (Schein 1988; Wässle and Boycott 1991; Da- the parafovea (Fig. 1.3C). The peak density of cey 1999). This 1:2:2 circuit provides a faithful 35,000/mm2 occurs about 1 mm from the foveal capture of visual information in a spot the size center (Curcio and Allen 1990). In the perifovea of a foveal cone, which underlies the high visu- (Fig. 1.3D) the ganglion cell layer drops from al acuity of the primate fovea. Because each four thick at the central edge to one thick at the cone driving this circuit contains only M or L peripheral edge.At this point ganglion cell den- opsin, current anatomical and electrophysio- sity has dropped to >3000/mm2 at the nasal logical evidence also indicates that the midget and >2000/mm2 at the temporal perifoveal system transmits color vision for red and green edge, reflecting the higher cone density in nasal (Dacey 1999). Acuity declines with eccentricity retina.About 50% of all ganglion cells are found because cones are larger and spaced farther within 4.5 mm of the foveal center, which ac- apart, and midget ganglion cell dendritic fields counts for the large amount of central visual receive input from more than one midget bipo- centers devoted to foveal processing. lar which further increases the effective capture There has been considerable quantitative area (compare Fig. 1.7C with D). work done to test the hypothesis that foveal S cones are absent from the central 100 µm acuity depends on a 1 : 2 : 2 midget system.If this but then rise to their highest density in the par- is true, then there has to be a minimum of two afovea (Curcio et al. 1991). S cone pedicles are ganglion cells for every foveal cone. Counts smaller than L or M cones (Ahnelt et al. 1990) comparing cone and ganglion cell density in but show the same basic synaptic structures. human retina (Curcio and Allen 1990) find that However, S cones have their own specialized a 1:2 ratio is certainly supported, and a ratio of circuit (Fig. 1.7B) which transmits blue color in- 1:3 is possible for the fovea. This would allow formation (reviewed in Martin 1998; Dacey additional foveal cone output to other types of 1999; Calkins 2001). The invaginating or “on” ganglion cells. Counts in Macaca (Schein 1988; 12 Anita Hendrickson

Wässle et al. 1990) and marmoset retina (Wild- Throughout foveal development the FAZ re- er et al. 1996) also support a 1:3 ratio out to the mains at or near adult diameter, suggesting that perifovea. it remodels as the foveal pit matures and wid- ens. Computer modeling of pit development (Springer and Hendrickson 2003) indicates that 1.1.4 this neurovascular relationship may be crucial Vascular and Glial Specializations for pit formation (see below). of the Fovea The adult primate peripheral retina has a large population of which are main- The primate fovea is characterized by the ab- ly found in the nerve fiber and ganglion cell sence of blood vessels in its center, forming the layers, where they often are seen adjacent to foveal avascular zone (FAZ). The FAZ outlines ganglion cell axons or blood vessels (Distler et the foveal slope (Figs. 1.1, 1.2, 1.4) and is al. 1993; Gariano et al. 1996a, 1996b). In both fe- 500–600 µm in diameter in humans (Mansour tal and adult retina, astrocytes can be identified et al. 1993) and monkeys (Weinhaus et al. 1995; by immunocytochemical labeling for glial fi- Provis et al. 2000). The blood vessels on the fo- brillary acidic protein (GFAP; Distler et al. 1993; veal slope are mainly capillaries, and they form Gariano et al. 1996a, 1996b). Astrocytes origi- a cone around the foveal pit (Snodderly et al. nate from stem cells near and within the optic 1992; Provis et al. 2000, also see Fig. 1.4). Most of disc (reviewed in Chan-Ling 1994; Provis 2001), the primate central retina contains four main but also proliferate within the fetal retina itself laminar plexuses (Iwasaki and Inomata 1986; (Sandercoe et al. 1999). Fetal astrocytes are Snodderly and Weinhaus 1990; Snodderly et al. closely related to immature blood vessels and, 1992; Gariano et al. 1994; Provis 2001). Starting especially at the front of the vascular advance, from the vitreal surface of the retina, the inner- label for the angiogenic factor vascular endo- most plexus lies within the nerve fiber layer thelial (Provis et al. 1997). Al- and the next within the ganglion cell layer. The though astrocytes are present throughout vas- two outer plexuses are found at the inner and cular development around the fovea (Distler outer borders of the inner nuclear layer. The and Kirby 1996; Provis et al. 2000), they rapidly is avascular across the reti- disappear after birth. This disappearance does na. Near the foveal slope, there are no capillar- not seem to involve elevated cell death in the as- ies in the nerve fiber or ganglion cell layer; in- trocyte population, but may reflect a migration stead these are replaced by a single out of the fovea, a downregulation of GFAP, or plexus at the border of the ganglion cell and in- the disappearance of -promoting fac- ner plexiform layers (Figs. 1.2, 1.4B). This plexus tors (Distler et al. 2000). Given the wide range is the major visible component of the FAZ of functions now ascribed to astrocytes, this (Snodderly et al. 1992; Provis et al. 2000). The raises a question as to whether their absence outer capillaries around the fovea are concen- from the fovea may make this region more vul- trated along the outer plexiform layer (Figs. 1.2, nerable to insults. 1.4A). Snodderly et al. (1992) show that vascular Microglia are marrow-derived glial cells coverage or “screening of light” of the photore- which are present within all layers of the adult ceptor layer by blood vessels is 0 in the FAZ,but primate retina. Several types are present which then rises rapidly to 30% at 1 mm and 45% at may be associated with neurons or with blood 2 mm. One functional reason for the FAZ and vessels, and some of these are antigen-present- foveal pit may be to provide an optically clear ing cells (reviewed in Provis et al. 1996; Penfold path to the foveolar cones. et al. 2001; Provis 2001).Whether or not there is Developmental studies in Macaca monkey a unique distribution or population of micro- show that the fovea is avascular throughout de- glia around the human fovea has not been de- velopment (Engerman 1976; Gariano et al. termined. 1994), and that the FAZ forms coincident with Müller cells originate from the same progen- initial pit formation (Provis et al. 2000). itor stem cells which give rise to all of the reti- Chapter 1 Organization of the Adult Primate Fovea 13 nal neurons (reviewed in Fischer and Reh cells which contain pigment granules. 2003). Muller cells are found throughout the EM reconstructions (Borwein et al. 1980) show retina and they are the only cell type which that cone outer segments do not reach the PE. spans all retinal layers. Their cell bodies form a Instead, PE microvilli lacking pigment granules single layer in the inner nuclear layer between surround the outer third of the outer segment the amacrine and bipolar cell bodies. Inner while its inner portion is covered by microvilli processes extend like pillars to form the end from the cone inner segment and from sur- feet abutting the internal limiting membrane rounding Müller cells. This close relationship and thinner, multiple processes form the outer facilities the phagocytosis of cone outer seg- limiting membrane made up of complex junc- ments by the PE, recycling of visual pigments, tions between Müller cells and photoreceptors inactivation of free radicals produced during (Polyak 1941; Yamada 1969; Krebs and Krebs phototransduction, and bidirectional transport 1991; Distler and Dreher 1996).Müller cells have of metabolites ( 2003; Thompson and Gal been implicated in multiple roles including 2003). There is some evidence that the complex neurotransmitter recycling, glycogen metab- interphotoreceptor matrix between photore- olism, K+-ion buffering and growth-factor pro- ceptors and PE has a different molecular com- duction. Recent evidence indicates that a subset position in the fovea and peripheral retina may be retinal stem cells which can give rise to (Hollyfield et al. 2001). neurons during retinal (Fischer Several studies have determined the numer- and Reh 2003). ical relationship between foveal photoreceptors Müller cells are particularly important in the and the overlying PE. It has been noted (Street- fovea because the foveola is formed only by en 1969; Robb 1985) that central PE cells tended cones and Müller cell processes (Yamada 1969). to get smaller and more closely packed during Quantitative EM finds that the number of fo- human retinal development, similar to the veal cones is matched 1:1 by Müller cell trunks packing pattern found in foveal cones during (Burris et al. 2002), suggesting that the fovea development (Yuodelis and Hendrickson 1986). could be especially vulnerable to Müller cell in- A striking change in the cone/PE relationship jury. Processes from several Müller cells form a over time has been confirmed in a quantitative glial basket around each cone pedicle which developmental study in macaque monkeys channels glutamate, the cone neurotransmitter, (Robinson and Hendrickson 1995) which toward the synaptic contacts at its base and al- counted both PE and the underlying photore- so effectively isolates each pedicle from gluta- ceptor density. These authors have found that mate released from neighboring cones. Müller in general PE cell density over the fovea rises, cells in peripheral retina lose GFAP labeling by while peripheral PE density falls during devel- birth and only regain it under pathological or opment. For instance, foveal PE cell densities stressful conditions (Milam et al. 1998). In con- rise from around 2500/mm2 at midgestation to trast, Müller cell processes within the fovea re- over 5500/mm2 in the adult. At midgestation, tain some GFAP immunogenicity throughout one perifoveal PE cell and one foveal PE cell life (Gariano et al. 1996b; Distler et al. 2000; each covers 5 cones. After birth the number of Provis et al. 2000), suggesting that their cellular cones/PE cell changes rapidly, so that in the environment is different. adult fovea each foveal PE cell covers 30–35 cones, a PE cell in the perifovea still covers about 5 cones, and one in the periphery covers 1.1.5 1–2 cones.A recent study (Snodderly et al.2002) Pigment Epithelium Numerical Relationships which counted PE cell density but not photore- with Foveal Photoreceptors ceptors, estimates that each foveal PE cell cov- ers 20 cones, while peripheral cells cover 1.5 The pigment epithelium (PE) over the fovea cones. A study in humans which counted both does not differ obviously from that of the more PE and photoreceptors (Panda-Jones et al. peripheral retina. It is a single layer of cuboidal 1996) has found a similar foveal 1:20 ratio, al- 14 Anita Hendrickson

though it should be noted that their earlier not affect foveal formation, because this separ- photoreceptor counts for human fovea are sig- ation occurred well before Old World monkeys nificantly lower than Curcio et al. (1991) or Os- developed trichromatic color vision 10–15 mil- terberg (1935). All of these studies indicate that lion years ago (Jacobs 1998; Kremers et al. 1999; each adult foveal PE cell has to support a much Ahnelt and Kolb 2000). However, detailed anal- higher number of metabolically active cones ysis using cell-specific markers and cell counts than does an individual peripheral PE cell. show that there are subtle differences between However, if the cones + rods per PEcell are con- these three primate groups. Peak cone density sidered, perifoveal PE cells also have a total of averages around 200,000/mm2 in Macaca 35 photoreceptors, mainly rods. This may pro- (Packer et al. 1989; Wikler and Rakic 1990; Rob- duce differential stresses on the PE which vary inson and Hendrickson 1995) and marmoset with the regional differences in photoreceptor (Troilo et al. 1993; Wilder et al. 1996), as well as ratio. human (Curcio et al. 1990), but marmoset has a much higher peripheral cone density than ei- ther (Martin and Grünert 1999). A few rods are 1.2 present within 50 µm of the foveal center in Anatomy of the Old Macaca (Packer et al. 1989), which has not been and New World Monkey Fovea: seen in normal humans. No detailed foveal What Are the Differences studies have been reported for the great apes. with Human Foveas? The most striking difference in both periph- eral and foveal retina between primate species Experimental approaches to find a cure for hu- is for S cone distribution. Humans, Macaca man blinding eye diseases such as adult macu- monkey and New World Cebus monkey have an lar dystrophy require a model for preliminary S cone mosaic in which single S cones are con- testing of new therapeutic approaches. The Old sistently separated by other photoreceptors World macaque (Macaca) monkey has long (Szél et al. 1988; Wikler and Rakic 1990; Curcio been used as a model for human retina (Polyak et al. 1991; Bumsted et al. 1997; Martin and Gru- 1941; Boycott and Dowling 1969; Wässle and nert 1999; Calkins 2001). Statistical measures of Boycott 1991; Dacey 1999), while others have whether this mosaic is ordered or not have suggested that the New World marmoset is also shown that the Macaca monkey S cone mosaic appropriate (Troilo et al. 1993; Wilder et al. 1996; is somewhat ordered,while that in the human is Jacobs 1998; Martin 1998). The fovea of a young random (Curcio et al. 1991; Martin et al. 2000; adult Macaca monkey is shown in Fig. 1.8A and Roorda et al. 2001). On the other hand, the mar- the fovea of a young adult marmoset in Fig.1.8B. moset monkey has a clustered pattern of S cone A major difference between New and Old distribution in which S cones can be immediate World monkeys is that most New World mon- neighbors and the clusters have random order keys, including marmosets, only have one type (Martin et al.2000).Humans consistently lack a of M or L cone combined with S cones, making significant number of S cones within the cen- them dichromats. Similar to humans, all known tral 100 µm of the foveola (Figs. 1.5E, 1.9C,D), Old World monkeys have both M and L cones although the size and shape of the S-cone-free combined with S cones, giving them trichro- region varies between individuals (Curcio et al. matic vision (Jacobs 1998). However, this differ- 1991; Bumsted and Hendrickson 1999). Prelimi- ence does not seem to have affected foveal anat- nary studies in our laboratory find that chim- omy in any obvious manner as shown in Fig. panzees and orangutans have a small S cone- 1.8A,B. This marked similarity in foveal organ- free zone, similar to humans. In New World ization between monkeys strongly argues that monkeys, the Cebus monkey also has an S- the fovea had evolved for high visual acuity well cone-free foveolar center, while the same study before New and Old World monkeys were sep- found that marmoset has S cones across the fo- arated 30 million years ago. This also argues vea (Martin and Grünert 1999). Several groups that the presence of one or two M/L genes does have studied Macaca monkey foveas with Chapter 1 Organization of the Adult Primate Fovea 15

Fig. 1.8A–C. Comparison of primate foveal morphology bodies (R) in the Macaca fovea (A) are found at the sa- in young adult primates. A is a semithin plastic section me point where the foveal slope begins. Arrows in A and from a glutaraldehyde-fixed Old World Macaca mon- B indicate neurons in the floor of the foveola in both key. B is a paraffin section from a Carnoy-fixed New monkeys. Note that the tarsier fovea in C has relatively World marmoset monkey.C is a semithin plastic section little packing of the outer nuclear layer over the deepest from a glutaraldehyde-fixed tarsier retina. The retinal part of the pit and the OPL is thin. The tarsier GCL is no layers are indicated on the right in A. The first more than three deep on the slope, its thickest point slightly different results. Using immunocyto- from the foveal center but do not give a dimen- chemical labeling with different antibodies to S sion; while Martin and Grunert (1999) report a opsin, Szél et al. (1988) and Wikler and Rakic 50-µm-wide S-cone-free zone which is similar (1990) report that a few S cones are missing to the rod-free zone (Packer et al. 1989). Bum- 16 Anita Hendrickson

Fig. 1.9A–D. Comparison of S cone distribution in two point of highest cone density. Note that S cones are different Macaca monkeys (A, B) and human (C, D).Ad- found across the foveal centers of both macaques. Both jacent counting fields 10× 10 µm were sampled system- humans have a clear S cone-free area around the star, atically across the fovea after the S cones were labeled but the individual on the right has an S cone-free region with an antibody to S opsin. Each dot represents a field which is much less regular than the one on the left. containing one to three S cones. The star indicates the (Modified from Bumsted and Hendrickson 1999)

sted and Hendrickson (1999) have used both 1992; Packer et al.1996) and more recently in vi- immunocytochemistry and in situ hybridiza- vo using adaptive optics (Roorda et al. 2001). tion and find no S-cone-free zone in either fetal These studies find that M and L cones are clus- or adult Macaca monkeys compared with hu- tered in a random pattern, similar to humans. mans (Fig. 1.9A,B). All of these authors agree However, a marked difference in L-to-M ratio that the highest density of S cones is found ad- between macaque and human was noted in jacent to the fovea in both humans and ma- these studies, and also in studies using a rtPCR caques, where their percentage ranges between molecular approach (Deeb et al. 2000; McMa- 5 and 15%. The highest S cone density in mar- hon et al.2001).Macaca and talapoin Old World mosets is in the foveola, where it reaches monkeys have a 1M : 1L ratio near the fovea, and 10,000/mm2 (Martin and Grünert 1999). molecular studies of a larger sample of Macaca L and M cones have been identified in sever- and baboon retinas show this is slightly M al monkey species by their selective wavelength biased. A 1:1 ratio is in marked to the absorption in vitro (Mollon and Bowmaker typical 2:1 L-biased human central ratio deter- Chapter 1 Organization of the Adult Primate Fovea 17 mined by the same methods. The other differ- establish the characteristic cone-dominated ence is that this 1 : 1 ratio remains stable across circuits within the fovea. the Macaca retina (McMahon et al. 2001), while It also is significant that foveal dimensions, the human periphery contains mainly L cones peak cone density, and relative position in the (Hagstrom et al. 1998). temporal retina remain constant, despite a wide The midget bipolar/ganglion cell system has range of eye and retinal size (Packer at al. 1989; been extensively analyzed in marmoset and Curcio and Hendrickson 1991; Martin and Macaca retina using anatomical and electro- Grünert 1999; Franco et al. 2000). This strongly physiological methods (reviewed in Dacey suggests that the forces that create a fovea are 1999; Martin 1998). Although details are less localized to a small region. A constant dimen- well documented for human (Dacey 1999), cur- sion also suggests that optimal visual acuity is rent concepts and the original Golgi analysis of achieved using a relatively fixed number of Polyak (1941) suggest that the midget system neurons. Given the large amount of visual thal- synaptic connections are very similar near the amus and cortex that is devoted to existing pri- fovea in all three primates. mate foveas, having larger foveas may not be Detailed histological studies following label- possible without enlarging the beyond its ing or section of one find that there cranial capacity. is a slight vertical overlap of visual fields at the Macaca fovea (Stone et al. 1973; Fukuda et al. 1989). Some of the central nasal ganglion cells 1.3.1 do not cross and some in temporal retina do Midget Ganglion Cells cross, while the majority take the opposite course. This creates overlap of the visual fields This ganglion cell type is generally considered within the central 1°. In recent psychophysical to be found only in primate retinas (Fig. experiments on humans with visual field de- 1.7A,C,D) and, in its most characteristic form, fects, a similar overlap was found (Reinhard only around the fovea where it makes up 80% of and Trauzettel-Klosinski 2003). This probably the ganglion cell layer (Boycott and Dowling is the basis of “macular sparing” described in 1969; Wässle and Boycott 1991; Dacey 1999; human visual field studies. Kremers et al. 1999). In central retina the mid- get ganglion cell dendritic tree is slightly larger than a single bipolar axon terminal and it re- 1.3 ceives synapses from a single midget bipolar What Are the Anatomical Requirements (Fig. 1.7C). One midget bipolar in turn receives to Create a Fovea? input from a single M or L cone. This arrange- ment guarantees that a midget ganglion cell re- The following anatomical and neuronal aspects ceives input from a single cone and also that it of foveal organization seem to be distinctive for can convey the M or L wavelength-selectivity of the primate retina compared with the “area that cone. In more peripheral retina (Fig. 1.7D), centralis” of many mammals (Ahnelt and Kolb the dendritic field is larger and it receives input 2000). Undoubtedly there are molecular mark- from several bipolar axons, but the 1:1 relation- ers which are equally characteristic, and must ship between midget bipolar cell and M or L underlie the differences between foveal and pe- cone changes very little until the far periphery ripheral retina, but no molecules unique to the (Kolb 1970). Because of the random nature of M creation of a fovea have yet been identified. and L cone distribution (Roorda et al. 2001), a However, the fact that a fovea with a thick gan- peripheral midget ganglion cell will receive in- glion cell layer and a photoreceptor layer free of put from several cones which are both larger rods can be identified in humans at fetal week and more widely spaced, and this input will al- 11 and in monkeys at fetal day 50 (Hendrickson so be a variable mixture of M and L. This degra- 1992) strongly suggests that these molecules dation of the midget system with increasing ec- must act at a very early stage of development to centricity explains both lower visual acuity and 18 Anita Hendrickson

poorer red/green color vision in the periphery little different from many other mammals, in- (Dacey 1999, but see Martin et al. 2001). cluding mice (Ahnelt and Kolb 2000). It there- In the afoveate owl monkey retina, there are fore appears that a major driving factor toward ganglion cells which fit the description of the a fovea is a drastic increase in cone density con- midget type, but even in the most central por- fined to a small central region. tion they have a large dendritic field and are Most of the high foveal cone density is more like peripheral midget ganglion cells formed by M or L cones, although peak S cone found in macaque or Cebus retina (Silveira et densities are close to the foveal center as well in al. 1994; Kremers et al. 1999). This suggests that macaque and human and at its center in mar- both a true midget ganglion cell as well as a moset (Curcio et al. 1991; Martin and Grünert high central cone density is missing from the 1999; Calkins 2001). Human foveolas lack S afoveate owl monkey retina. Interestingly, an- cones, as does the tarsier (Hendrickson et al. other nocturnal primate, the tarsier, has a fovea 2000), but monkey foveolas in general have with a relatively high cone density, but a low some to many S cones (Martin and Grunert ganglion cell density (Fig. 1.8C). This suggests 1999; Bumsted and Hendrickson 1999). Thus that the appearance of the midget ganglion cell the presence or absence of S cones does not may be an essential step in of the pri- seem to affect foveal formation, although it mate fovea. should be pointed out that the afoveate owl monkey totally lacks S cones in its retina (Jacobs 1998). Likewise, both trichromatic 1.3.2 (Fig. 1.8A) and dichromatic (Fig. 1.8B) monkeys High Cone Density and Types of Foveal Cones have virtually identical foveas,so the number of opsins seem irrelevant (Jacobs 1998; Kremers et Most mammalian retinas show an uneven dis- al. 1999). However, across primate species, the tribution of cones, with some specialized re- greatest pressure to develop the characteristic gion, the “area centralis” containing a peak of high foveal cone density appears to be within cone density (reviewed in Ahnelt and Kolb the M and L cone population(s). 2000). Human (Curcio et al. 1990), macaque monkey (Packer et al. 1989; Wikler and Rakic 1990), and marmoset monkey (Martin and 1.3.3 Grünert 1999) retinas have a foveal cone peak Absence of Rods density around 200,000/mm2, which is 2 orders of magnitude higher than peripheral cone den- In all primate retinas in which rod topography sity. However, the actual area that has the high- has been described, rods are absent from the est density contains no more than 10,000 cones center of the fovea (Fig. 1.5D), although the rod- out of a total of 4.6 million in a human retina free zone is slightly smaller in monkeys than (Wässle et al.1990; Curcio et al.1990).This small humans (Packer et al.1989; Curcio et al.1990).A number emphasizes the critical role played by rod-free zone in both monkey and human reti- foveolar cones and that the loss of even a small nas is present from the earliest stages of devel- number can significantly affect central vision. opment in which rods can be identified by mo- The nocturnal tarsier (Tarsius spectrum) has lecular markers (Swain et al. 2001; Bumsted- a fovea with a narrow pit (Fig. 1.8C) and a peak O’Brien et al. 2003). In contrast, in human albi- cone density between 50,000 and 85,000/mm2 nos, rods are found throughout the fovea, cones (Hendrickson et al. 2000; Hendrickson, unpub- are large, loosely packed and immature in lished work), the lowest yet found in primate shape,and the foveal pit is poorly formed or ab- foveas. The nocturnal owl monkey Aotes lacks a sent (Fulton et al. 1978; Mietz et al. 1992). This fovea and has a peak cone density of 7000/mm2 suggests that the molecular changes causing al- (Wikler and Rakic 1990). It also should be not- binism allow rods to form within the fovea ed that peripheral cone density in all primates which could interfere in some unknown way ranges between 2000 and 4000/mm2,which is with foveal pit development. A note of caution Chapter 1 Organization of the Adult Primate Fovea 19 should be inserted as to whether the presence The diameter of the fetal FAZ is similar to the of rods directly interferes with pit formation. adult FAZ, suggesting that it remodels in a dy- The nocturnal tarsier has a very high density of namic fashion as the foveal pit develops. Be- rods in central retina (Hendrickson et al. 2000; cause the adult FAZ outlines the foveal slope, Hendrickson, unpublished work). Our prelimi- this also strongly suggests that some mole- nary studies of the tarsier fovea show that it has cule(s) within the fetal fovea mark out the fu- a narrow but deep foveal pit and minimal evi- ture foveal zone and repel astrocytes, blood dence of cone-packing into a foveola (Fig.1.8C). vessels, and ganglion cell axons from this re- Although cone density may reach 85,000/mm2 gion (Provis et al. 2000; Provis 2001). Computer over the pit, rods are still present at significant modeling based on actual histological sections density throughout the foveal cone mosaic. of developing macaque fovea shows that, be- cause the FAZ is more elastic than surrounding vascularized retina, it is more easily deformed 1.3.4 by retinal growth. These models indicate that Striate Cortex Expansion unequal deformation within the FAZ combined with eye growth-induced retinal stretch causes The overall expansion in the area of primary the foveal pit to form (Springer and Hendrick- visual cortex in primates is due in large part to son 2004).If blood vessels fill the fovea,no pit is the amount of cortex devoted to the high den- formed. In turn, tensile forces generated within sity of ganglion cells around the fovea. A simi- the retinal layers around the developing pit lar expansion is seen in the lateral geniculate help pack cones into the foveal center (Springer nucleus, the thalamic relay to striate cortex 1999). These models conclude that the FAZ is a (Wässle et al. 1990; Kremers et al. 1999). In hu- major factor in foveal pit formation and may al- man retinas the peak ganglion cell density of so be important for cone packing. All retinas 35,000/mm2 occurs about 1 mm from the foveal that have a foveal pit have a FAZ (Wolin and center and 50% of all ganglion cells are found in Massopust 1967). Human albino retinas and hy- the central 4.5 mm (Curcio and Allen 1990). In poplastic foveal syndromes have a poorly macaque monkeys, 40% of V1 contains the rep- formed pit and a low cone-packing density, and resentation of the central 5°, or the 1.5 mm of these patients often have aberrant blood vessels retina surrounding the foveal pit (Tootell et al. in the fovea (Spedick and Beauchamp 1986; Ol- 1988). It is possible that other mammals started iver et al. 1987; Barbosa-Carneiro et al. 2000). If to develop foveas but failure of their striate cor- these vessels were present during foveal devel- tex to expand sufficiently to process the input opment, this could account for both the poor from this large number of neurons negatively pit formation and low cone density. However, affected subsequent foveal evolution. because these foveas are in the proper position on the retina but contain inappropriate cells such as rods, some earlier molecular mecha- 1.3.5 nisms must be disturbed while others are Vascular Specializations maintained. Thus it is likely that no single gene, develop- The primate retina shows a unique vascular de- mental event, or set of neurons gives rise to a velopmental pattern (Gariano et al. 1994, 1996a, fovea.A number of genes must work together to 1996b; Provis et al. 2000; Provis 2001). As blood first fix the foveal position on the retina and vessels growing across the inner retina from the then generate a large number of cones with optic disc approach the site of the future fovea, their specialized inner retinal neurons in this they diverge to surround it, leaving a FAZ position. During this phase it is critical that a (Figs. 1.1, 1.2, 1.4). It is important to emphasize large number of midget bipolar and ganglion that blood vessels invade the foveal region at a cells are generated to create the 1:2:2 high-acu- time when it contains all retinal layers and has ity circuit. Other genes probably cause central not yet started to form a pit (Provis et al. 2000). visual centers to generate additional neurons to 20 Anita Hendrickson

handle foveal processing. Next it is probable Calkins DJ (2001) Seeing with S cones. Prog Retin Eye that unknown molecules must remain to ex- Res 20 : 255–287 Carroll J, Neitz J, Neitz M (2002) Estimates of L : M cone clude astrocytes and blood vessels, and prob- ratio from ERG flicker photometry and genetics. J ably ganglion cell axons, to form the FAZ and Vis 2 : 531–542 begin pit formation. Finally a combination of Chan-Ling T (1994) Glial, neuronal and vascular inter- molecular changes within the foveal neurons actions in the mammalian retina. Prog Retin Eye Res and pit formation causes the central cones to 13 : 357–389 Chun M-H, Grünert U, Martin PR, Wässle H (1996) The remodel so that they become very thin and el- synaptic complex for cones in the fovea and in the ongated, which facilitates packing into the fove- periphery of the macaque monkey retina.Vis Res 36 : ola. All of these steps are necessary, but almost 3383–3395 none of them are understood in any molecular Curcio CA, Allen KA (1990) Topography of ganglion or functional detail. However, a better under- cells in human retina. J Comp Neurol 300 : 5–25 Curcio CA, Hendrickson AE (1991) Organization and standing of foveal dynamics might aid in un- development of the primate photoreceptor mosaic. raveling changes in aging which lead to foveal Prog Retin Res 10 : 89–120 degeneration. Curcio CA,Sloan KR,Kalina RE,Hendrickson AE (1990) Human photoreceptor topography. J Comp Neurol 292:497–523 Curcio CA, Allen KA, Sloan KR, Lerea CL, Hurley JB, References Klock IB, Milam AH (1991) Distribution and mor- phology of human cone photoreceptors stained with Ahnelt P,Kolb H (2000) The mammalian photoreceptor anti-blue opsin. J Comp Neurol 312 : 610–624 mosaic-adaptive design. Prog Retina Eye Res 19 : Dacey DM (1999) Primate retina: cell types, circuits and 711–777 color opponency. Prog Retin Eye Res 18 : 737–763 Ahnelt P,Keri C, Kolb H (1990) Identification of pedicles Deeb SS, Diller LC, Williams DR, Dacey DM (2000) of putative blue-sensitive cones in the human retina. Interindividual and topographical variation of L : M J Comp Neurol 293 : 39–53 cone ratios in monkey retinas. J Opt Soc Am A Opt Barbosa-Carneiro L, Peret M, Bichara A, Coscarelli G, Imag Sci Vis 17 : 538–544 Peret P (2000) Isolated foveal hypoplasia. Invest Delori FC, Goger DG, Hammond BR, Snodderly DM, Ophthalmol Vis Sci 41 : S571 Burns SA (2001) Macular pigment density measured Bird AC (2003) The Bowman Lecture. Towards an by autofluorescence spectrometry: comparison with understanding of age-related macular disease. Eye reflectometry and heterochromatic flicker photome- 17 : 457–466 try. J Opt Soc Am A Opt Imag Sci Vis 18 : 1212–1230 Borwein B, Borwein D, Medeiros J, McGowan JW (1980) Distler C, Dreher Z (1996) Glia cells of the monkey reti- The ultrastructure of monkey foveal photoreceptors, na. II. Müller cells.Vis Res 36 : 2381–2394 with special reference to the structure, shape, size Distler C, Kirby MA (1996) Transience of astrocytes in and spacing of the foveal cones. Am J Anat 159 : the newborn macaque monkey retina.Eur J Neurosci 125–146 8:847–851 Boycott BB, Dowling JE (1969) Organization of the pri- Distler C,Weigel H, Hoffman K-P (1993) Glia cells of the mate retina: light microscopy. Philos Trans R Soc monkey retina. I. Astrocytes. J Comp Neurol 333 : Lond B Biol Sci 255 : 109–184 134–147 Bumsted K, Hendrickson A (1999) Distribution and de- Distler C, Kopatz K, Telkes I (2000) Developmental velopment of short-wavelength cones differ between changes in astrocyte density in the macaque perifo- Macaca monkey and human fovea. J Comp Neurol veal region. Eur J Neurosci 12 : 1331–1341 403 : 502–516 Dowling JE (1965) Foveal receptors of the monkey reti- Bumsted K, Jasoni CL, Szél A, Hendrickson A (1997) na: fine structure. Science 147 : 57–59 Spatial and temporal expression of cone opsins dur- Engerman RL (1976) Development of the macular circu- ing monkey retinal development. J Comp Neurol lation. Invest Ophthalmol 15 : 835–840 378 : 117–134 Fischer AJ, Reh TA (2003) Potential of Muller glia to be- Bumsted-O’Brien K, Schulte D, Hendrickson A (2003) come neurogenic retinal progenitor cells. Glia 43 : Expression of photoreceptor-associated molecules 70–76 during human fetal . Mol Vis 9 : Franco ECS,Finlay BL,Silveira LCL,Yamada ES,Crowley 401–409 JC (2000) Conservation of absolute foveal area in Burris C, Klug K, Ngo IT,Sterling P,Schein S (2002) How New World monkeys. A constraint on eye size and Müller glial cells in macaque fovea coat and isolate conformation. Brain Behav Evol 56: 276–286. the synaptic terminals of cone photoreceptors. J Fukada Y,Sawai H,Watanabe M,Wakakuwa K,Morigiwa K Comp Neurol 453 : 100–111 (1989) Nasotemporal overlap of crossed and uncrossed Cajal S Ramon Y (1892) La rétine des vertébrés. Cellule retinal ganglion cell projections in the Japanese mon- 9 : 121–255 key (Macaca fuscata). J Neurosci 9:2353–2373 Chapter 1 Organization of the Adult Primate Fovea 21

Fulton AB, Albert DM, Craft JL (1978) Human : Kremers J, Silveira LCL, Yamada ES, Lee BB (1999) The light and electron microscopy. Arch Ophthalmol 96 : ecology and evolution of primate color vision. In: 305–310 Gegenfurtner KR, Sharpe LT (eds) Color vision: Gariano RF, Iruela-Arispe ML, Hendrickson AE (1994) from genes to perception. Cambridge University Vascular development in primate retina: comparison Press, Cambridge, UK, pp 123–142 of laminar plexus formation in monkey and human. Mansour AM, Schachat A, Bodiford G, Haymond R Invest Ophthalmol Vis Sci 35 : 3442–3455 (1993) Foveal avascular zone in mellitus. Gariano RF, Iruela-Arispe ML, Sage EH, Hendrickson Retina 13 : 125–128 AE (1996a) Immunohistochemical characterization Martin PR (1998) Colour processing in the primate reti- of developing and mature primate retinal blood ves- na: recent progress. J Physiol (Lond) 513 : 631–638 sels. Invest Ophthalmol Vis Sci 37 : 93–103 Martin PR, Grünert U (1999) Analysis of the short wave- Gariano RF, Sage EH, Kaplan HJ, Hendrickson AE length-sensitive “blue” cone mosaic in the primate (1996b) Development of astrocytes and their rela- retina: comparison of new world and old world tion to blood vessels in fetal monkey retina. Invest monkeys. J Comp Neurol 406 : 1–14 Ophthalmol Vis Sci 37 : 2367–2375 Martin PR, Grünert U, Chan TL, Bumsted K (2000) Spa- Goldberg ME (2000) The control of gaze. In: Kandel ER, tial order in short-wavelength-sensitive cone photo- Schwartz JH, Jessell TM (eds) Principles of neural receptors: a comparative study of the primate retina. science,4th edn.McGraw-Hill,New York,pp 782–800 J Opt Soc Am A Opt Imag Sci Vis 17 : 557–567 Hagstrom SA, Neitz J, Neitz M (1998) Variations in cone Martin PR, Lee BB, AJ, Solomon SG, Ruttiger L populations for red-green color vision examined by (2001) Chromatic sensitivity of ganglion cells in the analysis of mRNA. Neuroreport 9 : 1963–1967 peripheral primate retina. Nature 410 : 933–936 Haverkamp S, Grünert U, Wässle H (2000) The cone McMahon C, Neitz J, Dacey DM, Neitz M (2001) Re- pedicle, a complex synapse in the retina. Neuron 27 : versed L : M cone ratio in baboon indicates that pig- 85–95 ment gene order does not determine relative cone Helmholtz H (1924) Treatise on physiological optics, the number. Invest Ophthalmol Vis Sci 44 : 72S sensation of vision (JPC Southall, translator). Opti- Mietz H, Green WR,Wolff SM,Abundo GP (1992) Foveal cal Society of America hypoplasia in complete oculocutaneous albinism. Hendrickson A (1992) A morphological comparison of Retina 12 : 254–260 foveal development in man and monkey. Eye 6 : Milam AH, Li ZY, Fariss RN (1998) Histopathology of 136–144 the human retina in pigmentosa. Prog Retin Hendrickson A, Djajadi HR, Nakaura L, Possin DE, Saju- Eye Res 17 : 175–205 thi D (2000) Nocturnal tarsier retina has both short Missotten L (1965) The ultrastructure of the human ret- and long/medium-wavelength cones in an unusual ina. Arsia, Brussels, Belgium topography. J Comp Neurol 424 : 718–730 Missotten L (1974) Estimation of the ratio of the cones Hogan, MJ, Alvarado JA, Weddell JE (1971) of and neurons in the fovea of the human retina. Invest the Human Eye. Saunders, Philadelphia Ophthalmol 13 : 1045–1049 Hollyfield JG, Rayborn ME, Nihiyama K, Shadrach KG, Mollon JD, Bowmaker JK (1992) The spatial arrange- Miyagi M, Crabb JW, Rodriquez IR (2001) Interpho- ment of cones in the primate fovea. Nature 360 : toreceptor matrix in the fovea and peripheral retina 677–679 of the primate Macaca mulatta; distribution and Nathans J, Thomas D, Hogness DS (1986) Molecular ge- glyoforms of SPACR and SPACRCAN. Exp Eye Res netics of human color vision: the genes encoding 72 : 49–61 blue, green and red pigments. Science 232 : 193–202 Iwasaki M, Inomata H (1986) Relation between superfi- Oliver MD, Dotan SA, Chemke J,Abraham FA (1987) Iso- cial capillaries and foveal structures in the human lated foveal hypoplasia. Br J Ophthalmol 71 : 926–930 retina. Invest Ophthalmol Vis Sci 27 : 1698–1705 Osterberg G (1935) Topography of the layer of rods and Jacobs GH (1998) and seeing – lessons cones in the human retina. Acta Ophthalmol [Suppl from natural experiments. Invest Ophthalmol Vis Sci VI] : 1–102 39 : 2205–2216 Packer O, Hendrickson AE, Curcio CA (1989) Photore- Kolb H (1970) Organization of the outer plexiform layer ceptor topography of the retina in the adult pigtail of the primate retina: electron microscopy of Golgi- macaque (Macaca nemestrina). J Comp Neurol 288 : impregnated cells. Philos Trans R Soc London B Biol 165–183 Sci 258 : 261–283 Packer OS, Williams DR, Bensinger DG (1996) Photo- Kolb H (1994) The architecture of functional neural cir- pigment transmittance imaging of the primate pho- cuits in the retina. The Proctor Lecture toreceptor mosaic. J Neurosci 16 : 2251–2260 1994. Invest Ophthalmol Vis Sci 35 : 2385–404. Panda-Jones S, Jonas JB, Jakobczyk-Zmija M (1996) Ret- Krebs W, Krebs IP (1989) Quantitative morphology of inal pigment epithelial cell count, distribution and the central fovea in the primate retina. Am J Anat correlations in normal human eyes. Am J Ophthal- 184 : 225–236 mol 121 : 181–189 Krebs W, Krebs I (1991) Primate retina and : at- Penfold PL, Madigan MC, Gillies MC, Provis JM (2001) las of fine structure in man and monkey. Springer- Immunological and aetiological aspects of macular Verlag, New York degeneration. Prog Retin Eye Res 20 : 385–414 22 Anita Hendrickson

Polyak SL (1941) The retina.University of Chicago Press, Snodderly DM, Weinhaus RS (1990) Retinal vasculature Chicago, IL of the fovea of the squirrel monkey Saimiri sciureus: Provis JM (2001) Development of the primate retinal three-dimensional architecture, visual screening vasculature. Prog Retin Eye Res 20 : 799–821 and relationships to the neuronal layers. J Comp Provis JM, Diaz CM, Penfold PL (1996) Microglia in hu- Neurol 297 : 145–163 man retina: a heterogeneous population with dis- Snodderly DM, Weinhaus RS, Choi JC (1992) Neural- tinct ontogenies. Perspect Dev Neurobiol 3 : 213–221 vascular relationships in central retina of macaque Provis JM, Leech J, Diaz CM, Penfold PL, Stone J, Keshet monkeys (Macaca fascicularis). J Neurosci 12 : E (1997) Development of the human retinal vascula- 1169–1193 ture: cellular relations and VEGF expression.Exp Eye Snodderly DM, Sandstrom MM, Leung IY-F, Zucker CL, Res 65 : 555–568 Neuringer M (2002) Retinal pigment epithelial cell Provis JM, Diaz CM, Dreher B (1998) Ontogeny of the distribution in central retina of rhesus monkeys. In- primate fovea: a central issue in retinal development. vest Ophthalmol Vis Sci 43 : 2815–2818 Prog Neurobiol 54 : 549–580 Spedick MJ, Beauchamp GR (1986) Retinal vascular and Provis JM, Sandercoe T, Hendrickson AE (2000) Astro- abnormalities in albinism. J Pediatr cytes and blood vessels define the foveal rim during Ophthalmol 23 : 58–63 primate retinal development. Invest Ophthalmol Vis Springer AD (1999) New role for the primate fovea: a Sci 41 : 2827–2836 retinal excavation determines photoreceptor deploy- Reinhard J, Trauzettel-Klosinski S (2003) Nasotemporal ment and shape.Vis Neurosci 16 : 629–636. overlap of retinal ganglion cells in humans: a func- Springer A, Hendrickson A (2004) Development of the tional study. Invest Ophthalmol Vis Sci 44 : 1568–1572 primate fovea. 1. Use of finite element analysis mod- Robb RM (1985) Regional changes in retinal pigment els to identify mechanical variables affecting pit for- epithelial cell density during ocular development. mation.Vis Neurosci 21 : 53–62 Invest Ophthalmol Vis Sci 26 : 614–620 Stone J, Leicester J, Sherman M (1973) The naso-tempo- Robinson SR, Hendrickson A (1995) Shifting relation- ral division of the monkey’s retina. J Comp Neurol ships between photoreceptors and pigment epithe- 150 : 333–348 lial cells in monkey retina: implications for the de- Streeten BW (1969) Development of the human retinal velopment of retinal topography. Vis Neurosci 12 : pigment epithelium and the posterior segment.Arch 767–778 Ophthalmol 81 : 383–394 Röhrenbeck J, Wässle H, Boycott BB (1989) Horizontal Swain PK, Hicks D, Mears AJ, Apel IJ, Smith JE, John SK, cells in the monkey retina: immunocytochemical Hendrickson AE, Milam AH, Swaroop A (2001) Mul- staining with antibodies against calcium binding tiple phosphorylated isoforms of NRL are specifical- . Eur J Neurosci 1 : 407–420 ly expressed in photoreceptors. J Biol Chem 276 : Roorda A, Williams DR (1999) The arrangement of the 36824–36830 three cone classes in the living human eye. Nature Szél A, Diamanstein T,Röhlich P (1988) Identification of 397 : 520–522 the blue sensitive cones in the mammalian retina by Roorda A, Metha AB, Lennie P, Williams DR (2001) anti-visual pigment antibody. J Comp Neurol 273 : Packing arrangement of the three cone classes in 593–602 primate retina.Vis Res 41 : 1291–1306 Thompson DA, Gal A (2003) metabolism in Sandercoe TM, Madigan MC, Billson FA, Penfold PL, the retinal pigment epithelium: genes, Provis JM (1999) Astrocyte proliferation during de- and diseases. Prog Retin Eye Res 22 : 683–703 velopment of the human retinal vasculature. Exp Eye Tootell RBH, Switkes E, Silverman MS, Hamilton SL Res 69 : 511–523 (1988) Functional anatomy of macaque striate cor- Schein SJ (1988) Anatomy of macaque fovea and spatial tex. II. Retinotopic organization. J Neurosci 8 : densities of neurons in foveal representation.J Comp 1531–1568 Neurol 269 : 479–505 Troilo D, Howland HC, Judge SJ (1993) Visual optics and Sharpe LT, Stockman A, Jägle H, Nathans J (1999) Opsin retinal cone topography in the common marmoset genes, cone photopigments, color vision and color (Callithrix jacchus).Vis Res 33 : 1301–1310 blindness. In: KR Gegenfurtner, Sharpe LT (eds) Col- Wässle H, Boycott BB (1991) Functional architecture of or vision – from genes to perception. Cambridge the mammalian retina. Physiol Rev 71 : 447–480 University Press, Cambridge, UK, pp 3–51 Wässle H, Grünert U, Röhrenbeck J, Boycott BB (1990) Silveira LCL, Yamada ES, Perry VH, Picanco-Diniz CW Retinal ganglion cell density and cortical magnifica- (1994) M and P retinal ganglion cells of diurnal and tion factor in the primate.Vis Res 30 : 1897–1911 nocturnal New World monkeys. Neuroreport 5 : Weinhaus RS, Burke JM, Delori FC, Snodderly DM 2077–2081 (1995) Comparison of with Silveira LCL, Lee BB, Yamada ES, Kremers J, Hunt DM microvascular anatomy of macaque retinas. Exp Eye (1998) Postreceptoral mechanisms of colour vision Res 61 : 1–16 in new world primates.Vis Res 38 : 3329–3337 Chapter 1 Organization of the Adult Primate Fovea 23

Wikler KC, Rakic P (1990) Distribution of photorecep- Xiao M, Hendrickson A (2000) Spatial and temporal ex- tor subtypes in the retina of diurnal and nocturnal pression of short, long/medium or both opsins in primates. J Neurosci 10 : 3390–3401 human fetal cones. J Comp Neurol 425 : 545–559 Wilder HC, Grünert U, Lee BB, Martin PR (1996) Topog- Yamada E (1969) Some structural features of the fovea raphy of ganglion cells and photoreceptors in the centralis in the human retina. Arch Ophthalmol 82 : retina of a New World monkey: the marmoset Cal- 151–155 lithrix jacchus.Vis Neurosci 13 : 335–352 Yuodelis C, Hendrickson A (1986) A qualitative and Wolin LR, Massopust LC Jr (1967) Characteristics of the quantitative analysis of the human fovea during de- ocular of the primate. J Anat 101 : 693–699 velopment.Vis Res 26 : 847–855