Ординатура / Офтальмология / Английские материалы / The Retina and its Disorders_Besharse, Bok_2011

Population Properties and Gap-Junctional

Coupling

The A- and B-type cells form populations that completely cover the retinal surface and thus provide their signals at all points of the retina. Within each population, the cells form a rather regular mosaic, that is, an even tiling of the retinal surface. The horizontal cells, similar to many other neurons, decrease their population density from the central to peripheral retina and conversely increase the size of the individual cells such that the dendritic overlap, or coverage factor, remains approximately constant across the retina. In cat, the density of A-type cells decreases from 800 mm–2 to 100 mm–2, and that of B-type cells from 2300 mm–2 to 300 mm–2. The coverage factor is approximately four for each type; therefore, each cone can, on average, contact eight horizontal cells. In rabbit,

the central–peripheral density gradient is shallower (550 mm–2 to 250 mm–2 for the A-type and 1375 mm–2 to

400 mm–2 for the B-type); the coverage factor is approximately six for the A-type and 8–10 for the B-type, somewhat larger than in cat. Several staining methods can be used to stain entire horizontal cell populations. In many mammals, neurofibrillar stains and immunocytochemical staining of the neurofilament proteins specifically reveal the A-type population (Figure 4(a)), but in horse, the B-type population is specifically stained. Antibodies against the calcium-binding protein calbindin (CaBP 28 kDa) stain both horizontal cell populations in various mammals (Figure 4(b)).

Mammalian horizontal cells are electrically coupled by gap junctions. The coupling is homotypical; there are no gap junctions between A- and B-type cells (Figure 5). The gap junctions of A-type cells in rabbit are composed from connexin 50 (Cx50) subunits, and are larger than

those of the B-type cells where the connexin is Cx57. The gap junctions of B-type cells in mouse are also formed by Cx57; therefore, the cell-type-specific connexins may be conserved across species. The rod connectivity is similar through the coupling of B-type axon terminals; notably, this coupling is segregated from that of the B- type dendrites (Figure 5). Thus, with respect to electrical coupling, the horizontal cells form three separate networks, even though A-type and B-type dendrites largely share their synaptic input and output partners, the cones. The gap junctions are regulated by the ambient light level in a triphasic manner. At intermediate light levels, the horizontal cells are strongly coupled and their signals spread over large distances; in bright and in very low light, the gap junctions are closed and the horizontal cells act as smaller inhibitory units. Dopamine and other neuromodulators are involved in this regulation. Horizontal cell coupling is thought to play an important role in photoreceptor adaptation to different ambient light levels as it serves to collect light information over large areas of the retina.

Diversity of Morphology and Connectivity across Species

None of the basic features listed so far has provided compelling arguments to explain why there should be two horizontal cell types in mammals. Are there really

Figure 4 Flat views of population stains of horizontal cells.

(a)Neurofibrillar staining of the A-type population in rabbit and

(b)calbindin immunostaining of the A- and B-type populations in horse. In both images, the arrowheads indicate A-type somata. The scale bar ¼ 50 mm for (a) and 100 mm for (b).

ax

Figure 5 Gap-junctional coupling of rabbit B-type cells. After injection of neurobiotin into a B-type soma (asterisk), the dye spreads to neighboring B-type somata and dendritic trees (some arrowed) via gap junctions between dendrites, but not to neighboring B-type axon terminals or A-type cells. The dye that has diffused intracellularly through the axon (ax) to the axon terminal system of the injected cell then spreads to neighboring axon terminals via gap junctions (right side, arrowheads), and thereon diffuses to some B-type somata through their axons. This shows that gap-junctional coupling is type-specific (homotypic), and that coupling is segregated between B-type dendrites and axons. The scale bar ¼ 100 mm. Image kindly provided by David I. Vaney.

two types in every mammal? Which morphological differences are of functional significance? Are there differences in their specific connection with the cones and, thus, in chromatic processing? A- and B-type horizontal cells have been identified in a range of orders including primates, carnivores, lagomorphs, rodents, and ungulates; both types are present in rod-dominated and conedominated retinas. On this basic pattern, an unexpected diversity in morphology and connectivity is superimposed, which has led to modifications in the general definition of mammalian A- and B-type cells.

Variations of Shape

Old World primates including humans were perceived early on as deviating from the general mammalian pattern. Both the H1 cell (B-type equivalent) and the H2 cell have an axon (Figures 6(a) and 7). Hence, H2 was regarded as unique to the primate retina. The axon of the H1 cell connects to rods, while that of the H2 cell connects to cones; the dendrites of both types only synapse with cones. As the H2 axon has no rod contacts, this cell can be equated to the A-type of other mammals. However, H2 cells have finer dendrites than H1 cells, the reverse of the presumed characteristic distinction between the A-type and B-type. Cells with the same morphologies also exist in New World monkeys. Moreover, the cone contacts of H1 and H2 cells in both Old World and New World primates differ from those of the B- and A-type cells of other mammals (see below).

In line with the overall more fine-grained processing in primate retina, particularly near the fovea, H1 and H2 cells have higher densities and smaller dendritic fields than the horizontal cells of other mammals. In macaque, H1 and H2 densities near the fovea are 18 400 mm–2 and 4,600 mm–2, respectively, dropping to 1000 mm–2 and 500 mm–2, respectively, in the peripheral retina. A central H1 cell only contacts 6–7 cones and a peripheral one, 40–50 cones. H2 cells have larger dendritic fields than H1 cells at corresponding locations, conforming to the A-type/B-type differences in other mammals.

In artiodactyls (ox, sheep, pig, and deer), the B-type cells have a robust dendritic tree, while the A-type dendritic tree is delicate (Figures 8(c) and 7); with regard to this, they resemble primates. The B-type has a single axon ending in an axon terminal system. The A-type has no axon; however, sometimes, one or a few dendritic processes extend beyond the perimeter of the dendritic field. Our evidence indicates that they are conventional dendrites connected to cones. The suggestion is that the artiodactyl A-type represents an intermediate shape between the roughly symmetric A-type dendritic fields of many mammals and the singular asymmetry of the primate H2 cell’s axon, thus placing primate H2 cells at one end of a spectrum of A-type morphologies.

(a)

(b)

(c)

Figure 6 Primate horizontal cells and their cone contacts.

(a)Golgi-stained macaque H1 and H2 cells in flat view, focused on the dendritic trees and terminal aggregates (ax, axon).

(b)Plexus of macaque H1 cells, stained by a neurobiotin injection in one of the cells; tracer spread to the neighboring H1 cells occurred through the gap junctions. The H1 cells form dense terminal clusters at most cone pedicles (yellow, presumed

M- and L-cones), but nearly completely miss the three presumed S-cones (blue). (c) Similarly neurobiotin-labeled macaque H2 cells. They not only strongly innervate the three presumed S-cones (blue), but also contact the other (M and L) cones. The scale bar ¼ 25 mm. Images kindly provided by Dennis Dacey.

In perissodactyls (horse, ass, mule, and zebra), the B-type also has a robust dendritic tree. The single axon is very long, straight, and unusually thick (Figures 8(d) and 7). The dendrites of the A-type are very fine and sparsely branched, and there is no indication of an axon (Figure 7). The most interesting feature of this A-type cell, however, is its selective connection to the S-cones (see below).

The cone-dominated retina of the tree shrew (Tupaia belangeri, Scandentia) has particularly unusual A-type cells (Figure 9). They are large and have stout radial

Carnivora

Lagomorpha

Rodentia (some)

Marsupialia

Rodentia

Muridae

Scandentia

Tupaia

Primates

Artiodactyla

Perissodactyla

Figure 7 Schematic drawings to show interordinal variations in mammalian A- and B-type horizontal cell morphology. The diagram shows the basic branching patterns but not the synaptic terminals. Interruptions on B-type cells’ axons indicate that the axons are longer than drawn. Adapted from Figure 5 in Peichl, L., Sandmann, D., and Boycott, B. B. (1998). Comparative anatomy and function of mammalian horizontal cells. In: Chalupa, L. M. and Finlay, B. L. (eds.) Development and Organization of the Retina: From Molecules to Function, pp. 147–172. New York: Plenum Press. With kind permission of Springer Science and Business Media.

primary dendrites that rarely branch until the periphery of the dendritic field, where some ramify into unique bushy arborizations. The first description of these cells termed them multiaxonal because the arborizations were reminiscent of B-type axon terminal systems. We then showed that all connections of these cells, including those at the peripheral arborizations, are with cones. The cells thus conform to the basic mammalian A-type connectivity and can be interpreted as a further variety of A-type shape. This shape is not overtly associated with the high cone density in the tree shrew retina because the H2 cell (A-type equivalent) of the equally cone-dominated ground squirrel retina has a rather conventional shape. The B-type cells of the tree shrew are small and have a conventional dendritic tree; however, the axon is very sparsely branched

ax

Figure 8 Species variations in horizontal cell morphology; Lucifer yellow-injected cells in flat view. (a) Guinea pig A-type;

(b) gerbil B-type; (c) pig A-type; and (d) horse B-type. ax, axon. All cells shown at the same magnification and scale bar ¼ 50 mm.

B-type

A-type

100 m

Figure 9 Drawings of Lucifer yellow-injected A- and B-type cells from the cone-dominated retina of the tree shrew. The B-type cell has a conventional dendritic tree and a sparsely branched axon terminal system, whereas the A-type cell has a unique branching pattern. Adapted from Figure 4 in Peichl, L., Sandmann, D., and Boycott, B. B. (1998). Comparative anatomy and function of mammalian horizontal cells. In: Chalupa, L. M. and Finlay, B. L. (eds.) Development and Organization of the Retina: From Molecules to Function, pp. 147–172. New York: Plenum Press. With kind permission of Springer Science and Business Media.

and has only a few terminals (Figure 9). This is to be expected because less than 10% of the tree shrew’s photoreceptors are rods.

The rodents are the most diverse of the mammalian orders, and their horizontal cells have provided the biggest surprise. The muroid species rat, mouse, gerbil, and Syrian

hamster possess only one type, the axon-bearing B-type (Figures 8(b) and 7); intracellular injections and population analysis gave no evidence for a further type of horizontal cell. So far, the muroid rodents are the only known instance where the basic mammalian pattern of two horizontal cell types has not been found. In some other rodent groups, for example, sciurids and caviomorphs, both types are present (Figure 8(a)). The absence of A- type cells does not appear to be associated with nocturnality since the gerbil has active phases at both day and night and possesses a rather high cone proportion of 10– 20%. Rat and mouse have a lower cone proportion (1% and 3%, respectively), which is, however, comparable to the cone proportions found in cat and rabbit (2–4%). Apparently, there is no correlation between a low cone/ rod ratio and the absence of the A-type.

The morphology of the B-type axon terminal system also varies significantly across mammalian species. Most species have a thin B-type axon of a few hundred microns length that meanders randomly; in horse, the axon is thick, straight, and a few millimeters long. In cat, rabbit, and rat, the axon terminal system is rather densely branched and appears to innervate most of the rods in its field (as many as 3000 in cat). In other species such as primates or horse, the branching is less dense or even sparse. Here, only a minority of the rods present are innervated by any one terminal system. However, as a population, the overlapping axons of several cells ensure full coverage of the rods.

The existence of a third type of horizontal cell has been claimed for primates, the rabbit, and the South American opossum on the basis of individual cells that markedly differed from the other two types in morphology and presumed connectivity. However, other studies have concluded that such cells are encompassed in the normal variation of the standard two types; the most parsimonious interpretation is that these cells are extreme individuals. If horizontal cells with new or unusual morphological features are to be classified as a new type, it should also be demonstrated that this type exists as a population and adequately covers the retina.

The species variations in horizontal cell morphology add to the recognition that the basic blueprint of the mammalian retina is more flexible than is commonly assumed (Figure 7). Currently, there is little clarity regarding the meaning of these variations. The absence of the A-type in murids is of practical significance since genetically modified mouse strains are heavily used to elucidate the general principles of mammalian retinal wiring and function. Researchers have to be aware that mouse data on the contribution of horizontal cells to retinal processing may not be readily transferable to other species. On the other hand, mice may provide key insights into the effects that one versus two horizontal cell types have on the properties of ganglion cell receptive fields.

Species with Selective Cone Contacts

Most mammals are cone dichromats with a majority of L-cones and a minority of S-cones. In Old World primates and man, the mammalian L opsin gene has diverged into separate genes for the M (green) and L (red) cone opsins, making these species cone trichromats with refined color vision. Most New World monkeys are cone dichromats by genotype; however, some species were shown to have an L opsin polymorphism that results in a trichromatic phenotype in many females, while the males are dichromats. Hence, the retina of primates was an interesting place to investigate whether the horizontal cells are chromatically selective and might contribute to color processing.

Observations on Golgi-stained human and monkey horizontal cells had indicated that H1 dendrites specifically avoid or undersample the S-cones, whereas H2 dendrites contact all cones and H2 axons exclusively contact S-cones. In a seminal study, Dennis Dacey and his colleagues recorded the responses of H1 and H2 cells to chromatic stimuli in the isolated living macaque retina and demonstrated that H2 cells show the same hyperpolarization at all wavelengths, whereas H1 cells hyperpolarize to M- and L-cone stimuli but do not respond to S-cone stimuli. The recorded cells were injected with the tracer neurobiotin, labeling all members of patches of H1 or H2 cells around an injected cell through the homotypic gap junctions. The population of H1 cells innervates the majority of the cones, but hardly contacts the small fraction of presumed S-cones (Figure 6(b)). The H2 population, on the other hand, connects to all cones, but makes particularly numerous contacts with the small fraction of presumed S-cones (Figure 6(c)). Studies combining dye-labeling of individual horizontal cells and cone opsin labeling confirmed this connectivity pattern for macaque, orangutan, and chimpanzee. They showed that only close to 15% of the H1 cells contact S-cones, but then only sparsely; the H2 cells contact all cones within reach and have more synapses with each S-cone than with each of the M- and L-cones contacted. The axon of the H2 cell definitely contacts S-cones, but whether it does so exclusively has yet to be determined. Despite the differences in cone connectivity, the physiological recordings indicate that neither the H1 nor the H2 cell is involved in creating spectrally opponent receptive fields of bipolar and ganglion cells.

New World monkeys (where there are dichromatic and trichromatic individuals) have the same cone connectivity pattern of H1 and H2 cell dendrites as macaque and man. This suggests that the special horizontal cell connectivity with S-cones is not correlated with the evolution of trichromacy in the Old World primates. Notably, both horizontal cell types of trichromatic primates are nonselective in their M-cone and L-cone contacts. Apparently, when the red/green processing pathway evolved from a

presumably dichromatic early primate retina, this did not involve alterations in horizontal cell connectivity.

A differential cone connectivity of horizontal cells was also found in the dichromatic horse (Figure 3). Its A-type cells have particularly large and sparsely branched dendritic fields with few and widely spaced terminal aggregates. In one Lucifer yellow-filled A-type cell, counterstained with an S-cone marker, all but one of its 45 contacts were with S-cones. Further cells need to be studied to confirm this finding. However, the A-type, similar to the B-type, is a consistently occurring cell population that covers the horse retina (Figure 4(b)). Therefore, the horse (and other equids) may possess an A-HC that is selectively connected to S-cones, while the B-type connects to both types of cone.

A similarly S–cone-selective horizontal cell may be present in the cone-dominated retinas of the dichromatic sciurids. The red squirrel and ground squirrel possess an axon-bearing H1 cell (B-type) and an axonless H2 cell (A-type). The density of dendritic terminal aggregates on the H1 cell is high enough to contact all cones present. In contrast, the terminal aggregates on H2 dendrites are spaced so far apart that they can only contact a small fraction of the cones, which presumably are S-cones. It would be worthwhile and now feasible to determine, experimentally, if an S-cone selective horizontal cell is present.

Conclusions and Open Questions

The principal horizontal cell dichotomy of an axon-bearing B-type, which serves rods and cones, and a commonly axonless A-type that only serves cones, holds for most mammals. Thus, the A-type and B-type cells represent basic components of the mammalian retinal blueprint, suggesting an indispensable function. The B-types are necessary in all mammals as they all have cones and rods. However, why is there an additional A-type in the cone pathway? What functional differences are there between the A-type and the B-type? Modeling suggests that the two types with their different dendritic field sizes and spatial summation properties can explain the assumed receptive field characteristics of the cones, and, hence, the receptive field organization of bipolar cells and ganglion cells. The requirements for temporal processing may be another reason for having more than one horizontal cell type. Indeed, horizontal cells in cat differ in their flicker-response properties. However, despite the absence of the A-type, rat and mouse have ganglion cells with center/surround receptive field organizations very similar to rabbit and cat.

Mammalian horizontal cells are rather diverse across species. This diversity encompasses morphological features as well as details of cone connectivity (Figures 7 and 3). Thus, some amendments to the textbook characterizations are necessary:

1.Dendritic thickness and dendritic branching pattern are not defining characteristics of either type, unless the order of mammal is specified. Within a given species, the two types differ in their dendritic morphology, which suggests some physiological difference. Does it matter whether the finer dendrites are on the B-type (as in cat and rabbit) or on the A-type (as in primates and artiodactyls)?

2.Across species and orders, the A-type is more variable than the B-type. A-type variability includes axon-like processes in primates, an S-cone preference in primates and horse, and a complete lack of the A-type in some rodents. B-type variability includes dendritic and axonal branching patterns and the near avoidance of S-cones by primate H1 cells.

There is no obvious correlation between these horizontal cell variations and the phylogenetic distance of the corresponding mammalian orders. For example, Carnivora and Lagomorpha are phylogenetically less close than Primates and Scandentia, but the former have very similar horizontal cell morphologies, while those of the latter differ significantly. The peculiarity of a B-type/H1 cell with fine dendrites and an A-type/H2 cell with stout dendrites is shared by artiodactyls, perissodactyls, and primates. Within any one taxon, horizontal cell features are commonly more conserved across species.

There is no obvious correlation between these horizontal cell variations and specific adaptations to different visual requirements and other retinal specializations. Different cone/rod ratios, indicative of a nocturnal or diurnal lifestyle, do not predict the presence of one or two horizontal cell types. Primates with highly developed trichromatic color vision have cone-selective horizontal cells, but so does the dichromatic horse. In addition, the cone selectivity of H1 and H2 cells is the same in New World and Old World primates, which have different levels of color vision.

See also: Cone Photoreceptor Cells: Soma and Synapse; Information Processing: Bipolar Cells; Information Processing: Horizontal Cells; Morphology of Interneurons: Bipolar Cells; The Physiology of Photoreceptor Synapses and Other Ribbon Synapses; Rod Photoreceptor Cells: Soma and Synapse.

Further Reading

Ahnelt, P. and Kolb, H. (1994). Horizontal cells and cone photoreceptors in human retina: A Golgi-electron microscopic study of spectral connectivity. Journal of Comparative Neurology 343: 406–427.

Boycott, B. B., Peichl, L., and Wa¨ssle, H. (1978). Morphological types of horizontal cell in the retina of the domestic cat. Proceedings of the Royal Society (London) B 203: 229–245.

Chan, T. L. and Gru¨nert, U. (1998). Horizontal cell connections with short wavelength-sensitive cones in the retina: A comparison

between New World and Old World primates. Journal of Comparative Neurology 393: 196–209.

Dacey, D. M., Lee, B. B., Stafford, D. K., Pokorny, J., and Smith, V. C. (1996). Horizontal cells of the primate retina: Cone specificity without spectral opponency. Science 271: 656–659.

Hack, I. and Peichl, L. (1999). Horizontal cells of the rabbit retina are non-selectively connected to the cones. European Journal of Neuroscience 11: 2261–2274.

Mills, S. L. and Massey, S. C. (1994). Distribution and coverage of A- and B-type horizontal cells stained with neurobiotin in the rabbit retina. Visual Neuroscience 11: 549–560.

Peichl, L. and Gonza´lez-Soriano, J. (1994). Morphological types of horizontal cell in rodent retinae: A comparison of rat, mouse, gerbil and guinea pig. Visual Neuroscience 11: 501–517.

Peichl, L., Sandmann, D., and Boycott, B. B. (1998). Comparative anatomy and function of mammalian horizontal cells. In Chalupa, L. M. and Finlay, B. L. (eds.) Development and Organization of the Retina: From Molecules to Function, pp. 147–172. New York: Plenum Press.

Perlman, I., Kolb, H., and Nelson, R. (2003). Anatomy, circuitry, and physiology of vertebrate horizontal cells. In Chalupa, L. M and

Werner, J. S. (eds.) The Visual Neurosciences vol. I, pp. 369–394. Cambridge, MA: The MIT Press.

Smith, R. G. (2008). Contributions of horizontal cells. In Masland, R. H. and Albright, T. (eds.) The Senses: A comprehensive reference vol. 1, pp. 341–349. Amsterdam: Elsevier.

Wa¨ssle, H., Peichl, L., and Boycott, B. B. (1978). Topography of horizontal cells in the retina of the domestic cat. Proceedings of the Royal Society (London) B 203: 269–291.

Wa¨ssle, H., Dacey, D. M., Haun, T., et al. (2000). The mosaic of horizontal cells in the macaque monkey retina: With a comment on biplexiform ganglion cells. Visual Neuroscience 17: 591–608.

Relevant Website

http://webvision.med.utah.edu – Webvision: The Organization of the Retina and Visual System.

Glossary

Retinitis pigmentosa – A group of inherited disorders characterized by progressive loss of photoreceptors.

Sublamina of inner plexiform layer – Inner plexiform layer is divided into sublamina a (the distal sublamina) and sublamina b (the proximal sublamina). Sublamina a is further divided into two strata, 1 and 2, whereas sublamina b is divided into three strata, 3–5.

Introduction

The canonical flow of visual information in the retina is for light stimuli to be transduced into neurochemical signals by the rod and cone photoreceptors in the outer retina and then passed through the synaptic layers (outer plexiform layer, OPL; inner plexiform layer, IPL) to ganglion cells which then transmit them to the rest of the brain via the optic nerve. Interplexiform neurons are a unique class of retinal amacrine cells that form an intraretinal feedback pathway transmitting adaptational visual signals in the opposite direction, from the inner retina back to the outer retina (Figure 1). They receive their synaptic input in the IPL and make their output via interplexiform processes terminating in the OPL. These intraretinal feedback neurons are present in most vertebrate retinas, and in the majority of species these neurons secrete the neurotransmitter dopamine.

Dopaminergic interplexiform cells (IPCs) exert widespread influence on the physiology and function of the retina, reconfiguring retinal circuits and altering the processing of visual signals in the retina, by initiating slow and sustained changes in the physiology of retinal neurons and synapses. In particular, retinal dopamine has been found to act on electrical synapses, or gap junctions, to restrict the flow of visual signals in retinal neural networks at the level of photoreceptors, horizontal cells, and amacrine cells. In addition to the direct synaptic contacts from interplexiform processes in the outer plexiform layer, dopaminergic retinal neurons exert influence throughout the retina by volume transmission, via perfusion of dopamine beyond synaptic zones. Dopamine has been found to affect all major classes of retinal neurons, from photoreceptors to

ganglion cells. An overall effect on retinal function of these feedback signals from dopaminergic IPCs is to enhance signaling in cone pathways and to decrease the signaling on rod pathways, optimizing retinal circuitry for photopic visual processing during periods of relatively high light levels. However, roles for dopamine in dark-adapted retinal responses have also been shown. Thus, dopamine and IPCs play key roles in optimizing retinal function over the wide dynamic range of light intensities encountered in the visual environment.

Secretion of dopamine by IPCs is controlled by two factors: background illumination and an intrinsic daily clock in the retina, enhancing cone signals in conditions of bright light and during the day, and rod signals in conditions of dim light and during the night. In addition to this adaptational effect on retinal neural networks, dopamine secreted from IPCs has trophic effects on photoreceptor survival and eye growth that are associated with eye diseases. In the following, we discuss the anatomy, physiology, and significance for human visual health of dopaminergic IPCs.

Morphology of Dopaminergic

Interplexiform Neurons

Dopaminergic amacrine (DA) cells are a subpopulation of amacrine cells whose cell bodies lie in the innermost cell row of the inner nuclear layer (INL). DA cell processes ramify extensively in the outermost layer of the IPL where they are both presynaptic to and postsynaptic to other neurons. The total number of DA cells averages 500 per retina, although it varies slightly in different species. IPCs are a class of DA cells in which additional fine processes arising either from the cell body or from one of the dendrites ascend through the INL to the OPL where they form an output plexus for the secretion of dopamine in the outer retina. The IPC was first described in the retina of cat by Gallego in 1971, and was extensively studied by Dowling and co-authors. The percentage of DA-IPCs among DA cells is species dependent. In goldfish, Cebus monkey, and mouse retinas, almost all DA cells are likely to be DA-IPCs, whereas in rabbit, turtle, salamander, and human retinas, interplexiform processes are rarely observed on DA neurons. Approximately 50% of DA cells are DA-IPCs in the rat, Xenopus and Rhesus monkey retinas.

Morphology and Distribution

Antibodies against tyrosine hydroxylase, the rate-limit- ing enzyme of catecholamine synthesis, are commonly used to detect DA cells for characterization of their localization, shape, size, dendritic arborization, and distribution (Figure 2). In order to enable targeting of living DA cells for morphological, functional, and molecular analysis, transgenic mouse lines have been created in which the DA cells are labeled with chemical or fluorescent protein reporters, driven by the tyrosine hydroxylase gene promoter. DA cells have three descriptive components to their morphology: soma, dendrites, and axon-like process. The morphology of dopaminergic IPCs is similar to that of other DA cells except for the presence of an additional axon-like process ascending to the OPL.

Soma

Somata of DA cells are either round or ovoid with diameters of 12–15 mm and areas of 100–150 mm2 (Figure 2). In most species, DA cell somata are regularly and sparsely distributed in the innermost aspect of the INL across the entire extent of the retina. The distribution of the DA cells varies across the retina in the rat, with the peak density in the superior temporal quadrant and the lowest density in the inferior nasal quadrant. Occasionally, DA cells are displaced into the ganglion cell layer and, in this case, there is no evidence that these displaced amacrine cells have interplexiform processes.

Dendrites

Two to six primary dendrites arise from the cell body, and branch 4–6 times in the outermost aspect of the OFF

sublamina (stratum 1) of the IPL. These branches radiate symmetrically from the soma and follow a straight somatofugal direction. The thickness of the primary dendrites is up to 4.5 mm and decreases at each bifurcation. The primary dendrites are usually smooth, whereas terminal branches exhibit varicosities (up to 0.5 mm in diameter). The long axis of the dendritic field of individual DA cells is approximately 800 mm. The dendritic fields of adjacent DA cells overlap extensively with a coverage factor of 2–4, forming a dense plexus in stratum 1 of the inner plexifom layer (Figure 2). Dendrites from teleost DA cells spread proximally throughout the IPL. In contrast, in mammalian retinas, only occasional processes from the plexus in stratum 1 run deeper into other stratums.

Axon-like fine process

Axon-like processes are quite distinct from the dendrites of DA cells. They are thin, straight, long, and sparsely branched at close to right angles. Teleost DA-IPC axonlike processes form a pronounced plexus in the OPL. DA-IPCs in Cebus monkey have dense plexuses of fine processes in both the outer and inner plexifom layer. In other species, DA cell processess form an extensive dense

network in the IPL, whereas the ascending axon-like processes either do not branch or branch within clusters in the OPL. For instance, in the mouse retina, two to three axon-like processes arise either from the soma or from the primary dendrites. Each axon-like process, with multiple successive bifurcations, covers the extensive area of the retina: total length up to 10–25 mm. Thus, axon-like processes can overrun each other and, together with overlapping dendrite networks, form a particularly dense plexus of dopaminergic processes in stratum 1 of the IPL. This process network forms small rings around the origin of the primary dendrites of AII amacrine cells at the INL–IPL margin. Axon-like processes bear varicosities that are thought to be sites of dopamine release. Compared to the dense plexus in the IPL, the OPL plexus of fine dopaminergic processes is more loosely arranged. Axon-like processes also traverse the INL to the OPL, sometimes forming clusters of fine processes (Figure 2).

Synaptic Input to DA Neurons

Dopaminergic IPC processes extend widely into the outer and the IPLs. There is anatomical evidence that DA neurons receive synaptic input in the IPL from bipolar

and amacrines cells, as well as centrifugal fibers arising from brain nuclei, and there is functional evidence for retrograde input to DA neurons from ganglion cells.

Input from bipolar cells and amacrine cells

DA cell interplexiform processes are presynaptic to horizontal and bipolar cells in the OPL, and there is no evidence that interplexiform processes are postsynaptic to any cells in the OPL. Electron microscopy of tyrosine hydroxylase immunoreactivity in rhesus monkey, cat, and rabbit has shown that DA cells receive synaptic input from both bipolar cells and inhibitory amacrine cells in mammalian retinas. The size, structure, and position of the bipolar to DA neuron synapses in the IPL suggest that they are from bistratified bipolar cells. Bistratified bipolar cells have also been found in the mouse, fish, turtle, and salamander retinas. In addition, close apposition of bipolar and DA neuron processes has also been observed in deeper layers of the IPL. Therefore, light information generated by photoreceptors may reach DA cells directly from bipolar cells or indirectly from bipolar cells through inhibitory amacrine cells (Figure 3).

Input from intrinsically photoreceptive ganglion cells

Intrinsically photoreceptive ganglion cells (ipRGCs) are endogenously photosensitive because they express melanopsin, a photopigment first described in photosensitive dermal melanophores of the frog. They comprise 1–2% of

retinal ganglion cells forming a photosensitive network in the inner retina. The ipRGC dendritic processes ramify in the OFF layer of the IPL where they have close contact with DA neuron processes (Figure 3). Physiological experiments have suggested that DA cells receive synaptic input from ipRGCs (below). Whether DA cells are postsynaptic to the dendrites of ipRGCs is under investigation.

Input from centrifugal fibers

Many vertebrate retinas receive input from other parts of the brain via retinopetal axons. The efferent input to the retina originates from the ventral thalamus in reptiles, the olfactory bulb in fish, the isthmo-optic nucleus in birds, and the tuberomammillary nucleus of the posterior hypothalamus in mammals. In fish centrifugal fibers make synapses directly onto DA-IPCs. Histamine has been localized to retinopetal axons in the guinea pig, monkey, and rat, and evidence suggests that DA cells receive input from histaminergic centrifugal fibers in mammalian retinas. Thus, IPCs likely form a conduit for feedback to the retina from the rest of the brain as well as for intraretinal feedback signals.

Physiology of Dopaminergic

Interplexiform Neurons

Dopamine Reconfigures Retinal Circuits

DA neurons comprise the central neuromodulatory system of the retina, forming an intraretinal feedback