Layers of the Retina

The 10 layers of the retina are shown in Figure 8-2. The retina begins with the pigmented epithelium (layer 1), which is just inside the choroid. The pigment cells have tentacle-like processes that extend into the photoreceptor layer (layer 2) and surround the outer segments of the rods and cones. These processes prevent transverse scatter of light between photoreceptors. In addition, they serve a mechanical function in maintaining contact between layers 1 and 2 so that the pigmented epithelium can (1) provide nutrients and remove waste from the photoreceptors; (2) phagocytose the ends of the outer segments of the rods, which are continuously shed; and (3) reconvert metabolized photopigment into a form that can be reused after it is transported back to the photoreceptors.

Figure 8-2 Layers of the retina. Light impinging on the retina comes from the top of the figure and passes through all the superficial layers to reach the photoreceptor rods and cones.

Light rays that originate from different parts of the visual target map onto the photoreceptor array of layer 2 in a point-to-point fashion. Retinal glial cells, known as Müller cells, play an important role in maintaining the internal geometry of the retina. Müller cells are oriented radially, parallel to the light path through the retina. The outer ends of Müller cells form tight junctions with the inner segments of the photoreceptors. The numerous connections made between Müller cells and the inner segments give the appearance of a continuous layer, the outer limiting membrane (layer 3 of the retina).

Inside the external limiting membrane is the outer nuclear layer (layer 4 of the retina) that contains the cell bodies and nuclei of the rods and cones. The next layer of the retina (layer 5) is called the outer plexiform layer. It contains synapses between the photoreceptors and retinal interneurons, including the bipolar cells and horizontal cells, whose cell bodies are found in the inner nuclear layer (layer 6 of the retina). This layer also contains the cell bodies of other retinal interneurons (the amacrine and interplexiform cells) and the Müller cells.

The next layer is the inner plexiform layer (layer 7 of the retina). It contains synapses between the retinal neurons of the inner nuclear layer, including the bipolar and amacrine cells, and the ganglion cells. Layer 8 of the retina is the ganglion cell layer. As previously mentioned, the ganglion cells are the output cells of the retina; it is their axons that transmit visual information to the brain. These axons form the optic fiber layer (layer 9 of the retina), pass along the vitreous surface of the retina while avoiding the fovea, and enter the optic disc, where they leave the eye in the optic nerve. The portions of the ganglion cell axons that are in the optic fiber layer remain unmyelinated, but the axons become myelinated after they reach the optic disc. The lack of myelin where the axons cross the retina is a specialization that helps permit light to pass through the inner retina with minimal distortion.

The innermost layer of the retina is the inner limiting membrane (layer 10 of the retina). This layer is formed by the end-feet of Müller cells.

Structure of Photoreceptors: Rods and Cones

Each rod or cone photoreceptor cell is composed of a cell body (in layer 4), an inner segment and an outer segment that extend into layer 2, and a synaptic terminal that projects into layer 5 (Fig. 8-3). The outer segments of cones are not as long as those of rods, and they contain stacks of disc membranes formed by infoldings of the plasma membrane. The outer segments of rods are longer, and the stacks of membrane discs float freely in the outer segment after having disconnected from the plasma membrane when formed at the base. Both sets of discs are rich in photopigment molecules, but the greater photopigment density of rods partly accounts for their greater sensitivity to light. A single photon can elicit a rod response, whereas several hundred photons may be required for a cone response.

Figure 8-3 Rods and cones. The drawings at the bottom show the general features of a rod and a cone. The insets show the outer segments.

The inner segments of the photoreceptors are connected to the outer segments by a modified cilium that contains nine pairs of microtubules, but it lacks the two central pairs of microtubules found in most cilia. The inner segments contain a number of organelles, including numerous mitochondria.

The photopigment is synthesized in the inner segment and incorporated into the membranes of the outer segment. In rods, the pigment is inserted into new membranous discs, which are then displaced distally until they are eventually shed at the apex of the outer segment. There, they are phagocytozed by cells of the pigmented epithelium. This process determines the rod-like shape of the outer segments of rods. In cones, the photopigment is inserted randomly into the membranous folds of the outer segment, and shedding, comparable to that seen in rods, does not take place.

Regional Variations in the Retina

The macula lutea is the area of central vision and is characterized by a slight thickening and a pale color. The thickness is due to the high concentration of photoreceptors and interneurons, which are needed for high-resolution vision. The pale color is a consequence of the fact that both optic nerve fibers and blood vessels are routed around it.

The fovea, which is a depression in the macula lutea, is the region of the retina that has the very highest visual resolution. Correspondingly, the image from the fixation point is focused on the fovea. The retinal layers in the foveal region are unusual because several of them appear to be pushed aside into the surrounding macula. Because light can reach the foveal photoreceptors without having to pass through the inner layers of the retina, both image distortion and light loss are minimized. The fovea has cones with unusually long and thin outer segments. This cone shape permits high packing density. In fact, cone density is maximal in the fovea, and this high density provides for high visual resolution, as well as high quality of the image (Fig. 8-4).

Figure 8-4 This graph plots the density of cones and rods as a function of retinal eccentricity from the fovea. Note that cone density peaks at the fovea, rod density peaks at about 20 degrees eccentricity, and no photoreceptors are found at the optic disc.

(Data from Cornsweet TN: Visual Perception. New York, Academic Press, 1970.)

The optic disc lacks photoreceptors and therefore lacks photosensitivity. Thus, the optic disc is a “blind spot” in the visual surface of the retina. A person is normally unaware of the blind spot, both because the corresponding part of the visual field can be seen by the contralateral eye and because of the psychological process in which incomplete visual images tend to be completed perceptually.

Visual Transduction

Light energy must be absorbed for it to be detected by the retina. Light absorption is accomplished by the visual pigments, which are located in the outer segments of the rods and cones. The pigment found in the outer segments of rods is rhodopsin, or visual purple (so named because it has a purple appearance after green or blue light have been absorbed), and it absorbs light best at a wavelength of 500 nm. Three variants of visual pigment are found in cones, and these cone pigments absorb best at 419 nm (blue), 533 nm (green), or 564 nm (red). However, the absorption spectrum of visual pigments is broad so that they overlap considerably (Fig. 8-5).

Figure 8-5 The spectral sensitivity of the three types of cone pigments in the human retina is shown. Note that the curves overlap and that the so-called blue and red cones actually absorb maximally in the violet and yellow range, respectively.

(Data from Squire LR et al [eds]: Fundamental Neuroscience. San Diego, CA, Academic Press, 2002.)

Rhodopsin is formed when a retinal isomer, 11-cis retinal, is combined with a glycoprotein known as opsin. When rhodopsin absorbs light, it is “boosted” to a higher energy state. This boost causes a series of chemical changes that lead to isomerization of 11-cis retinal to all-trans retinal, release of the bond with opsin, and conversion of retinal to retinol. Separation of all-trans retinal from opsin causes bleaching of the visual pigment; that is, the pigment loses its purple color.

In darkness, photoreceptors are slightly depolarized (around −40 mV) because cGMP-gated Na⁺ channels (Fig. 8-6, A) in their outer segments are open, thereby increasing g_Na and driving the membrane potential toward the Na⁺ equilibrium potential. This net influx of Na⁺ results in a continuous current, called the dark current. As a consequence of this constant depolarization, the neurotransmitter glutamate is tonically released at the rod cell’s synapses. Intracellular [Na⁺] is kept at a steady-state level by the pumping action of Na⁺,K⁺-ATPase.

Figure 8-6 A, Drawing of a rod. The flow of current in the dark is indicated, as well as the Na⁺ pump. B, Sequence of the second messenger events that follow the absorption of light. cGMP, cyclic guanosine monophosphate; GC, guanylate cyclase; GTP, guanosine triphosphate; PDE, phosphodiesterase; Rh, rhodopsin; T, transducin.

When light is absorbed, the photoisomerization of rhodopsin activates a G protein called transducin (Fig. 8-6, B). This G protein, in turn, activates cyclic guanosine monophosphate phosphodiesterase, which is associated with the rhodopsin-containing discs, hydrolyzes cGMP to 5′-GMP, and lowers the cGMP concentration in the rod cytoplasm. The reduction in cGMP leads to closing of the cGMP-gated Na⁺ channels, hyperpolarization of the photoreceptor membrane, and a reduction in the release of transmitter. Thus, cGMP acts as a “second messenger” to translate reception of a photon by the photopigment into a change in membrane potential.

The extraordinary sensitivity of rods, which can signal the capture of a single photon, is enhanced by an amplification mechanism such that photoactivation of only one rhodopsin molecule can activate hundreds of transducin molecules. In addition, each phosphodiesterase molecule hydrolyzes thousands of cGMP molecules per second. Similar events occur in cones, but the membrane hyperpolarization occurs much more quickly than in rods, and requires thousands of photons.

Thus, in all photoreceptors, capture of light energy leads to (1) hyperpolarization of the photoreceptor and (2) a reduction in the release of transmitter. Note that with the very short distance between the site of transduction and the synapse, this transmitter modulation is accomplished without the generation of an action potential.

Visual Adaptation

Adaptation permits the retina to adjust its sensitivity to large changes in ambient lighting, such as you experience when entering a darkened movie theater or, later, leaving to encounter afternoon sunlight. Light adaptation is associated with a reduction in the amount of rhodopsin and the resulting reduced photosensitivity. In bright light, 11-cis retinal is isomerized into the all-trans form, which then splits from the opsin. To regenerate the rhodopsin, the all-trans retinal is transported to the retinal pigmented cell layer to be reduced to retinol, isomerized, and esterified back to 11-cis retinal. It is then transported back to the photoreceptor layer, taken up by outer segments, and recombined with opsin to regenerate the rhodopsin. Light adaptation, which occurs rapidly, within seconds, favors cone vision because the rhodopsin in rods bleaches (separates from its opsin) more readily than the cone pigments do.

The regeneration of photopigment is also involved in dark adaptation, a process that results in an increase in visual sensitivity. Cones adapt more rapidly to darkness than rods do, but their adapted threshold is relatively high. Thus, cones do not function when the ambient light level is low. By contrast, rods adapt to darkness slowly, as their sensitivity increases. Within 10 minutes in a dark room, rod vision is more sensitive than cone vision.

Dark adaptation is very familiar to moviegoers, who must wait several minutes after entering the darkened theater before they can see an empty seat. Although the theater is dark and rod vision is operative, visual acuity is low and colors are not distinguished (this is called scotopic vision). When the movie is projected, however, cone function resumes (this is called photopic vision), and visual acuity and color vision are restored.

Color Vision

The three visual pigments in the cone outer segments have opsins that differ from the opsin found in rhodopsin. As a result of these differences, the three types of cone pigments absorb light best at different wavelengths. Although the cone pigments have maximum efficiency closer to violet, green, and yellow wavelengths, they are referred to as blue, green, and red pigments, respectively (Fig. 8-5).

According to the trichromacy theory, these differences in absorption efficiency are presumed to account for color vision because a suitable mixture of three colors can produce any other color. However, a neural system must also exist for the analysis of color brightness because the amount of light absorbed by a visual pigment, as well as the subsequent response of the cell, depends on both the wavelength and the intensity of the light (Fig. 8-5). Two or three of the cone pigments may absorb a particular wavelength of light, but the amount absorbed by each will differ according to their efficiencies at that wavelength. If the intensity of the light is increased (or decreased), all will absorb more (or less), but the ratio of absorption among them will remain constant. Consequently, there must be a neural mechanism to compare the absorption of light of different wavelengths by the different types of cones for the visual system to distinguish different colors. At least two different kinds of cones are required for color vision. The presence of three kinds decreases the ambiguity in distinguishing colors when all three absorb light, and it ensures that at least two types of cones will absorb most wavelengths of visible light.

The opponent process theory is based on observations that certain pairs of colors seem to activate opposing neural processes. Green and red are opposed, as are yellow and blue, as well as black and white. For example, if a gray area is surrounded by a green ring, the gray area appears to acquire a reddish color. Furthermore, a greenish red or a bluish yellow color does not exist. These observations are supported by findings that neurons activated by green are inhibited by red. Similarly, neurons excited by blue may be inhibited by yellow. Neurons with these characteristics are found both in the retina and at higher levels of the visual pathway and seem to serve to increase our ability to see the contrast between opposing colors.

Retinal Circuitry

A diagram of the basic circuitry of the retina is shown in Figure 8-7. Photoreceptors (R) synapse on the dendrites of bipolar cells (B) and horizontal cells (H) in the outer plexiform layer. The horizontal cells make reciprocal synaptic connections with photoreceptor cells, are electrically coupled to other horizontal cells, and receive input from interplexiform cells (I). Bipolar cells synapse on the dendrites of ganglion cells (G) and on the processes of amacrine cells (A) in the inner plexiform layer. Amacrine cells connect with ganglion cells, other amacrine cells, and interplexiform cells.

Figure 8-7 Basic retinal circuitry. The arrow at the left indicates the direction of light through the retina. A, amacrine cells; B, bipolar cells; G, ganglion cells; H, horizontal cells; I, interplexiform cells; R, photoreceptors.

Several features of this circuitry are noteworthy. Input to the retina is provided by light striking the photoreceptors. The output is carried by axons of the retinal ganglion cells to the brain. Information is processed within the retina by the interneurons. The most direct pathway through the retina is from a photoreceptor to a bipolar cell and then to a ganglion cell (Fig. 8-7). More indirect pathways that provide for intraretinal signal processing involve photoreceptors, bipolar cells, amacrine cells, and ganglion cells, as well as horizontal cells to provide lateral interactions between adjacent pathways. Interplexiform cells allow interactions to occur from the inner to the outer retina.

Contrasts in Rod and Cone Pathway Functions

Rod and cone pathways have several important functional differences, based partly on differences in their phototransduction mechanisms and partly on retinal circuitry. As described previously, rods have more photopigment and a better signal amplification system than cones do, and there are many more rods than cones. Thus, rods function better in dim light (scotopic vision), and loss of rod function results in night blindness. In addition, all rods contain the same photopigment, so they cannot signal color differences. Furthermore, because many rods converge onto individual bipolar cells, thereby resulting in very large receptive fields, rods cannot provide high-resolution vision. Finally, in bright light most rhodopsin is bleached, so rods no longer function under photopic conditions.

Cones have a higher threshold to light and thus are not activated in dim light after dark adaptation. However, they operate very well in daylight. They provide high-resolution vision because only a few cones converge onto individual bipolar cells in the cone pathways. Moreover, no convergence occurs in the fovea where the cones make one-to-one connections to bipolar cells. As a result of the reduced convergence, cone pathways have very small receptive fields and can resolve stimuli that originate from sources very close to each other. Cones also respond to sequential stimuli with good temporal resolution. Finally, cones have three different cone photopigments. Thus, they can discriminate relative spectral content independent of absolute intensity and therefore provide for color vision. Loss of cone function results in functional blindness; rod vision is not sufficient for normal visual requirements.

Synaptic Interactions

The distances between retinal components are short. Hence, modulated transmitter release and postsynaptic potentials are sufficient for most of the activity in retinal circuits, and action potentials are not required in most of the interneurons. Only the ganglion cells and some amacrine cells generate action potentials. It is unclear why amacrine cells have action potentials, but ganglion cells must generate them to transmit information over the relatively long distance from the retina to the brain.

Although receptor potentials in photoreceptors are hyperpolarizing, synaptic potentials in the retina can be either hyperpolarizing or depolarizing. Hyperpolarizing events reduce neurotransmitter release from the synaptic terminals of a retinal interneuron, whereas depolarizing events increase neurotransmitter release.

Receptive Field Organization

The receptive field of an individual photoreceptor is small and circular. Light in the receptive field will hyperpolarize the photoreceptor cell and cause it to release less neurotransmitter. The receptive fields of photoreceptors and retinal interneurons determine the receptive fields of the retinal ganglion cells onto which their activity converges. The characteristics of the receptive fields of retinal ganglion cells constitute an important step in visual information processing because it is this processed information about visual events that is conveyed to the brain.

A bipolar cell that receives input from a photoreceptor can have either of two types of receptive fields, as shown in Figure 8-8. Both are described as having a center-surround organization in which the light that strikes the central region of the receptive field either excites or inhibits the cell, whereas the light that strikes the annular region that surrounds the central portion has the converse effect. The receptive field with a centrally located excitatory region surrounded by an inhibitory annulus is called an on-center, offsurround receptive field (Fig. 8-8, A). Bipolar cells with such a receptive field are described as “on” bipolars. The other type of receptive field has an off-center, on-surround arrangement, which characterizes “off” bipolars (Fig. 8-8, F).

Figure 8-8 The receptive fields of on-center (A) and off-center (F) bipolar and ganglion cells. B and G show the responses of these ganglion cells to central (upper record) and peripheral (lower record) spots. Also shown are responses to central (C and H), surround (D and G), and diffuse whole field (E and J) Illumination in their receptive fields. The on-center and off-center bipolar cells providing input to these ganglion cells have similar receptive fields, but wheras ganglion cells increase or decrease their spike frequency, bipolar cells depolarize or hyperpolarize, without generating action potentials.

(Redrawn from Squire LR et al [eds]: Fundamental Neuroscience. San Diego, CA, Academic Press, 2002.)

The receptive fields of bipolar cells depend on input from photoreceptors and from horizontal cells. The neurotransmitter used in the retinal pathway from photoreceptor cells to bipolar cells and to horizontal cells is the excitatory amino acid glutamate. Excitatory amino acids depolarize “off” bipolar cells, as well as horizontal cells, through the activation of ionotropic glutamate receptors. These are called “off” bipolars because when light is removed from the receptive field center, the photoreceptor is depolarized and releases more glutamate to depolarize the bipolar cell. In contrast, “on” bipolar cells have metabotropic glutamate receptors that close their Na⁺ channels, and thus “on” bipolars are depolarized by turning the light on because the reduced release of glutamate results in more influx of Na⁺.

In other words, if the neurotransmitter tonically released by the photoreceptor hyperpolarizes the bipolar cell, absorption of light will hyperpolarize the photoreceptor and thereby reduce its release of the neurotransmitter; the “on” bipolar cell will be depolarized (disinhibited) and thus excited. On the other hand, the neurotransmitter tonically released by the photoreceptor depolarizes the “off” bipolar cell, and it will be hyperpolarized (disfacilitated) by central illumination.

The central property of bipolar cell receptive fields is due to only a few directly connected photoreceptors. The antagonistic surround response is due to light impinging on adjacent photoreceptors, which changes the activity of horizontal cells. This pathway through the horizontal cells results in a response that is opposite in sign to that produced directly by the photoreceptors that mediate the center response. The basis for this is that horizontal cells, like “off” bipolars, are hyperpolarized in the light and, because they are electrically coupled to each other by gap junctions, have very large receptive fields. Darkness in the periphery of a bipolar cell’s receptive field (such as an annulus that does not affect the photoreceptors to which it is directly connected) will depolarize neighboring photoreceptors and horizontal cells. Depolarized horizontal cells release GABA onto central (and peripheral) photoreceptor terminals, reducing their release of glutamate. Thus, when a darkness surrounds central illumination, there is increased excitation of on-center bipolars. There is a complementary effect on off-center bipolars when a bright annulus surrounds a central dark spot (Fig. 8-8).

Bipolar cells may not respond at all to large or diffuse areas of illumination, covering both the receptors that cause the surround response and those responsible for the center response because of the opposing actions from the center and surround. Thus, bipolar cells may not signal changes in the intensity of light that strikes a large area of the retina. On the other hand, a small spot of light moving across the receptive field may sequentially and dramatically alter the activity of the bipolar cell as the light crosses the receptive field from surround to center and then back again to surround. This demonstrates that bipolar cells respond best to the local contrast of stimuli and function as contrast detectors.

Amacrine cells receive input from different combinations of on-center and off-center bipolar cells. Thus, their receptive fields are mixtures of on-center and off-center regions. There are many different types of amacrine cells, and they may use at least eight different neurotransmitters. Accordingly, the contributions of amacrine cells to visual processing are complex.

Ganglion cells may receive dominant input from bipolar cells, dominant input from amacrine cells, or mixed input from amacrine and bipolar cells. When amacrine cell input dominates, the receptive fields of ganglion cells tend to be diffuse, and they are either excitatory or inhibitory. Most ganglion cells are dominated by bipolar cell input and have a center-surround organization, similar to that of bipolar cells (Fig. 8-8).

P, M, and W Cells

Experiments have shown that in primates, retinal ganglion cells can be subdivided into three general types called P cells, M cells, and W cells. P and M cells are fairly homogeneous groups, whereas W cells are heterogeneous. P cells are so named because they project to the parvocellular layers of the LGN, whereas M cells project to the magnocellular layers of the LGN. P and M cells have center-surround receptive fields; hence, they are presumably controlled by bipolar cells. W cells may also have center-surround receptive fields, but many have large, diffuse receptive fields (which corresponds to extensive dendritic fields) and slowly conducting axons, and they respond poorly to visual stimuli. They are probably influenced chiefly through amacrine cell pathways, but less is known of them than of M and P cells.

Several of the physiological differences among these cell types correspond to morphological differences (Table 8-1). For example, P cells have small receptive fields (which corresponds to smaller dendritic trees) and more slowly conducting axons than M cells do. In addition, P cells show a linear response in their receptive field; that is, they respond with a sustained, tonic discharge of action potentials to maintained light but do not signal shifts in the pattern of illumination as long as the overall level of illumination is constant. Thus, a small object entering a P cell’s central receptive field will change its firing, but continued movement within the field will not be signaled. P cells respond differently to different wavelengths of light. Because there are blue, green, and red cones, many combinations of color properties are possible, but in fact P cells have been shown to have only opposing responses to red and green or to blue and yellow (a combination of red and green). They may have center-surround antagonism in which one color excites the center while the other inhibits the surround (or vice versa), or one color might excite the entire receptive field while another inhibits it (e.g., R+G− describes a cell that is excited by red and inhibited by green). These mechanisms can greatly reduce the ambiguity of color detection caused by the overlap in cone color sensitivity and may provide a substrate for the opponency process observations.

Table 8-1 Properties of Retinal Ganglion Cells

M cells, on the other hand, respond with phasic bursts of action potentials to the redistribution of light, such as would be caused by the movement of an object within their large receptive fields. M cells are not sensitive to differences in wavelength but are more sensitive to luminance than P cells are.

Thus, the output of the retina consists primarily of ganglion cell axons from (1) sustained, linear P cells with small receptive fields that convey information about color, form, and fine details and (2) phasic, nonlinear M cells with larger receptive fields that convey information about illumination and movement. Both come in on-center and off-center varieties.