visual angle, CPD) at the central retina but not so much better than that of rodents in the peripheral retina (at eccentricities greater than around 12 deg) (28). The highest limit of visual acuity set by the C57BL mouse retina is 50‐fold worse than in humans at 0.6 CPD (9,29). However, visual pattern discrimination is not only determined by visual acuity but also by contrast sensitivity. The peak contrast sensitivity of mice is surprisingly good at around ~2 % compared to that of humans at ~0.5 % (30,31). Despite the differences between the mouse and the human eye, the structure and function of the retina in mice and humans is similar, especially at periphery (7,31).
2.2 FUNCTIONAL ARCHITECTURE OF THE MAMMALIAN RETINA
The mammalian retina displays 10 distinct laminar layers under a light‐microscope (Figs. 2
& 3, note the internal limiting membrane is missing in both images, it limits the nerve fiber layer from the vitreous). Although the light enters from the RGC side of the retina, that side is referred to as inner retina (proximal retina) and photoreceptor side as outer retina (distal retina). The neural retina is circumscribed by retinal pigment epithelium (RPE) on the photoreceptor side (Fig. 3.), and nerve fiber layer (NFL) on the RGC side of the retina.
Energy and nutrient supplies of the retina reside immediately below the RPE and at the NFL (32). Choriocapillaries below the RPE, i.e. choroid, receive massive blood flow and they are vital for the high metabolic demand of the outer retina (33). The actual retinal blood flow enters from the central retinal artery from the optic nerve head forming capillaries that nourish the inner retina (32). The blood flow in the choroid is manifold compared to the blood flow inside the retina. The primary energy source to the retina is provided by glucose. The retina has a high rate of anaerobic glycolysis even under basal physiological conditions, but it can switch to oxidative metabolism on demand (19).
Photoreceptors are metabolically very active due to maintenance of dark current, i.e.
continuous repolarization after depolarization of cell membranes in the dark. In addition, continuous phagocytosis of photoreceptor outer segment (POS) discs by the RPE, and their renewal, uses a lot of energy. Because energy requirements are high, oxygen consumption is also high. The capillary blood flow in the retina has been measured at 60 ml/min/100 g of tissue in primates, being similar to the blood flow in the brain (34). In the choroid, the blood flow is very high at 2000 ml/min/100 g of tissue (35), because the oxygen must diffuse from here to the inner segments of photoreceptors where their mitochondria are located. Oxygen usage by photoreceptors is 3‐4 times higher than in other CNS neurons (19). Thus hypoxic state may impact the retina very quickly (36).
Fig. 3. Schematic image of the retina and rod and cone structures. A: The retinal pigment epithelium (RPE) has numerous supportive functions for the retina, one of the most important ones being continuous phagocytosis of photoreceptor outer segments. From the apical side of RPE start photoreceptors (rods and cones) that convert light to electrical signals. The photoreceptors synapse with bipolar and horizontal cells (modulatory cells providing lateral inhibition) at the outer plexiform layer, and the bipolar cells synaptically transmit the signal further to ganglion cells (or amacrine cells in primary rod-pathway and then to ganglion cells) at the inner plexiform layer. Ganglion cell axons run along the nerve fiber layer and finally form the optic nerve. Müller cells are radial glial cells that are present throughout the entire retina and have a crucial supportive function. B: The rod and cone photoreceptors consist of outer segment (OS), where the visual transduction (see Fig. 4) takes place, connected to inner segment via connecting cilium (CC). The inner segment contains the cell machinery of photoreceptors including mitochondria. The nuclei (N) of photoreceptors are situated at the outer nuclear layer of the retina. Ribbon synaptic terminals functionally connect the photoreceptors to the interneurons. Calyceal processes are found in primate photoreceptors but are absent in mice (37). BB: basal body of the connecting cilium (Figure adapted from (38)).
The RPE is the outermost retinal layer, consisting of a monolayer of pigmented hexagonal cells (19). The basal side of the RPE cell is adjacent to the choroid while the apical side faces the neural retina. The basal aspect of RPE contains numerous infoldings and is adherent to its basement membrane forming a part of Bruch´s membrane of the choroid.
Therefore, the attachment of RPE and the choroid is vivid. The apical side of the RPE comprises microvilli that extend into the photoreceptor outer segments (Fig. 3). However, this subretinal space is loose compared to RPE‐Bruch´s membrane interface and does not contain intercellular junctions. The RPE has many functions such as fostering the retina and choriocapillaries (reviewed in (39)). It forms part of the blood‐retinal barrier (notably similar to choroid plexus in the ventricles of the brain), which selectively controls movement of nutritients and metabolites into the retina, and on the other hand, allows waste products out of the retina (40). The most important functions of RPE (in regard to normal vision) are maintenance of visual cycle (visual cycle explained in Fig. 4) and phagocytosis of shed POS. In mice, 10 % of POS is shed daily and RPE needs to phagocytose and degrade it out of the way for renewal. This mission renders RPE cells one of the most active phagocytosing cell types in the body (41). POS phagocytosis by RPE follows the circadian rhythm, being most active at the beginning of circadian cycle at light‐
onset (42,43). POS phagocytosis can be divided into five different phases: 1. recognition and attachment; 2. ingestion; 3. phagosome formation; 4. phagosome fusion with lysosome; and
finally 5. digestion of POS (44). It is well appreciated that failure at any phase of the POS phagocytosis may lead to retinal degeneration (45). In addition, it has been recently shown that phagocytosis of POS is crucial for the visual cycle at RPE via a novel noncanonical form of autophagy (46).
Fig 4. A graph of visual cycle i.e.
phototransduction. The visual cycle is responsible for the first-in-order visual reaction and regeneration of visual pigment for continuum. Reaction a: 11-cis-retinal (11cRAL) diffuses from the RPE to photoreceptor outer segments (OS) and couples with opsin to generate rhodopsin (Rh). Reaction b: absorption of light photons by rhodopsin leads to isomerization of the chromophore from the 11-cis to the all-trans form (atRAL).
Reaction c: atRAL is reduced to all-trans-retinol (atROL) by all-trans-retinal-spesific dehydrogenases (all-trans-RDH).
Reaction d: atROL diffures to RPE where it is esterified by retinol acyltransferase (LRAT, lecithin) to all-trans-retinyl-ester (atRE). Reaction e: RPE-spesific 65 kDA protein (RPE65) catalyzes the isomerization of atRE to 11cROL. Reaction f: when 11cROL is oxidized back to 11cRAL by 11-cis-RDH, the 11cRAL is ready for diffusion back to the OS and the visual cycle is complete.
Notably, a failure in any step of the visual cycle has been shown to induce retinal degeneration (47). IPM, interphotoreceptor matrix; IRBP, inter-photoreceptor retinol binding protein (Figure adapted from (48)).
The POS consists of membranous discs where visual pigment molecules are located within the disc membrane (19). The outer and inner segments of photoreceptors are connected via the cilium (Fig. 3b). The inner segment is the power plant of the photoreceptor containing lots of mitochondria. The inner segment contains the cell nucleus, and on the other side of the nucleus, synaptic terminals connect photoreceptors to interneurons (21). Adjustable sensivity of the photoreceptors enable vision to function within an impressive dynamic range of over 10‐log units (49). Mammalian and amphibian rods can detect even single photons but they saturate at bright light (50). Still, vision works even at very bright sunlight thanks to rapid regeneration of visual pigments and adaptational mechanisms at the cone pathway (51). Cones are 100 times less sensitive than rods (52), but on the other hand, they do not saturate easily due to their extremely fast pigment regeneration and dark‐adaptation (51).
Unlike most sensory systems where appropriate stimulation causes sensory receptors to depolarize, the photoreceptors act by hyperpolarization and subsequent change in neurotransmitter release onto postsynaptic terminals. Visual sensation commence once absorption of light changes the conformation of retinal and activates (rhod)opsin (see Fig. 4) (21). Activation of opsin stimulates the G‐protein transducin, which then activates phosphodiesterase enzyme (PDE). Once activated, PDE hydrozylases cyclic guanosine monophosphate (cGMP). The decrease in cytosolic cGMP leads to closure of cyclic nucleotide gated ion‐channels preventing influx of Na+ and Ca2+ ions, and thereby hyperpolarizes photoreceptors. The G‐protein cascade substantially potentiates the photoresponse: many transducin molecules are activated with a single 11‐cis‐retinal
photoisomerization and each PDE enzyme hydrozylases more than one cGMP molecule. As a result, absorption of a single photon by a rod leads to closure of approximately 200 ion channels corresponding to about 2 % of all the channels open in each rod in darkness.
The retina modifies the photoresponse amplification magnitude at prevailing levels of illumination (21). This phenomenon is known as light adaptation. As levels of illumination increase, the photoreceptor sensitivity decreases preventing the receptors from saturating, and thereby they increase the range of light intensities at which they may operate. The concentration of Ca2+ in POS is a key player in the light‐induced modulation of photoreceptor sensitivity. Light‐induced closure of ion channels leads to a net decrease of POS Ca2+ concentration that causes many changes in the phototransduction cascade. These changes tend to reduce the receptor sensitivity to light. For instance, the lowered Ca2+
concentration increases activity of guanylate cyclase that synthetizes cGMP, leading to an elevated levels of cGMP. Likewise, the decreased levels of Ca2+ increase the affinity of cGMP‐channels for cGMP diminishing the impact of light‐induced reduction in cGMP levels. The light adaptation mechanism driven by the Ca2+ concentration is not the only light adaptational mechanism in the retina. Retinal sensitivity at background light levels is also neurally modulated. A key neurotransmitter here is dopamine (53). Many dopamine driven physiological mechanisms lead to an increased signal flow through cone circuits and reduced signal flow through rods. In Parkinson´s disease, a reduction in retinal dopamine levels may result in reduced visual contrast sensitivity. In a healhy retina, the light adaptation is a rather fast process: the retina habituates from complete darkness to bright light levels in 5‐10 minutes in humans (19). Vice versa, adaptation from bright sunlight to complete darkness (dark adaptation) is slower and may take 30 minutes.
The inverted mechanism by which photoreceptors act continues at photoreceptor–interneuron synapse. When a photoreceptor is hyperpolarized, activated, the release of its neurotransmitter glutamate into the synaptic cleft at the outer plexiform layer decreases (54). This modulates activity at the interneuron level, in bipolar and horizontal cells (52). Bipolar cells are second‐order visual transmitting cells, whereas horizontal cells regulate the signal transmission between photoreceptors and bipolar cells by lateral inhibition (Fig. 5). Bipolar cells can be divided into two major classes: rod bipolar and cone bipolar cells. Cone bipolar cells form subclasses: ON bipolar cells that depolarize and OFF bipolar cells that hyperpolarize to increments in light levels. Rod bipolar cells are always ON cells. On the other hand, rods have been shown to signal through OFF cone bipolar cells (55). The canonical mechanism by which bipolar, horizontal and ganglion cells make retina possible to detect differences in light increments, contours, is illustrated in Fig.
5.
Fig. 5. A schematic drawing of bipolar-ganglion cell center-surround organization. Figure A.
represents ON-center cell activation and figure B. OFF-center cell activation (Figure adapted from (56)).
Bipolar cells contact RGCs and amacrine cells at the inner plexiform layer (IPL) (52). RGCs are the only output neurons of the retina and they mimic typical CNS neurons composed of a cell body, dendrites and an axon, that is unmyelinated (4). Immediately posterior to the eye globe (in fact already posterior to lamina cribrosa in humans), the RGC axons form the optic nerve that is myelinated by oligodendrocytes, similarly to other nerve fiber tracts in the CNS. Excitation of RGCs at the IPL is modulated in two ways by amacrine cells: 1.
feedforward inhibition from amacrine cell synapses directly onto RGCs dendrites, or 2.
feedback inbihition where amacrine cells contact axon terminals of bipolar cells (52).
Amacrine cells exert their action largely by two fast neurotransmitters: gamma‐
aminobutyric acid (GABA) and glycine. A fundamental difference between cone and rod bipolar cells signaling is that cone bipolars cells make direct synaptic connections to RGCs whereas rod bipolar cells synapse first with AII amacrine cells that then synapse with RGC (55). The visual transmission cascade discussed above is a mere simplification of reality.
Although there are only five classes of neuronal cells in the retina (photoreceptors, bipolar, horizontal, amacrine and RGCs), there are numerous subclasses of bipolar, amacrine and RGCs. Recent estimates suggest that there are at least 11 different subtypes of cone bipolar cells (57), 40 amacrine cell subtypes (58), and 32 RGC subtypes in the mouse retina (1).
Indeed, even the mouse retina is capable of high computation of visual stimuli ranging from direction selectivity to polarity sensitivity.
2.2.1 Retinal neurotransmitters
Neurotransmitter composition seems to be similar in the retina and in the brain. Glutamate is the primary excitatory neurotransmitter throughout the vertical pathways of the retina, whereas GABA is the main inhibitory neurotransmitter (32). ON‐center bipolar cells signal via metabotropic glutamate channel (mGluR6) whereas OFF‐center bipolar cells signal via ionotropic AMPA and kainate receptors (59). Amacrine cells and horizontal cells in most vertebrate retinas exert their inhibitory action mostly via GABA (60). Another typical inhibitory neurotransmitter, glycine, is found in most small‐field types of amacrine cells (32). Modulatory monoamine neurotransmitters, dopamine and serotonin, are also found in amacrine cells (32,61,62), as well as acetylcholine (63). In fact, the two classes of dopamine receptors, D1‐ and D2‐class, are both found in mammalian retina fairly abundantly. D1‐
receptors are expressed in bipolar, ganglion and horizontal cells (64), whereas D2‐receptors are found in the synapse between horizontal cells, bipolar cells, and photoreceptors (65).
Thus, maybe not so surprisingly, drugs acting on monoamine and cholinergic neurotransmitter systems alter retinal function (66,67), and several psychiatric disorders are associated with abnormal ERG responses (66). In addition to above mentioned neurotransmitters, adenosine (68), nitric oxide (69) and several neuropeptides (32,70‐72) contribute more or less to neurotransmission within the retina.
2.2.2 Retinal glial cells
There are three main classes of glial cells in the mammalian retina: Müller cells, astrocytes and microglial cells (73). Müller cells are the most abundant glial cells in the retina, comprising 90 % of total retinal glia (73). Müller cells are radially oriented such that they pass through the retina from its inner vitreal border all the way to distal end of outer nuclear layer. Astrocytes are mainly located horizontally in the nerve fiber layer and they colocalize with the blood vessels in the inner nuclear layer. Indeed, the distribution of retinal astrocytes is always correlated with distribution of blood vessels and thereby astrocytes have a role in forming retina‐blood barrier. All retinal glial cell classes are integrated with the retinal neurons, allowing correct functioning of the retina and providing structural support. They are all phagocytic cells and thereby respond to immunological insult, and they interact with neurons and modulate the synapses. They contribute to neurotransmission by releasing certain neurotransmitters and trophic factors, and they are important buffers for K+ ions. In addition, the surrounding glial cells have an imperative role in supplying different nutritients to retinal neurons as the neurons in the retina are highly specialized and their metabolic demands are highly specific. A specific role of Müller cells is maintenance of cone‐spesific visual cycle, which enables much more rapid chromophore supply and pigment regeneration than the canonical RPE based visual cycle and enables cones to function rapidly even under extremely bright light conditions
(51).
A few decades ago, it was still believed that the inverted orientation of the retina, where light would traverse through many cell layers before hitting the target, was an example of poor evolutional design (74). More recently the claim was proven wrong. It has been shown that Müller cells are wavelength‐dependent guides, guiding green‐red spectrum of the visible light onto cones and blue‐purple part of the spectrum onto nearby rods (75). Thus, Müller cells, which span vertically through the retina, enhance wavelength
targeting in the retina and are crucial for visual optics. That is why the inverted structure of the retina is necessary.
2.3 FUNCTIONAL ORGANIZATION OF THE PRIMARY VISUAL PATHWAYS