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

As mentioned above, the kinesin KIF3A is located at the ribbon, but since ATP-driven molecular motors are not needed for vesicle release from the ribbon, the role of KIF3A at the ribbon is unclear. However, kinesin motor proteins interact with plant homologs of CtBP, indicating that the interaction of kinesins with the major structural protein in ribbon synapses is highly conserved. It has also been proposed that KIF3A may assist with circadian changes in ribbon structure.

Interactions among synaptic proteins help maintain the structure of the ribbon. Interactions between ribeye and the cytomatrix scaffold protein bassoon anchor ribbons to the active zone. Bassoon also tethers ribbons in hair cells and pinealocytes. Bipolar cells lack bassoon but possess a related protein, piccolo. In photoreceptors, piccolo is located further up the ribbon than bassoon.

Interactions between cytoskeletal proteins, membranespanning dystroglycan proteins, and extracellular matrix pikachurin proteins are important for maintaining contacts between photoreceptor ribbon synapses and their postsynaptic targets. Disrupting these interactions leads to reductions in the electroretinogram (ERG) b-wave (indicating a reduction in ON bipolar cell responses) in patients with muscular dystrophy.

Interactions between ribeye and Unc119, a protein that is highly expressed in photoreceptors, may help localize Ca2þ channels to the base of the ribbon. Unc 119 can bind to both ribeye and the calcium-binding protein CaBP4. CaBP4 binds in turn to Ca2þ channels in photoreceptor terminals. Mutations in Unc 119 lead to cone/rod degeneration.

Photoreceptor Calcium Channels

Glutamate release from photoreceptors requires an influx of Ca2þ through Ca2þ channels. However, unlike conventional synapses that use N- and P-type calcium channels, photoreceptors and other ribbon synapses rely on dihydropyridine-sensitive, L-type calcium channels. L- type calcium channels are classified by their a1 pore forming subunits into CaV1.1-CaV1.4 subtypes. Mutations in CaV1.4 (also known as a1F) cause incomplete congenital stationary night blindness and antibodies to CaV1.4 label mammalian rod terminals, suggesting that CaV1.4 is the principal subtype in rods. Rods and longwavelength sensitive cones also appear to possess CaV1.3 channels but CaV1.3 antibodies do not label short-wave- length-sensitive cones, suggesting that they possess a different channel subtype.

Calcium channels cluster beneath the ribbon. Freeze fracture electron micrographs from mammalian cones show clusters of 500 polyhedral transmembrane particles, each with a central dimple, beneath the arciform density of each ribbon. Sites of calcium influx co-localize with

ribeye-binding peptides and antibodies to L-type calcium channels co-localize with antibodies to bassoon and ribeye.

Properties of single-calcium channels recorded from amphibian rod photoreceptors are similar to those of L-type calcium channels in other tissues. Single-CaV1.4 channels expressed in HEK293 cells showed a tiny singlechannel conductance and extremely low open probability. However, it is unlikely that these properties are retained in vivo since they imply an unrealistically large number of channels per ribbon (>15 000).

Photoreceptor calcium currents (ICa) exhibit a sigmoidal voltage dependence (Figure 2). When measured under the same experimental conditions, the voltage dependence of ICa (calcium current) in rods and cones of different species are remarkably similar. Typically, photoreceptor ICa activates above –60 mV and is fully activated around –20 mV. ICa attains about a third of its

−38 mV

Figure 2 Influence of light-evoked changes in membrane potential on cone calcium currents. (a) Response of a salamander cone to a bright flash of light. (b) Calcium current averaged from eight salamander cones evoked by a ramp voltage protocol (0.5 mV ms 1). By convention, the influx of positively charged Ca2þ ions into cones is shown as negative or inward current. The dark potential of the cone in (a) is denoted by the dashed line and the reduction in ICa caused by the hyperpolarizing light response is shown by the arrow.

peak amplitude at the dark resting potential (ca. –40 mV). The hyperpolarizing light response of a rod or cone photoreceptor diminishes ICa and thereby diminishes Ca2þ-dependent release. The sigmoidal voltage dependence of ICa contributes to response compression at higher intensities. As illustrated in Figure 2, the first 10 mV of membrane hyperpolarization during a cone light response causes a much greater decrease in ICa than the next 10 mV. This diminished responsiveness at more hyperpolarized potentials is sufficiently pronounced at rod synapses that it has been described as response clipping. There is evidence that Ca2þ influx through cGMP-gated cation channels can stimulate synaptic release from cones but the role that these channels play in release under normal physiological conditions is not clear.

Photoreceptor ICa shows limited and slowly developing inactivation involving both voltageand calciumdependent mechanisms. Limited inactivation is important for sustaining synaptic release in darkness when photoreceptors are continuously depolarized. Although the amplitude of ICa declines slowly in darkness, changes in ICa produced by brief changes in illumination mirror the sigmoidal voltage dependence of ICa.

Along with pore-forming a1 subunits, calcium channels possess accessory b and a2/d subunits. Knockout of b2 subunits almost completely abolishes both the ERG b-wave and staining for CaV1.4 in the outer plexiform layer indicating that b2 subunits are the predominant subtype at photoreceptor synapses. Mutations in a2/d type 4 subunits lead to disordered ribbons, reduced scotopic b-waves, absent photopic b-waves, and a human cone dystrophy, suggesting that this accessory subunit is associated with calcium channels in photoreceptors, particularly cones.

The calcium-binding protein, CaBP4, is closely associated with photoreceptor calcium channels. When heterologously expressed in the presence of CaBP4, the voltage dependence of CaV1.3 and CaV1.4 ICa is shifted to more negative potentials, similar to the voltage dependence of ICa in photoreceptors. By shifting activation to more positive potentials, mutations in CaBP4 reduce the amplitude of ICa in the normal physiological range and thereby reduce synaptic output from photoreceptors. The reduction in rod output accompanying CaBP4 mutations causes congenital stationary night blindness.

Role of Intracellular Ca2þ in Release

Free calcium levels in the photoreceptor synapse are tightly regulated by a variety of mechanisms to maintain synaptic output and prevent calcium overload during the continual influx of Ca2þ in darkness.

One way to remove Ca2þ from the cytoplasm is to pump it into the endoplasmic reticulum using sarcoand endoplasmic reticulum ATPases (SERCA). SERCA 2A predominates in photoreceptor terminals. The release of calcium from these sequestered stores by calciuminduced calcium release (CICR) can amplify calcium entry through L-type calcium channels. CICR in retina is mediated by a retina-specific variant of the type 2 ryanodine receptor. In rods, CICR amplifies synaptic release and increases the likelihood of the simultaneous fusion of multiple vesicles. CICR also contributes to synaptic release from ribbon synapses in vestibular hair cells. Calcium imaging studies show evidence for CICR in cone cell bodies but it does not appear to contribute to synaptic release. Immunohistochemical studies suggest IP3 receptors, which mediate the release of intracellular Ca2þ in many cell types, may be present in cone terminals. However, there is no physiological evidence for IP3-mediated release of calcium in photoreceptors.

Depletion of calcium from intracellular stores triggers the opening of calcium-permeable channels in the plasma membrane to facilitate store refilling. Store-operated calcium entry can influence synaptic release by regulating basal calcium levels in photoreceptor terminals.

Calcium can also be removed from the cytoplasm by pumping it out of the cell. Calcium is removed from outer segments by a Na/Ca exchanger whereas extrusion from inner segments and synaptic terminals rely more on plasma membrane calcium ATPases (PMCA). PMCA2 antibodies label photoreceptor terminals and PMCA2 knockout mice show significant reductions in rod-driven responses, suggesting that this subtype is particularly important in regulating calcium levels in rod terminals.

Calcium buffering by cytoplasmic proteins provides a much more rapid way to reduce free calcium levels than extrusion. The principle calcium buffers are calbindin, calretinin, and parvalbumin although many signaling proteins (e.g., calmodulin, synaptotagmin, and CaBP4) also bind calcium. There is considerable species variability, but cones typically possess the fast, low mobility buffer calbindin whereas the higher mobility buffers calretinin and parvalbumin are less common in photoreceptors.

Supplementing these mechanisms, large calcium increases in rods and cones can be buffered by mitochondrial uptake.

Physiology of Release at Photoreceptor

Synapses

Photoreceptors hyperpolarize to light and decreases in light intensity cause photoreceptors to depolarize. Depolarization increases the open probability of Ca2þ channels clustered beneath the ribbon. The opening of Ca2þ

channels increases [Ca2þ]i, stimulating fusion of vesicles at the base of the ribbon. The calcium sensors in bipolar and hair cells have a low affinity for calcium requiring >10 mM calcium to stimulate exocytosis. Release from bipolar cells and most other CNS neurons exhibits high cooperativity consistent with the binding of as many as five Ca2þ ions required for release. The release mechanism employed by photoreceptors differs from these synapses by showing a much higher affinity for Ca2þ whereby submicromolar calcium levels can stimulate release. This high affinity is consistent with the possible involvement of synaptotagmin III, which has a higher affinity for calcium than synaptotagmin I or II. In addition to higher Ca2þ affinity, release from photoreceptors shows lower cooperativity for Ca2þ binding (N 3).

Ca2þ channels are close to the ribbon and thus opening of a channel will expose nearby release sites to high [Ca2þ]. Opening only a few channels is sufficient to stimulate fusion of a vesicle. Increasing the number of active Ca2þ channels increases the number of active Ca2þ microdomains and this in turn increases the number of active release sites. Because the number of active release sites increases linearly with ICa amplitude, there is a linear relationship between ICa and release at hair cell synapses. This mechanism may also contribute to linearity between ICa and release at photoreceptor synapses although linearity at this synapse is also promoted by use of a sensor with a low cooperativity for calcium binding (N 3).

Synaptic release at ribbon synapses involves two components: a transient burst of release stimulated by abrupt membrane depolarization and the slower, sustained release that accompanies maintained depolarization. In bipolar cells, as with many other neurons, fast release requires very high levels of Ca2þ whereas slow sustained release is triggered by low Ca2þ levels. The separate control of fast release by low Ca2þ affinity sensors and slow release by high Ca2þ affinity sensors is consistent with the idea that fast and slow release in bipolar and hair cells may occur at ribbon and nonribbon sites, respectively.

Unlike bipolar cells, fast and slow release from photoreceptors exhibit the same high affinity for Ca2þ, suggesting that both components of release occur at the same site. Thus, sustained release from photoreceptors appears to be predominately due to continued release of vesicles from the ribbon, albeit at lower rates than those attained during fast transient release. Because of the high affinity for Ca2þ exhibited by the release apparatus in photoreceptors, micromolar levels of Ca2þ present at the base of the ribbon in darkness are sufficient to stimulate fusion of vesicles at the base of the ribbon almost immediately after docking. As a consequence, the base of the cone ribbon is largely devoid of vesicles in darkness. This means that in darkness, calcium channel openings often occur beneath empty release sites. The rate of sustained release in darkness is

therefore not directly controlled by the stochastic opening of individual Ca2þ channels, but by the rate at which vesicles are delivered and readied for release at the base of the ribbon.

Release rates decline when photoreceptors hyperpolarize, allowing vesicles to be replenished at release sites along the base of the ribbon. With a sufficiently long and bright flash of light, the entire readily releasable pool of vesicles can be replenished. When the cone depolarizes at light offset, the rapid release of this replenished pool of vesicles can evoke a large off response in second-order neurons.

Photoreceptors release vesicles continuously at a rate of 10–20 vesicles per ribbon per second in darkness. Cones can respond to light intensities spanning a 10 000fold range, but this sustained release rate can encode only 10–20 distinguishable levels of steady light if synaptic release exhibits Poisson release statistics. If sustained release is controlled by the rate of vesicle delivery down the ribbon rather than the stochastic openings of individual calcium channels, this will make the rate of sustained release more regular. Regularization allows discrimination of a greater number of light levels than predicted for a Poisson release process. The high rates of release from cones that can be attained at light offset allow for the encoding of up to 100 distinguishable light decrements. This may account for psychophysical results showing a greater sensitivity to decrements than increments of light.

Rods exhibit slower release kinetics than cones, roughly matched to the slow kinetics of rod light responses. Rod and cone synapses have similar ribbons, ICa with similar properties, and similarly rapid, high affinity calcium sensors, suggesting that differences in Ca2þ handling and buffering may be responsible for rod/cone differences in release kinetics.

The continuous release of vesicles from photoreceptor synapses in darkness is balanced by compensatory endocytosis of vesicles. Photoreceptors rely largely on clathrin-mediated endocytosis whereas bipolar cells and hair cells rely more on bulk retrieval of large endosomes. Visualization of single vesicles at bipolar cell terminals by TIRF microscopy show that the vast majority of vesicles undergo full collapse during fusion indicating that kiss- and-run retrieval of fully formed vesicles is minimal at this synapse.

Disease-Related Mutations in Synaptic

Proteins at the Photoreceptor Synapse

Given that all visual information must pass through the photoreceptor synapse, it is not surprising that mutations in synaptic proteins of photoreceptors can produce visual deficits. For example, rod–cone dystrophies can be caused by mutations in Rab3 interacting protein (RIM1),

UNC-119, or Ca2þ channel a2/d subunits. Congenital stationary night blindness can be caused by mutations in the rod CaV1.4 Ca2þ channel or CaBP4. Misregulation of glutamate release by photoreceptors and bipolar cells may also contribute to excitotoxic damage in neurodegenerative diseases of the retina.

See also: Cone Photoreceptor Cells: Soma and Synapse; Rod Photoreceptor Cells: Soma and Synapse.

Further Reading

Choi, S. Y., Borghuis, B. G., Rea, R., et al. (2005). Encoding light intensity by the cone photoreceptor synapse. Neuron 48: 555–562.

Daiger, S. P., Sullivan, L. S., and Browne, S. J. (2009). RetNet – Retinal Information Network. http://www.sph.uth.tmc.edu/retnet (accessed July 2009).

DeVries, S. H., Li, W., and Saszik, S. (2006). Parallel processing in two transmitter microenvironments at the cone photoreceptor synapse. Neuron 50: 735–748.

Dowling, J. E. (1987). The Retina: An Approachable Part of the Brain.

Cambridge, MA: Harvard University Press.

Heidelberger, R., Thoreson, W. B., and Witkovsky, P. (2005). Synaptic transmission at retinal ribbon synapses. Progress in Retinal and Eye Research 24: 682–720.

Jackman, S., Choi, S.-Y., Thoreson, W. B., et al. (2009). Role of the synaptic ribbon in transmitting the cone light response. Nature Neuroscience 12: 303–310.

Kolb, H., Fernandez, E., and Nelson, R. (2009). Webvision: The Organization of the Retina and Visual System. http://webvision.med. utah.edu (accessed July 2009).

Krizaj, D. and Copenhagen, D. R. (2002). Calcium regulation in photoreceptors. Frontiers in Bioscience 7: 2023–2044.

LoGiudice, L. and Matthews, G. (2007). Endocytosis at ribbon synapses. Traffic 8: 1123–1128.

Prescott, E. D. and Zenisek, D. (2005). Recent progress towards understanding the synaptic ribbon. Current Opinions in Neurobiology 15: 431–436.

Rodieck, R. (1998). The First Steps in Seeing. Sunderland, MA: Sinauer.

Sterling, P. and Matthews, G. (2005). Structure and function of ribbon synapses. Trends in Neuroscience 28: 20–29.

Thoreson, W. B., Rabl, K., Townes-Anderson, E., and Heidelberger, R. (2004). A highly Ca2þ-sensitive pool of vesicles contributes to linearity at the rod photoreceptor ribbon synapse. Neuron

42: 595–605.

tom Dieck, S. and Branstatter, J. H. (2006). Ribbon synapses of the retina. Cell and Tissue Research 326: 339–346.

Glossary

Chromophore – As used here, a small molecule that when bound to a protein causes the complex to absorb light at visible or near-visible wavelengths. The chromophore for visual pigments, in which the protein component is opsin, is either 11-cis retinal or a very similar molecule.

Circular polarization – As used here, a type of polarization of light in which the electric vector, or e-vector, rotates one full circle for each wavelength

traveled by the light, thus describing a circle as seen from the wave front or a helix as seen from the side. Circular polarization can be either right-handed or left-handed, depending on the direction of rotation. Dichroism – The property of a substance to absorb polarized light of one e-vector orientation more strongly than of other orientations, thus transmitting linearly polarized light.

Dielectric – Refers to chemical compounds or substances that do not conduct electricity. Water and most biological molecules are dielectric.

e-Vector – The electrical vector of an electromagnetic wave. For polarized light, the e-vector orientation is usually taken to be the plane of polarization.

Linear polarization – Sometimes called plane polarization. Refers to light in which the e-vectors of the constituent photons are all oriented on the same axis, or in the same plane.

Microvillus – A membranous protrusion from a cell surface shaped like a tiny tube, typically only a few cell membrane thicknesses in radius.

Polarized light – The light in which the e-vector lies in a plane (for linearly polarized light) or rotates through a full circle once for each wavelength (circularly polarized light).

Rayleigh scattering – A type of scattering of electromagnetic energy caused by interactions of the energy with particles much smaller than the wavelength of the energy. Rayleigh scattering produces the blue color of the sky and also produces a celestial polarization pattern by scattering of sunlight. Specular reflection – Reflection as from a mirror, where the reflected ray leaves the surface at the same angle that the incident ray arrived. Specular reflection is typical of shiny surfaces; examples in nature include shiny leaves, insect cuticle, wet skin, or the surface of smooth water.

Light is made up of streams of photons, the elementary particles that carry electromagnetic energy. Each of these photons can be thought of as a miniature electromagnetic wave, which has a single wavelength related to the energy it carries (the distance the photon travels from one energy maximum to the next, inversely proportional to the photon’s frequency) and a single plane within which the electrical energy vibrates – the polarization angle, properly called the e-vector (for electrical vector) angle. Note that since the energy is electromagnetic, there are both electrical vectors and magnetic vectors present, normal to each other. For consistency, throughout this article reference is made only to the e-vector. Therefore, a beam of light, containing countless photons, is characterized by its intensity (the number of photons delivered per unit time), its spectrum (the distribution of wavelengths of all the photons in the beam), and its polarization (the distribution of the planes of vibration, or e-vector angles, of all the photons in the beam).

The most common form of polarization, linear (or plane) polarization, has two descriptors: the overall e-vector angle, which is the mean angle of all planes of vibration of the constituent photons, and the degree of polarization, which is the fraction of energy of all photons vibrating within the plane of the e-vector angle. Of course, in a typical beam consisting of photons of mixed wavelengths, these polarization parameters generally vary with wavelength, creating a polarization spectrum. In this article, only linearly polarized light is discussed unless otherwise noted.

Like many vertebrates, humans are not generally aware of light’s polarization properties, but the visual systems of most animals perceive light’s polarization and use this ability to regulate their behavior. To help us understand what visualizing polarization would be like, the polarization properties of light can be analogized to its color properties. The spectrum of light produces the sensation of color, with a perceived hue (the predominant wavelength of constituent photons) and purity or saturation (the overall distribution of wavelengths around that of the hue itself). Hue is therefore analogous to the e-vector (the predominant angle of polarization of constituent photons) and saturation to the degree of polarization (the distribution of angles around this). In fact, polarization fields are often portrayed as images using false colors where angle is coded into hue and degree of polarization into saturation. Such a display can also include the coding of overall brightness as the intensity at each point, to provide a complete description of the polarized-light field.

Polarized Light in Nature

There are no natural sources of polarized light of known biological significance. Nevertheless, linearly polarized

Transmission

Reflection

Scattering

Figure 1 The three most common ways by which linearly polarized light is created either in nature or in the laboratory. At the top part of the figure, the polarization is produced by transmission through some dichroic material, with its preferential plane of transmission symbolized by the vertically parallel lines. Light emerging from a perfect dichroic material becomes fully linearly polarized. Dichroic polarizers are relatively rare in nature, and account for only a minor fraction of the polarized light observed in natural light fields. The middle part of the figure illustrates polarization by reflection from a smooth, dielectric surface. At a particular angle, known as Brewster’s angle, the reflected light is fully polarized parallel to the surface. Most biological surfaces are dielectric, as is the surface of water, so much light reflected from shiny natural surfaces is highly polarized. The bottom section of the figure illustrates polarization induced by scattering. When the scattering angle is orthogonal to the axis of the ray being scattered, the scattered light is fully polarized at an e-vector angle perpendicular to the plane containing the original ray and the scattered ray.

light is abundant in natural scenery. Light can become polarized in many ways, but the most important processes in nature are through differential absorption, differential reflection, or differential scattering (Figure 1).

Some natural or artificial transparent materials preferentially transmit one e-vector plane while absorbing others, usually because of aligned molecules within the material. This property, known as dichroism, is not particularly common in biological systems, but there are important exceptions, including the inherent dichroism of visual pigment molecules described later. Reflection of light from dielectric surfaces produces polarization parallel to the surface (Figure 1). Therefore, bodies of water and many surfaces in natural scenery reflect horizontally polarized light. Rayleigh scattering from molecules and suspended particles in air produces a well-known pattern of polarization in the sky. Scattering-induced polarization varies with the scattering angle, being greatest (often near 100% polarization) for scattering perpendicular to the axis of the incoming ray (Figure 1). As a result, skylight polarization reaches its maximum in a band that stretches across the sky at 90 to the sun. The axis of the e-vector of the scattered ray is perpendicular to the plane defined by the incoming ray and the scattered ray, such that the band of maximum sky polarization has its e-vectors oriented tangentially to the great circle 90 from the sun’s position. At dawn or dusk, this band stretches vertically across the celestial hemisphere (Figure 2). Since Rayleigh scattering is most effective at short wavelengths, skylight polarization is strongest in the ultraviolet. Scattering from water molecules and very small particles suspended in natural waters also produces polarization (Figure 3), although it rarely reaches the very high degrees of polarization seen in the sky. Light scattering in water is optically different from the processes operating in air, and polarization in

water typically reaches its maximum value at blue-green wavelengths.

Thus, in the sky and underwater, scattering of incoming light produces partial polarization that varies with solar position and direction of view, and reflection of light from the air–water interface or from shiny surfaces (e.g., leaves, wet surfaces, animal skin, scales, or cuticle) produces strong polarization in geometrically favorable circumstances. If a terrestrial animal has polarized-light vision, the sky presents a reliable pattern useful for navigation, but in contrast, the chaotic and unpredictable pattern of polarized-light reflection produces false, pointillistic images that can mask or taint the true colors and locations of objects. Consequently, as described later, photoreceptors in animals that would normally be sensitive to the polarization of light are sometimes structurally modified to destroy polarization sensitivity.

The situation is almost always simpler in water than in air, particularly at depths greater than a few meters. Due to refraction at the air/water interface, illumination from the sun or moon is confined to within 46 of overhead position. The resulting polarization field, while variable to some extent, has horizontally oriented e-vectors much of the time, and the degree of polarization is almost always lower than in air. The pointillistic reflection of polarized light from objects is virtually gone underwater, as the refractive index gradient between water and most natural objects is much lower than in air, such that there is little specular reflection of light (required to produce polarization from dielectric surfaces). The predictable surround, typically low degree of polarization, and minimal polarizedlight reflective noise would seem to make polarization vision in water of little utility, yet many aquatic animals have excellent polarization sensitivity. Currently, it is not

always clear what biological advantages are provided by such visual abilities.

Polarization Sensitivity and Polarization

Vision

Polarization Responses of Photoreceptor Cells

Light is absorbed in visual photoreceptors of all animals by molecules of visual pigment, which consist of a chromophore (derived from vitamin A or a close chemical analog) linked to a protein, termed opsin. Just as each visual pigment molecule has its characteristic absorption spectrum, which ultimately determines the spectral sensitivity of the photoreceptor within which it resides, it also has an inherent polarization sensitivity. This exists because the chromophore itself is dichroic, absorbing preferentially when the e-vector of an incident photon is parallel to the long axis of the molecule. Since the chromophores of visual pigment lie nearly parallel to the membranes of photoreceptors, when light arrives perpendicular to these membranes the incident photons are likely to be polarized parallel to the absorption axes of some of the visual pigment chromophores. The way in which these chromophores are aligned within the photoreceptor cell as a whole determine whether or not the receptor responds differentially to polarized light, and thus whether it has inherent polarization sensitivity (Figure 4).

In the rod and cone photoreceptors of vertebrate retinas, photoreceptive membranes are arranged in a series of parallel layers, either in flattened disks (rods) or lamellae formed from folded membrane sheets (cones). Since light generally strikes these layers normal to their surfaces, it encounters visual pigments that are arrayed at

Side view

(as light arrives)

Surface view (as light arrives)

Side view

End view

Figure 4 An illustration to show absorption of polarized light by vertebrate rod photoreceptors (left; cone photoreceptors would have similar properties) and by microvillar photoreceptors like those of arthropods or cephalopod mollusks (right). In life, light arrives normal to the flat surfaces of rod disks and encounters randomly oriented chromophores of visual pigment lying within the disk membrane (symbolized by double-headed arrows to indicate the preferred axis of polarization for best absorption). Since the orientation is fully random, there is no preferential absorption of any given e-vector angle. If light were to arrive from the side of the disk, it would encounter chromophores that are restrained to angles near that of the membranes themselves, favoring the absorption of horizontally polarized light. In microvillar photoreceptors (right), light arrives orthogonal to the long axis of each microvillus, and encounters visual pigment chromophores that are oriented roughly parallel to the axis of the microvillus. Thus, the microvillus as a whole preferentially absorbs light polarized parallel to its axis. If light were to impinge on the microvillus from the end, it would encounter chromophores arrayed at all possible angles around the circumference of the microvillus, and no preferred absorption orientation would exist.

all possible orientations (Figure 4, top-left). Consequently, rods and cones rarely have an overall polarization response, even though the individual molecules of visual pigment are dichroic. The situation would be very different if light impinged on rods or cones from the side (i.e., normal to the long axes of their outer segments). It would then meet chromophores lying in the planes of the membrane layers, and all chromophores would preferentially absorb light polarized nearly parallel to the membrane. Note that while the individual chromophores have random arrangements in the membrane’s plane, and thus absorb light from this direction with varying effectiveness (suggested by the variable lengths of the double-headed arrows), they always absorb light polarized in the membrane’s plane most effectively.

The photosensitive membranes of photoreceptor cells of arthropods (crustaceans, insects, etc.) and cephalopods (octopus, squid, and cuttlefish) are constructed from bundles of microvilli. In each microvillus, for reasons that are not yet fully understood, the molecules of visual pigments are arranged such that their chromophores are roughly

parallel to the axis of the microvillus. The microvilli typically extend out perpendicular to the axis of the receptor cell as a whole, such that light arrives perpendicular to each microvillus. In this orientation, each microvillus preferentially absorbs light polarized parallel to its axis, such that if microvilli are arranged parallel throughout the receptor as a whole, the cell will be polarizationsensitive. Note that this property requires no other cellular specializations, and as a result, almost all microvillar photoreceptor cells have some level of polarization sensitivity.

Polarization sensitivity

Having receptor cells that respond differentially to polarized light is only the first requirement for polarization sensitivity at higher levels of neural analysis. For the nervous system to be able to analyze light’s polarization, sets of photoreceptors with different preferred polarization orientations must be compared, typically through opponent processing. This type of analysis is like that of color vision, where sets of photoreceptors with differential spectral sensitivity are compared for color processing. Recall that polarization of light has three attributes: intensity, degree of polarization, and polarization angle. Thus, for full awareness of light’s polarization at a given point in the visual field, independent inputs from three receptor sets must be analyzed. Interestingly, few animals do this; in almost all cases, only two receptor sets with orthogonal microvilli are compared. This is reasonably effective in practice, because natural polarization tends to be predictable, such that if the receptor sets are appropriately oriented, the polarization is well analyzed. Two-channel polarization analysis can be extended to full polarization sensitivity if the receptors are rotated relative to the stimulus, although this has rarely been observed in practice.

The animal groups for which the mechanisms of polarization sensitivity are best understood are the insects, the crustaceans, and the cephalopod mollusks (octopus and squid). All of these have microvillar photoreceptors, and in most species the receptors are arranged orthogonally. Crustaceans and insects have compound eyes that are particularly well designed for analyzing polarized light. Each unit of the eye contains a group of photoreceptor cells, such that each unit of the compound eye can potentially serve as an independent polarization detector. For two-axis polarization sensitivity, subsets of receptors in each group have orthogonal microvilli. In insects, these subsets often extend into a central, fused photoreceptor from four sides, with microvilli entering from opposite sides abutting near the center. Viewed from the tip of the receptor group, the overall arrangement ends up having two cell sets with horizontal microvilli and two with vertical microvilli (Figure 5). Crustaceans have a similar system, but here the orthogonal sets of microvilli exist in successive layers, making the entire composite

Figure 5 A schematic diagram of the structure of typical polarization-sensitive photoreceptors such as would be found in the compound eyes of insects. The receptor is viewed in a section as seen on the axis of the photoreceptor group. In life, these receptors would form a bundle of cells arranged in a circle and together forming a tall cylinder with the microvilli arranged in a smaller cylinder running down its middle. Each cell forms a section of the overall receptor (a single cell in this diagram can represent either one receptor cell or two cells lying side-by-side with parallel microvilli). Note that cells on opposite sides of the receptor extend parallel microvilli toward the junction in the center, and thus have parallel polarization sensitivity. Since two sets of receptors exist, with either horizontally oriented or vertically oriented microvilli, the receptor group as a whole can provide information for two-axis polarization analysis. Receptors of crustaceans are similar to this, but each layer of the joint photoreceptor contains microvilli from only one subset of cells, with either horizontally arrayed or vertically arrayed orientations. Successive layers of the receptor contain microvilli from the other subset of cells, and thus the receptor cylinder has stacks of mutually orthogonal microvillar layers which can contribute to two-axis polarization analysis. Photoreceptors of cephalopods are somewhat different from these, since they are arrayed continuously side-by-side throughout a retina, but each junction of four cells forms a set of microvilli organized like those in the center of the diagrammatic insect photoreceptor illustrated here. Cephalopod receptor cells also form parallel microvilli on the opposite side of each cell. Consequently, each cell contributes to two junctions of microvilli, one on each side of the cell. Again, with two primary axes of orientation of microvilli, separate cells can contribute to two-axis polarization sensitivity.

photoreceptor like a pile of a large number of circular segments, each having microvilli orthogonal to the segments immediately above and below. Finally, the cephalopods (which do not have compound eyes, but instead have a single lens eye structured much like a vertebrate camera eye) arrange their microvillar receptors such that each cell has microvilli on two opposing sides (like a twosided toothbrush). The mosaic of cells forms junctions similar to what is pictured in the center of Figure 5, except that each of the four cells in this figure would form another junction on the opposite side with yet other cells. In all these cases, the cells with parallel microvilli viewing one point in space join to form one polarization channel, and those with microvilli orthogonal to these join for the opponent channel.

Vertebrate polarization sensitivity is more difficult to explain, and it is fair to say that we are still not able to account for it satisfactorily. Nevertheless, there is no doubt at all that some vertebrates sense light’s polarization. Recall that end-on stimulation of rods and cones is unlikely to produce any differential sensitivity to the plane of polarization, because chromophores are randomly oriented for such light (Figure 4). If vertebrate photoreceptor cell outer segments were slanted relative to the axes of impinging rays of light, this would confer some polarization sensitivity. It appears that in at least some fishes, the outer segments of some classes of cones lie on their sides, tilting their lamellae vertically in the retina. If all cones of a given class lay parallel, or were organized into orthogonal classes, this could permit the retina as a whole to achieve an overall polarization sense. There is recent evidence that some rod or cone classes are measurably dichroic to end-on illumination. The origin of this dichroism is unclear, but it could be caused by parallel tilting of the rod disks or cone lamellae.

Polarization vision

If an animal has polarization sensitivity, it can obviously respond in some way to a polarization stimulus. As described later, these responses are frequently hardwired and inflexible, and the polarization sense that drives them does not correspond to what is normally conceived of as vision, which implies a perception of space, form, and individual objects. The term polarization vision refers to a polarization sense analogous to color vision, whereby animals visualize polarization attributes of features within the overall field of view and use polarization variations to enhance the visibility, contrast, or features of particular objects. In principle, an animal that is capable of polarization vision perceives the visual world as a pattern varying in polarization features among receptive fields. While polarization sensitivity is most useful for orientation or for organizing simple responses, polarization vision offers the potential to direct complex behavior such as predation, camouflage generation or breaking, and signal detection. There is only weak evidence of this ability in some vertebrates, but many species of both arthropods and cephalopods probably use true polarization vision in ways that are discussed later.

Disentangling polarization and color sensitivity

Many – perhaps most – animals that are sensitive to polarized light also have color vision. This presents both perceptual and sensory-processing challenges, as it is generally not desirable to mix these visual modalities. For example, if a receptor cell that contributes to a perceptual color channel has some residual polarization sensitivity, color appearance will be altered by stimuli that contain polarized light. This is most often a problem for

animals with microvillar photoreceptors due to their inherent polarization bias. The cephalopods have firmly dealt with this issue by discarding color vision entirely – the great majority of octopuses, squids, and cuttlefishes have only a single spectral receptor class in their retinas, restricting vision entirely to the intensity and polarizational domains. Many crustaceans have reached a similar solution, devoting nearly all of their receptors to polarizedlight reception. Some crustacean species, however, separate color and polarization processing, using a single spectral class for polarization analysis while reserving a set of other polarization-insensitive classes for color vision. This solution is used, for example, by stomatopod crustaceans, also known as mantis shrimps. Insects also commonly separate color-sensitive from polarizationsensitive receptors, isolating their polarization receptors to just one part of the visual field and often using only ultraviolet receptors for polarization analysis. Some insect species destroy polarization sensitivity in photoreceptors by twisting the entire receptor group around its long axis. In a few cases, surprisingly, insects unify polarization and color perception in the same receptor cells, interpreting some stimuli by combining these two modalities into a single signal. Some butterfly species, for instance, examine potential ovoposition sites in this way. In vertebrates, however, as with other aspects of polarized-light photoreception, it is unknown how (or even if) polarized-light processing is kept separate from color processing. This could be a difficult problem, as it is thought that some vertebrates use different spectral types of cones to sense different polarization planes, a technique that immediately must mix color and polarization information at the first level of light detection.

The Contributions of Polarized-Light Perception to Behavior

Sensing polarized light seems strange to us, but for most animals it is as fundamental to their visual perception as color vision is to humans. Indeed, as will be described shortly, in many animals polarized-light perception plays similar roles to those assumed by color vision, and it can even work together with color vision to improve visual interpretation of stimuli. However, there are many situations where polarized light is used for special purposes unique to this modality. Among these are water surface detection and skylight navigation.

Water surfaces reflect horizontally polarized light, as illustrated in Figure 1. This is why sunglasses with polarizing lenses make it easier for fisherman to see fish – the lenses are oriented to block horizontally polarized light, reducing the glare from the water’s surface and clarifying the visibility of objects in the water itself. Many flying insects, including adult water beetles and mayflies, use the

reflected polarization in the opposite way – their eyes are adapted to respond strongly to large expanses of horizontal polarization below the horizon, which in nature invariably correspond to water surfaces. The insects respond by diving into the water, or by alighting on its surface to ovoposit. This simple response is extremely reliable in nature, but can lead to disastrous consequences for the insects today, when many shiny, horizontal surfaces are man-made. Parking lots, oil ponds, and even the painted surfaces of cars and other manufactured objects induce the same response, which in these cases is frequently lethal.

Scattering of sunlight in the clear sky produces a highly reliable pattern of polarization (Figure 2), recognized by the visual systems of many insects (including bees, ants, and crickets) as well as by other arthropods including some spiders and crustaceans. The pattern is used for navigation, as it is a perfect indicator of the current position of the sun, persisting even when the sun is not visible behind an obscuring object or landscape feature, or when the sun is hidden by clouds. Thus, navigation is possible even on quite cloudy (but not wholly overcast) days. Most insects that navigate using skylight polarization devote a small region of the compound eye, called the dorsal rim, to perceiving the pattern, and most require only a small patch of clear sky to orient. Navigation using the location of the sun or skylight polarization patterns is not simple, as the solar position drifts through the sky with changing dynamics throughout the seasons, and insects must be able to compensate for these changes each day as they manage their foraging excursions. Some insects, including dung-foraging scarab beetles, use skylight polarization created by moonlight to navigate during their nocturnal rambles. In an interesting vertebrate example, migrating birds are thought to use skylight patterns of polarization at twilight to calibrate their magnetic compasses.

The tasks described so far are analogous to map senses, simple reflexes, or other perceptual abilities that are not strictly visual in the sense that we humans understand it. In other words, these types of abilities do not examine features or objects in the outside world except in very general ways. However, there are animals that actually see patterns of polarization in a fashion that is quite analogous to the way that we perceive the external world – they use polarization vision to recognize objects, to enhance contrast of prey, or to see signals of conspecific animals. Some animals, in fact, can be trained to discriminate objects that we see as identical but that differ in the patterns of polarization that they reflect or transmit. Octopus and mantis shrimps learn such tasks.

Near the water’s surface, the skylight polarization pattern penetrates and is therefore available as an orientation cue. Deeper than this, underwater polarization is only rarely usable for navigation (although it can be used to orient vertical migration) because it is frequently weak,