Philosophers have long argued whether we perceive or truly observe our surroundings, but in a world fraught with challenge and opportunity the selective advantage of a visual system is evident throughout nature. The evolution of the retina has allowed us to detect the light from a star, contrast images in the desert sun and discern subtle movements in a bustling city street. It has taken hundreds of millions of years to refine the structure and function of the human retina. Yet the original blueprint was so successful that the fundamental architecture of the retina has been conserved in all species from fish to primates. Not only does the retina relay information about light, contrast and movement, but all the more remarkably it is able to do so whether in the twilight or sunlight through a process known as adaptation. Yet only recently have the intricate cellular processes that underlie adaptation been unraveled, revealing how the information neurons carry is altered in response to an ever changing environment. Such developments take us a few steps closer to one of the goals of top neuro-ophthalmologists such as Ronald Burde: the design of an artificial retina. But the retina is more to science than a study of vision, it has become a window to understanding how cells communicate with one another and the world outside.

Light from our surroundings is collected by the eye through a regulated aperture called the iris and focused as an inverted image onto the retina. It is this simple physical arrangement that is captured in the design of the camera, and yet science is far from designing a light detection system as sensitive and as detailed in its resolution as the retina. Contrary to popular conception, the retina is in fact an integral part of the brain and has provided a focus for the study of the nervous system since Ramon Cajal first described the architecture of the tissue around the turn of the century. Since his pioneering work science has tried to relate the structure of the retina to its function, and more recently to understand the biochemical and biophysical basis of light adaptation. As an electrical circuit the retina is perhaps one of the best understood parts of the nervous system. The function of the retina, as for any given circuit, is defined by the precise arrangement of chemical switches and electrical contacts that are formed between individual nerve cells, or neurons. The resulting neuronal networks serve to direct sensory information to centers that first integrate and then pass on the processed information to the appropriate command centers that control the muscles, glands and organs. Following this general plan the retina consists of a layer of specialised sensory neurons called photoreceptors that capture the light image and convert it into electrical information which is then transmitted 'vertically' to a layer of bipolar relay neurons. It is at the level of the bipolar cells that two parallel pathways of information flow are created. One population of bipolar cells is excited when the intensity of illumination is increased and the other responds when it is decreased, and the two populations are referred to as On and Off bipolar cells respectively. The On and Off bipolar cell populations channel information about changes in light intensity onto a third layer of neurons known as ganglion cells. The ganglion cell layer serves as a center for the integration of light information in the retina, where sub-populations of On and Off ganglion cells respond with a burst of spikes in response to increases or decreases in the intensity of light respectively, depending upon the nature of the bipolar cell from which they receive their input.

So why did evolution take on the apparent expense of creating distinct On and Off channels for the processing of light information in the retina ? This arrangement confers two advantages, in the generation of a signal as an object moves across the field of view and in the enhancement of contrast between areas of relative darkness and light. Ganglion cells integrate all the information they receive from On and Off channels and encode their computed output in the form of electrical spikes which travel along the optic nerve to another relay center in the mid-brain, the lateral geniculate nucleus of the thalamus. From this point light information is directed to the visual cortex and other centers in the brain for further processing and image reconstruction, completing the description of the 'through' pathway for information flow. One of the reasons for our high level of visual acuity is our ability to contrast between areas of relative darkness and light. This capacity is achieved in part by the lateral processing of light information within the retina. The light signal from the photoreceptor is not delivered exclusively to the bipolar cell layer, as information also flows to a layer of horizontal cells which lie between the bipolar and photoreceptor cell layers. Horizontal cells are believed to spread a radiating wave of inhibition to surrounding bipolar cells by releasing the inhibitory transmitter gamma-aminobutyric acid, or GABA upon stimulation. The signal sent by the horizontal cell layer to the bipolar cell is in effect the reverse of the information about the amount of available light that it receives from the photoreceptor. So where the photoreceptor senses light, the horizontal cell network informs the surrounding On and Off bipolar cell pathways that they are relatively dark and vice-versa, providing us with an exaggerated impression of contrast between regions of light and shade in our surroundings. In more philosophical terms, where we see black and white there may in fact only be shades of gray. The artist Claude Monet took the idea of contrast one step further, and was renowned for his striking use of contrast enhancement between objects in his paintings. Through an absence of natural shading in his paintings he was able to create a dramatic impression of contrast, in effect imitating the function of our retina.

Further lateral processing occurs in the inner retina, where a heterogeneous population of amacrine cells receive information directly from the bipolar cell layer and radiate a range of both inhibitory and excitatory outputs onto the receptive terminals of surrounding ganglion cells and the output terminals of the bipolar cell layer. In addition to the vertical and lateral processing of light information, a population of amacrine cells called interplexiform cells feeds information about the amount of light flowing in the retinal circuit back onto the receptive terminals, or dendrites of horizontal and bipolar cells in the outer retina by releasing the hormone dopamine. Dopamine diffuses widely and changes the sensitivity of retinal neurons to light to co-ordinate the process of adaptation. Another important role that has been proposed for amacrine cells is in the detection of movement. A population of amacrine cells in the salamander receives its information from both On and Off channels, and fires a volley of spikes in response to the beginning and end of a light stimulus, such as a target moving across the field of view, and it beleived that an equivalent population of transient amacrine cells may exist in the mammalian retina.

It is only very recently that researchers have been able to describe the intricate cellular pathways that are involved in the detection and amplification of light signals, and to elucidate the biochemical pathways that modify the size and shape of the responses of individual neurons to light. In the first step in phototransduction, the process of converting light energy into electrical energy, individual light particles, or photons are captured by pigments in the outer segments of photoreceptors. The rod photoreceptor has been the more extensively studied of the two and is specialised for nocturnal, or scotopic vision. When we refer to cones we tend to think of their familiar role in color vision. Populations of red, green and blue cone photoreceptors are present at high density in the specialised foveal region of the retina, each containing a specific pigment that absorbs light over a narrow band of the spectrum of visible light. However, the principal role of the cone photoreceptor is in the measurement of light intensity rather than wavelength, and to this end the molecular design of cones enables them to operate at higher background light intensities, whereas rod photoreceptors are specialised for the detection of very low light levels. Although cones do not operate efficiently in the twilight and require the assistance of rod photoreceptors, their design allows them to distinguish changes in light intensity over a large range of background light levels to confer daylight, or photopic vision.

Light information from cones passes directly into the On and Off bipolar cell channels, whereas information from rod photoreceptors in the mammalian retina is sent exclusively to a population of 'rod' bipolar cells, which in other respects resembles the cone On bipolar cell in the nature of its responses to light. This would at first glance suggest that the rod pathway only serves to detect the presence of light rather than generating On and Off channels to gain the advantages of contrast and movement detection. Ultimately however, the rod pathway does separate the night-time picture into distinct On and Off channels that the visual system can interpret, and does this by feeding the output into the On and Off channels created by the cone bipolar cells, thus avoiding the loss of resolution and spatial complications that the creation of two additional rod channels would incur. The rod bipolar cell funnels all of its information about increases in light intensity into the AII amacrine cell, a specialised neuron that then feeds this information directly into the On and Off channels. In response to light the AII amacrine cell simultaneously decreases information flow through the Off channel, whilst increasing information flow in the On channel, and this feat of contradiction is made possible because the AII amacrine cell employs both chemical and electrical methods of information transfer within a single type of neuron. Upon illumination the AII cell releases the inhibitory transmitter glycine onto the output terminals of the cone Off bipolar cell and passes excitatory information directly into the cone On bipolar cell because the two cells are electrically coupled as if the two were wires connected by a closed switch. This electrical switch is created by the presence of conductive pores known as gap junctions adjoining pairs of cells which allow charged ions to pass freely between cells, and so excitation in one neuron is transmitted directly into another without the need for the release of a chemical intermediary. Gap junctions also connect pairs of horizontal and AII amacrine cells and assist in the lateral processing of light information through these electrically coupled networks. In a far-reaching recent discovery Stephen Mills and Stephen Massey showed that the hormone dopamine, released by certain amacrine cells in response to light, closes down the gap junctions between pairs of AII amacrine cells as well as those between pairs of horizontal cells, and dopamine does this by increasing levels of the intracellular messenger cyclic adenosine monophosphate (cyclic AMP) in these neurons. Mills and Massey were further able to demonstrate that another messenger molecule released from cells in the retina in response to increases in background light, the infamous nitric oxide, specifically closes gap junctions that electrically couple the AII amacrine cell and the cone On bipolar cell. The closure of this gap junction has the interesting consequence of shutting down the pathway for night vision. These exciting findings show that the retina literally 'rewires' itself from nocturnal to diurnal vision through the release of specific hormones in response to an increase in the ambient level light, as part of the integral process of adaptation.

Many people are familiar with the idea that most neurons encode electrical information in the form of spikes, or action potentials. Bipolar cells release the excitatory transmitter glutamate from their presynaptic terminals onto the postsynaptic dendrites of amacrine and ganglion cells in response to changes in light intensity. Glutamate binds to a receptor on the postsynaptic membrane where it opens pores which allow positively charged sodium ions to flood into the dendrites, making the ganglion or amacrine cell more conductive in terms of information flow. These pores are better known as ion channels and are found in all types of cell. Ion channels are frequently selective for certain ions, such as chloride, sodium or potassium, and control the distribution of these charged ions across the cell membrane determining the voltage, or potential across the cell membrane. As the membrane potential determines the state of activity in cells, especially neurons, they play a central role in cell signaling in almost all processes. The influx of sodium ions triggered by the binding of glutamate to its receptor results in a run-away spike depolarization, or action potential because a population of voltage-sensitive sodium channels are rapidly activated by the depolarization once the membrane potential has crossed a certain threshold. The depolarizing phase of the action potential is then quickly reversed by a reflexive outward movement of potassium ions returning the membrane potential a less excited state. This wave of membrane depolarization propagates along the length of the output process of the neuron to the presynaptic terminals of the neuron, where it opens excitable calcium channels. The resulting influx of calcium ions into the terminal causes the release of transmitter from specialised membrane storage compartments, which are triggered to fuse with the pre-synaptic membrane and release their contents into the synaptic cleft by the entry and binding of calcium to special sensor proteins in the terminal.

The opening of excitable sodium and potassium channels to generate the depolarizing waveform known as an action potential was first described by the Nobel Laureates Alan Hodgkin and Andrew Huxley, but since their pioneering efforts we are now aware that information is not only encoded in the frequency of these action potentials, but may also be encoded in the delay and the shape of the action potential. Whilst ganglion cells and most amacrine cells encode and pass on light information received from bipolar cell pathways through the generation of action potentials, the cells of the outer retina are unusual in that they possess no excitable population of sodium channels that are opened by membrane depolarization. Photoreceptors, bipolar cells and horizontal cells respond to light with only slow, graded changes in their membrane potential, and such graded changes in membrane potential migrate more slowly and less efficiently to the terminal. It took some time before Craig Jahr, George Ayoub and David Copenhagen were able to show that information flow from photoreceptor terminals is also mediated by the release of glutamate onto the receptive dendrites of bipolar and horizontal cells. Illumination of the photoreceptor causes a graded reduction in the rate of release of glutamate from the photoreceptor terminal. David Attwell and colleagues subsequently revealed that it is this reduction in glutamate release that causes the Off bipolar cell become less conductive of light information and On bipolar cells to become more conductive. The observations by Attwell, Wilson, Scott Nawy and David Copenhagen that the nature of the responses to light and glutamate in the On bipolar cell are the opposite of those of the Off bipolar cell suggested that glutamate activates very different signal transduction pathways in the two types of bipolar cell.

Action potentials stimulate transmitter release from pre-synaptic terminals in an explosive 'all-or-none' fashion, and little release of transmitter tends to occur before the arrival of an action potential, whereas the graded potentials of photoreceptors and bipolar cells cause a graded change in the rate of continuous release of glutamate from the terminals of these neurons that is in some way proportional to the size of the light stimulus. The encoding of light information in these neurons is therefore entirely relative. In the dark-adapted retina a few photons of starlight can stimulate a small yet perceptible change in the release of glutamate from the terminals of photoreceptors and bipolar cells, whereas a comparatively larger increase in light intensity may not elicit a perceptible response in daylight. This is the underlying principle of light adaptation, where the retina switches from a detector and amplifier of light to a circuit that is comparative, measuring relative changes in light intensity. Cells in the outer retina are therfore not so much designed to pass on information about the absolute intensity of light at a point in the field of view, but to be able to make measurements of relative changes in light intensity throughout a range of background levels of light.

We can perhaps now begin to appreciate how the process of phototransduction serves to encode changes in the intensity of light into electrical signals that represent the language of the nervous system. The outer segments of rod photoreceptors contain the light-capturing pigment rhodopsin, which is composed of the chromophore 11-cis-retinal chemically bound to the protein opsin. Rhodopsin is present in the membrane discs of the outer segment, spanning the membrane seven times in all. Selig Hecht showed as long ago as 1938 that the absorption of single photon of light by a human rod cell is sufficient to cause a perceptible one milliVolt change in the potential across the photoreceptor membrane. This was later shown through the classical experiments of George Wald and co-workers to result from a photoisomerization, or geometrical change in the structure of 11-cis-retinal to trans-retinal, converting the energy of light into atomic motion. This atomic motion is in turn transmitted through the chemical bond linking retinal to opsin, causing a structural change in its protein partner within milliseconds. The activated form of the protein, metarhodopsin is then in the right 'shape' or conformation to activate a universal cellular signaling mechanism which translates the atomic motion of receptors into changes in the energy stored in chemical bonds. This mechanism is provided by 'G'-proteins, which are in essence a type of 'chemical switch' that couples signals received by receptors in the cell membrane into biochemical processes that take place both inside the cell and elsewhere in the cell membrane. G-proteins are found anchored to the membrane when unstimulated and are formed from two components; an alpha subunit and a seemingly inseparable combination of beta and gamma subunits. Metarhodopsin causes the alpha subunit to exchange the nucleotide guanosine diphosphate (GDP) for guanosine triphosphate (GTP) which contains an extra high energy bond between its second and third phosphate molecules. The presence of this additional negatively charged phosphate bound to the alpha subunit commands a change in the conformation of the alpha subunit which causes it to lose its affinity for the beta-gamma subunit. The two components then separate to form a bifurcating, or forked signaling pathway. The alpha subunit, now in its stimulated GTP-bound state is then able to activate the enzyme phosphodiesterase by pulling away its inhibitory subunit.

The phosphodiesterase of the photoreceptor cleaves the high energy phosphate bond that is formed within the second messenger molecule cyclic guanosine monophosphate, or cyclic GMP which is produced by the action of the enzyme guanylate cyclase upon GTP. A second messenger is any ion, molecule or protein that mediates the transfer of information within a cell following the arrival of a signal carried by a primary messenger to the cell surface, which may be a hormone, transmitter or in the case of the photoreceptor, light. The regulation of cGMP levels in the photoreceptor reflects the essence of many second messenger pathways where a metabolic building block such as the nucleotide component of GTP used in the assembly of DNA, is tranformed through a one-step dead-end biochemical pathway into a short-lived messenger molecule that is itself a highly specific and carefully regulated signal. This is elegantly demonstrated by the process of converting GTP into cyclic GMP through the action of a guanylyl cyclase and then back into the 'linear' non-messenger form guanosine monophosphate (GMP) by cleaving the phosphate bridge formed by the phosphate molecule by the action of a phosphodiesterase enzyme, the entire process being driven by the release of free energy stored in the bonds formed between phosphate molecules. The concentration of cGMP becomes the critical signal informing the visual system as to relative changes in light intensity in the photoreceptor at a point in time, and is itself precisely regulated by the balance of phosphodiesterase and guanylate cyclase enzyme activities in the photoreceptor. However, if changes in the cGMP concentration provide an almost instantaneous measure of changes in relative light intensity, other signaling pathways must have evolved to ensure that the sensitivity of elements that detect and control the concentration of cGMP are continuously adapted, or modulated to the absolute background level of light.

But what role do the beta-gamma subunits play in the light response? It was once thought that they served only as anchors for alpha subunits in the membrane, but they have now been implicated as major players in a number of signaling pathways. In the photoreceptor dissociated beta-gamma subunits migrate to activate rhodopsin kinase which is instrumental in the termination of the light response. Rhodopsin kinase belongs to another class of enymes involved in cell signaling called protein kinases which participate in another ubiquitous 'dead end' signaling pathway of cell metabolism. Protein kinases facilitate the transfer of the terminal phosphate molecule from adenosine triphosphate (ATP), the universal energy currency of the cell, onto specific amino acids of the target protein. These are amino acids which possess 'water-like' hydroxyl (-OH) groups and are surrounded by a specific signal sequence of amino acids that are recognised by the kinase before the transfer of a phosphate molecule onto the amino acid can occur, a process known as phosphorylation. Protein kinases are classified into two broad families according to whether they phosphorylate the hydroxyl groups of the amino acids serine or threonine, like rhodopsin kinase, or the larger amino acid tyrosine, an event typical of many pathways involved in the regulation of cell growth and division. The transfer of a highly negatively charged phosphate molecule onto a target protein has several effects. Firstly it dramatically alters interactions between charged residues within the target protein resulting in a change in its conformation and consequently its activity, and secondly it alters its interactions with other proteins either as a result of these conformational changes or from direct interactions with the added phosphate groups. In some proteins phosphorylation alters the affinity of a protein for the cell membrane, which may sometimes lead to a change in its location within the cell.

This does not fully explain how rhodopsin kinase inactivates metarhodopsin. Once rhodopsin kinase has performed multiple phosphorylations at the terminal portion of metarhodopsin, another protein aptly named arrestin binds to metarhodopsin to prevent it from further stimulating G-protein activity. This in cell talk is known as a feedback pathway, where the duration of a signal is regulated by downstream elements in the pathway; in this case the receptor that stimulates the G-protein is itself switched off by an element downstream of G-protein activation. The light response finally ends for metarhodopsin when it is regenerated by the replacement of the spent trans-retinal chromophore with fresh 11-cis-retinal, whereupon arrestin can no longer bind to phosphorylated rhodopsin and leaves. A protein phosphatase, the reverse component of the 'dead-end' kinase signaling pathway, is then able to strip the phosphates off the terminal portion of rhodopsin in preparation for another cycle. As for the alpha subunit, it possesses an intrinsic time switch which is triggered by its interaction with the cGMP phosphodiesterase, resulting in the cleavage of the terminal phosphate of its bound GTP molecule in a second example of a feedback pathway. This returns the alpha subunit to its inactive GDP-bound conformation and allows it to reassociate with beta-gamma subunits, and the G-protein cycle is ready to start again.

But how is the concentration of cyclic GMP finally translated into an electrical signal ? It seemed reasonable to predict that the photoreceptor possesses a sensor that couples changes in the cyclic GMP concentration to the observed changes in membrane potential. Fesenko and colleagues were the first to be able to show that cGMP binds directly to a population of ion channels causing them to open. These cGMP-regulated channels were shown to be selectively permeable to calcium and sodium ions, which in the absence of light-stimulated phosphodiesterase activity flood into the cell and depolarize the photoreceptor membrane. This depolarization is in turn is propagated and translated into an increase in the rate of glutamate release from the photoreceptor terminal. Conversely, in response to light phosphodiesterase reduces the cyclic GMP concentration, cyclic GMP-regulated channels close, the intracellular face of the membrane becomes more negatively charged, or hyperpolarized and there is a subsequent decrease in glutamate release from the photoreceptor terminal. So how does a single photon generate as much as a milliVolt change in membrane potential ? The answer lies in the remarkable amplification of the light signal by the elements of the phototransduction cascade. A metarhodopsin receptor activated by a single photon has been estimated to activate about five hundred G-proteins during its lifetime. In turn, each phosphodiesterase enzyme activated by a G-protein can cleave more than four thousand molecules of cGMP for each second of its active lifetime. The resulting fall in the cGMP concentration due to the activation of this biochemical cascade is sufficient to close hundreds of cation channels and evoke a membrane hyperpolarization as large as a milliVolt, which we sense as a reduction in the rate of release of glutamate from the photoreceptor terminal.

But what happens to the calcium ions that flood into the photoreceptor outer segment through the cyclic GMP-regulated channel ? Calcium in its free ionic form is normally kept at very low concentrations inside cells, in fact as much as one hundred thousand times lower than the external concentration. In the photoreceptor this gradient is maintained by a membrane transport protein that continuously exports calcium and potassium ions from the cell interior in exchange for entry sodium ions from outside the cell. Following the influx of calcium ions through the cGMP-regulated channels specialised intracellular proteins bind to free calcium ions with high affinity to prevent the internal calcium concentration from reaching levels that will destroy the cell. King-Wai Yau and Nakatani were the first to show that this membrane transport protein continues to export calcium ions after the cGMP-regulated channels that mediate calcium entry have closed in response to light, which Peter McNaughton and colleagues subsequently showed causes a fall in the intracellular calcium concentration which is a critical signal in adaptation.

But at what points in the phototransduction cascade is the degree of amplification, or sensitivity to light controled to provide a mechanism for adaptation to lower background levels of light ? The first clue was provided by the observation that after a prolonged period in the dark calcium levels in the unstimulated rod photoreceptor become elevated due to the continuous influx of calcium through GMP-regulated channels. The team of Mark Gray-Keller, Arthur Polans, Kris Palczewski and Peter Detwiler demonstrated that the calcium-binding protein recoverin prolongs the duration of the light response by preventing rhodopsin kinase from phosphorylating metarhodopsin. Denis Baylor and Leon Lagnado showed that in addition to recoverin, a number of other calcium-regulated factors combine to amplify light-stimulated increases in phosphodiesterase activity as much as thirty-fold. However Koch and Stryer, and later Yiannis Koutalos and colleagues provided evidence that increases in free calcium that occur in dark-adapted rod photoreceptors also inhibit the production of cyclic GMP by guanylate cyclase through another unidentified calcium-binding protein, decreasing the size of the current carried through cyclic GMP-regulated channels. Lastly, but by no means finally, Yi-Te Hsu and Robert Molday showed that calcium binds to a ubiquitous calcium-regulated protein known as calmodulin, causing a reduction in the affinity of the channel for cyclic GMP. This in turn results in a decrease in the sensitivity of the channel to changes in cyclic GMP concentration, and may serve to extend the range of cGMP concentrations over which the channel sensor operates. Combining the actions of these various calcium-binding proteins of the rod photoreceptor into the larger picture of adaptation, Yiannis Koutalos and King-Wai Yau were able to make a number of predictions about the consequences that a fall in intracellular free calcium levels will have upon the degree of amplification of the light signal in broad daylight. The decrease in free calcium levels will increase the cGMP concentration in the rod photoreceptor by both inhibiting phosphodiesterase activity and stimulating guanylate cyclase activity. More cation channels will open due to the elevated cGMP concentrations and their increased affinity for cGMP, because calmodulin is no longer activated by calcium. The combined effect of these decreases in calcium during light adaptation is to reduce the sensitivity of the rod photoreceptor to a given increase in light intensity, because each excited metarhodopsin protein hydrolyses a smaller fraction of the available pool of cyclic GMP molecules due to the reduced level of gain in the phototransduction cascade and the increased cGMP concentration. The net advantage that all of these biochemical adjustments confer is to trade a reduction in the absolute sensitivity of the rod photoreceptor to light for an increase the range of background light levels over which the photoreceptor is able to detect a change in light intensity.

Is calcium the only second messenger that plays a role in light adaptation in the photoreceptor? It would appear not. The guanylate cyclase present in the photoreceptor has a built-in sensor that detect increased nitric oxide levels, and when this sensor binds to nitric oxide it stimulates an increase in cGMP production. A team led by Steve Barnes and Dmitry Kurrenny provided evidence that nitric oxide opens both cGMP-regulated cation and excitable calcium channels in rod photoreceptors. Their findings strongly suggest that nitric oxide production in response to increased levels of background light serves to amplify changes in the rate of glutamate release from the rod photoreceptor by enhancing the rate of calcium entry into the terminal. Further, increases in cyclic GMP concentration stimulated by nitric oxide will help to extend the range of light intensities over which the photoreceptor is responsive to changes in light intensity.

Now that we have seen that adaptation to light involves both the closure of gap junctions and changes in the sensitivity of photoreceptors to light, are the bipolar cells that create the On and Off channels modulated during adaptation ? To answer this question we need to understand the nature of the glutamate receptors present on the two types of bipolar cells that regulate the passage of information through the On and Off channels of the retina. Neurons have evolved two general types of receptor for the transmission of information at chemical synapses. The fastest of the two forms of receptor-mediated transmission at chemical synapses occurs when the transmitter substance opens an ion channel directly. As both the ion channel and the region that binds to transmitter are parts of the same assembly of protein subunits, conformational changes in the receptor are transmitted directly into the rapid opening of an ion channel. The second type of receptor transfers information from the transmitter into the stimulation of a G-protein pathway which then regulates one or more ion channels either directly or via an intermediate messenger such as cyclic GMP. Signals passing through such a metabotropic receptor pathway are therefore slower than those transmitted via receptor-operated ion channels, although metabotropic receptor pathways present more potential levels of regulation for the fine tuning of the synaptic response which may have been a key factor in their evolution. At the synapse between the Off bipolar cell and the photoreceptor, glutamate binds to a receptor-operated channel which is permeable to positively charged ions, or cations, and the resulting influx of these ions causes the post-synaptic membrane of the dendrites to become depolarized. The resulting wave of membrane depolarization travels to the bipolar cell terminal, resulting in the opening of calcium channels and an increase in the rate of glutamate release onto the receptive dendrites of amacrine and ganglion cells. Receptor-operated channels that selectively allow the entry of negatively-charged chloride ions into the cell are also present in most neurons in the retina. Such receptor-operated chloride channels bind to the inhibitory transmitters GABA or glycine that are released from horizontal and amacrine cells, and the resulting influx of negatively-charged chloride ions into neurons limits the extent of excitation in neurons bearing these receptors. In nature ying and yang are inseparable, and where excitatory receptor-operated channels are found, others that are inhibitory must be present to restore the balance, and in the brain the absence of such a balance can result in an epilepsy.

The first important clue that adaptation to light might occur in the channel created by the Off bipolar cell was provided by Gilbertson and Wilson, who showed that the receptor-operated channel of the Off bipolar cell is of an unusual type. This channel is composed of an arrangement of protein subunits that allow not only sodium ions, but also calcium ions to enter into the dendrites. This is a property shared by glutamate receptor-operated channels of the hippocyclic AMPal region of the brain which mediate long-lasting changes in the passage of synaptic information through the admission of calcium into the dendrites, a process that is believed to be central to the higher processes of learning and memory. As the concentration of intracellular calcium provides a hinge around which cell processes swing in almost all cells, it is tempting to propose a role for calcium in adaptation. Logic directs us to the glutamate-operated receptor-channel as the most likely molecular target for calcium-mediated changes in the amplification of light signals in the Off bipolar cell. Reinforcing this proposal Greg Maguire and Frank Werblin were recently able to show that dopamine increases the size of the cation current passing through the glutamate receptor-channel. In an elegant study they revealed that dopamine binds to its membrane receptor at the surface of the Off bipolar cell and stimulates a class of G-protein whose liberated alpha subunits migrate to activate the enzyme adenyate cyclase which converts ATP into cyclic adenosine monophosphate (cyclic AMP) in a second messenger pathway that is analogous to the guanylate cyclase-phosphodiesterase pathway. The second messenger cyclic AMP diffuses to the regulatory subunits of protein kinase A which have a high affinity for the messenger molecule. The change in conformation in the regulatory subunits that follows the binding of cyclic AMP causes the liberation of the catalytic subunits of protein kinase A which are then free to phosphorylate the glutamate receptor-operated channel. This phosphorylation increases the size of the cation current in response to a given discharge of glutamate from the photoreceptor. By acting through the cyclic AMP second messenger pathway dopamine amplifies the magnitude of receptor responses in the Off bipolar cell to light signals received from the photoreceptor. Thus dopamine effectively increases gain in the light signal through the Off channel in daylight, which may compensate for the parallel decrease in the amplification of the light response in the photoreceptor.

This brings us at last to the mystery of the On bipolar cell, whose responses to light are a curious inversion of those of the photoreceptor and the Off bipolar cell. The conundrum of this 'sign-inverting synapse' was solved by Craig Jahr, Gertrude Falk, Scott Nawy and Richard Shiells who were able to show that the glutamate receptor of the On bipolar cell activates a G-protein cascade which appears at first glance to be identical to the pathway in the photoreceptor. Binding of glutamate to the mGluR6 receptor stimulates a G-protein cascade leading to the stimulation of another cGMP-cleaving phosphodiesterase and the closure of another population of cation channels that appear to opened by increases in the cyclic GMP concentration. The genetic make-up of this glutamate receptor was subsequently described by Nakajima and Nakanishi and the receptor was found to be present exclusively in the On bipolar cell of the retina. The glutamate receptor of the On bipolar cell was the sixth member of the family of metabotropic glutamate receptors which exert their effects through G-proteins to be cloned, and hence it was given the abbreviation mGluR6. Taking advantage of the unique presence of this receptor in the On bipolar cell, Masu and Nakanishi bred a strain of mice in which both copies of the mGluR6 receptor gene had been 'knocked out' or deleted at the level of the germ cell. In these mice all responses to light in the On channel were abolished, but responses carried by the Off channel were left unscathed, and despite significant visual impairments these mice were in fact far from blind. These intriguing experiments support the theory that the On and Off channels were created for the generation of contrast and the detection of movement, as neither channel on its own appears essential for sight. The idea that the receptor pathways of the On bipolar cell and photoreceptor persisted were essentally similar persisted until Pedro de la Villa, Takashi Kurahashi and Akimichi Kaneko demonstrated that the properties of the cyclic GMP-regulated channel of the On bipolar cell were fundamentally different from those of the photoreceptor. A further contradiction arose from the experiments of Rhodri Walters and Scott Nawy which showed that although there was much evidence for a G-protein receptor kinase to terminate the response to glutamate, there was in fact no evidence for the binding of an arrestin protein.

But how is the On channel adapted to changes in background light levels ? There seemed to be little food for thought until Masayuki Yamashita and Heinz Wassle reported that the channel current that is suppressed by glutamate in the On bipolar cell was very sensitive to the concentration of calcium ions outside the cell. These findings were further explained when Rhodri Walters and Scott Nawy showed that current flow through the cyclic GMP-regulated channel was diminished if phosphorylation by a protein kinase activated by the binding of calcium-calmodulin was prevented. Moreover, cGMP-regulated channels quickly close even in the presence of high concentrations of cyclic GMP if the phosphate groups on the channel are stripped off by a phosphatase. Therefore during light adaptation the size of the cation current carried through the channel may increase because calcium ions enter the cell, most probably through the channel itself, and subsequently activate the calcium-calmodulin-regulated protein kinase which then phosphorylates and opens more cyclic GMP-dependent channels. Again, as in the case of the photoreceptor and the Off bipolar cell, calcium appears to play a pivotal role in mediating adaptation in the On bipolar cell. But what role does dopamine play in adaptation in the On channel of light information processing ? Walters and Nawy recently showed that phosphorylation by protein kinase A prolongs the active life-time of the receptor after the receptor binds to glutamate. This effectively increases the efficiency of coupling of the mGluR6 receptor to G-protein stimulation and activation of the cGMP phosphodiesterase, prolonging the closure of the cGMP-regulated channels and the duration of the light signal in the On pathway. In fact for the mGluR6 receptor to be able to close cGMP-dependent channels at all, it appears that it must first be phosphorylated by protein kinase A. This may not be the complete story of modulation of the mGluR6 receptor in adaptation, as Walters and Nawy were also able to show that lowering the concentration of free calcium ions inside the cell using the rapid calcium-capturing compound BAPTA, resulted in the uncoupling of the receptor from the phosphodiesterase, implicating a further role for calcium ions in receptor regulation.

Putting the On bipolar cell into the general picture of adaptation, we may predict that increases in free calcium and cyclic AMP levels in response to daylight will increase both the size of the current carried through the cyclic GMP-dependent channel and also the ability of the metabotropic glutamate receptor to suppress it, as the degree of coupling of the mGluR6 receptor to the cyclic GMP-phosphodiesterase is enhanced by both protein kinase A and elevations in the levels of cellular calcium. However, as increased levels of calcium appear to open more cyclic GMP-dependent channels during daylight, the potential number of channels that are available for closure in response to glutamate will also increase, extending the range of background light intensities over which the On-bipolar cell can measure changes in light intensity. But this does not explain why distinct populations of rod and cone bipolar cells evolved, one specialised for nocturnal and the other for diurnal vision. The rod On bipolar cell is very effective in the detection of photoreceptor signals generated in response to changes in light intensity at very low light levels, and Jonathon Ashmore and Gertrude Falk were able to show that light responses may be amplified as much as two-hundred fold at the synapse between the rod-bipolar cell and photoreceptor. In contrast, evolution may have directed the cone On bipolar cell to change roles from a high-gain amplifier at low background light levels to a variable amplifier that is regulated by second messengers such as calcium and cyclic AMP which allow the cone On bipolar cell to adjust to a range of background light levels.

In summary the retina has neatly evolved into a self-regulating neural circuit that can be studied in isolation, which converts the energy of light into electrochemical signals that can be deciphered by higher centers in the visual system. The photoreceptor captures and amplifies the energy of light and passes on the information encoded as changes in the rate of glutamate release from its terminals onto a layer of bipolar relay neurons. Bipolar neurons both amplify the signal and separate it into parallel On and Off channels in response to increases and decreases in light intensity respectively. Bipolar cells then pass photoreceptor signals that have been shaded by the output of surrounding horizontal cells onto the ganglion cell layer which integrates outputs from both bipolar and amacrine cells. The ganglion cell layer converts information about incident light from the graded analogue signal it receives from the bipolar cell into a digital signal which is then transferred at speed to the visual cortex for reconstruction. The retina has employed three principal second messenger pathways to rewire its circuitry from the efficient detection of low level changes in light intensity at night, to one that measures rapid changes in light intensity against the higher background light levels of the day. The release of dopamine from amacrine cells into the retina as background light levels increase, leads to the amplification of signals from the photoreceptor in both On and Off bipolar cells through the elevation of cyclic AMP levels. In addition dopamine shuts down gap junctional communication between horizontal and amacrine cells which appears to be of importance in enhancing signal resolution at low light intensities. Nitric oxide may increase both the amplification of light responses in the photoreceptor and the range of light intensities over which the photoreceptor is able to operate. In addition nitric oxide closes down the channel of night vision at the level of the AII amacrine cell. A third intracellular messenger calcium alters the sensitivity of the photoreceptor and bipolar cells to changes in light intensity to provide a rapid channel-mediated barometer of ambient light levels. Increases in intracellular calcium levels reduce the sensitivity of the photoreceptor to changes in light intensity and increase the level of gain at the synapse between the bipolar cell and the photoreceptor. In both types of cell increases in cellular calcium levels appear to extend the range of background light levels over which the circuitry can detect changes in light intensity in exchange for a reduction in the absolute sensitivity to light, or amplification of the light signal.

What other fundamental questions remain to be answered about mechanisms of light adaptation in the retina? After the adage of the goose and the gander, one might reasonably predict that cyclic GMP levels in the On bipolar cell will increase in response to nitric oxide, and conversely that photoreceptor responses to light may be somehow modulated by dopamine. And if the gap junctional communication between horizontal cells is regulated by dopamine, what role is there for calcium and nitric oxide in regulating adaptation in these cells ? If the analogue output of the photoreceptor, horizontal cell and bipolar cell circuitry of the outer retina is dramatically altered during adaptation, how is the digitising output of the amacrine and ganglion cells modulated by these second messenger pathways ? It is already known that responses to light and GABA of ganglion cells are modulated by dopamine. Further, nitric oxide-sensitive forms of guanylate cyclase are present in certain populations of amacrine and ganglion cells, suggesting that this second messenger pathway modulates the excitability of these neurons in response to an increase in background light levels. To date over twenty types of amacrine cell have been described, each with its own distinct pattern of wiring and cargo of neurotransmitter, and their respective roles in the processing of light information is poorly understood. Perhaps one of the greatest challenges facing neuroscience is the functional characterization of these highly specialised cells that are involved in the primary processing of light information.

In conclusion we are learning many fundamental lessons about cell signaling and inter-cellular communication from the retina. We are beginning to gain an insight into how molecules, proteins and membranes come together to form electrical circuits that are more sensitive than any that man has so far devised. The retina is not only a window to determining the how higher centers of the nervous system may function, it is a bridge towards understanding the interaction between mind and environment. By discovering how sensory information is detected and encoded in the retina, reconstructed in the visual cortex and ultimately converted into changes at the level of the genes themselves, we hope to understand the processes of the nervous system that form the basis of memory and experience and define our very being.