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

Log threshold elevation

5

4

3

2

1

0

Time in dark (min)

in humans and other mammals, measured by retinal densitometry; (2) normal human dark adaptation behavior (as in Figure 5); and (3) the slowed regeneration of pigment and the slowed dark adaptation that is characteristic of a number of diseases that affect the photoreceptors and/or retinal pigment epithelium.

See also: Light-Driven Translocation of Signaling Proteins in Vertebrate Photoreceptors; Phototransduction: Adaptation in Cones; Phototransduction: Inactivation in Rods; Phototransduction: Phototransduction in Rods; Phototransduction: The Visual Cycle.

Further Reading

Cameron, A. M., Mahroo, O. A. R., and Lamb, T. D. (2006). Dark adaptation of human rod bipolar cells measured from the b wave of the scotopic electroretinogram. Journal of Physiology 575: 507–526.

Krispel, C. M., Chen, C-K., Simon, M. I., and Burns, M. E. (2003). Novel form of adaptation in mouse retinal rods speeds recovery of phototransduction. Journal of General Physiology 122: 703–712.

Lamb, T. D. and Pugh, E. N., Jr. (2004). Dark adaptation and the retinoid cycle of vision. Progress in Retinal and Eye Research

23: 307–380.

Lamb, T. D. and Pugh, E. N., Jr. (2006). Phototransduction, dark adaptation, and rhodopsin regeneration. The Proctor Lecture.

Investigative Ophthalmology and Visual Science 47: 5138–5152. Nikonov, S., Lamb, T. D., and Pugh, E. N., Jr. (2000). The role of steady

phosphodiesterase activity in the kinetics and sensitivity of the lightadapted salamander rod photoresponse. Journal of General Physiology 116: 795–824.

Pugh, E. N., Jr. and Lamb, T. D. (2000). Phototransduction in vertebrate rods and cones: Molecular mechanisms of amplification, recovery and light adaptation. In: Stavenga, D. G., de Grip, W. J., and Pugh, E. N., Jr (eds.) Handbook of Biological Physics, Vol. 3, Molecular Mechanisms of Visual Transduction, ch. 5, pp. 183–255. Amsterdam: Elsevier.

Pugh, E. N., Jr., Nikonov, S., and Lamb, T. D. (1999). Molecular mechanisms of vertebrate photoreceptor light adaptation. Current Opinion in Neurobiology 9: 410–418.

Tamura, T., Nakatani, K., and Yau, K-W. (1991). Calcium feedback and sensitivity regulation in primate rods. Journal of General Physiology 98: 95–130.

Glossary

Arrestin – A protein that selectively binds light-activated phosphorylated photopigment and blocks further signal transduction. Cones express two subtypes, arrestin1 and arrestin4 (often termed rod and cone arrestins, respectively).

Cone opsins – Light receptors, consisting of the protein part (opsin) and 11-cis-retinal covalently attached via Schiff base to a lysine in the seventh transmembrane domain. All opsins are members of superfamily of G-protein-coupled receptors (GPCRs), the largest family of signaling proteins in animals (mammals have 1000 different GPCRs). GCAP – Guanylyl cyclase activating protein is a member of the superfamily of EF-hand-containing calcium-binding proteins. Cones express two homologs, GCAP1 and GCAP2, which in the calcium-liganded form inhibit and in magnesium-liganded form enhance the activity of retinal guanylyl cyclase (retGC).

GRKs – G-protein-coupled receptor kinases that specifically phosphorylate active forms of their cognate receptors. Cones express rhodopsin kinase (systematic name: GRK1) and a cone-specific form GRK7. However, mice do not have GRK7; therefore, photopigments in mouse cones and rods are phosphorylated by a single isoform, GRK1.

Phosphodiesterase (PDE) – The photoreceptorspecific cyclic guanosine monophosphate (cGMP) PDE, PDE6. Cone PDE6 is a heterotetramer, consisting of two identical catalytic a0-subunits and two inhibitory g-subunits. PDE rapidly hydrolyzes cGMP upon its activation by transducin, when its catalytic activity approaches the theoretical limit set by the rate of cGMP diffusion.

RetGC – Retinal guanylyl cyclase is structurally related to receptor guanylyl cyclases. Cones predominantly express RetGC1, in contrast to rods that express RetGC1 and RetGC2 at comparable levels.

RGS9-1 – Photoreceptor-specific short isoform of the regulator of G-protein signaling 9 expressed in both rods and cones. It interacts with the complex of the guanosine triphosphate (GTP)-liganded active a-subunit of transducin with PDEg and facilitates its intrinsic GTPase activity, thereby directly inactivating

transducin and indirectly PDE. Cones express much more RGS9-1 than rods.

Transducin – Photoreceptor-specific heterotrimeric G protein that couples to light-activated opsins. Its a-subunit belongs to Gi/o family. All types of cones express the same a-subunit that is different from the rod variant.

Rod photoreceptors are often described as a marvel of molecular engineering, which creates an impression that cones are just noisier and less-sensitive rods. In fact, as light sensors, cones are just as amazing: their adaptability gives cones a much wider dynamic range covering more than seven orders of magnitude of light intensity without saturation. Cones begin to function in the light of the full moon reflected from objects in the night and are still adequate for a direct look at the sun. Mostly for technical reasons, the biochemistry of cone photoreceptors, particularly the molecular mechanisms underlying adaptation, is not as well studied as the signaling in rods. The assumption that the signaling and shutoff mechanisms in cones and rods are qualitatively similar is often used to fill the gaps in our knowledge of cone biochemistry. To avoid repetition, here we emphasize known differences between the cone and rod inactivation mechanisms.

Cone Signaling Cascade

Cone opsins are closely related to rhodopsin and belong to the same branch of the G-protein-coupled receptor superfamily. Gene duplication events in early vertebrate evolution produced five groups of light receptors: rhodopsins and four classes of cone opsins. Mammals lost half of cone opsin classes, retaining only two. Light activates cone opsins via induced isomerization of 11-cis-retinal covalently attached to a lysine in the seventh transmembrane domain. Cone opsins use the same 11-cis-retinal as rhodopsin, but have very different spectral sensitivity, with maxima ranging from 360 nm (ultraviolet) to 575 nm (red). Spectral tuning of covalently linked retinal is achieved by changing its environment in the retinalbinding pocket of opsin. Light-activated cone opsins couple to cone transducin, which, in turn, activates the cone subtype of phosphodiesterase 6 (PDE6). Subsequently,

a decrease of cytoplasmic cyclic guanosine monophosphate (cGMP) reduces the influx of Naþ and Ca2þ through the cone variant of cGMP-gated channels, resulting in cell hyperpolarization. Similar to rods, the decrease in Ca2þ concentration and consequent replacement of bound Ca2þ with Mg2þ converts guanylate cyclase activating proteins (GCAPs) from inhibitors to activators of retinal guanylate cyclase (RetGC). The latter replenishes cGMP lost during the light response, which opens the channels, thereby restoring cytoplasmic Ca2þ to the original levels. Thus, the activation and deactivation mechanisms employed by rods and cones are quite similar. However, subtle differences at every step of the pathway, including differences in the subtypes of signaling proteins involved, their expression levels, and the geometry of the cell, result in striking functional specialization of the two types of photoreceptors.

Shutoff of the Light-Activated Cone

Opsins

As far as signaling is concerned, the key difference between rhodopsin and cone opsins is lower thermal stability of the latter. Because cone opsins spontaneously activate with much higher probability than rhodopsin and readily release retinal even in the dark, cones generate noise that is orders of magnitude higher than in rods. This rules out the detection of signals below the noise level (e.g., a few photons) and makes even completely dark-adapted cones pre-desensitized and ready to operate at light levels they can detect. Loose attachment of the retinal to opsin also results in a significantly faster spontaneous decay of the light-activated cone opsin. For example, mouse cone S-opsin transgenically expressed in rods lacking arrestin was estimated to decay 40 times faster than rhodopsin coexpressed in the same cell. However, this spontaneous decay with a time constant of 1.3 s is still much slower than the rate of recovery in cones.

Evolution equipped cones with a more elaborate (and presumably more efficient) photopigment shutoff machinery than that found in rods. The opsin inactivation is accelerated by pigment phosphorylation followed by arrestin binding. In most species, cones express two G-protein receptor kinase (GRK) subtypes: GRK1 (shared with rods) and cone-specific GRK7. It is likely that the co-expression of GRK7 with higher enzymatic activity accelerates opsin phosphorylation in cones. However, it should be noted that mice and rats are rare exceptions: these nocturnal rodents have only GRK1 in both types of photoreceptors. Cones also express two arrestin subtypes, arrestin1 and cone-specific arrestin4 (formerly known as rod and cone arrestins, respectively). Arrestin1 is present in cones at 50-fold molar excess over arrestin4. A recent study in knock-out animals shows

that both arrestins contribute comparably to the shutoff of the photopigment in cones. It is not entirely clear why cones express two arrestin subtypes, rather than a higher level of one subtype, especially considering that the cone opsin transgenically expressed in mouse rods is rapidly and efficiently deactivated by rod arrestin1.

Two functional differences between these arrestins provide some clues. Arrestin1 has high propensity to selfassociate, cooperatively forming dimers and tetramers at physiological concentrations. Even in the dark-adapted rod, where the outer segment contains a small fraction of the total arrestin1, most of arrestin1 is a tetramer. It has been unambiguously shown that only monomeric arrestin1 is an active rhodopsin-binding species; therefore, oligomers appear to be storage forms. In contrast, cone-specific arrestin4 does not self-associate at physiologically relevant concentrations; therefore, the whole complement of arrestin4 present in cones is an active monomer. Recent estimates of their expression and arrestin1 self-association constants suggest that dark-adapted cones have in the outer segment 60 mM of arrestin4 and 30 mM (5% of the total) of monomeric arrestin1 ready to bind phosphorylated opsin at any time, in addition to a huge backup supply of arrestin1 oligomers.

The second important difference lies in the stability of the arrestin complex with phosphorylated opsin. Arrestin1 forms very stable complexes that take the bound molecule of phosphopigment out of the game for a long time. This is important in the rod to ensure the fidelity of the shutoff. In order to release completely inactive rhodopsin upon dissociation, arrestin1 must stay bound until metarhodopsin II (Meta II) slowly decays and likely until it is regenerated with 11-cis-retinal. In contrast, arrestin4 forms fairly transient complexes with phosphorylated cone opsins, likely to ensure the quick return of the opsin back into the active pool. This is important for cone photoreceptors that function at a high rate of pigment bleaching. Although we do not know with certainty why cones express both arrestin subtypes, one scenario appears to provide a plausible explanation. Given the concentrations of the two arrestins in cones, in moderately bright light arrestin4 likely has an advantage, so that the majority of phosphorylated cone opsin would be rapidly recycled to the signalingcompetent pool. Increasing levels of illumination inducing massive pigment bleaching would force the cell to draw increasingly on the virtually inexhaustible supply of arrestin1, which forms long-lived complexes with the opsin. The recent finding that arrestin1 plays a more prominent role in cone recovery after very bright flashes is consistent with this model. The formation of arrestin1opsin complexes would take larger and larger fraction of the pigment out of action for a relatively long time, possibly serving as one of the mechanisms of light adaptation. Even though a cone-specific visual cycle involving

Mu¨ller glia provides 11-cis-retinal faster than the canonical retinal pigment epithelium-based visual cycle supplying rods, very bright light bleaches cone pigment faster than it can be regenerated. This loss of functional opsin was proposed to reduce light capture, acting as a mechanism of adaptation. It is entirely possible that in bright light, both incomplete regeneration of opsin and its binding by arrestin1 cooperate to limit the active pool, thereby reducing light sensitivity of cones.

Overall, cones combine less-stable photopigment with more sophisticated machinery of its inactivation (Figure 1). These factors apparently contribute to faster shutoff at the opsin level and likely provide cone-specific mechanisms for light adaptation.

Inactivation of Transducin and PDE

It is generally accepted that the activation of cone transducin by cone opsins and that of cone PDE6 by the guanosine triphosphate (GTP)-liganded a-subunit of cone transducin proceeds in similar ways to corresponding processes in rods. All three subunits of cone transducin differ from their rod counterparts, but the significance of this specialization is uncertain. In fact, cone S-opsin transgenically expressed in mouse rods efficiently activates the signaling cascade coupling to rod transducin. Cone PDE6 is an a02g2 heterotetramer, in contrast to the abg2 version in rods, but the functional significance of the use of different catalytic subunits remains to be elucidated. There is one biochemical difference that undoubtedly contributes to the much faster inactivation of transducinPDE6 complex in cones: 10-fold higher level of the regulator of G-protein signaling 9-1 (RGS9-1) expression. It has been convincingly shown that the deactivation at this step rate limits the recovery kinetics in rods, and that the level of RGS9-1, which accelerates self-inactivating GTPase of transducin a-subunit, sets the speed of transducin-PDE6 inactivation. Thus, the shutoff at the opsin and transducin-PDE6 level in cones is much faster than corresponding processes in rods; however, it is still not clear which step is rate limiting in cone recovery.

Restoration of cGMP and Intracellular

Calcium Level

Similar to the situation in rods, cone activation results in a drop in the intracellular Ca2þ concentration due to the closure of the cGMP-gated channels mediating the bulk of Ca2þ entry. In order to return to the initial state after opsin and PDE6 are fully inactivated, cones need to restore cytoplasmic cGMP hydrolyzed by PDE6. Cones and rods use the same negative-feedback mechanism that

translates the drop in Ca2þ resulting from the reduction in the cGMP level into a signal to make more cGMP. Ca2þ dissociates from GCAPs when its cytoplasmic concentration drops in the light. The replacement of lost Ca2þ by Mg2þ converts GCAPs from inhibitors to activators of RetGC. The generated cGMP opens the channels, and the consequent increase in cytoplasmic Ca2þ stops further cGMP synthesis. Cones apparently express the same combination of GCAP1 and GCAP2 (which differ in their Ca2þ sensitivity) as rods. The functional significance of the predominance of the RetGC1 isoform in cones (in contrast to similar levels of RetGC1 and RetGC2 in rods) is not clear.

Several important differences between rods and cones are known to be responsible for much faster cone recovery. First, the rate of recovery depends on the absolute amounts of cGMP and Ca2þ that need to be replenished. Here cones hold an obvious advantage due to the much smaller volume of their outer segments: the hydrolysis or synthesis of the same absolute amount of cGMP leads to a more significant change in its concentration. Similarly, the closure of the same fraction of cGMP-gated channels leads to a more profound drop in intracellular Ca2þ in cones. However, geometry is only part of the story. The channel expressed in cones has a different subunit composition and ion preference. About 35% of the inward current via the cone cGMP-gated channel is carried by Ca2þ, whereas in rods this fraction is only 20%. Thus, the closure of the same fraction of channels upon PDE6 activation results in a substantially greater change in the absolute number of Ca2þ ions entering the cell. Increased Ca2þ influx in cones is balanced by its accelerated extrusion via Naþ/Kþ–Ca2þ-exchanger, so that the turnover of Ca2þ in cone outer segments is more rapid. The combination of faster constitutive extrusion, larger fraction of the current carried by Ca2þ through cGMP-gated channels, and much smaller outer segment volume greatly increases the rate of Ca2þ drop in response to light stimulus, speeding up RetGC activation and cGMP resynthesis. High intracellular Ca2þ reduces the sensitivity of the channels to cGMP, so that when the intracellular Ca2þ drops, the channels become more sensitive to the cytoplasmic cGMP and therefore reopen faster. This mechanism operates in both types of photoreceptors, but it is more powerful in cones, further contributing to accelerated recovery.

Conclusions

Cone photoreceptors use essentially the same molecular mechanisms of signal shutoff at the opsin level as rods. At this step, cones achieve much higher speed of inactivation by employing, in addition to GRK1 and arrestin1

used by rods, cone-specific GRK7 and arrestin4. The presence of two GRKs speeds up the phosphorylation of light-activated opsin, whereas the expression of two arrestin subtypes with very different functional characteristics

likely results in a gradual switch from rapidly reversible arrestin4 interaction with phospho-opsin at moderate light levels to semi-irreversible binding of arrestin1 in very bright light. Inactivation at the transducin/PDE

level is accelerated by a 10-fold higher expression of RGS9-1 in cones. Two key features ensure faster recovery in cones than in rods. Faster Ca2þ turnover due to higher influx through cone-specific cGMP-gated channels and efflux via Naþ/Kþ–Ca2þ-exchanger generate greater net changes in the number of Ca2þ ions in the outer segment when the same fraction of the channels is closed. Due to much smaller outer-segment volume, the same net change in the number of cGMP molecules or Ca2þ ions produces greater changes in the concentration of these second messengers. Rapid response and recovery gives cones better temporal resolution than rods. The high speed of activation and inactivation in combination with more powerful adaptation mechanisms (many of which still need to be elucidated at the molecular level) allows cones to function in a broad range of light levels without saturation.

See also: Phototransduction: Adaptation in Cones; Phototransduction: Inactivation in Rods; Phototransduction: Phototransduction in Cones; Phototransduction: Phototransduction in Rods; Phototransduction: Rhodopsin.

Further Reading

Cote, R. H. (2006). Photoreceptor phosphodiesterase (PDE6):

A G-protein-activated PDE regulating visual excitation in rod and cone photoreceptor cells. In: Beavo, J. A., Francis, S. H., and Houslay, M. D. (eds.) Cyclic Nucleotide Phosphodiesterases in Health and Disease, pp 165–193. Boca Raton, FL: CRC Press.

Dizhoor, A. M., Olshevskaya, E. V., and Peshenko, V I. (2006). Calcium sensitivity of photoreceptor guanylyl cyclase (RetGC) and congenital photoreceptor degeneration: Modeling in vitro and in vivo. In:

Philippov, P. P. and Koch, K.-W. (eds.) Neuronal Calcium Sensor Proteins, pp 203–219. New York: Nova Science Publishers, Inc.

Gurevich, V. V., Hanson, S. M., Gurevich, E. V., and Vishnivetskiy, S. A. (2007). How rod arrestin achieved perfection: Regulation of its availability and binding selectivity. In: Kisselev, O. and Fliesler, S. J. (eds.) Signal Transduction in the Retina. Methods in Signal Transduction Series, pp 55–88. Boca Raton, FL: CRC Press.

Hanson, S. M., Van Eps, N., Francis, D. J., et al. (2007). Structure and function of the visual arrestin oligomer. European Molecular Biology Organization Journal 26: 1726–1736.

Knox, B. E. and Solessio, E. (2006). Shedding light on cones. The Journal of General Physiology 127: 355–358.

Korenbrot, J. I. and Rebrik, T. I. (2002). Tuning outer segment Ca2þ homeostasis to phototransduction in rods and cones. Advances in Experimental Medicine and Biology 514: 179–203.

Nikonov, S. S., Brown, B. M., Davis, J. A., et al. (2008). Mouse cones require an arrestin for normal inactivation of phototransduction. Neuron 59: 462–474.

Glossary

Arrestin (also known as S-antigen, 48-kDa protein, and rod or visual arrestin; systematic name: arrestin1) – A protein that selectively binds light-activated phosphorylated rhodopsin and blocks further signal transduction.

Guanylyl cyclase activating protein (GCAP) –

A member of the superfamily of EF-hand-containing calcium-binding proteins. Rods express two homologs, GCAP1 and GCAP2, which in calciumliganded form inhibit and in magnesium-liganded form enhance the activity of retinal guanylyl cyclase (retGC).

Phosphodiesterase (PDE) – Photoreceptorspecific cGMP phosphodiesterase, PDE6. Rod PDE6 is a heterotetramer, consisting of two nonidentical catalytic subunits (a- and b-) and two inhibitory g-subunits. PDE6 rapidly hydrolyzes cyclic guanosine monophosphate (cGMP) upon its activation by transducin. In fully activated state, its catalytic activity approaches the limit set by the rate of cGMP diffusion.

RetGC – It is structurally related to receptor guanylyl cyclases. Rods express comparable levels of two homologs, RetGC1 and RetGC2.

Regulator of G-protein signaling 9 (RGS9-1) –

Photoreceptor-specific short isoform of the regulator of G-protein signaling 9 expressed in both rods and cones. It interacts with the complex of the guanosine triphosphate (GTP)-liganded active a-subunit of transducin with PDEg and facilitates its intrinsic GTPase activity, thereby directly inactivating transducin and indirectly PDE.

Rhodopsin – Light receptor, consisting of the protein part (opsin) and 11-cis-retinal covalently attached via Schiff base to a lysine in the seventh transmembrane domain. A member of the superfamily of G-protein- coupled receptors (GPCRs), also known as seven transmembrane domain receptors (7TMRs), the largest family of signaling proteins in animals (mammals have 1000 different GPCRs).

Rhodopsin kinase (RK) (systematic name: GRK1) –

It is a member of the G-protein-coupled receptor kinase (GRK) family expressed in both rods and cones.

Transducin – Photoreceptor-specific heterotrimeric G protein that couples to light-activated rhodopsin. Its a-subunit belongs to Gi/o family.

As light sensors, vertebrate rod photoreceptors are a remarkable evolutionary achievement: rods yield amazingly low noise despite the presence of 108–109 molecules of the light receptor rhodopsin, and demonstrate singlephoton sensitivity and a dynamic range of seven orders of magnitude of light intensity. This level of perfection is achieved through several unique structural and biochemical adaptations. The rod outer segment (OS) is a specialized signaling compartment containing rhodopsin molecules tightly packed in disks. It is separated from the inner segment (IS), which is a mitochondria-rich power station providing huge amounts of energy. Several soluble signaling proteins move between the two compartments depending on the illumination, ensuring their on-demand delivery to the OS. The OS concentrations of transducin (Td) and arrestin, the proteins that transmit and shut down rhodopsin signaling, respectively, change by at least 10-fold. An important functional feature of the rod is that every biochemical step in the pathway between photon capture and the change in synaptic output has its own dedicated shutoff mechanism.

What Needs to Be Inactivated: Overview of the Signaling Cascade

Rhodopsin activation by a photon of light is the first step in visual signaling. Due to extremely high concentration of its cognate G protein, Td, and rapid diffusion of both active rhodopsin (Rh*) and Td in the plane of the disk membrane, Rh* activates a molecule of Td every few milliseconds, generating 50–100 active Td (Td*) during its lifetime. These events occur in the two-dimensional space on the cytoplasmic surface of disk membranes. Each Td* binds the inhibitory g-subunit of cyclic guanosine monophosphate (cGMP) phosphodiesterase (PDE6), turning the enzyme on. Each molecule of active PDE6 hydrolyzes several cGMP molecules per millisecond, producing a rapid drop in the cGMP concentration in the three-dimensional cytoplasmic space. This results in closure of cGMP-gated Na+/Ca2+ channels on the

plasma membrane. In rods, light activation of a single rhodopsin translates into the hydrolysis of 100 000 cGMP molecules. The channels are heterotetramers, with each subunit carrying a cGMP-binding site in its C-terminal domain. Highly cooperative cGMP binding to the four sites in the channel greatly increases its response to the change in cGMP concentration. The decrease of the inward current hyperpolarizes the rod, reducing neurotransmitter release in its output synapse.

Several features of the rod signaling machinery bring its light sensitivity within the range of its physical limit: the detection of single photons. First, the concentration of signaling molecules in the rod OS is orders of magnitude higher than in normal cells: 3 mM rhodopsin (compared to low nanomolar concentrations of related receptors elsewhere), 0.3 mM Td, 60 mM PDE, and so on. Second, all three signaling proteins involved have much lower basal activity than their counterparts in other cells. This results in an incredibly low noise level, making signal-to-noise ratio favorable for the detection of even an extremely weak signal. Third, very efficient shutoff mechanisms at every step of the pathway rapidly terminate the signaling, allowing for an exquisite subsecond temporal resolution of mammalian rods.

Shutoff of the Light-Activated Rhodopsin

Rhodopsin is a prototypical G-protein-coupled receptor. In contrast to 1000 other members of this superfamily, it has virtually no basal activity, because it is effectively suppressed by the covalently attached inverse agonist, 11-cis-retinal. Retinal is converted by light into the alltrans form, which is a potent agonist of rhodopsin. The fact that it remains covalently attached to the receptor (in contrast to other GPCRs where bound and free agonists are in dynamic equilibrium) ensures a powerful burst of signaling. Through a series of short-lived photoproducts, light-activated rhodopsin reaches the Metarhodopsin II (Meta II) state, which is an active form (Rh*) that couples to Td. Meta II is in equilibrium with the two other states, Meta I and Meta III, which are believed to be inactive, or at least considerably less active than Meta II. Ultimately, all-trans-retinal dissociates, yielding empty protein opsin, which has orders of magnitude lower ability to activate Td than Meta II. However, this spontaneous deactivation of rhodopsin is inadequate as a shutoff mechanism for two reasons. At physiological temperatures, rhodopsin decay takes about a minute, which would greatly compromise temporal resolution. Moreover, the activity of opsin, which is much higher than that of dark rhodopsin, would generate considerable noise, compromising rod sensitivity. Therefore, rods use a sophisticated two-step mechanism to achieve rapid and complete rhodopsin deactivation.

First, light-activated rhodopsin is phosphorylated by rhodopsin kinase (RK). Similar to Td, RK is activated by binding to Rh*. Therefore, it selectively phosphorylates the active form of rhodopsin. It should be noted that at low light levels, RK was reported to phosphorylate multiple rhodopsin molecules for each light-activated one, likely by targeting neighboring inactive rhodopsins in the crowded disk membrane. In mammals, RK activity is believed to be held in check by its interaction with Ca2+- loaded recoverin. As a result, RK is fully unleashed only after a brief delay, which allows Td activation to continue until the Ca2+ concentration in the rod actually drops. However, low affinity of recoverin for Ca2+ (KD 5 mM) is the weak point of this model. Current estimates of the Ca2+ concentrations in the darkand light-adapted mouse OS are 250 nM and 25 nM, respectively, so that only a very small fraction of recoverin would be Ca2+-loaded in either. In addition, unlike RK, recoverin predominantly localizes in the inner segment. Still, rods express 50 molecules of recoverin for each RK, so a relatively small change in the Ca2+ occupancy of a fraction of recoverin present in the OS could conceivably play a role in RK regulation.

Phosphorylation per se reduces, but does not abolish the ability of rhodopsin to activate Td. In the next step, arrestin binds active phosphorylated rhodopsin (P-Rh*), shielding its cytoplasmic tip and precluding further Td interaction. Arrestin apparently remains bound until rhodopsin decays to opsin, and very likely even longer, until opsin is regenerated with 11-cis-retinal to the truly inactive dark rhodopsin. Arrestin has several dedicated phosphate-binding residues and other elements that specifically interact with light-activated rhodopsin independently of its phosphorylation state. These partial interactions mediate relatively low-affinity binding to dark P-Rh and unphosphorylated Rh*, respectively. Arrestin elements participating in these interactions also serve as sensors, allowing arrestin to test the functional state of the rhodopsin molecule it encounters and then quickly dissociate from its low-affinity targets, dark Rh, dark P-Rh, or Rh*. In contrast to all other forms, P-Rh* simultaneously engages both sets of elements. This turns the two sensors on at the same time, allowing the arrestin transition into a high-affinity rhodopsin-binding state. This transition involves a global conformational change in arrestin, which mobilizes additional arrestin elements for the interaction. Thus, arrestin works as a molecular coincidence detector, swinging into action only when the rhodopsin molecule it encounters is both active and phosphorylated. The model of sequential multisite interaction readily explains exquisite arrestin selectivity, that is, manifold difference in arrestin binding to Rh* and dark P-Rh on the one hand, and to its preferred target P-Rh* on the other. The salt bridge between positively charged Arg175 and negatively charged Asp296, which is one of the

intramolecular interactions holding arrestin in its basal state, was identified as the main phosphate sensor in arrestin. Rhodopsin-attached phosphates bind Arg175 and neutralize its charge, thereby breaking the salt bridge and facilitating arrestin transition into its active conformation. The reversal of either charge by targeted mutagenesis yields mutants with reduced need for rhodopsinattached phosphates that bind Rh* with much higher affinity than wild-type protein.

Rhodopsin has multiple phosphorylation sites in its C- terminus. The issue of the number of rhodopsin-attached phosphates necessary for high-affinity arrestin binding was resolved only recently. Studies performed in vitro with rhodopsin carrying defined number of phosphates and in vivo with mice expressing rhodopsin mutants with different number of sites show that rhodopsin multi-phos- phorylation is required. A single rhodopsin-attached phosphate does not appreciably increase arrestin affinity, two somewhat enhance the binding, and three phosphates are necessary for high-affinity interaction in vitro and for the rapid shutoff of photoresponse in vivo. Whereas arrestin binding does not further increase when Rh* has more than three phosphates, the presence of additional phosphorylation sites on Rh accelerates the shutoff of the photoresponse in vivo. This likely reflects the kinetic effect of the abundance of sites that remain available to RK on partially phosphorylated rhodopsin. For example, rhodopsin with three sites would have only one possible RK target left after the incorporation of two phosphates, whereas rhodopsin with six sites would still have four available targets at the same level of phosphorylation. Thus, supernumerary sites would ensure that the magic number of three phosphates per rhodopsin is achieved faster.

Inactivation of Td and PDE

Td is a prototypical heterotrimeric G protein consisting of a-, b-, and g-subunits. In the inactive state, the abgtrimer has guanosine diphosphate (GDP) in the nucleo- tide-binding site of the a-subunit. In this state, lipid modifications of both a-(N-terminal myristoyl) and g- (C-terminal farnesyl) subunits provide a fairly strong membrane anchor. This restricts the Td diffusion to the plane of the disk membrane and enforces the orientation favorable for Rh* interaction, thereby maximizing its chances of encountering active rhodopsin and being activated by it. The Td interaction with Rh* opens its nucle- otide-binding pocket, whereupon GDP promptly falls out and is immediately replaced by GTP simply because the latter is much more abundant in the cytoplasm. The GTP-liganded a-subunit dissociates from Rh* and bgdimer. Tda-GTP binds the inhibitory g-subunit of cGMP PDE, greatly increasing PDE activity by relieving

the inhibition. Importantly, the separation of the two parts of Td heterotrimer dramatically weakens their membrane anchoring, so that active Tda-GTP can jump off the disk where it was generated by Rh* and activate PDE on neighboring discs, spreading the signaling in three dimensions. Due to its very high catalytic activity (kcat2000 s 1 per subunit), active PDE rapidly reduces cGMP concentration in its vicinity, which leads to the closure of cGMP-gated channels and hyperpolarization of the rod within milliseconds of rhodopsin activation by light (Figure 1).

Similar to other heterotrimeric G proteins, Tda has GTPase activity, which serves as a built-in self-inactivation mechanism. However, the intrinsic GTPase of free Tda is very slow. Interaction of Tda with PDE g-subunit increases the activity of its GTPase. The interaction of Tda–GTP–PDEg complex with rod-specific GTPase activating protein (GAP) increases the GTPase activity even further. GAP consists of the short isoform of RGS9, Gb5 (homolog of G-protein b-subunits), and another protein that provides membrane anchor for the complex, RGS9 anchoring protein (R9AP). Low basal GTPase of Tda gives it time to diffuse around searching for PDE to activate without losing the signal in the transmission. The dramatic acceleration of the GTPase activity of Tda by the PDE and GAP ensures that the signal is terminated quickly after it is received by PDE, improving the temporal resolution of the photoreceptor cell. The recent finding that the expression of the GAP complex in rods increases the rate of the response shutoff in a dose-dependent manner convincingly demonstrated that the inactivation of Tda–GTP– PDEg complex is the rate-limiting step in this process. These elegant experiments also revealed that when this step is maximally accelerated, the recovery kinetics becomes dominated by some other process with the time constant of 80 ms. This number sets the upper limit for the next slowest step, which could be one of the following: the average lifetime of active Rh*; the release of PDEg from Tda-GDP; reassociation of PDEg with PDE catalytic subunits; or even the time free Tda-GTP spends searching for PDE and/or docking to it.

Resynthesis of cGMP and Restoration of

Calcium Level

Obviously, to return to its initial state and become ready to respond to the next photon with the same vigor, the rod photoreceptor needs to do more than just turn off Rh* and all Td and PDE molecules activated by it. The response leaves, in its wake, substantially reduced cytoplasmic cGMP concentration and very low intracellular calcium due to the closure of the cGMP-gated channels that are responsible for the bulk of Ca2+ entry into the OS. Photoreceptors are equipped with an

ingenious negative-feedback mechanism that translates the drop in Ca2+ resulting from the reduction in cGMP level into a signal to replenish it. In photoreceptors, cGMP is synthesized by retinal guanylyl cyclases (retGCs). RetGCs are related to a family of hormoneregulated guanylyl cyclases, such as atrial natriuretic factor receptor, which have extracellular hormone-binding domain connected via a single transmembrane helix to the intracellular guanylyl cyclase domain. Similar to these receptors, retGCs are dimeric, with each monomer

equipped with a catalytic domain and an extracellular domain. Interestingly, Mg2+ and GTP are bound by two different subunits forming the active catalytic site. As far as we know, the extracellular domain of retGCs neither binds any ligands nor participates in the enzyme regulation. Instead, the activity of retGCs is tightly regulated by their interaction via intracellular elements with GCAPs. GCAPs, as well as recoverin, are members of the neuronal calcium sensor protein branch of the superfamily of calcium-binding proteins containing EF hands (that includes calmodulin).