Molecular Mechanisms Characterizing Cone Photoresponses


  • This invited paper is part of the Symposium-in-Print: Photobiology in Asia

*email: (Shuji Tachibanaki)


In the vertebrate retina, rods mediate twilight vision and cones mediate daylight vision. Their photoresponse characteristics are different. The light-sensitivity of a cone is 102–103 times lower than that of a rod. In addition, the photoresponse time course is much faster in cones. The mechanism characterizing cone photoresponses has not been known mainly because of the difficulty in isolating cones in large quantities to perform biochemistry. Recently, we developed a method to purify cones from carp retina using a density gradient, which made it possible to analyze the differences in the molecular mechanism of phototransduction between rods and cones. The results showed that signal amplification in cones is less effective, which explains the lower light-sensitivity of cones. The results also showed that visual pigment phosphorylation, a quenching mechanism of light-activated visual pigment, is much more rapid in cones than in rods. The rapid phosphorylation in cones is attributed to a very high total kinase activity in cones. Because of this high activity, cone pigment is readily phosphorylated even at very high bleaching levels, which probably explains why cone photoresponses recover quickly. Based on these findings, the molecular mechanisms of the differences in the photoresponse characteristics between rods and cones are outlined.


We can see the external world in the presence of various intensities of light. To detect light signals, there are two types of photoreceptor cells, rods and cones (for a review, see 1–3), in the vertebrate retina. Although both types of the cells convert light signals to neural signals, their characteristics of the photoresponses are different at least in two aspects (4–7): one is the difference in the sensitivity to light, and the other is the difference in the duration of a photoresponse elicited by a brief nonsaturating light flash (flash response).

The light sensitivity of a rod is higher than that of a cone. Thus, rods mediate scotopic (twilight) vision and cones mediate photopic (daylight) vision. Each rod or cone covers about a 103-fold range of light-intensity, but the sensitivity of a rod is 102–103 times higher. Therefore, using rods and cones, we can cover the range of light-intensity over 106-fold. In addition, both rods and cones adapt to environmental light and extend their responsive ranges of light by changing their sensitivities (8,9). For this, light signals over a ∼1010-fold range can be converted to neural signals with the rod and cone system. In turtle cones, it has been reported that there is no compelling evidence for saturation (7).

The duration of a flash response in a cone is much shorter than that of a rod. In other words, in cones, the on- and the off-timing of a photoresponse coincides well with that of a light stimulus. This characteristic contributes to the improved time resolution in the detection of light in cones over rods. Thus, the motion detection is easier in cones (daylight vision) than in rods (twilight vision).

Although Purkinje described the presence of the two mechanisms of photoreception in 1823, known today as the rod and cone system, little is known about the molecular bases in the differences of the photoresponse characteristics between rods and cones. One of the reasons for this was that no one had succeeded in obtaining cones in large enough quantities to carry out biochemical studies. However, we recently succeeded in purifying cones from carp (Cyprinus carpio) retina which made it possible to investigate the molecular mechanisms of the differences in the photoresponse characteristics between rods and cones. In this review, elucidated mechanisms are shown based on our recent results (10–12).

Materials and methods

Purification of rods and cones from carp retina.  All manipulations were carried out in complete darkness with the aid of an infrared image converter (NVR 2015, NEC) under illumination of >800 nm light. Carp (Cyprinus carpio), 25–30 cm in length, were dark-adapted in a light-tight tank for >3 h before use, and the retina was dissected after pithing in darkness at night.

The surface of the retina was gently brushed and tapped using a paintbrush, and the photoreceptors taken off the retina were suspended in a Ringer’s solution (119.9 mM NaCl, 2.6 mM KCl, 0.5 mM CaCl2, 0.5 mM MgCl2, 0.5 mM MgSO4, 1 mM NaHCO3, 16 mM glucose, 0.5 mM NaH2PO4, 4 mM Hepes, pH 7.5). The suspension was filtered through a nylon mesh to eliminate large fragments of retinal tissues. The filtrate containing isolated photoreceptors was layered on the top of a stepwise Percoll gradient (see Fig. 3) and centrifuged for 20 min at 10 000 g. Cells at the interfaces were collected and mixed with the same volume of the Ringer’s solution to reduce the density of Percoll. The cells were then sedimented by a gentle centrifugation (first at 600 g for 12 s and then at 3000 g for 4 s). To eliminate contamination by red blood cells, the sedimented cells were additionally suspended in a K-gluconate buffer (K-gluc buffer; 115 mM K-gluconate, 2.5 mM KCl, 2 mM MgCl2, 0.2 mM EGTA, 0.1 mM CaCl2, 1 mM DTT, 10 mM Hepes, pH 7.5), and then gently sedimented again. Red blood cells were disrupted and removed in the supernatant. The resultant cells were frozen immediately in liquid nitrogen and stored at –80°C. To prepare rod and cone membranes, the frozen cells were thawed, and washed twice with the K-gluc buffer. The membranes were quantified spectrophotometrically by measuring the amount of visual pigments contained in the membranes.

Figure 3.

 Purification of rods and cones from carp retina. Photoreceptor cells were brushed off the retina, and then rods and cones were purified with a Percoll stepwise gradient. Modified from fig. 1 in Ref. (10). Copyright (2001) National Academy of Sciences USA.

Electrophysiological measurements.  The photoresponses of photoreceptor cells were measured with suction electrodes as described (13). The outer segment of a mechanically dissociated carp red-sensitive cone or rod was sucked into an electrode, and flash responses were recorded.

Transducin activation assay.  A rod or cone membrane suspension (15 μL) was mixed with 10 μL of the K-gluc buffer containing [35S] GTPγS, GDP and EGTA. The sample contained 0.3–3 μM visual pigment, 5 μM [35S] GTPγS, 5 μM GDP, and 0.8 mM EGTA. A light flash was given to the sample 30 s after the mixing, and the sample was incubated for a desired time. The reaction was terminated by adding 200 μL of the K-gluc buffer containing 20 μM cold GTPγS and 10 μM GDP. The sample was then filtered immediately through a nitrocellulose membrane. The rod or cone membranes containing activated transducin with bound [35S] GTPγS were trapped on the nitrocellulose membrane. Then the free GTPγS was washed out from the nitrocellulose membrane with 600 μL of the K-gluc buffer containing 25 mM MgCl2. The amount of activated transducin was quantified with an image analyzer (BAS 2000, Fuji) by measuring the radio activity of [35S] GTPγS bound to activated transducin on the nitrocellulose membrane.

PDE assay.  The activity of cGMP phosphodiesterase (PDE) was measured using the pH assay method (14). Rod or cone membranes were suspended in the K-gluc buffer containing 0.5 mM GTP, 5 mM cGMP and 0.8 mM EGTA. The pH drop as a result of hydrolysis of cGMP by activated PDE was monitored with a small combination glass electrode (MI-410, Microelectrodes). The range of the pH drop during a measurement was less than 0.1 pH unit.

Phosphorylation assay.  Rod or cone membranes (15 μL) were mixed with 10 μL of the K-gluc buffer containing [γ-32P] ATP, GTP and EGTA. The final concentrations of these chemicals were 0.1 or 1 mM ATP, 0.5 mM GTP and 0.8 mM EGTA. The phosphorylation reaction was initiated by irradiating the sample by a test flash. After incubation for a desired time, the reaction was terminated by adding 150 μL of 10% (wt/vol) trichloroacetic acid. To measure the early time course of phosphorylation with good time resolution, we used a rapid-quench apparatus shown in Fig. 4. The sample was then centrifuged (20 000 g for 10 min) and the precipitated membranes were washed twice with the K-gluc buffer. The resultant membranes were subjected to SDS-PAGE. Incorporated amount of 32P into the visual pigment band was quantified by an image analyzer (BAS 2000).

Figure 4.

 (a) Schematic drawing of the rapid-quench apparatus. (b) The color changes of rhodopsin is the result of denaturation after adding TCA. Protein denaturation was found to be complete 50–70 ms after the addition of TCA. (c) Phosphorylation time course in rod and cone membranes. (Inset) Normalized carp red-sensitive cone photoresponses (n = 4) elicited by a light flash of which intensity was similar to that used in the phosphorylation measurement. (a and c), modified from figs. 7 and 1, respectively in Ref. (11). Copyright (2005) National Academy of Sciences USA.

Expression and purification of recombinant GRK1 and GRK7 in Sf9 cells.  In carp rods and cones, a few subtypes of rod visual pigment kinase and cone pigment kinase are expressed (12). Among them, the major subtypes of carp rod kinase (GRK1A-1a) and cone kinase (GRK7-1a) were expressed in Sf9 cells and purified based on the method reported previously (15) with some modification. Briefly, the harvested cells were suspended and sonicated in a Hepes buffer (115 mM K-gluconate, 2.5 mM KCl, 2 mM MgCl2, 1 mM DTT, 10 mM Hepes, pH7.5) supplemented with 1 mM CaCl2. The lysate was centrifuged (27000 g for 20 min), and the supernatant containing the kinase was collected. The supernatant was applied to an NHS-activated Sepharose column in which nonacylated S-modulin was covalently attached to the Sepharose beads. The column was pre-equilibrated by the Hepes buffer supplemented with 1 mM CaCl2. The GRKs were retained in the column at 1 mM CaCl2 because of the Ca2+-dependent binding to S-modulin, and then the column was washed with the Hepes buffer supplemented with 1 mM CaCl2 first and then the Hepes buffer supplemented with 1 mM CaCl2 plus 500 mM NaCl. Finally, bound GRKs were eluted at low Ca2+ conditions by the Hepes buffer supplemented with 5 mM EGTA, and they were concentrated in the K-gluc buffer.

Results and discussion

Rods and cones in vertebrate retina

Rods and cones of carp are shown in Fig. 1a. Both cells consist of outer segment, inner segment, nucleus region and synaptic terminal. The name of the cell type, a rod or cone, comes from the shape of the outer segment. In the outer segment, all enzymes and channels necessary for generation of a photoresponse are contained. To increase the surface area of the membrane, which is the site to convert a photon signal to a chemical signal, ∼1000 disk membranes are packed in the outer segment of a rod, and in a cone, the cell membrane is invaginated hundreds of times. The photoreceptive proteins (called visual pigments) are one of the G protein-coupled receptor proteins (16,17), and are contained in the disk membranes in rods and in the invaginated membranes in cones. Because of the repeated stacks of membrane structure and high density of visual pigments in the membranes, the density of the visual pigment is so high to reach ∼3 mM in the outer segments (18). This assures efficient capture of incident light entering into the retina.

Figure 1.

 Characteristics of rods and cones. (a) A rod and a cone brushed off carp retina. The sketch shows the structure of each cell. OS, outer segment; IS, inner segment; N, nucleus region; T, synaptic terminal region. The arrows indicate the direction of the light incidence to the cell in the retina. (b) Families of photoresponses at different flash intensities of a rod (left) and a red-sensitive cone (right). The arrows indicate the timing of the delivery of a light flash. (c) Flash intensity-response relations of rods (circles, n = 2) and red-sensitive cones (triangles, n = 4). (b and c), modified from fig. 1 in Ref. (10). Copyright (2001) National Academy of Sciences USA.

Photoresponses of rods and cones. Fig. 1b shows a family of carp rod or red-sensitive cone photoresponses elicited by a series of light flashes of various intensities. When the intensity of a light flash increased, the amplitude and the duration of a flash response increased. The results showed that the flash response is terminated much more quickly in cones than in rods. This brief flash response in cones is one of the bases why our daylight vision shows high time-resolution. The flash intensity–response relation was obtained from the kind of results shown in Fig. 1b. It was shown that the light-sensitivity was ∼103 times lower in red-sensitive cones than in rods (Fig. 1c). The high sensitivity of rods ensures twilight vision.

Mechanism of photoresponse generation.  The molecular mechanism to generate photoresponses in vertebrate rods has been extensively investigated and is now well understood (1–3,9). The mechanism is called phototransduction cascade and is made up of relayed enzymatic reactions, where several signaling proteins are involved (Fig. 2, open arrows). The cascade is one of the G protein-linked signal transduction cascades and probably is the best understood system because the easiness of purification of rods has been a great advantage to carry out biochemical studies. The cascade is triggered by the activation of rod visual pigment, called rhodopsin, with photon absorption. Visual pigment, a photoreceptive protein, contains 11-cis-retinal as the chromophore, and the absorption of a photon induces cis-trans isomerization of the chromophore. A structural change in the chromophore from 11-cis to all-trans form then induces the structural change in the protein moiety of visual pigment. After subsequent thermal reactions, an active intermediate (called meta II intermediate) is formed (19). This active form of visual pigment then activates transducin, a visual trimeric G protein, by substitution of GTP for GDP. Activated transducin in turn activates cGMP phosphodiesterase (PDE), and as a result, cGMP in the cytoplasm is hydrolyzed, and the cGMP concentration is decreased. There are cGMP-gated cation channels in the plasma membrane of the outer segment, and the decrease in the cGMP concentration induces the closure of the channels. As a consequence, the membrane hyperpolarizes. In the phototransduction cascade, the light signal is greatly amplified at two steps. The first step is at the G protein activation. A single activated rhodopsin molecule can activate 100–200 transducin molecules per second (20). The second step is at the hydrolysis of cGMP by PDE. Each PDE hydrolyzes thousands of cGMP molecules per second. There is no amplification in the activation of PDE by transducin: one transducin molecule activates only one catalytic subunit of a PDE molecule. As a result of these amplification processes, absorption of a photon causes hydrolysis of ∼106 cGMP molecules, and a single photon can hyperpolarize the outer segment of a rod. The hyperpolarization causes reduction in the transmitter release at the synaptic terminal, and the light signal is transmitted to secondary neurons.

Figure 2.

 Schematic drawing of the phototransduction mechanism (ON reactions, indicated in open arrows; OFF reactions, indicated in filled arrows) in photoreceptor cells. A number near an arrow shows the efficiency of the reaction in carp cones relative to that in carp rods.

Mechanism of photoresponse termination.  To terminate a photoresponse, the hydrolysis of cGMP must cease, and therefore, the activated enzymes in the phototransduction cascade (visual pigment, transducin and PDE) must be inactivated (Fig. 2, filled arrows). Activated visual pigments are inactivated by a phosphorylation reaction (21). In rods, visual pigment rhodopsin is phosphorylated by rhodopsin kinase (GRK1), one of the G protein-coupled receptor kinases (GRKs) (22). The phosphorylation itself reduces the catalytic activity of activated rhodopsin (23), but subsequent binding of arrestin to the phosphorylated rhodopsin completely shuts off the activity of rhodopsin (24). It is well known that this rhodopsin inactivation by phosphorylation and subsequent binding of arrestin are the essential mechanisms to terminate the hydrolysis of cGMP (25). To terminate a photoresponse completely, in addition to inactivation of the enzymes responsible for hydrolysis of cGMP, restoration of cGMP is necessary. Retinal guanylate cyclase is responsible for this restoration.

Phototransduction mechanism in cones.  The phototransduction cascade and the mechanisms of response termination are well investigated in rods, and are now regarded as a model system of G-protein-coupled receptor signaling. On the other hand, little is known about cones. As known, rods can be highly purified from the retina in large quantities to carry out biochemistry. However, cones are much less abundant than rods in the retina in general, and the method of purification of cones had not been established. Thus, biochemical approaches were limited in the study on the phototransduction mechanism of cones. Because of the progress in molecular biology, however, it has been known that there are rod and cone versions of visual pigment, transducin, PDE, cGMP-gated channel and GRK, for example (Fig. 2). For this reason, it is thought that the phototransduction mechanisms are similar in rods and cones. However, because the sensitivity and the photoresponse time course are different between rods and cones, the phototransduction reactions in cones must be different quantitatively from those in rods.

In some of the phototransduction proteins, quantitative differences between the rod and the cone version have been studied. For example, it has been shown that the lifetime of the active intermediate of visual pigment (meta II intermediate) is shorter in cones than in rods (19). However, it has also been shown that a cone pigment expressed in a rod does not affect the photoresponse of this rod (26). This result indicates that the measurement of the activity of certain cone-type enzymes is not enough to understand the molecular mechanisms that characterize cone photoresponses. It is important to measure the efficiencies of the reactions in the cone phototransduction cascade, and compare the result with those of the corresponding reactions in rods. To perform this, we tried to obtain purified cones in large quantity to carry out biochemical measurements. In addition, rods and cones were obtained from the same animal species so that direct comparisons were possible.

Purification of rods and cones from carp retina

For the purification of rods and cones, we used the retina of carp (Cyprinus carpio). During the course of previous studies, we realized that cones sediment more quickly than rods in the suspension of carp photoreceptor cells detached from the retina. This fact suggested that the buoyant density of a cone is higher than that of a rod. Based on this finding, we tried to separate cones from rods by using a stepwise density gradient method as shown in Fig. 3 (10).

Rods and cones were detached from the retina into a Ringer’s solution by tapping the surface of the photoreceptor side of a retina with a paintbrush. Rods and cones were then separated by a stepwise Percoll gradient. At the light interface (45/60% (w/v) interface, Fig. 3b), rods were trapped, and at the heaviest interface (75/90% (w/v) interface, Fig. 3c), cones were sedimented. From each interface, rods or cones were collected. Most of the purified rods and many of the purified cones had outer and inner segments. Some of the cones lacked outer segments, although they were present at the heaviest interface concomitantly along with cones having both outer and inner segments. This fact indicated that the inner segment of a cone is heavier than that of a rod. In purified rods, there was little contamination. Purified cones were contaminated with red blood cells in our earlier study (10). However, in our later study, it was found that these contaminated cells can be removed by washing the sample with the K-gluc buffer (11). In carp retina, red-, green-, blue-, and UV-sensitive cones are expressed. As our purified cone preparation contained red-, green-, and blue-sensitive cone pigments with a ratio of 3: ∼1: ∼1 (10), our measured results shown below were mainly those of red cones.

Comparison of the reaction efficiencies in the phototransduction cascade between rods and cones

By using purified rods and cones, we compared two reactions in the phototransduction cascade: (1) transducin activation by visual pigment, and (2) PDE activation through transducin activation by visual pigment (10).

Transducin activation in rods and cones was measured by quantifying the amount of GTPγS bound to activated transducin. The result showed that transducin is activated by an activated visual pigment at the rate of ∼50 molecules/s in rods. In the case of cones, the rate was less than 2 molecules/s. The result showed that the activation of transducin is about 25 times less effective in cones (Fig. 2). This means that the light signal is not amplified so much in cones, while the signal is strongly amplified in rods at this stage. As stated above, our measured result was mostly that of red cones. However, based on the content of three types of the pigment (red: green; blue = 3: ∼1: ∼1), the contribution of green and blue pigment was not negligible. This fact, in turn, implies that the transducin activation by green-pigment and the activation by blue pigment are both significantly less effective than that by rod pigment.

PDE activation per activated visual pigment was found to be about 250 times less efficient in cones (not shown), which is partly because the transducin activation is about 25 times less effective in cones. The result therefore indicated that PDE activation by an activated transducin molecule is about 10 times less effective in cones (Fig. 2). The low efficiencies in the transducin activation and PDE activation in cones cause the low signal amplification and therefore explain low sensitivity of a cone. It is shown that rod visual pigment evolved from cone visual pigment (27), and it is thought that rods evolved from cones. The results suggested that the efficiencies of transducin activation and PDE activation, and thus the signal amplification, increased during the evolution of rods to acquire high sensitivity.

Visual pigment inactivation in rods and cones

Comparison of the efficiencies of visual pigment phosphorylation between rods and cones.  Recently, in cones, we measured the time course of visual pigment phosphorylation, a quenching mechanism of light-activated visual pigment. However, the time course of the phosphorylation was too fast to measure with a manual operation in our previous study (10). Thus, we constructed a rapid-quench apparatus shown in Fig. 4a (11). Phosphorylation reactions were started by giving a light flash to the sample in a test tube, and were terminated by adding trichloroacetic acid (TCA). The timing of the delivery of TCA after a light flash was controlled by a computer that opens an electric valve to supply a nitrogen gas pressure (5 kg/cm2) to a syringe containing TCA. It takes 30 ms for the actual acid delivery to the sample from the opening of the valve. The delay was subtracted in our data analyses. The time necessary to quench the reaction after adding TCA was calibrated by observing color changes accompanied by denaturation of rhodopsin (Fig. 4b): TCA was added to a suspension of rod membranes at time 0 in the dark, and a photograph was taken by giving a light flash at various time intervals after adding TCA. By visual observation of the color change from purple to orange, acid denaturation of rhodopsin was found to be complete 50–70 ms after adding TCA. This delay was not subtracted in our data analyses because of relative uncertainty of the time required for the color change. Fig. 4c shows the phosphorylation time course in rod and cone membranes (11). As shown, the time course in cones is very rapid. The initial rate of the reaction in cones was 4.8 Pi incorporated into an activated visual pigment per second (Pi/R*/s) in cones and 0.09 Pi/R*/s in rods. The phosphorylation rate was 50 times faster in cones (Fig. 2). In Fig. 4c, inset shows the four carp red-sensitive cone photoresponses elicited by a light flash of similar light intensity used in the phosphorylation measurement. The photoresponses started to recover at 0.3 s after the flash, and at the time of this recovery, about two phosphates were incorporated into an activated visual pigment. It is known that two or more phosphates should be incorporated to quench an activated visual pigment (28). Thus, the result is the first demonstration that the time course of visual pigment phosphorylation coincides with the photoresponse recovery.

Molecular bases of rapid phosphorylation of cone visual pigment.  The molecular mechanisms responsible for the rapid phosphorylation in cones were further investigated (11). There were two possible mechanisms that accounted for this rapid reaction. The phosphorylation on visual pigment is an enzymatic reaction by a G protein-coupled receptor kinase (GRK). It was possible that cone visual pigment was phosphorylated more easily than rod visual pigment. It was also possible that the GRK activity in cones was much higher than that in rods. To determine which was the case, we measured the phosphorylation on rod pigment by a kinase in cone membranes (cone kinase) and the phosphorylation on cone pigment by a kinase in rod membranes (rod kinase). For this measurement, a rod membrane preparation containing intact rod pigment but lacking the rod kinase activity (rod pigment membranes) and a rod membrane preparation retaining the kinase activity but lacking the rod pigment activity (rod kinase membranes) were prepared. Cone pigment membranes and cone kinase membranes were similarly prepared. Rod pigment membranes were fused with cone kinase membranes to measure the phosphorylation on rod pigment by cone kinase. We prepared four types of fused membranes, each containing either rod or cone pigment and either rod or cone kinase. When the phosphorylation was measured in the four types of the fused membranes, it was found that the phosphorylation was rapid in the samples containing the cone kinase activity regardless of whether the visual pigment was a rod-type or a cone-type (Fig. 5). This result clearly indicated that the rapid phosphorylation in cones is the result of high kinase activity in cones. Further analyses showed that the kinase activity is potentially about 100 times higher in a unit volume of the outer segment in cones. One important point in this study was that the initial rate of phosphorylation on rod pigment was similar to that on cone pigment no matter whether the enzyme was rod kinase or cone kinase (Fig. 5). This means that the rod and cone visual pigment are equally good substrates for each kinase.

Figure 5.

 Rod and cone visual pigment phosphorylation by rod kinase or cone kinase. Rod and cone pigment membranes were mixed with either rod or cone kinase membranes, and the time course of visual pigment phosphorylation was measured. Each sample contained rod pigment and cone kinase (open circles), cone pigment and cone kinase (open triangles), rod pigment and rod kinase (closed circles), or cone pigment and rod kinase (closed triangles). The time courses of the phosphorylation in native membranes are shown by the dashed line (cone) and the dotted line (rod). Adapted from fig. 2 in Ref. (11). Copyright (2005) National Academy of Sciences USA.

Molecular bases of high kinase activity in cones.  Because the kinase activity measured in Fig. 5 was the total activity of the kinase, there were two possible mechanisms that accounted for the high kinase activity in cones. One was that the expression level of kinase was rather higher in cones than in rods. The other was that the specific activity of a single cone kinase molecule was higher than that of a rod kinase molecule. To distinguish which was the case, the expression levels of rod kinase and cone kinase were measured. There are a few subtypes of rod kinase and cone kinase in carp (12). We raised rod kinase-specific antibodies against a peptide region highly conserved in the major rod kinase subtypes and also raised cone kinase-specific antibodies with a similar strategy. Using these antibodies together with the known amounts of purified recombinant rod and cone kinases, we measured the expression levels of the total amount of rod and cone kinase with western-blot analysis. The result showed that the content of rod kinase was ∼0.004 per rod visual pigment (12 μM, assuming that the visual pigment content is 3 mM) in agreement with previous studies (29), and that of cone kinase was ∼0.04 per cone visual pigment (∼120 μM). This result indicated that the high kinase activity in cones is partially as a result of high expression level of the kinase in cones, but that 10 times higher expression level is not enough to explain the potential total kinase activity difference (100-fold). Simple calculation suggested that the specific activity of a single cone kinase molecule is estimated to be about 10 times higher than that of a single rod kinase molecule. To confirm the appropriateness of this estimation, in an independent study, we expressed and purified recombinant rod and cone kinase in Sf9 cell, and measured the specific activity of each kinase (11). The result showed that the expressed cone kinase has about 20 times higher activity (9.3 ± 3.3 Pi/kinase/s) than expressed rod kinase (0.41 ± 0.16 Pi/kinase/s). The calculated and measured specific activities of the cone kinase are the highest among the GRKs so far known. High content of a highly potent enzyme is the machinery that ensures rapid shut off of a cone photoresponse in daylight.

Complexity of expression of GRK1 and GRK 7 in rods and cones.  As described above, the high expression level of a highly active enzyme in cones explains well the rapid turn-off of a cone photoresponse. However, this mechanism may not always explain briefer photoresponses observed in cones in some of the mammals. The visual pigment kinase in rod is rhodopsin kinase or GRK1 (G protein-coupled kinase 1). GRK1 is expressed in rods of all of the animal species including carp so far examined. The kinase in carp cones is GRK7, another member of GRKs (12). The expression level and the specific activity of carp GRK7 are higher than those of carp GRK1 as shown above. In mouse, however, GRK7 is not expressed, but instead, GRK1 is expressed in cones (30). In human cones, both GRK1 and GRK7 are expressed (31). Yet, the cone photoresponses in these animals are briefer than rod responses. All these complexities of the expression pattern of GRK1 and GRK7 among animal species may indicate that (1) higher expression level of GRKs in cones than in rods, and/or (2) higher activity of the cGMP supply after light flash in cones, contributes to briefer photoresponses in cones in these mammals. Alternatively, visual pigment phosphorylation may not be the major rate-limiting step in these animals. It will be interesting to examine the activity and the expression level of the visual pigment kinase(s) in cones of these animals.

Other mechanisms required for photoresponse termination in rods and cones.  To terminate phototransduction, it is also needed to inactivate both activated transducin and activated PDE in the phototransduction cascade. We found that PDE inactivation after light activation is several times faster in cones than in rods (Fig. 6). This fact suggests that the inactivation of PDE itself and also the transducin inactivation are faster in cones than in rods. It is reported that the expression level of RGS9, a GTPase activating protein that accelerates the inactivation of transducin, is higher in cones than in rods (32). All mechanisms seem to contribute to the lower light sensitivity and the faster time course of the photoresponse in cones.

Figure 6.

 PDE activation and inactivation in rod and cone membranes. (a) cGMP hydrolysis was monitored in rod (left) and cone (right) membranes with the pH assay method as a function of time after light flashes of various intensities. An arrow indicates the timing of the test flash, and an arrowhead indicates the onset of a bright steady illumination to observe the maximum hydrolysis rate. To measure the light-induced changes in PDE activity, dark PDE activity was always subtracted. (b) First derivatives determined from the data shown in a). (c) Half-life of activated PDE as a function of the flash intensity. The PDE half-life was determined as the time required for the recovery of the PDE activity to 50% of the peak activity. Modified from Fig. 4 in Ref. (10). Copyright (2001) National Academy of Sciences USA.

Summary and conclusion

Rods and cones respond to light differently, although their molecular mechanisms are qualitatively similar. As described in this review, the efficiencies of the reactions in the phototransduction cascade were measured in both rods and cones, and it was found that they are quantitatively different.

We studied molecular bases of some of the differences, but not all of them. For example, the faster visual pigment phosphorylation in cones was explained by the high expression level of highly active kinase, but neither the molecule responsible for the lower activation efficiency of transducin nor the molecule responsible for the lower activation efficiency of PDE in cones was identified. Furthermore, the reaction of cGMP synthesis in cones was not measured. We are still in the middle of these studies. As has been shown in the reaction of visual pigment phosphorylation, further study will reveal which molecular species are responsible for the difference in the photoresponse characteristics between rods and cones.

Acknowledegements— This work was supported by grants from the Japan Society for the Promotion of Science (JSPS) to S.K. (15370068) and S.T. (06770114), Human Frontier Science Program to S.K., and Senri Life Science Foundation to ST. Y.S.-M. and D.A. are supported by JSPS Research Fellowships for Young Scientists.