Dephosphorylation during Bleach and Regeneration of Visual Pigment in Carp Rod and Cone Membranes*

Background: Light-activated visual pigment is inactivated by phosphorylation and then regenerated. In this regeneration cycle, phosphates incorporated should be removed. Results: Dephosphorylation was more effective in cones than in rods and did not show substrate preference during the cycle. Conclusion: Phosphorylated visual pigment is dephosphorylated constantly in the regeneration cycle in both rods and cones. Significance: A mechanism of visual pigment regeneration is known. On absorption of light by vertebrate visual pigment, the chromophore, 11-cis retinal, is isomerized to all-trans retinal to activate the phototransduction cascade, which leads to a hyperpolarizing light response. Activated pigment is inactivated by phosphorylation on the protein moiety, opsin. Isomerized all-trans retinal is ultimately released from opsin, and the pigment is regenerated by binding to 11-cis retinal. In this pigment regeneration cycle, the phosphates incorporated should be removed in order that the pigment regains the capability of activating the phototransduction cascade. However, it is not clear yet how pigment dephosphorylation takes place in the regeneration cycle. First in this study, we tried to estimate the dephosphorylation activity in living carp rods and cones and found that the activity, which is present mainly in the cytoplasm in both rods and cones, is three times higher in cones than in rods. Second, we examined at which stage the dephosphorylation takes place; before or after the release of all-trans retinal, during pigment regeneration, or after pigment regeneration. For this purpose we prepared three types of phosphorylated substrates in purified carp rod and cone membranes: phosphorylated bleaching intermediate, phosphorylated opsin, and phosphorylated and regenerated pigment. We also examined the effect of pigment regeneration on the dephosphorylation. The results showed that the dephosphorylation does not show substrate preference in the regeneration cycle and suggested that the dephosphorylation takes place constantly. The results also suggest that, under bright light, some of the regenerated visual pigment remains phosphorylated to reduce the light sensitivity in cones.

In the vertebrate retina there are two types of photoreceptor cells, rods and cones. In both cells visual pigment (R), 2 composed of a chromophore, 11-cis retinal, and a protein moiety, opsin, absorbs light to isomerize 11-cis retinal to all-trans retinal to induce a conformational change in opsin. Once all-trans retinal is formed, the pigment is activated to lead to activation of an enzymatic cascade known as the phototransduction cascade to evoke a hyperpolarizing light response (1,2). Activated pigment is finally bleached through bleaching intermediates and is decomposed into all-trans retinal plus opsin so that the pigment should be regenerated for rods and cones to maintain their ability to detect light. Regeneration of the pigment is attained by binding of opsin to 11-cis retinal supplied with mechanisms known as the retinoid cycle or the visual cycle (3). In this way visual pigment is recycled (visual pigment regeneration cycle). In the phototransduction cascade, the activated pigment (R*) is inactivated by phosphorylation with G proteincoupled receptor kinases and/or binding of arrestin (4,5). For the pigment to fully recover its ability to activate the phototransduction cascade, the phosphates incorporated into opsin should be removed.
It is well known that the sensitivity to light is higher in rods than in cones, and thus, rods mediate night vision, whereas cones mediate daylight vision. It is known that rods are easily saturated with weak light. However, cones are not saturated even under very bright light that can bleach potentially almost all cone pigment in Ͻ0.1 s (6), which strongly suggests that visual pigment having the capability of activating the phototransduction cascade is regenerated in cones very effectively. In fact, very effective supply of 11-cis retinal has been known to be present only in cones (7-10), but little is known about removal of the phosphates incorporated into opsin in rods or cones.
Removal of phosphates (dephosphorylation) from phosphorylated opsin has been shown in living animals (11)(12)(13)(14). Although biochemical studies have been made previously (15,16), it has not been known clearly at which stage in the regeneration cycle dephosphorylation takes place. Previous suggestions were that dephosphorylation takes place after decay of the light-activated protein (15) or after regeneration of visual pigment (13). Most of these studies were made in rods, and little is known about the dephosphorylation reaction in cones.
Although it has been suggested that the dephosphorylation is faster in cones than in rods (14), comparison was made between zebrafish cones and mouse rods in that study. In the present study, therefore, we first compared the rate of dephosphorylation between rods and cones in the same animal species using membranes and cytoplasmic proteins of purified rods and cones obtained from carp retina. Second, to examine at which stage in the regeneration cycle dephosphorylation takes place most effectively, we prepared phosphorylated substrates at various stages and measured dephosphorylation activities in these preparations; they were phosphorylated bleaching intermediates produced after a light flash (BI-P), phosphorylated opsin (Opsin-P), and phosphorylated and regenerated visual pigment (VP-P), all in carp rod and cone membranes. Furthermore, we measured dephosphorylation activities in BI-P in the presence of 11-cis retinal to examine the effect of pigment regeneration.

Experimental Procedures
Preparation of Membranes and Cytoplasmic Proteins of Rods and Cones-Carp (Cyprinus carpio) rods and cones were purified as described previously (17). Animal care was conducted according to the institutional guidelines (approval number FBS-14-005). Purified cells were washed first with Ringer's solution (119.9 mM NaCl, 2.6 mM KCl, 0.5 mM CaCl 2 , 0.5 mM MgSO 4 , 1 mM NaHCO 3 , 16 mM glucose, 0.5 mM NaH 2 PO 4 , 4 mM Hepes, pH 7.5) and then with potassium-gluconate buffer (K-gluc buffer; 115 mM potassium gluconate, 2.5 mM KCl, 2 mM MgCl 2 , 0.2 mM EGTA, 0.1 mM CaCl 2 , 1 mM DTT, 10 mM Hepes, pH 7.5), both by mild centrifugation (600 ϫ g for 12 s and then 3000 ϫ g for 4 s). The washed cells were frozen at Ϫ80°C until use. All of the manipulations were performed in complete darkness with the aid of an infrared image converter (NVR 2015, NEC).
To prepare rod and cone membranes and their cytoplasmic proteins, frozen purified rods and cones were thawed, suspended in 1 ml of K-gluc buffer, and centrifuged (100,000 ϫ g, 20 min). Then the supernatant containing cytoplasmic proteins was collected. In this report, "cytoplasmic proteins" means soluble proteins found in the cytoplasm of a cell and possibly together with those present in subcellular organelles. These proteins were obtained by freeze-thaw to minimize the solution volume. The remaining precipitate was resuspended in 1 ml of K-gluc buffer and centrifuged again to obtain the membranes and the supernatant. The membranes precipitated were suspended in K-gluc buffer and used as rod or cone membranes. Visual pigments in these membranes were quantified spectrophotometrically. Our purified cone membranes contain all of the cone pigment types but mainly red-sensitive pigment (molar ratio of the pigments, red:green:blue:UV ϭ 3:1:1:negligible) (18). The supernatant was combined with that obtained in the previous centrifugation, and the resultant cytoplasmic proteins were concentrated using Vivaspin 20 (M r 10,000 cutoff, GE Healthcare). We confirmed that our cytoplasmic proteins contain Ͼϳ80% of soluble proteins that can be obtained with repetitive freeze-thaws. The amount of the cytoplasmic proteins was quantified by Bradford method using bovine serum albumin (BSA) as a molar standard. In some of the studies in cone membranes (Figs. 1, 2 and 5), we used very concen-trated cytoplasmic proteins (Ͼ20 mg/ml). This was because the content of cytoplasmic proteins per single cells was much larger in cones than in rods (see "Results and Discussion"). In other studies (Figs. 3 and 4), we used cone cytoplasmic proteins at a low concentration (2 mg/ml), which was similar to that of rod cytoplasmic proteins used in the measurement in rod membranes. The effect of the cytoplasmic proteins varied slightly depending on the batches prepared.
Quantification of Visual Pigments in Rod and Cone Membranes-Concentrations of visual pigments in rod and cone membranes were determined spectrophotometrically. Carp rod and cone pigments contain A 2 -type 11-cis retinal. To quantify the A 2 -type visual pigment in carp, we obtained A 1 -type rod pigment and A 2 -type rod pigment, both regenerated from opsin in rod membranes. Assuming that the molar extinction coefficient at the absorption maximum (⑀ max ) of A 1 -type rod pigment is 40,000 and that the same amounts of rod pigment are regenerate by the addition of an excess amount of A 1 -type or A 2 -type 11-cis retinal, we found that the ⑀ max of A 2 -type rod pigment is 32,000, in agreement with the value reported previously (19). The ⑀ max of A 2 -type red-sensitive cone pigment was assumed to be 38,400 based on a previous study that the ⑀ max of red-sensitive cone pigment is 1.2 times higher than that of rod pigment (20). Pigment concentrations in carp rod and cone membranes were determined based on these ⑀ max values.
Preparation of Photoreceptor Membranes Containing Opsin-P and VP-P-To prepare rod membranes containing Opsin-P or VP-P, rod membranes in K-gluc buffer containing 0.5 M rhodopsin, 1 mM [␥-32 P]ATP, 0.5 mM GTP, and 0.6 mM EGTA in K-gluc buffer was exposed to yellow light (Ͼ410 nm) from a 150 watt tungsten/halogen lamp for 10 min at a distance of 8 cm at room temperature to phosphorylate bleaching intermediates of rhodopsin in the membranes. Then, to bleach rhodopsin fully and to remove isomerized all-trans retinal completely, the membranes were illuminated for another 10 min with the same illumination in the presence of 10 mM NH 2 OH. We confirmed that rhodopsin was fully bleached with these illuminations. Then the membranes were washed once with K-gluc buffer containing 1% (w/v) BSA and 50 mM NH 2 OH to remove unreacted [␥-32 P]ATP and all-trans retinal, with K-gluc buffer containing 1% BSA, and finally with K-gluc buffer twice, all by centrifugation at 100,000 ϫ g for 10 min, and the membranes were resuspended in K-gluc buffer. These membranes containing phosphorylated-and all-trans retinal-free opsin were used as the source of Opsin-P and VP-P (see below).
To obtain the membranes containing VP-P, a 10ϫ molar excess of 11-cis retinal (A 1 -type) dissolved in ethanol was added to the membranes containing phosphorylated-and all-trans retinal-free opsin. After the membranes were incubated for 30 min at 20°C in the dark, the membranes were washed once by K-gluc buffer containing 1% BSA and 50 mM NH 2 OH to remove excess 11-cis retinal and successively washed by K-gluc buffer containing 1% BSA once. The membranes were additionally washed twice with K-gluc buffer to replace the buffer solution and were used as the membranes containing VP-P. As a result, 102 Ϯ 16% (n ϭ 3, mean Ϯ S.D. throughout this study) of rhodopsin was regenerated. We prepared rod membranes contain-ing Opsin-P in a similar way as we prepared VP-P; instead of 11-cis retinal, all-trans retinal was added to the membranes containing phosphorylated-and all-trans retinal-free opsin. Prepared Opsin-P and VP-P were used immediately after their preparation. It took ϳ3.5 h to prepare Opsin-P and VP-P in rod membranes, and the phosphorylation levels were similar in these preparations at the beginning of the measurement of phosphorylation levels or of dephosphorylation: 2.8 Ϯ 0.2 phosphates were present per Opsin-P (n ϭ 3) and 2.2 Ϯ 0.1 phosphates per VP-P (n ϭ 12). The maximum level of the phosphorylation in our native rod membranes was three per R* (17), so that most of the phosphates incorporated into the pigment were retained in Opsin-P and VP-P in rod membranes.
Cone membranes containing Opsin-P or VP-P were prepared similarly as rod membranes but with some modifications. To phosphorylate cone pigments, cone membranes were illuminated with yellow light (Ͼ410 nm) from the 150-watt tungsten/halogen lamp for 5 min at room temperature. To bleach cone pigments further and to remove isomerized all-trans retinal, the membranes were illuminated with the yellow light for another 1 min in the presence of 100 M NH 2 OH (NH 2 OH concentration was reduced compared with that used for rod membranes because cone pigments are slightly sensitive to NH 2 OH). The bleach level of the pigment was 94.5 Ϯ 3.2% (n ϭ 3). Regeneration level of the pigments by the addition of 11-cis retinal was 101 Ϯ 13% (n ϭ 3). It took ϳ2.5 h to prepare Opsin-P and VP-P in cone membranes, and the phosphorylation levels were similar in Opsin-P and VP-P: 2.5 Ϯ 0.1 phosphates were present per Opsin-P (n ϭ 3) and 2.5 Ϯ 0.1 phosphates per VP-P (n ϭ 12). Because the maximum phosphorylation level was three phosphates per R* in native cone membranes (17), most of the phosphates were retained in Opsin-P and VP-P also in cone membranes.
Measurements of Phosphorylation Levels in BI-P, Opsin-P, and VP-P-We measured dephosphorylation activities at three different stages of the pigment regeneration cycle: BI-P, Opsin-P, and VP-P. In the measurement of dephosphorylation activity in BI-P, time course of phosphorylation levels were measured in rod or cone membranes after a light flash. These membranes in K-gluc buffer containing 0.5 M visual pigment, 1 mM [␥-32 P]ATP, 0.5 mM GTP, and 0.6 mM EGTA were preincubated for 30 s in the dark and illuminated with a light flash bleaching 65% of rod pigment or 80% of cone pigments. After incubation for a desired time, a 20-l aliquot was withdrawn and mixed with 100 l of quenching buffer (200 mM KCl, 100 mM NaF, 5 mM adenosine, 200 mM EDTA in 250 mM sodium phosphate buffer, pH 7.2) based on a method reported previously (11). After centrifugation (20,000 ϫ g for 15 min), the precipitate was washed twice with K-gluc buffer and suspended in SDS sample buffer (62.5 mM Tris-HCl, 2.3% (w/v) SDS, 10% (w/v) glycerol, 0.0025% bromphenol blue, 1.25% (w/v) 2-mercaptoethanol, pH 6.8). After SDS-PAGE, the amount of 32 P present in the band of visual pigment was quantified using an image analyzer (BAS 2500, Fuji Film). In both rod and cone membranes, the amount of visual pigment was quantified by Coomassie Brilliant Blue staining after SDS-PAGE using BSA as a molar standard. The band of cone pigment in our SDS-PAGE was too broad and blurred in our Coomassie Brilliant Blue staining at the quantity of cone membranes in an aliquot withdrawn. In our purified cones, F 1 ATPase, which is a membrane-bound protein in the inner segment membranes, is abundantly present and could be a marker protein of the quantity of cone cells. For this reason the ratio of the amount of F 1 ATPase ␤ subunit to that of cone pigment was determined each time when cone membranes were prepared, and the amount of cone pigments in an aliquot withdrawn was estimated from the amount of F 1 ATPase ␤ subunit in the same aliquot with Coomassie Brilliant Blue staining after SDS-PAGE. When the effect of the cytoplasmic proteins was examined, the proteins were added to rod or cone membranes at the desired concentrations. Phosphorylation levels in Opsin-P and VP-P were measured similarly as in BI-P.
It should be mentioned that we measured phosphorylation levels in BI-P in the dark after a light flash. In this case, probably phosphorylation and dephosphorylation were both taking place because ATP was present during incubation and because R* should be present for a certain period of time after a light flash. Our attempt to wash out free [␥-32 P]ATP by centrifugation to terminate phosphorylation and to measure only dephosphorylation was not effective, because the measurement of early time course of phosphorylation levels was not possible. In contrast, in the measurement of phosphorylation levels in Opsin-P and VP-P, only dephosphorylation was measured. This is because ATP was not added during incubation in these measurements. It has been shown that the dephosphorylation time course is dependent on how many phosphates are incorporated and from which phosphorylation sites a phosphate is removed (13), which means that the dephosphorylation time course is complex when the pigment is multiply phosphorylated. In our study we prepared BI-P, Opsin-P, and VP-P by bleaching visual pigment at least 65%, and under these conditions these substrates are all multiply phosphorylated. For this reason, dephosphorylation time course would be composed of multiple exponential functions (see below).
When the phosphorylation level in BI-P, Opsin-P, or VP-P decreased during incubation, this decrease was fitted to an exponential function to determine the rate of dephosphorylation. However, a single exponent did not give good fitting, which is probably because of the reason mentioned above. For a practical reason, therefore, we used a double exponential function as a measure of a dephosphorylation reaction.
where t is the time after the start of the measurement, P(t) is the number of phosphates bound per R*(P i /R*), per opsin (P i /opsin), or per pigment present (P i /R) at time t, P 1 and P 2 are practical initial numbers of phosphates expressed in the unit of P i /R*, P i /opsin, or P i /R in the fast component and the slow component, respectively, and k 1 and k 2 are the rate constants of the fast and the slow component, respectively. In the actual determination of the rate of dephosphorylation, we obtained the initial rate of dephosphorylation by calculating P 1 k 1 ϩ P 2 k 2 after fitting. In this study the rate of dephosphorylation shown is the initial rate of dephosphorylation obtained in this way. When the rates are compared under different conditions such as the OCTOBER 2, 2015 • VOLUME 290 • NUMBER 40 measurements in the presence and absence of the cytoplasmic proteins, P 1 and P 2 were fitted so as to be the same under different conditions. Dephosphorylation activities are slightly different among the batches of rod or cone membranes prepared. It could be because elution of phosphatase(s) in membranes differs in each preparation. As stated above, dephosphorylation activities induced by the cytoplasmic proteins, too, are slightly different in different batches. For these reasons we used same batches of membranes and cytoplasmic proteins in a single set of study.

Dephosphorylation of Visual Pigment in Rods and Cones
Cytoplasmic proteins obtained in this study contain proteins that possibly affect phosphorylation and/or dephosphorylation of the pigment. Among them, S-modulin (or recoverin) could inhibit pigment phosphorylation. However, the Ca 2ϩ concentration in K-gluc buffer we used is low, so that inhibition by S-modulin would be negligible (21). Phosphorylation of R* is inhibited by the GTP-free form of transducin, but it is not inhibited in the presence of GTP␥S, an analogue of GTP (22). To minimize the effect of transducin, therefore, we added GTP to rod or cone membranes when the effect of transducin is likely present. There is an additional reason that probably allows us to exclude the possibility of the contribution of transducin in the effect of the cytoplasmic proteins. First, transducin is present in our rod and cone membranes, not in the cytoplasmic proteins. It is because these membranes were prepared in the dark and in the absence of GTP. Second, a regulator of transducin, RGS9, is also present in the membranes (23). As we described above, transducin may minimally contribute to the inhibition of phosphorylation in our study, but its effect is always present no matter whether the cytoplasmic proteins are added or not. Therefore, the effect observed in the presence of the cytoplasmic proteins is surely caused by some proteins, but it is not by transducin. Arrestin is one of the proteins in the cytoplasmic proteins and could inhibit dephosphorylation (24) by binding firmly to phosphorylated pigment. This possibility is discussed under "Results and Discussion." Measurements of Phosphorylation Levels in the Presence of Pigment Regeneration in Rod and Cone Membranes-Dephosphorylation activities were also measured in the presence of pigment regeneration. Rod and cone membranes were illuminated by a light flash of Ͼ410-nm light similarly as in the dephosphorylation measurement in BI-P, but to examine the effect of pigment regeneration the study was made in the presence of 11-cis retinal. Under this condition, rod pigment in rod membranes or cone pigments in cone membranes were bleached, but 11-cis retinal is not isomerized to other isomers with this light. Then we could expect pigment regeneration in the dark after the light flash in both rod and cone membranes. Phosphorylation levels were measures as in BI-P.
Measurement of Pigment Regeneration-To compare the time course of dephosphorylation and that of pigment regeneration, we measured time course of pigment regeneration in rod and cone membranes spectrophotometrically under the conditions similar to that used for the measurement of dephosphorylation activities. In this pigment regeneration measurement, NH 2 OH was not added, so that late bleaching intermediates such as metarhodopsin III could affect the spectrophotometrical quantification of the regenerated pigment. It is not known whether such intermediates are present for a relatively long time after a light flash in the presence of 11-cis retinal. To minimize the contribution of metarhodopsin III ( max , ϳ450 nm) and that of N-retinylidene-phosphatidylethanolamine ( max , ϳ450 nm), if any, in the case of rod pigment, regenerated rod pigment ( max , 500 nm) was quantified at 550 nm using a rapid scanning spectrophotometer, USB4000 equipped with DT-MINI-2-GS light source (Ocean Optics). The data of the spectrum in the visible region was collected at the desired time.
In the case of cone pigments, regeneration of cone pigment is very fast, so that the spectrum was recorded every 250 ms. To minimize the effect of metarhodopsin III and that of N-retinylidene-phosphatidylethanolamine and also for a technical reason that only the red-sensitive pigment can be quantitated in a mixture of various types of cone pigments, regenerated red-sensitive cone pigment was quantified at 570 nm in cone membranes. The bleach of the pigment by measuring light was Ͻ0.3% during the measurements of rod and cone pigment regeneration.

Results and Discussion
Dephosphorylation in BI-P in Rod and Cone Membranes-To compare the rate of dephosphorylation between rods and cones, we first measured time-dependent changes in pigment phosphorylation levels in the presence of ATP after a light flash in rod and cone membranes. In rod membranes, the phosphorylation level of light-activated visual pigment continuously increased (Fig. 1A, open circles), and apparent dephosphoryla- tion did not take place up to 250 min after a light flash. Although the phosphorylation level is determined by the balance between phosphorylation and dephosphorylation, the result seemed to suggest that the dephosphorylation activity in carp rod membranes is low. Because phosphatase activity on phosphorylated pigment has been reported to localize in the cytoplasmic fraction in bovine rod outer segments (15,25,26), we next measured the dephosphorylation reactions in the presence of the cytoplasmic proteins prepared from purified carp rods. In the presence of these proteins (4 mg/ml), the pigment phosphorylation level in rod membranes started to decrease at 10 min and half-decreased at ϳ150 min after the flash (ϩcyto, filled circles in Fig. 1A), which indicated that the phosphatase activity is present in the cytoplasm in rods in agreement with previous studies. One noticeable point was that visual pigment phosphorylation was facilitated in the presence of the cytoplasmic proteins (compare the initial increases in the phosphorylation levels between the data shown by open and filled circles). However, we did not study this issue further. Fig. 1B shows dephosphorylation in cone membranes. In cone membranes, in the absence of the cytoplasmic proteins, dephosphorylation was detected at 2 min after a light flash (open triangles in Fig. 1B) and half-decreased at ϳ50 min, both much earlier than those in rod membranes. The result, therefore, indicated that phosphatase activity in cone membranes is higher than that in rod membranes in the measurement in Fig.  1. When the cytoplasmic proteins prepared from purified cones were added (ϩcyto, filled triangles in Fig. 1B) at a concentration of 39 mg/ml, dephosphorylation was speeded up. Interestingly, phosphorylation of the pigment in cone membranes was not facilitated by the cytoplasmic proteins, in contrast to their effect in rod membranes. Instead, pigment phosphorylation was inhibited in the presence of the cytoplasmic proteins (compare the initial increases in phosphorylation levels between the data shown by open and filled triangles in the inset). It could be due to very rapid dephosphorylation, which we could not measure in the present study. Although the effect of the cytoplasmic proteins on pigment phosphorylation is interesting, we did not study this issue further and concentrated to the study on dephosphorylation in phosphorylated pigment.
The rate of dephosphorylation in bleached intermediate (BI-P) in the study shown in Fig. 1 was difficult to determine correctly. This was because phosphorylation in addition to dephosphorylation was taking place, and relative contribution of each of the reactions was difficult to determine. However, under the assumption that the phosphorylation can be neglected at times after the peak of phosphorylation (Ͼ10 min in rod membranes and Ͼ5 s in cone membranes), the initial rates of dephosphorylation in the presence of the cytoplasmic proteins (4 mg/ml in rod membranes and 39 mg/ml in cone membranes) were 0.019 Ϯ 0.009 P i dissociated per R* per min (P i /R*/min) (n ϭ 3) in rod membranes and 0.42 Ϯ 0.16 P i /R*/ min (n ϭ 3) in cone membranes. These values are the lower limits of the dephosphorylation activities in rod and cone membranes under the conditions employed.
The increase in dephosphorylation activities in the presence of the cytoplasmic proteins is probably due to the presence of phosphatase(s) in these proteins, not the factors that increase the phosphatase activity. It is because even after treatment of rod or cone membranes with 5 M urea to denature the intrinsic phosphatase activities completely, the addition of the cytoplasmic proteins increased the dephosphorylation activity.
Estimation of Dephosphorylation Activities in Opsin-P in Living Rods and Cones-As shown above, cytoplasmic proteins prepared from our purified rods and cones contained the activity of dephosphorylation of BI-P, which is probably attained by phosphatases in the outer segment. We tried to estimate this activity in living cells. For this purpose, we determined the concentration of the cytoplasmic proteins containing outer segment proteins at a concentration similar to that in living cells (CCP/OSP). For this purpose we first quantified the amount of cytoplasmic proteins present in known numbers of rod or cone cells and obtained the content of cytoplasmic proteins in a rod and a cone as an average: they were 7.2 Ϯ 0.5 pg/rod (n ϭ 3) and 188 Ϯ 59 pg/cone (n ϭ 3). Our purified rods and cones consist of the outer segment and the inner segment and are devoid of nucleus and terminal regions (for example, see Fig. 2

.3 in Ref. 1).
The higher protein content in a cone would be due to a much larger size of the cone inner segment than the rod inner segment in carp because the inner segment contains mitochondria, which probably contain a large amount of soluble proteins.
To obtain estimates of the CCP/OSP, we assumed that the cytoplasmic volume in the outer segment of a rod and a cone is half of the outer segment envelope volume (27), which is the volume determined from the outer dimension of an outer segment. In case responsible proteins for dephosphorylation are localized only in the outer segment, the CCP/OSP can be calculated by dividing the above protein content in a single rod or cone (7.2 pg/rod and 188 pg/cone) by half of the outer segment envelope volume. In this case, based on our volume measurements from the outer dimension of the outer segment of a rod and a cone (Table 1), we obtained the value of 390 Ϯ 30 mg/ml for rod cytoplasmic proteins and 6.5 Ϯ 2.0 g/ml for cone cytoplasmic proteins as the CCP/OSPs.
In the case where responsible proteins for dephosphorylation distribute both in the outer segment and the inner segment evenly, we need to consider the cytoplasmic volume of the inner segment. When we assume that the cytoplasmic volume of the inner segment is a half of the envelope volume as in the outer segment, based on our estimation of the inner segment volume (Table 1), we obtained the CCP/OSP of 297 Ϯ 105 mg/ml in a rod and 959 Ϯ 316 mg/ml in a cone. We also estimated the CCP/OSP in the case where the cytoplasmic volume in the inner segment is 9 ⁄ 10 of the inner segment envelope volume. In this case, the concentrations were 250 Ϯ 20 mg/ml in rods and 570 Ϯ 180 mg/ml in cones. Although it is difficult to estimate the CCP/OSP correctly, we believe that our above estimation gives us the range of the CCP/OSP that enables us to consider the in vivo effect of the cytoplasmic proteins on dephosphorylation: 250 -390 mg/ml in rod membranes and 570 mg/ml to 6.5 g/ml in cone membranes.
To evaluate the phosphatase activity in living rods and cones, dephosphorylation time course was measured in membranes containing Opsin-P in the presence of the cytoplasmic proteins of various concentrations in rod ( Fig. 2A) and cone (Fig. 2B) membranes. In these sets of measurements, dephosphorylation activities in the absence of the cytoplasmic proteins in rod membranes (0 mg/ml in Fig. 2A) and in cone membranes (0 mg/ml in Fig. 2B) were similar, which is probably because phosphatase activity in cone membranes was eluted significantly in this study. From curve fitting to the double exponential function, the values of the initial rate of dephosphorylation were determined (see "Experimental Procedures"). In Fig. 2C, each initial rate in rod (open circles) or cone membranes (filled triangles) was plotted against the concentration of the cytoplasmic proteins, and the results were fitted with the Michaelis-Menten equation. The estimated K m and V max in rod membranes were 2.7 mg/ml of the rod cytoplasmic proteins and 0.16 P i dissociated from opsin per min (P i /opsin/min), respectively. In cone membranes they were 17 mg/ml and 0.47 P i /opsin/min. Our estimates of the CCP/OSP, the concentration of the cytoplasmic proteins to exert their in vivo effects, were 250 -390 mg/ml in rod membranes and 570 mg/ml to 6.5 g/ml in cone membranes (see above) and were Ͼ100 and Ͼ35 times higher than the K m values in rod and cone membranes, respectively. For this reason, it is probable that dephosphorylation reaction exerts its maximum activity (V max ; 0.16 P i /opsin/ min in rod membranes and 0.47 P i /opsin/min in cone membranes) in living rods and cones. Half-times of dephosphorylation expected from the V max values of the initial rate of dephosphorylation were 4.3 min in rod membranes and 1.5 min in cone membranes. However, the actual half-times of dephosphorylation we obtained in Fig. 2 were much longer than these, probably because Opsin-P was multiply phosphorylated. The actual half-times in rod membranes (ϳ64 min at 50 mg/ml cytoplasmic proteins in Fig. 2A) and in cone membranes (ϳ13.5 min at 150 mg/ml cytoplasmic proteins in Fig. 2B) are broadly similar to those observed in living animals, ϳ50 min in mouse rods (12,13) and 4 min in zebrafish (14).
In Fig. 1 we measured dephosphorylation after giving a light flash in BI-P and estimated the lower limit of the phosphorylation rate in the presence of the cytoplasmic proteins. In Fig. 2, we measured dephosphorylation in Opsin-P in the presence of the cytoplasmic proteins. At a similar concentration of the added cytoplasmic proteins, we can compare the rate of dephosphorylation at two different stages in the visual pigment regeneration cycle. In rod membranes, the initial dephosphorylation rates were 0.019 Ϯ 0.009 P i /R*/min minimally in BI-P (4 mg/ml cytoplasmic proteins in Fig. 1A) and 0.052 Ϯ 0.016 P i /opsin/ min (n ϭ 3) in Opsin-P (2 mg/ml cytoplasmic proteins in Fig.  2A). In cone membranes, they were 0.42 Ϯ 0.16 P i /R*/min in BI-P (39 mg/ml in Fig. 1B) and 0.37 Ϯ 0.04 P i /opsin/min (n ϭ 3) in Opsin-P (50 mg/ml in Fig. 2B). The rates are not so different between BI-P and Opsin-P in either rod or cone membranes. The result, therefore, suggests that the dephosphorylation rate is not so different between BI-P and Opsin-P.
In a separate study we measured the initial rate of dephosphorylation activity in rod membranes with rod cytoplasmic proteins and also the rate in rod membranes with the cone cytoplasmic proteins, both at 2 mg/ml. The initial rate was higher in the presence of the cone proteins than the rod proteins (0.056 Ϯ 0.006 P i /opsin/min with cone proteins and 0.035 Ϯ 0.003 P i /opsin/min with rod proteins, n ϭ 3). Taking the difference in the K m values into consideration (2.7 mg/ml for the rod proteins and 17 mg/ml for the cone proteins), the result suggests that phosphatase(s), not Opsin-P, is responsible for the higher dephosphorylation activity in cones.
Measurements of Dephosphorylation in VP-P-Dephosphorylation was measured in VP-P in the dark and compared to that in Opsin-P in both rod (Fig. 3, A and B) and cone membranes  (n ϭ 3). C, dose-dependent effect of the cytoplasmic proteins on the initial dephosphorylation rate constant (see "Experimental Procedures") in rod and cone membranes. Initial rates of dephosphorylation (P i /opsin) obtained in A and B are plotted against the protein concentration of the cytoplasmic proteins in rod (open circles) and cone (filled triangles) membranes. Each data point is shown as the mean Ϯ S.D. (n ϭ 3), except that obtained at 150 mg/ml protein concentration in cone membranes; it was examined only one time because of limitation of availability of cone material. The results were fitted to the Michaelis-Menten equation for rod (dotted curve) and cone (solid curve) membranes. The fitted K m and V max values in rod membranes were 2.7 mg/ml and 0.16 P i released from an opsin molecule per min (P i /opsin/min), respectively, and those in cone membranes were 17 mg/ml and 0.47 P i /opsin/min. Data at 0.4 and 10 mg/ml in rod membranes and those at 0.4 mg/ml in cone membranes are not shown in A and B, respectively, to avoid crowding. (Fig. 3, D and E) in the absence (Fig. 3, A and D) and presence (Fig. 3, B and E) of the cytoplasmic proteins (2 mg/ml). As seen in Fig. 3, A, B, D, and E, dephosphorylation was very similar in Opsin-P and VP-P in both rod and cone membranes. The initial rates of dephosphorylation are summarized in Fig. 3, C and F, for rod and cone membranes, respectively. Similarly as in BI-P, in both Opsin-P and VP-P, the dephosphorylation rate was higher in the presence of the cytoplasmic proteins (Fig. 3, B and  E). Because the concentration of the cytoplasmic proteins in cone membranes was much lower in the study in Fig. 3D (2 mg/ml) than that used in Fig. 1 (39 mg/ml), the rate of dephosphorylation in the presence of the cytoplasmic proteins in cone membranes (0.03-0.05 P i /opsin or R/min in Fig. 3F) was much lower than that we showed in Fig. 1 (0.42 P i /R*/min). In Figs. 1  and 3, concentrations of the cytoplasmic proteins in rod membranes were similar (4 mg/ml in Fig. 1 and 2 mg/ml in Fig. 3), so that the rates were similar (0.019 P i /R*/min in Fig. 1A and 0.02-0.03 P i /opsin or R/min in Fig. 3C).
Effect of Bleach on Dephosphorylation in VP-P-The effect of bleach on dephosphorylation in VP-P (open symbols in Fig. 4, A  and B and in D and E) was measured after giving a light flash in the absence of ATP in rod membranes (Fig. 4, A and B) and in cone membranes (Fig. 4, D and E) and in the absence (Fig. 4, A and D) and in the presence (2 mg/ml in both rod and cone membranes; Fig. 4, B and E) of cytoplasmic proteins. Their dephosphorylation time courses were compared with those obtained in VP-P kept in the dark (filled symbols in Fig. 4, A and B and in D and E) and fitted to a double exponential function to obtain the initial rates of dephosphorylation. The summarized result (Fig. 4, C and F) showed that there were no significant differences between dephosphorylation rates in VP-P kept in the dark and that bleached by a light flash in both the absence and presence of the cytoplasmic proteins (compare black and white bars in Fig. 4, C and F). The result showed that dephosphorylation in VP-P is not affected by light and, therefore, activation of the pigment.
Effect of Pigment Regeneration on Dephosphorylation in Rod and Cone Membranes-Up to here, we examined dephosphorylation in BI-P (Fig. 1), Opsin-P (Figs. 2 and 3), and VP-P (Figs.  3 and 4). In living cells, bleached pigment is regenerated with binding to 11-cis retinal supplied through a conventional retinoid cycle to rods and cones and through a cone-specific retinoid cycle to cones (3). It could be the case that pigment regeneration couples with the dephosphorylation reaction. To test this possibility, we examined the effect of pigment regeneration on a dephosphorylation reaction in BI-P. First, we examined the time course of pigment regeneration in rod and cone membranes. Fig. 5, A and D, shows the regeneration in rod and cone membranes in the presence of an excess amount of 11-cis retinal. For a technical reason, we measured the regeneration of red-sensitive cone pigment in cone membranes (see "Experimental Procedures"). Rod pigment was half-regenerated at ϳ120 s after a light flash (Fig. 5A), and red-sensitive cone pigment was half-regenerated at ϳ8 s after the flash (Fig. 5D).
Dephosphorylation was measured in the presence of pigment regeneration (ϩ11cRAL) in rod membranes (Fig. 5, B and C) and in cone membranes (Fig. 5, E and F) in the absence (Fig.  5, B and E) and presence of the cytoplasmic proteins (5 mg proteins/ml in rod membranes in Fig. 5C and 20 mg proteins/ml in cone membranes in Fig. 5F). As controls, we added all-trans retinal (ϩatRAL) instead of 11-cis retinal. Although pigment regeneration seems to inhibit phosphorylation in rods (most effectively shown in Fig. 5B), the addition of 11-cis retinal (filled symbols) and, therefore, pigment regeneration did not affect the dephosphorylation significantly in the absence (Fig.  5B) and presence (Fig. 5C) of the cytoplasmic proteins in rod membranes. Similarly, pigment regeneration did not affect the dephosphorylation in cone membranes (Fig. 5, E and F).
Effects of the Cytoplasmic Proteins on Dephosphorylation-In this study we found that the phosphatase activity on phosphorylated visual pigment is low in rod and cone membranes alone, and the addition of the cytoplasmic proteins extracted from purified rods and cones increased the activity (Fig. 1). The increase in the activity should be due to the presence of the phosphatase activity, not the presence of activating factors (see above). We added all of the cytoplasmic proteins in the outer segment and the inner segment. So far, it is not known whether phosphatases probably present in the rod and cone inner segment can dephosphorylate BI-P, Opsin-P, or VP-P, and this study did not add any information on this point. If phosphatases in the inner segment show activities on phosphorylated pigment, phosphatase activities in the outer segment of a rod and a cone are lower than those estimated in Fig. 2.
It has been reported that protein phosphatase 2A is present in bovine rod outer segments (15). Our control study showed that 20 M okadaic acid, an inhibitor of the concentration that suppresses protein phosphatase 1 (PP1), PP2A, PP2B, and PP5 (28), inhibited exogenously applied PP2A in BI-P in mouse rod membranes, but it was not effective on our carp rod or cone membranes in the presence of the cytoplasmic proteins. Furthermore, chelation of Mg 2ϩ by EDTA to inhibit PP2C was not effective. We also tried to separate the phosphatase activity, but we lost the activity during chromatographic separation. As a consequence, our attempt to identify the phosphatase(s) responsible for dephosphorylation in our study was not successful.
Arrestin has been reported to inhibit dephosphorylation (24). We measured dephosphorylation in the presence of the rod and cone cytoplasmic proteins, which contained rod arrestins (rArr1 and rArr2) and cone arrestins (cArr1 and cArr2), respectively. Carp rod and cone arrestins bind to bleached pigment no matter whether the pigment is phosphorylated or not (5), which suggests that arrestins bind essentially to bleached pigment. In agreement with this notion, it has been shown that bovine rod arrestin does not bind to VP-P in the dark (29). In our present study dephosphorylation activity was similar between VP-P kept in the dark and that bleached by light (Fig.  4) in the presence of the cytoplasmic proteins or arrestins in both rod and cone membranes. The result, therefore, suggests that arrestins do not affect dephosphorylation in rods or cones as implied previously in mouse rods (13) or that the inhibition of dephosphorylation by arrestin and the facilitation of dephos- phorylation by phosphatase(s) are almost equally balanced with the addition of the cytoplasmic proteins. A further study is required to determine which is the case.
Physiological Relevance of the Biochemical Measurements of Dephosphorylation and Pigment Regeneration-In the present study, half-times of dephosphorylation reaction in Opsin-P, for example, were ϳ64 min in rod membranes and ϳ13.5 min in cone membranes in the presence of the cytoplasmic proteins at the maximum concentrations we examined (50 mg/ml in rod membranes and 150 mg/ml in cone membranes; Fig. 2). These half-times are close to those obtained previously in living animals under the condition of high bleach levels as in this study: ϳ50 min in mouse rods (12,13) and 4 min in zebrafish cones (14). From these results we could say that our biochemical measurements of dephosphorylation reaction are physiologically relevant; the range of the time required for half-reduction of the phosphorylation levels is 50 -60 min in rods and 4 -15 min in cones in living animals at high bleach levels.
Our present study showed that pigment regeneration in membrane preparations is fast; half-times required for pigment regeneration were ϳ120 s in rod membranes and ϳ8 s in cone membranes (Fig. 5, A and D) even though 65 and 80% pigment were bleached in rod and cone membranes, respectively. In contrast, under similar high bleach conditions in rods and cones in living animals, half-times are much longer, and they have been reported to be 50 -60 min in mouse (12,13) and frog rods (30) and 22 s in frog red-sensitive cones (31). In our measurement in membranes, we added an excess amount of 11-cis retinal, which is probably the reason why the pigment regeneration in our biochemical study is much faster than that in living animals. In other words the supply of 11-cis retinal is probably limited so that the regeneration is slower in living animals.
We showed that pigment regeneration does not affect dephosphorylation in either rod or cone membranes (Fig. 5). However, it has been shown that dephosphorylation and pigment regeneration show similar time courses in rods in living mouse (12). This result could be explained by assuming that the supply of 11-cis retinal is the limiting step in living rods. Because of this limitation, pigment regeneration would be slow in living rods, so that the time course of pigment regeneration is similar to that of dephosphorylation. It is also reported that Rlbp Ϫ/Ϫ mice, in which pigment regeneration is expected to be slower than that in wild-type mice, showed normal dephosphorylation time course (12). This result is consistent with the above notion that the supply of 11-cis retinal is the limiting step of pigment regeneration in living rods.
Visual Pigment Dephosphorylation during Its Regeneration Cycle-In this study we showed that the rate of dephosphorylation is not different much among the stages in the visual pigment regeneration cycle in the presence of the cytoplasmic proteins; dephosphorylation activities were similar in BI-P, Opsin-P, and VP-P in both rod and cone membranes (Figs. 1, 3,  and 4). In addition, dephosphorylation activities were not affected by pigment regeneration (Fig. 5). All of these findings suggest strongly that pigment dephosphorylation takes place constantly in the pigment regeneration cycle. In vivo pigment dephosphorylation activities would be higher in cones than in rods (Fig. 2), which will probably be advantageous for cones to maintain the level of functional visual pigment even under bright light (but see below).
In the case of cones in living cold-blooded animals, pigment regeneration (half-time, 22 s in Ref. 31) seems to be faster than dephosphorylation (half-time, 4 -15 min, see above) at least by a factor of 10. This comparison strongly suggests that in cones, some of the regenerated pigments remain phosphorylated. This possibility has been suggested previously in living mouse rods (32), but the extent of the presence of VP-P could be much larger in cones, because time courses are similar in pigment regeneration and in dephosphorylation in rods (12), whereas pigment regeneration is much faster than dephosphorylation in cones as shown above. Our measurement of these time courses in rod and cone membranes (Fig. 5) supports this view. Population of VP-P would depend on both how much pigment is bleached in unit time and the rate of supply of 11-cis retinal, but surely it is high under bright light. It has been shown that bleached VP-P activates the phototransduction cascade much less effectively than the non-phosphorylated pigment (33). In agreement with this previous study, we observed a 3.7-times decrease in the rate of transducin activation in VP-P compared with that in regenerated and non-phosphorylated pigment (VP) in rod membranes; the rates were 20.5 Ϯ 8.1 transducin activated per R* per s (Tr*/R*/s) (n ϭ 3) in VP and 5.6 Ϯ 3.6 Tr*/ R*/s in VP-P (n ϭ 3). In cone membranes, the rate was 11ϫ lower in VP-P; they were 4.1 Ϯ 1.3 Tr*/R*/s in VP (n ϭ 3) and 0.37 Ϯ 0.23 Tr*/R*/s in VP-P (n ϭ 3). For this reason, VP-P should have some roles in desensitization in cones during light adaptation.
Author Contributions-H. Y. designed, performed, and analyzed the experiments and wrote the paper. S. T. and S. K. designed the study and analyzed the experiments and wrote the paper. All authors reviewed the results and approved the final version of the manuscript.