Evolutionary steps involving counterion displacement in a tunicate opsin

Significance

The visible light sensitivity of animal opsins is attributed to a protonated Schiff base linkage to retinal, which is stabilized by a negatively charged counterion. Glu113 and Glu181 serve as the counterion in vertebrate- and invertebrate-type opsins, respectively. Ascidian opsin, the sole invertebrate opsin found in the vertebrate visual opsin group, contains Glu113 and Glu181, which act as synergistic double counterions. Deleting either counterion changes the ascidian opsin’s molecular properties to resemble those of either vertebrate- or invertebrate-type opsins. These results strongly suggest that counterion position regulates opsin molecular properties beyond spectral tuning. Moreover, ascidian opsin is an apparent evolutionary intermediate leading to vertebrate visual opsin, which has acquired the high G protein activation efficiency optimized for the vertebrate vision.

Abstract

Ci-opsin1 is a visible light-sensitive opsin present in the larval ocellus of an ascidian, Ciona intestinalis. This invertebrate opsin belongs to the vertebrate visual and nonvisual opsin groups in the opsin phylogenetic tree. Ci-opsin1 contains candidate counterions (glutamic acid residues) at positions 113 and 181; the former is a newly acquired position in the vertebrate visual opsin lineage, whereas the latter is an ancestral position widely conserved among invertebrate opsins. Here, we show that Glu113 and Glu181 in Ci-opsin1 act synergistically as counterions, which imparts molecular properties to Ci-opsin1 intermediate between those of vertebrate- and invertebrate-type opsins. Synergy between the counterions in Ci-opsin1 was demonstrated by E113Q and E181Q mutants that exhibit a pH-dependent spectral shift, whereas only the E113Q mutation in vertebrate rhodopsin yields this spectral shift. On absorbing light, Ci-opsin1 forms an equilibrium between two intermediates with protonated and deprotonated Schiff bases, namely the MI-like and MII-like intermediates, respectively. Adding G protein caused the equilibrium to shift toward the MI-like intermediate, indicating that Ci-opsin1 has a protonated Schiff base in its active state, like invertebrate-type opsins. Ci-opsin1’s G protein activation efficiency is between the efficiencies of vertebrate- and invertebrate-type opsins. Interestingly, the E113Y and E181S mutations change the molecular properties of Ci-opsin1 into those resembling invertebrate-type or bistable opsins and vertebrate ancient/vertebrate ancient-long or monostable opsins, respectively. These results strongly suggest that acquisition of counterion Glu113 changed the molecular properties of visual opsin in a vertebrate/tunicate common ancestor as a crucial step in the evolution of vertebrate visual opsins.

Opsins are universal, photoreceptive animal proteins functionally characterized as light-sensitive G protein-coupled receptors (GPCRs) (1, 2). The opsins that have been identified to date are classified into several groups based on their amino acid sequences (3, 4). All opsins share certain structural elements, including seven transmembrane α-helices, and contain a light-absorbing chromophore, 11-cis-retinal, which is bound to Lys296 of opsin, based on the bovine rhodopsin numbering system, through a Schiff base linkage. Light causes isomerization of 11-cis- to all-trans-retinal, which causes opsins to form an active state that can couple with cognate G proteins. In visible light-absorbing opsins, the retinylidene Schiff base is protonated and stabilized by a nearby negatively charged residue (either glutamic acid or aspartic acid) called a counterion (Fig. S1). Invertebrate Gq-coupled visual opsins and invertebrate Go-coupled opsins (i.e., invertebrate-type opsins) have a glutamic acid counterion at position 181 (Glu181) (5). In contrast, glutamic acid at position 113 (Glu113) serves as the counterion in vertebrate visual opsin, rhodopsin, and cone visual pigments (vertebrate-type opsins) (6–8), which originated from the counterion displacement from positions 181 to 113 in the evolutionary history of vertebrate visual opsins, resulting in a change in photochemical properties and the acquisition of a highly efficient G protein activation mechanism in vertebrate visual opsins (5, 9). Thus, this counterion displacement from Glu181 to Glu113 increases the photosensitivity of vertebrate visual opsins, which is thought to be the molecular basis for the development of highly functional vertebrate eyes (10).

Fig. S1.

Phylogenetic relationship of opsins and the counterion. (A) Schematic drawing of the phylogenetic tree of opsins and their amino acid residues at positions 113 and 181. The glutamic acid highlighted in red acts as a counterion in each opsin group. (B) Retinal chromophore covalently binds to Lys296 in helix VII through a protonated Schiff base that is stabilized by a negatively charged residue (glutamic acid or aspartic acid) that is called a counterion. (C) Amino acid sequence alignment of bovine rhodopsin (bovine Rh) against Ci-opsin1. We used the bovine rhodopsin numbering system in this study.

Ascidians are tunicates that are the closest relatives of invertebrates to vertebrates (11). In their larval stage, they have a photoreceptive organ, the ocellus, which contains photoreceptor cells responsible for light-dependent regulation of swimming behavior (12, 13). Ci-opsin1 is an opsin expressed in photoreceptor cells of the larval ocellus (14). Although Ci-opsin1 is not a vertebrate opsin, it belongs to the vertebrate visual and nonvisual opsin group in the opsin phylogenetic tree (3, 14). This fact is in contrast to another chordate species, amphioxus, which has no opsins that can be placed in the vertebrate visual and nonvisual opsin group, despite the presence of at least 16 opsin-encoding genes in its genome (15). Thus, Ci-opsin1 may represent a key step in the evolution of vertebrate visual opsins.

Ci-opsin1 contains glutamic acid residues at positions 113 and 181 as candidates for counterions (14). Therefore, in the initial stage of our investigation, we examined which glutamic acid residue acts as a counterion in Ci-opsin1. Unexpectedly, we found experimental evidence that both of the glutamic acids act synergistically as counterions. The acquisition of synergistic counterions may explain why Ci-opsin1 exhibits photochemical properties and a G protein activation efficiency intermediate between the efficiencies of vertebrate visual and invertebrate-type opsins. More interestingly, replacing Glu113 with tyrosine, which is highly conserved in invertebrate-type opsins, changes the photochemical properties of Ci-opsin1 to those resembling invertebrate-type opsins. In contrast, replacing Glu181 with serine, which is conserved in vertebrate ancient/vertebrate ancient-long (VA/VAL) opsins, changes the photochemical properties of Ci-opsin1 into those resembling VA/VAL opsins. We discuss the molecular properties of Ci-opsin1 and the conversion of its properties into those resembling either invertebrate-type opsins or VA/VAL opsins in the context of opsin molecular evolution.

Results and Discussion

Function of Glu113 and Glu181 as Synergistic Counterions in Ci-opsin1.

As reported previously (12), the absorption maximum (λ_max) of the purified Ci-opsin1 after being reconstituted with 11-cis-retinal was 497 nm (Fig. S2A). Thus, Ci-opsin1 is a visible light-absorbing opsin. The extinction coefficient of Ci-opsin1 was determined to be 44,096 M⁻¹ cm⁻¹ at 497 nm (Fig. S2B) by the acid denaturation method (16, 17). Given that Ci-opsin1 has glutamic acids at positions 113 and 181, we attempted to determine which glutamic acid acts as a counterion by preparing single (E113Q and E181Q) and double (E113Q/E181Q) opsin mutants. We successfully generated recombinant forms of the single-mutant proteins, but failed to generate an active recombinant form of the double-mutant protein in cultured cells.

Fig. S2.

Absorption spectra of wild-type, E113D mutant, and E181D mutant of Ci-opsin1 and the extinction coefficient of Ci-opsin1. (A) Absorption spectrum of purified wild-type Ci-opsin1 reconstituted with 11-cis-retinal at pH 6.5 and 0 °C. (Inset) Difference spectrum in the presence of 10 mM NH₂OH before and after yellow light irradiation. The negative peak is the λ_max of the dark state (497 nm). (B) Estimation of the molar extinction coefficient of the dark state of Ci-opsin1 by the acid denaturation method. We measured the absorption spectra of Ci-opsin1 before (pH 6.5, curve 1) and after acid denaturation by the addition of HCl (pH 1.5, curve 2). We also measured the spectra of bovine rhodopsin before (pH 6.5, curve 3) and after acid denaturation (pH 1.6, curve 4). The molar extinction coefficient of the dark state of Ci-opsin1 is 44,096 M⁻¹ cm⁻¹ at 497 nm. (C and D) Absorption spectra of E113D and E181D mutants of Ci-opsin1 measured at pH 5.3, 0 °C. (Inset) Difference spectra in the presence of 2.5 mM NH₂OH before and after yellow light irradiation. λ_max values of the dark state of E113D and E181D are 470 and 481 nm, respectively.

The absorption spectrum of wild-type Ci-opsin1 was independent of the pH change of the sample (Fig. 1). In contrast, the E113Q mutant showed pH-dependent spectral changes like the E113Q mutant of bovine rhodopsin (6–8). Interestingly, the E181Q mutant also showed pH-dependent spectral changes, in contrast to the absence of pH-dependent spectral changes in bovine rhodopsin carrying the E181Q mutation (5, 18). Therefore, unlike bovine rhodopsin, both glutamic acids of Ci-opsin1 act as counterions to synergistically stabilize the protonated Schiff base in the dark state.

Fig. 1.

Absorption spectra of wild-type Ci-opsin1 and its E113Q and E181Q mutants. Absorption spectra of wild-type Ci-opsin1 and its E113Q and E181Q mutants were measured at three pH levels and 0 °C.

It should be noted that the spectral shifts induced by the mutations at these glutamic acids are different between Ci-opsin1 and other opsins that have been characterized to date. Many visible light-absorbing opsins, including bovine rhodopsin, showed almost the same λ_max when their counterions were removed (5, 6, 19), which can be explained by the contribution of a chloride ion in the sample as a surrogate counterion (20); however, the λ_max values of E113Q and E181Q mutants of Ci-opsin1 were blue-shifted to ∼464 nm and ∼485 nm, respectively. These considerable shifts in λ_max suggest that chloride ions in the sample cannot serve as a surrogate counterion, likely because counterion mutations can cause conformational changes near the retinal Schiff base in Ci-opsin1.

To test this possibility, we tried to remove chloride ions from the Ci-opsin1 sample by extensive dialysis; however, the Ci-opsin1 protein was denatured by this treatment. Therefore, we were unable to observe conclusive evidence concerning the role of chloride ions in Ci-opsin1 function. It also has been reported that in bovine rhodopsin, the replacement of counterion Glu113 with aspartic acid caused a ∼10-nm red shift in λ_max (20). This red shift was accounted for by a model that proposes that the separation of the negatively charged carboxyl group from the protonated Schiff base by the introduction of a smaller amino acid residue results in decreased electronic interaction between them. However, replacing either Glu113 or Glu181 with aspartic acid in Ci-opsin1 caused blue shifts to λ_max similar to those caused by replacement with glutamine (Fig. S2 C and D). These results suggest that removing one glutamic acid at either of these positions caused some conformational changes in the protein near the retinal Schiff base, resulting in the close approach of the other glutamic acid to the protonated Schiff base.

Characterization of the Ci-opsin1 Active State.

Given Ci-opsin1’s unique counterion system, we expected Ci-opsin1 to show somewhat different molecular properties from those of vertebrate visual and invertebrate-type opsins. We first investigated G protein activation efficiency of Ci-opsin1 using a fluorescence assay (21, 22). Because Ci-opsin1 is colocalized with Gi-type of G protein in the ocellus of ascidian larvae (23), we irradiated the mixture of purified Ci-opsin1, Gi, and GTPγS with yellow light and confirmed an increase in fluorescence due to Gi activation (Fig. 2A). The initial rate of Gi activation by Ci-opsin1 (0.6 Gi*/s) was 4.5-fold lower than activation by bovine rhodopsin (2.7 Gi*/s) (Fig. S3A), but more than 10-fold higher than activation by the invertebrate-type opsin, amphioxus rhodopsin (5). Therefore, it seems plausible that the photosensitivity of Ci-opsin1 increased over the course of evolution, given that ascidian larvae are thought to search actively for dark places near the sea bottom for attachment and metamorphosis after having floated on the sea surface against gravity (24, 25).

Fig. 2.

G protein activation and photochemical property of Ci-opsin1. (A, Left) Fluorescence increase derived from Gi activation by photoactivated Ci-opsin1. The purified opsin (340 nM Ci-opsin1) was mixed with 1.5 μM Gi and 1 mM GTPγS at pH 6.5. The mixture was irradiated with a yellow light flash at t = 0, and the fluorescence intensity was monitored (upper trace) at 0 °C. The mixture in the absence of Gi was irradiated and the fluorescence intensity was monitored as well (lower trace). (A, Right) Fluorescence increase derived from Gi activation by Ci-opsin1 photoproduct. The purified opsin (340 nM Ci-opsin1) was irradiated with a yellow light flash at pH 6.5. After a 40-min incubation, 1.5 μM Gi and 1 mM GTPγS were added at t = 0, and the fluorescence intensity was monitored (upper trace) at 0 °C. The purified opsin was irradiated, and after a 40-min incubation, the fluorescence change following the addition of GTPγS was also recorded (lower trace). (B) Photoreaction of Ci-opsin1 after yellow light irradiation. Spectra were measured in the dark (curve 1) and at 75 s (curve 2) and 225 s (curve 3) after yellow light irradiation at pH 6.5 and 0 °C. (C) Effect of Gi on the equilibrium between MI-like and MII-like intermediates of Ci-opsin1. Absorption spectra of the intermediates were obtained by subtracting the spectral components of the unreacted Ci-opsin1 in the irradiated sample from the spectra measured at 2,475 s after yellow light irradiation of Ci-opsin1 in the absence of Gi (curve 1), in the presence of 1.5 μM Gi (curve 2), and in the presence of 1.5 μM Gi and 1 mM GTPγS (curve 3).

Fig. S3.

Gi activation of bovine rhodopsin and acid denaturation of MII-like intermediate of Ci-opsin1. (A) Fluorescence increase derived from Gi activation by photoactivated bovine rhodopsin. Purified opsin (40 nM bovine rhodopsin) was mixed with 1.5 μM Gi and 1 mM GTPγS at pH 6.5. The mixture was irradiated with a yellow light flash at t = 0, and the fluorescence intensity was monitored (upper trace) at 0 °C. The mixture in the absence of Gi was irradiated and the fluorescence intensity was also monitored (lower trace). (B) Fluorescence increase derived from Gi activation by the photoproduct of bovine rhodopsin. The purified opsin (40 nM bovine rhodopsin) was irradiated with a yellow light flash at pH 6.5. After 40 min incubation, 1.5 μM Gi and 1 mM GTPγS were added at t = 0, and the fluorescence intensity was monitored (upper trace) at 0 °C. The purified opsin was irradiated, and after a 40-min incubation, the fluorescence change by the addition of GTPγS was monitored as well (lower trace). (C) Acid denaturation of the MII-like intermediate of Ci-opsin1. The absorption spectra of the photoproduct of Ci-opsin1 were measured at 40 min after yellow light irradiation (curve 1, pH 6.5), and the photoproduct was denatured by adding HCl (curve 2; final pH 1.9).

To identify the active intermediate of Ci-opsin1, we recorded absorption spectra after irradiating Ci-opsin1 with yellow light (Fig. 2B). Ci-opsin1 formed a mixture of two photoproducts with λ_max at ∼470 nm and ∼380 nm. Our acidification experiments verified that the ∼380-nm product is not all-trans-retinal free from the protein but rather is an intermediate (17) (Fig. S3C). Hereinafter, we refer to the two intermediates as MI-like and MII-like intermediates based on their λ_max. Time-resolved absorption spectroscopy demonstrated the conversion of the MI-like intermediate to the MII-like intermediate, and the subsequent formation of an equilibrium between the two intermediates. The time constant for the formation of the equilibrium was 312 s at 0 °C (Fig. S4 A and B). It should be noted that although the MII-like intermediate was formed slowly, the increase in fluorescence from Gi activation was observed immediately after light irradiation (Fig. 2A). Furthermore, the increase in fluorescence observed using the Ci-opsin1 sample incubated for 40 min after light irradiation was much smaller than that observed immediately after light irradiation (Fig. 2A and Fig. S3B). These results strongly suggest that the MI-like intermediate, rather than the MII-like intermediate, is the active state of Ci-opsin1 to interact with G protein. However, because of the limited amount of Ci-opsin1 available for the G protein activation assay, we could not conclude that the MII-like intermediate lacks the ability to activate Gi.

Fig. S4.

Photochemical property of Ci-opsin1. (A) Effect of Gi on the decay of the MI-like intermediate of Ci-opsin1. Changes in absorbance at 480 nm were plotted against the incubation time after yellow light irradiation. The decay rates were 351 s in the absence of Gi, 742 s in the presence of Gi, and 312 s in the presence of Gi and GTPγS. (B) pH dependency of the photoreaction of Ci-opsin1. Spectra were measured after yellow light irradiation at pH 5.2 (red closed circles), pH 5.8 (yellow open circles), pH 6.5 (blue closed squares), pH 7.2 (green open squares), and pH 7.5 (purple closed triangles) at 0 °C, and the changes in absorbance at 480 nm were plotted against the incubation time after irradiation. (C) pH dependency of the equilibrium between the MI-like and MII-like intermediates of Ci-opsin1. Spectra were measured after yellow light irradiation at pH 5.2, 5.8, 6.5, 7.2, and 7.5 and 0 °C (Fig. S4B), and difference spectra in the apparent equilibrium state were calculated by the SVD method (22). (Inset) The ratios of intermediates with the protonated Schiff base in the apparent equilibrium state of Ci-opsin1 and bovine rhodopsin. The plots were fitted using the Henderson–Hasselbalch equation. The pK_a values of Ci-opsin1 and bovine rhodopsin were 6.5 and 6.8, respectively.

It has been reported that either Gt or Gi in the irradiated bovine rhodopsin sample caused the MI-MII equilibrium to shift toward MII. This shift can be accounted for by the stabilization of MII through the interaction with G protein (called “extra-MII”) (26). Interestingly, Gi in the irradiated Ci-opsin1 sample caused the equilibrium to shift to the MI-like intermediate. This shift disappeared when GTPγS was added to the sample (Fig. 2C). We also observed a delay in the decay process of the MI-like intermediate in the presence of Gi (Fig. S4A). These results clearly indicate that Gi can interact with and stabilize the MI-like intermediate in Ci-opsin1. Therefore, Ci-opsin1 has the G protein-interacting state with a protonated Schiff base like invertebrate-type opsins (27). Next, we investigated the pH-dependent shift in the equilibrium between MI-like and MII-like intermediates (Fig. S4 B and C). Acidification of the sample caused the equilibrium to shift to the MI-like intermediate. This pH dependency is similar to that of invertebrate-type opsins and in contrast to that of vertebrate visual opsins (5, 28, 29).

Photochemical Reaction of the Ci-opsin1 Active State.

We next investigated the photochemical properties of the MI-like and MII-like intermediates. To monitor the photochemical reaction of the MI-like intermediate, we first irradiated Ci-opsin1 with yellow light at acidic pH to form a large amount of MI-like intermediate (curve 2 in Fig. 3A) before irradiating the sample with blue light (curve 3 in Fig. 3A). Blue light irradiation caused increased absorbance at wavelengths longer than 500 nm, suggesting that the MI-like intermediate could photoconvert to a red-shifted product, which is likely the original state of Ci-opsin1. To verify this, we first obtained the difference spectrum (curve 1 in Fig. S5) by subtracting curve 2 from curve 3 in Fig. 3A, which is the composite of two difference spectra, one derived from the photoreaction of the MI-like intermediate induced by the blue light and the other derived from the thermal conversion from the MI-like intermediate to the MII-like intermediate during the experiment.

Fig. 3.

Photoconversion back to the dark state of Ci-opsin1. (A) Photoreaction of the active state (MI-like intermediate) of Ci-opsin1 by blue light irradiation. Absorption spectra of Ci-opsin1 were measured in the dark (curve 1), after yellow light irradiation (curve 2), and after subsequent blue light irradiation (curve 3) at pH 5.3 and 0 °C. (Inset) Curve 4 is the blue light-dependent spectral change (see text for details). Curve 5 is the difference spectrum obtained by subtracting curve 1 from curve 2. Both spectra are shown after normalizing them by the peak absorbance at 523 nm. (B) Retinal configuration changes caused by yellow light and subsequent blue light irradiation of Ci-opsin1. Isomeric compositions of retinal before and after irradiation of Ci-opsin1 were estimated by HPLC analysis (Fig. S6A).

Fig. S5.

Blue light-dependent spectral change of the MI-like intermediate of Ci-opsin1. Curve 1 is the difference spectrum obtained by subtracting curve 2 from curve 3 in Fig. 3A. Curve 2 shows the spectral change involved in the conversion from MI-like to MII-like intermediates during the experiment, which is obtained from the difference spectrum between curves 2 and 3 in Fig. 2B. The calculation procedure is explained in detail in the text. Curve 3 is the spectral change induced solely by blue light irradiation, which was obtained by subtracting curve 2 from curve 1.

Because the difference spectrum between the MI-like and MII-like intermediates can be calculated by subtracting curve 2 from curve 3 in Fig. 2B, we next estimated the component of the conversion from the MI-like intermediate to the MII-like intermediate (curve 2 in Fig. S5) contained in curve 1 in Fig. S5. This estimation was based on the amount of MI-like intermediate produced by the initial blue light irradiation of the dark state and the amount of MII-like intermediate thermally produced during the experiment. The latter amount was calculated using a previously reported method (16) with modifications.

We next subtracted curve 2 from curve 1 in Fig. S5 to obtain the blue light-dependent spectral change (curve 3 in Fig. S5). Curve 4 is shown in Fig. 3A, Inset after normalizing curve 3 in Fig. S5 by the positive maximum (523 nm). We also obtained the difference spectrum (curve 5 in Fig. 3A, Inset) by subtracting the spectrum before irradiation from that measured after yellow light irradiation. We subsequently normalized the difference spectrum by the negative maximum at 523 nm. These two curves are mirror images of each other, which indicates that the MI-like intermediate can photoconvert back to the dark state of Ci-opsin1 on absorption of blue light. Furthermore, we verified that cis-trans and trans-cis isomerization of retinal chromophore occurred in these processes by extracting retinal chromophores (Fig. 3B and Fig. S6A).

Fig. S6.

Retinal configuration changes by light irradiation of wild-type Ci-opsin1 and its E113Y and E181S mutants. (A) Retinal configuration changes caused by yellow light and subsequent blue light irradiation of Ci-opsin1. (B) Retinal configuration changes triggered by blue light, subsequent orange light, and second blue light irradiation of E113Y mutant of Ci-opsin1. (C) Retinal configuration changes by yellow light and subsequent blue light irradiation of E181S mutant of Ci-opsin1. Retinal configurations before and after light irradiation were determined by HPLC after extracting the chromophore as retinal oximes: syn and anti forms of 11-cis-retinal oximes (11), 13-cis-retinal oximes (13), and all-trans-retinal oximes (AT).

We also investigated the photochemical properties of the MII-like intermediate. We incubated the Ci-opsin1 sample for 40 min after yellow light irradiation under pH 6.5 to accumulate the MII-like intermediate before irradiating the sample with UV light. However, UV light irradiation caused no significant spectral change (Fig. S7A). We also incubated the Ci-opsin1 sample for 500 s after yellow light irradiation at pH 7.5 to accumulate the MII-like intermediate, and UV light irradiation caused no significant spectral change thereafter (Fig. S7B). These data strongly suggest that the MII-like intermediate is photoinsensitive.

Fig. S7.

Photochemical property of MII-like intermediate of Ci-opsin1. (A) Absorption spectra were measured 40 min after yellow light irradiation (curve 1) and after subsequent UV light irradiation (curve 2) at pH 6.5 and 0 °C. (Inset) Spectral change caused by subsequent UV light irradiation (difference spectrum between curves 1 and 2). (B) Absorption spectra were measured at 500 s after yellow light irradiation (curve 1), and again after subsequent UV light irradiation (curve 2) at pH 7.5 and 0 °C. (Inset) Spectral changes caused by subsequent UV light irradiation (difference spectrum between curves 1 and 2).

Change in the Photochemical Properties of Ci-opsin1 Caused by Removal of a Counterion.

Ci-opsin1 has two glutamic acid residues that act synergistically as counterions and shows the mixed molecular properties of those of vertebrate visual and invertebrate-type opsins. Therefore, we examined whether the molecular properties of Ci-opsin1 can be changed by removing one of the counterions. For this purpose, we prepared a Ci-opsin1 mutant that has tyrosine at position 113, a residue that is conserved in most invertebrate-type opsins. We also prepared a mutant with serine at position 181, which is conserved in VA/VAL opsins, the closest monostable opsin group in the phylogenetic tree (3, 30). We successfully generated the recombinant E113Y and E181S mutant proteins after reconstitution with 11-cis-retinal.

The E113Y mutant registered a λ_max of 460 nm (curve 1 in Fig. 4A), which is ∼40 nm blue-shifted from that of the wild-type but similar to that of the E113Q mutant. Irradiation with blue light caused a slight red shift in λ_max with an increase in absorbance, indicating the formation of an active state (curve 2 in Fig. 4A). Subsequent irradiation with orange light caused a blue shift in λ_max with slightly decreased absorbance. Reirradiation with blue light produced a photoproduct with an absorption spectrum identical in shape to that produced after the initial blue light irradiation (curves 2 and 4 in Fig. 4A). The difference spectrum before and after orange light irradiation (curve 5 in Fig. 4A, Inset) is a mirror image of that before and after a second round of blue light irradiation (curve 6 in Fig. 4A, Inset). Furthermore, our analysis of the retinal configuration changes confirmed that the isomerization between 11-cis and all-trans forms occurred in these photoreactions (Fig. 4B and Fig. S6B). Therefore, the E113Y mutation caused the formation of a bistable pigment that exhibits the photoconversion between two stable states, the dark and active states, like invertebrate-type opsins (5, 31, 32). These results indicate that Glu113 is necessary for the formation of the intermediate that has a deprotonated Schiff base.

Fig. 4.

E113Y and E181S mutations changed the photochemical property of Ci-opsin1. (A and B) Bistable nature of the E113Y mutant. (A) Absorption spectra of the E113Y mutant were measured in the dark (curve 1), after blue light irradiation (curve 2), after subsequent orange light irradiation (curve 3), and after subsequent blue light irradiation (curve 4) at pH 6.0 and 0 °C. (Inset) Curve 5 is the difference spectrum obtained by subtracting curve 2 from curve 3. Curve 6 is the difference spectrum obtained by subtracting curve 3 from curve 4. (B) Retinal configuration changes by blue light, subsequent orange light, and subsequent blue light irradiations of the E113Y mutant. Isomeric compositions of retinal before and after light irradiation of E113Y mutant were estimated by HPLC analysis (Fig. S6B). (C and D) Monostable nature of the E181S mutant. (C) Absorption spectra of the E181S mutant were measured in the dark (curve 1), after yellow light irradiation (curve 2), and after subsequent blue light irradiation (curve 3) at pH 5.3 and 0 °C. (Inset) Difference spectrum before and after blue light irradiation (curve 4). Curve 4 is the difference spectrum obtained by subtracting curve 2 from curve 3. (D) Retinal configuration changes by yellow light and subsequent blue light irradiation of the E181S mutant. Isomeric compositions of retinal before and after irradiation of the E113Y mutant were estimated by HPLC analysis (Fig. S6C).

The λ_max of the E181S mutant was 468 nm (curve 1 in Fig. 4C), which is ∼30 nm blue-shifted from that of the wild-type. Irradiation with yellow light produced a photoproduct with λ_max at ∼445 nm (curve 2 in Fig. 4C), accompanied by retinal photoisomerization from the 11-cis to the all-trans form (Fig. 4D and Fig. S6C). Subsequent irradiation with blue light caused no spectral change (curve 3 in Fig. 4C) and no retinal configuration change (Fig. 4D and Fig. S6C), indicating that the photoproduct is photoinsensitive. These photochemical properties are similar to those of VA/VAL opsins (30).Therefore, removing one of the counterions in Ci-opsin1 through either the E113Y or E181S mutation converts it into either an invertebrate-type opsin (bistable opsin) or a VA/VAL opsin (monostable opsin), respectively. These results strongly suggest that counterion displacement contributes to a change in the molecular properties of opsins.

Evolutionary Significance of Ci-opsin1.

In this study, we determined that Ci-opsin1 and vertebrate visual opsin, which share Glu113 and Glu181, have different active states and have protonated and deprotonated intermediates, respectively. Thus, it has been suggested that during the evolutionary process from the common ancestor of these opsins, switching of the active state from the protonated (MI) to the deprotonated (MII) intermediate could lead to an increase in the ability of vertebrate visual opsins to activate G proteins. Vertebrate visual opsins contain a hydrogen-bonding network system that connects Glu113 and Glu181 through Ser186 and a water molecule (33, 34). The crystal structure of bovine rhodopsin shows that the distance between Glu113 and Ser186 is too far (∼4.1 Å) to form the hydrogen bond directly, and thus the hydrogen-bonding network is incomplete in the resting state. However, the counterion switch from Glu113 to Glu181 occurs in MI, which could be caused by conformational changes near Glu113 and Ser186 to complete the hydrogen-bonding network. These local movements may induce the formation of MII, which exhibits efficient G protein activation. However, we showed that the mutations at Glu113 and Glu181 have very different effects on the photochemical properties in Ci-opsin1 and vertebrate visual opsins.

Replacement of Glu113 or Glu181 prevented formation of the deprotonated intermediate in Ci-opsin1 (Fig. 4), whereas the Glu113 or Glu181 mutants of bovine rhodopsin also formed the deprotonated intermediate (MII) (35–37). Moreover, replacing Ser186 with Ala in Ci-opsin1 did not affect the photoreaction and even stabilized the MI-like intermediate (Fig. S8), whereas the corresponding replacement in bovine rhodopsin lost the formation of MI and caused the rapid formation of MII (38). Thus, we believe that Ci-opsin1 has a different hydrogen-bonding network, which includes Glu113, Glu181, and Ser186, from that of the vertebrate visual opsins. An investigation of the mechanism of formation of the hydrogen-bonding network is needed to understand the molecular evolution of vertebrate visual opsins.

Fig. S8.

Photochemical property of S186A mutant of Ci-opsin1. (A) Photoreaction of the S186A mutant of Ci-opsin1 following yellow light irradiation. Spectra were measured in the dark (curve 1) and 75 s (curve 2) and 40 min (curve 3) after yellow light irradiation at pH 6.5 and 0 °C. (B) Changes in absorbance at 480 nm plotted against incubation time after yellow light irradiation. The decay rate of the MI-like intermediate was 1,356 s.

Conclusion

We have determined that Glu113 and Glu181 together act as a unique double-counterion system in Ci-opsin1. Moreover, we found that this ascidian opsin exhibits 10-fold greater G protein activation efficiency compared with those of invertebrate-type opsins that have Glu181 as a single counterion. However, in contrast to vertebrate visual opsins that have Glu113 as a single counterion, Ci-opsin1 has a hydrogen-bonding network near two counterions that is insufficient to produce an active deprotonated intermediate. Thus, Ci-opsin1 is in an evolutionarily intermediate state between invertebrate-type and vertebrate visual opsins (Fig. 5). The mutational analysis at counterion positions strongly suggests that counterion displacement contributes to changes in the molecular properties of opsins. Therefore, we speculate that the acquisition of counterion Glu113 contributed to the change in molecular properties of a visual opsin from an ancestral chordate as a crucial step in the evolution of vertebrate visual opsin.

Fig. 5.

Intermediate property of Ci-opsin1 between those of invertebrate-type and vertebrate visual opsins.

Materials and Methods

Preparation of Pigments.

The cDNA of Ci-opsin1 (GenBank accession no. AB058682) was fused to a C-terminal sequence encoding the epitope recognized by the anti-bovine rhodopsin monoclonal antibody Rho1D4, ETSQVAPA. The fusion product was inserted into pUSRα or pMT4 vector (39). Ci-opsin1 cDNAs containing mutations were constructed using the In-Fusion Cloning Kit (Clontech). The plasmid was transfected into HEK293 cells using the calcium-phosphate method. After 2 d of incubation, transfected cells were collected by centrifugation and suspended in Buffer A (50 mM Hepes and 140 mM NaCl, pH 6.5), and 11-cis-retinal was added to the cell suspension to reconstitute the photoactive pigments. The cells were solubilized in Buffer A containing 1% dodecyl maltoside (DDM) and adsorbed to a Rho1D4 affinity column to purify the pigments. After the column was washed with Buffer B (50 mM Hepes, 140 mM NaCl, and 3 mM MgCl₂, pH 6.5) containing 0.02% DDM, the pigment was eluted by adding synthetic peptide with the epitope sequence. Photoreaction experiments and the Gi activation assay were performed in Buffer B containing 0.01% DDM. Absorption spectra of the dark state were measured in Buffer B containing 0.02% DDM.

Preparation of Giα and Gtβγ.

We obtained Gtβγ from bovine rod outer segments as described previously (40). The rat Giα subunit was expressed in Escherichia coli stain BL21 using the pQE6 vector containing Giα cDNA, and was purified as described previously (41). Purified Giα was mixed with equal amounts of purified Gtβγ.

Spectroscopic Measurements and HPLC Analysis.

Absorption spectra of the samples were recorded with a UV-visible spectrophotometer (Shimadzu UV-2450 and UV-2400). Samples were kept at 0 °C using a cell holder equipped with a circulation system that uses temperature-controlled water to analyze the thermal reaction of the pigments in detail (5, 30). The samples were irradiated with either yellow light through a Y-51 cutoff filter (Toshiba), blue light through a KL-43 bandpass filter (Toshiba), UV light through a UVD-36 glass filter (Asahi Technoglass), or orange light through a O-54 cutoff filter (Toshiba) from a 1 kW halogen lamp (Master HILUX-HR; Rikagaku).

Chromophore configurations of the samples were analyzed by HPLC (LC-10AT VP; Shimadzu) equipped with a silica column (150 × 6 mm, A-012–2; YMC) as described previously (42).

G Protein Activation Assay.

Gi activation was measured by a fluorescence assay (21, 22, 43). Fluorescence changes were monitored using a laboratory-constructed photon counting system with some modifications (43). The opsins were irradiated with a yellow light flash generated by a combination of a short-arc xenon flash lamp (SA-200F; Nissin Electronic) with a Y-51 filter for Ci-opsin1 and bovine rhodopsin. Initial rates of Gi activation efficiencies were calculated as well, as described previously (22, 43). The concentrations of Gi and GTPγS were 1.5 μM and 1 mM, respectively. The assay was performed in Buffer B (pH 6.5) containing 0.01% DDM at 0 °C.