Introduction

Animals sense light by using opsins, photosensitive proteins belonging to the large family of G protein-coupled receptors (GPCRs). Opsins have a seven-transmembrane helix structure and bind to a retinal chromophore to form a light-sensitive pigment. Based on their phylogenetic relationships (Koyanagi and Terakita, 2014), eumetazoan animal opsins can be classified into eight groups with distinctive properties: vertebrate visual/non-visual opsins, opn3/TMT opsins, invertebrate Go-coupled opsins, cnidarian Gs-coupled opsins (cnidopsins), neuropsins (opn5), Gq-coupled visual pigments/melanopsins (opn4), peropsins, and retinochrome/RGR. Such diversity possibly underlies the diversification of light-dependent physiologies in animals. But, furthermore, this diversity also provides a range of potential optogenetic tools to manipulate intracellular G protein-mediated signaling (Koyanagi and Terakita, 2014).

Reef-building corals and sea anemones belong to the subphylum Anthozoa, which together with the subphylum Medusozoa constitute the phylum Cnidaria. Cnidarian animals possess multiple opsins categorised as part of the cnidarian Gs-coupled opsin group (cnidopsins), which regulate light-dependent processes (Koyanagi et al., 2008). For example, a member of this group is expressed in the ciliary type visual cells of the box jellyfish lens eyes (Koyanagi et al., 2008; Kozmik et al., 2008). Beyond these Gs-coupled cnidopsins, anthozoan animals have opsins phylogenetically distant from the other known eight groups, as well as opsins found exclusively in anthozoans (Feuda et al., 2012; Suga et al., 2008). These anthozoan-specific opsins (ASO) can be further classified into two groups, ASO-I and ASO-II (Gornik et al., 2020; Hering and Mayer, 2014; Mason et al., 2023; Picciani et al., 2018; Ramirez et al., 2016). Gornik et al. (2021) proposed that both ASO-I and ASO-II were present in the last common ancestor of Anthozoa and Medusozoa, but were lost secondarily in the Medusozoa lineage (Gornik et al., 2020). While it has been reported that both ASO-I and ASO-II are expressed in multiple tissues of sea anemones (Gornik et al., 2020; Suga et al., 2008) and corals (Levy et al., 2021), there is still a limited understanding of their molecular characteristics and physiological functions.

The members of the ASO-II group are not only phylogenetically unique but also display interesting features in their amino acid sequences. For instance, several of these opsins lack an amino acid residue conserved among typical opsins that is crucial for absorption of visible light (Gornik et al., 2020; Mason et al., 2023). While free retinal in solution has its absorption maximum (λ_max) in the ultraviolet (UV), this shifts to visible light when retinal is bound to a lysine residue in the transmembrane bundle of the opsin (usually at Lys296, numbering according to the bovine rhodopsin sequence) through a protonated Schiff base to form the pigment. Such protonated Schiff base is necessary to achieve sensitivity to visible light in opsin-based pigments (Pitt et al., 1955). However, the proton on the positively charged Schiff base is energetically unstable in the hydrophobic transmembrane environment. To stabilise this proton, a negatively charged residue, glutamic or aspartic acid, is situated near the Schiff base to act as a counterion. This counterion is essential for opsin-based pigments to absorb visible light, and the residues serving as the counterion are highly conserved across opsins (Terakita, 2005; Terakita et al., 2012). To date, three experimentally confirmed sites for the counterion have been identified in animal opsins: 94 in helix 2 (Gerrard et al., 2018), 113 in helix 3 (Nathans, 1990; Sakai et al., 2022; Sakmar et al., 1989; Zhukovsky and Oprian, 1989) and 181 in extracellular loop 2 (Nagata et al., 2019; Terakita et al., 2004, 2000). Remarkably, some opsins belonging to the ASO-II group lack glutamic or aspartic acid at any of these established counterion positions (Gornik et al., 2020; Mason et al., 2023). This absence raises the question of whether these opsins can absorb visible light, and if so, by what mechanism.

In this study, we investigate the spectroscopic properties of opsins in the ASO-II group isolated from the reef-building coral Acropora tenuis. Absorption spectra reveal that this group includes opsins sensitive to both UV and visible light. We then focus on a particular visible light-sensitive opsin within the ASO-II group (Antho2a) by spectroscopically analysing the protonated and deprotonated states of the Schiff base in the wild type and in single point mutants. By interpretating the spectroscopy data in the light of hybrid quantum mechanics/molecular mechanics (QM/MM) simulations, we demonstrate that a chloride anion (Cl⁻) serves as a counterion to the retinylidene Schiff base in animal opsins, specifically in visible light-sensitive opsins of the ASO-II group.

Results

Identification of Acropora tenuis opsins

We identified 17 opsins from the Acropora tenuis genome and transcriptome datasets by homology search, which included eight opsins in Gs-coupled cnidopsin group, one opsin in the ASO-I group, and eight opsins in the ASO-II group (Fig. S1; Fig. 1A). Full-length cDNAs of seven out of the eight opsins in the ASO-II group were isolated and cloned from adult or larval tissues of the coral (highlighted by bold letters in Fig. 1A). We failed to amplify one opsin in the ASO-II group (gene model ID: aten_s0263.g14) by RT-PCR possibly because of its little mRNA expression level. Amino acid sequence alignment shows that all the seven A. tenuis opsins in the ASO-II group lack a glutamic or aspartic acid at the established counterion positions 94, 113 or 181 (Fig. S2). These opsins also have no E(D)RY motif at the cytoplasmic end of helix 3, which is conserved throughout most class A GPCRs (Hofmann et al., 2009).

Fig. 1.

Absorption spectra of Acropora tenuis opsins in the ASO-II group

We expressed seven members of the ASO-II group in COS-1 cells and purified their recombinant pigments in detergent-solubilised conditions. We successfully obtained the absorption spectra of three (Antho2a, Antho2c and Antho2e) out of the seven members, which showed that Antho2a and Antho2c are visible light-sensitive opsins having with λ_max at 503 nm and 450 nm, respectively, whereas Antho2e is a UV-sensitive opsin with λ_max at ∼360 nm (Fig. 1B). We have previously reported that one opsin in the ASO-II group, acropsin 4 of the coral Acropora millepora, induces a light-dependent elevation of intracellular Ca²⁺ levels (Mason et al., 2023). Here, we showed that Antho2a, Antho2c and Antho2e evoked a similar light-dependent increase of Ca²⁺ levels in HEK293S cells (Fig. 1C).

Search for the counterion in Acropora tenuis opsins of the ASO-II group

Antho2a and Antho2c form visible light-sensitive pigments in the dark (Fig. 1B) despite the lack of a negatively charged counterion at any of the established positions (Fig. S2). To investigate how the protonated Schiff base is stabilised in these opsins, we studied in more detail Antho2a (λ_max = 503 nm), as it could be expressed well in cultured cells and was stable in detergent-solubilised conditions (Fig. 1B).

(a) Contribution of Glu292 to the absorbance spectra of the dark state and photoproduct of Antho2a

First, we searched for potential counterions at positions different from known established amino acid sites (91, 113, and 181) in the Antho2a sequence. Using the crystal structure of bovine rhodopsin (PDB ID: 1U19) as a template, we identified glutamic or aspartic acids located within 5Å of the Schiff base in Antho2a and other members in the ASO-II group. Notably, all A. tenuis opsins in this group contain a conserved glutamic/aspartic acid at position 292 (Fig. S2), positioned just one helix turn away from the retinal-binding residue Lys296. To determine whether Glu292 could function as the counterion in Antho2a, we mutated Glu292 to alanine and measured the absorption spectra. The absorption spectrum of the E292A mutant in the dark was nearly identical to that of wild type (Fig. 2A, B, curve 1), exhibiting a clear absorbance in the visible light region with only a slightly red-shifted λ_max (505 nm) at 140 mM NaCl and pH 6.5. This shows that a negative charge other than Glu292 may serve as a counterion in the dark state of wild type Antho2a.

Fig. 2.

We next investigated the spectroscopic properties of the photoproduct of wild type Antho2a and the E292A mutant. Upon irradiation of wild type Antho2a with orange light, the λ_max shifted from 503 nm in the dark to 476 nm in the photoproduct (Fig. 2A, curve 2; Fig. S3A, curves 2 and 3). This shift is due to the photoisomerisation of the 11-cis retinal chromophore to its all-trans form, converting almost 100% of the dark state to the photoproduct (Fig. S3B). The photoproduct remained stable for at least 5 minutes (Fig. S3A, curve 2 and curve 3), but it exhibited a gradual decay (Fig. S3C, D), indicating that purified Antho2a is neither strictly bleach resistant nor bistable (see also Fig. S3 legend). In contrast to the dark state, the photoproduct of the E292A mutant displayed two distinct absorption peaks in UV and visible light regions, at ∼370 nm and 476 nm respectively (Fig. 2B, curve 2). This suggests that the E292A mutation causes UV-light absorption due to a deprotonated Schiff base in the photoproduct. Additionally, altering the pH modified the ratio of absorbance between the ∼370 nm and 476 nm peaks in the E292A mutant (Fig. S4B, curves 2), with the UV-peak to the visible light-peak ratio increasing at higher pH levels (pH 7.5, Fig. S4B, curve 2). Conversely, the wild type did not exhibit an increase in UV absorbance under similar high pH condition (pH 7.5, Fig. S4A, curve 2). These results indicate that the Schiff base in the photoproduct of the Antho2a E292A mutant has a lower acid dissociation constant (pK_a) than that of the wild type, suggesting that Glu292 acts as the counterion in the photoproduct of Antho2a.

We then further explored the nature of the counterion in the dark state of Antho2a. Previous studies have shown that in the bovine rhodopsin E113A and E113Q mutants, as well as in the retinochrome E181Q mutant (referred to as “counterion-less” mutants), halide ions like Cl⁻ can act as “surrogate” counterions to stabilise the proton on the Schiff base. Consequently, these counterion-less mutants can still absorb visible light in the presence of Cl⁻ (Nathans, 1990; Sakmar et al., 1991; Terakita et al., 2000). To assess the potential role of Cl⁻ as a surrogate counterion in the dark state of the Antho2a E292A mutant, we performed spectroscopic analyses under Cl⁻-depleted conditions (Fig. 2C, D). We observed that the λ_max of the E292A mutant shifted to the UV region (Fig. 2D, curve 1). Unexpectedly, a similar shift in absorption to the UV region was also observed in the wild type under the Cl⁻-depleted condition (Fig. 2C, curve 1). These results indicate that, in the absence of Cl⁻, the Schiff base in both the wild type and the E292A dark states becomes deprotonated. The subsequent addition of Cl⁻ (final concentration: 140 mM NaCl) restored clear absorbance in the visible light region (Fig. 2C, D, curves 2), showing that Cl⁻ facilitates the protonation of the Schiff base of the dark state even in the wild type. In contrast, the photoproduct of the wild type exhibited no significant change in the ratio of UV to visible-light absorption peaks at pH 6.5 across NaCl concentrations from 0.28 mM to 800 mM (Fig. S5). The wild type consistently absorbed visible light under these NaCl conditions (curves 2 in Fig. S5A, B), suggesting that Cl⁻ has little impact on the Schiff base pK_a in the photoproduct of wild type Antho2a. However, the photoproduct of the E292A mutant exhibited a pH-dependent shift in the ratio of UV to visible-light absorption between pH 4.8 and pH 7.6, even at 800 mM NaCl, where the dark state predominantly absorbed visible light (Fig. S5C). This further supports that Glu292 serves as the counterion in the photoproduct of Antho2a.

(b) Effect of halide anions on λ_max values of the dark state of Antho2a

To obtain a further evidence supporting the Cl⁻ counterion in the dark state of Antho2a, we examined the impact of different halide anions on the absorption spectrum in the dark state of Antho2a, as observed in the bovine rhodopsin counterion-less mutant (Nathans, 1990; Sakmar et al., 1991). Antho2a readily absorbed visible light in the presence of bromide ion (Br⁻) and iodide ion (I⁻) as well as Cl⁻ and the λ_max of wild type Antho2a shifted depending on the halide solutions (503 nm in 140 mM NaCl; 506 nm in 140 mM NaBr; 511 nm in 140 mM NaI solutions; Fig. S6A). The E292A mutant showed a similar shift in λ_max (505 nm in 140 mM NaCl; 507 nm in 140 mM NaBr; 517 nm in 140 mM NaI solutions; Fig. S6B).

(c) Effect of Cl⁻ concentration on the pK_a of the protonated Schiff base of Antho2a

To further investigate the influence of Cl⁻ on the protonation state of the Schiff base in the dark state of Antho2a, we estimated the pK_a of the Schiff base by measuring the pH dependent-changes in the absorption spectra of Antho2a at different Cl⁻ concentrations. The pH-dependent equilibrium between the visible (protonated Schiff base) and UV (deprotonated Schiff base) forms revealed that their ratio changes with Cl⁻ concentration (Fig. S7A–E). A plot of the changes in absorbance at λ_max against pH (Fig. 3A) shows that in wild type Antho2a, the pK_a of the protonated Schiff base increases with higher Cl⁻ concentrations (7.3 at 0.28 mM NaCl, 8.0 at 2.8 mM, 8.8 at 28 mM, 8.8 at 140 mM, and 9.0 at 500 mM). We failed to determine the pK_a at 0 mM NaCl, as the observed λ_max in acidic conditions (pH < 6.5) were shorter than expected in Antho2a (503 nm), suggesting that a normal pigment was not produced under these conditions (Fig. S8A). Similarly, the Cl⁻ concentration also affected the pK_a of the protonated Schiff base in the E292A mutant (6.1 at 2.8 mM NaCl, 6.8 at 28 mM, 7.7 at 140 mM, and 8.9 at 500 mM) (Fig. 3B; Fig. S9). At 0 mM NaCl, the E292A mutant showed no visible light absorption, even under the most acidic conditions (pH 4.7), preventing the determination of its pK_a (Fig. S8B). Notably, at low Cl⁻ concentrations (2.8 mM NaCl), the wild type exhibited a higher pK_a than the E292A mutant (8.0 and 6.1, respectively). However, at 500 mM NaCl, the pK_a of the E292A mutant and wild type were comparable (9.0 and 8.9, respectively; Fig. 3A, B). These results suggest that Cl⁻, rather than Glu292, serves as the counterion in the dark state of Antho2a, while Glu292 facilitates the protonation of the Schiff base by the Cl⁻ counterion.

Fig. 3.

(d) Binding affinity of Cl⁻ to wild type Antho2a and the E292A mutant

To evaluate the Cl⁻ binding affinities of both wild type Antho2a and the E292A mutant, we measured changes in their absorption spectra by gradually increasing Cl⁻ concentrations at pH 6.5 and estimated the Cl⁻ dissociation coefficients (K_d). The relative absorbance in the visible region increased with higher Cl⁻ concentrations both in the wild type and the E292A mutant (Fig. S10A, B). By fitting the Hill equation to the experimental data (Fig. 3C), the dissociation constants (K_d) of Cl⁻ were determined to be 0.079 ± 0.010 mM for the wild type Antho2a and 12.7 ± 0.519 mM for the E292A mutant. This significant increase in the K_d value for the E292A mutant suggests that the Cl⁻ binding affinity is considerably reduced due to the mutation. Consequently, we suggest that while Glu292 does not act as a direct counterion, it plays a crucial role in facilitating Cl⁻ binding to Antho2a.

Structural modelling and QM/MM calculations of the dark state of Antho2a

To gain a deeper understanding of the environment surrounding the retinylidene Schiff base in the dark state of Antho2a, we performed QM/MM-based structural modelling of both the wild type Antho2a (with Glu292 either neutral or negatively charged) and the E292A mutant. The QM/MM geometry optimisation positioned the Cl⁻ ion close to the Schiff base (∼3Å) and near Glu292 (∼4.7Å), with Glu292 itself located in proximity to the Schiff base (∼3.3Å) (Fig. 4A, B). To validate these models, we calculated the QM/MM vertical excitation energies of the ground state geometries (Table S1).

Fig. 4.

For wild type Antho2a with a protonated neutral Glu292, the calculated λ_max using the CAM-B3LYP/cc-pVTZ level of theory was 503 nm (Fig. 4A), in good agreement with the experimentally observed value (503 nm; Fig. 2A). In contrast, the λ_max calculated with a deprotonated negatively charged Glu292 was blue-shifted to 415 nm (Fig. 4B), deviating significantly from the experimental value. Finally, the calculated λ_max for the E292A mutant was 499 nm (Fig. 4C), also in agreement with the experimental value (505 nm). To further substantiate these findings, we recalculated the excitation energies using the RI-ADC(2)/cc-pVTZ method. Although these λ_max values are blue-shifted compared to those calculated with the CAM-B3LYP method, they followed a similar trend. Both these computational methods have previously been employed to accurately calculate the excitation energies of rhodopsins (Church et al., 2021). These results strongly suggest that in the dark state of Antho2a, Glu292 is protonated and neutral at pH 6.5, and therefore, it does not function as the counterion.

Effect of the Glu292 mutation on the function of the photoproduct

The spectroscopy data indicate that Glu292 is involved in stabilising the protonated Schiff base by facilitating Cl⁻-binding in the dark state and also serves as a counterion in the photoproduct. This suggests that Glu292 significantly contributes to the visible light absorption of Antho2a. To explore additional roles of Glu292 in Antho2a, we measured the light-induced Ca²⁺ response in cultured cells expressing wild type Antho2a or the E292A mutant. Notably, cells expressing wild type Antho2a showed a ∼30-fold increase in Ca²⁺ levels upon light irradiation (Fig. 5, solid black circles), whereas cells expressing the Antho2a E292A mutant showed a smaller Ca²⁺ elevation (< 5-fold increase) (Fig. 5, red open circles). This indicates that the peak Ca²⁺ response in cells expressing wild type Antho2a was approximately nine times greater than in cells expressing the E292A mutant. This result, along with the crucial role of Glu292 in Cl⁻ binding in the dark state and as a counterion in the photoproduct, suggests that Glu292 plays also a role in G protein activation.

Fig. 5.

Cl⁻-dependent changes in the absorption spectra of the dark states of Antho2c and Antho2e

We tested whether Cl⁻ concentration affects the pK_a of the Schiff base in another visible light-sensitive opsin, Antho2c (λ_max = 450 nm, Fig. 1B). The pH-dependent equilibrium between UV- and visible-light absorbing forms was clearly observed at 0 mM or 0.093 mM NaCl, but not at 9.3 mM NaCl, where Antho2c stably absorbed visible light across the measured pH range (pH 4.8 to 7.2, Fig. 6A–C). Also, the ratio of UV to visible-light absorption increased with higher Cl⁻ concentrations at pH 6.5 (Fig. 6D). These results demonstrate that Cl⁻ serves as a counterion in the dark state of Antho2c, as it does in Antho2a. In contrast, wild type Antho2e continues to absorb UV light even at 1 M NaCl (Fig. 6E). Notably, Antho2e has an arginine at position 113, which corresponds to the counterion position in vertebrate visual opsins (Fig. S2). When this arginine is mutated to alanine (R113A), the mutant becomes sensitive to visible light (λ_max = ∼420 nm) in the presence of Cl⁻ (Fig. 6F), suggesting that Cl⁻ can serve as the counterion in the R113A mutant.

Fig. 6.

Discussion

In this study, we reveal for the first time the spectral properties of opsins in the ASO-II group from the coral Acropora tenuis, showing that their sensitivity spans from UV to visible light. Opsins in this group have a highly conserved Glu292 residue near the Schiff base, which can potentially stabilise the proton on the Schiff base. Indeed, our results show that the pK_a of the protonated Schiff base in the photoproduct (λ_max = 476 nm) of Antho2a is altered when Glu292 is substituted with alanine (Fig. 2A, B; Fig. S4), suggesting that Glu292 serves as the counterion of the photoproduct. Conversely, the dark state of Antho2a (λ_max = 503 nm) exhibits robust visible light absorption only in the presence of Cl⁻ at physiological pH, and the pK_a of the protonated Schiff base changes with Cl⁻ concentration in both wild type Antho2a and the E292A mutant. Furthermore, the pK_a for wild type and the E292A mutant are comparable in the presence of sufficient Cl⁻ (500 mM NaCl, Fig. 3A, B), supporting the conclusion that Cl⁻, and not Glu292, acts as the counterion of the dark state of A. tenuis Antho2a. We found that the type of halide anions in the solution affected the λ_max values of the dark state of Antho2a, which is consistent with the observed effect on the λ_max of the bovine rhodopsin counterion-less mutant, in which halide ion serves as a surrogate counterion (Nathans, 1990; Sakmar et al., 1991), further substantiating our hypothesis that a solute anion acts as the counterion in the dark state of Antho2a. Moreover, QM/MM calculations for the dark state of Antho2a suggest that Glu292 is protonated and neutral, further supporting the hypothesis that Glu292 does not serve as the counterion in the dark state. Along with the result showing Cl⁻ has little to no effect on the visible light absorption of the photoproduct (Fig. S5), unlike the wild type, we conclude that Cl⁻ and Glu292, respectively, act as counterions for the protonated Schiff base of the dark state and photoproduct of Antho2a. This represents a unique example of counterion switching from exogeneous anion to a specific amino acid residue upon light irradiation.

Our spectroscopic data also showed that the other visual light sensitive opsin, Antho2c, exhibited the Cl⁻ dependency on the Schiff base pK_a of the dark state, which suggesting that opsins in the ASO-II group may share a spectral tuning mechanism based on the Cl⁻ counterion. Interestingly, Antho2e had an arginine at position 113 and when it was mutated to alanine (R113A), the mutant showed sensitivity to visible light in the presence of Cl⁻ (Fig. 6F). We hypothesise that the positive charge of Arg113 disturbs interaction between the Cl⁻ and the protonated Schiff base or that it completely inhibits Cl⁻ binding, rendering Antho2e UV-sensitive.

It is noteworthy that although Cl⁻ has been reported to serve as a surrogate counterion in “counterion-less” mutants of animal opsins (such as E113Q in bovine rhodopsin and E181Q in retinochrome) (Nathans, 1990; Sakmar et al., 1991; Terakita et al., 2000), Antho2a is, to our knowledge, the first example in a wild type animal opsin that employs Cl⁻ as a counterion. Interestingly, within the microbial rhodopsin family, heliorhodopsin (TaHeR) incorporates a Cl⁻ into the Schiff base region under high Cl⁻ concentrations at pH 4.5. This Cl⁻ stabilises the protonated retinal Schiff base when its primary counterion, E108, is neutralised (Besaw et al., 2022). The observed 18 nm red shift at low pH is consistent with E108 protonation. The TaHeR E108A mutant shows the same λ_max under high Cl⁻ concentrations, further supporting the role of Cl⁻ as a counterion. At pH 8, the Schiff base proton is stabilised by the negatively charged E108 (Shihoya et al., 2019).

The E292A mutation in Antho2a drastically decreases the Cl⁻ binding affinity in the dark state at pH 6.5 (K_d: Antho2a WT = 0.079 mM, Antho2a E292A = 12.7 mM). Based on these results, supported by our QM/MM calculations of the dark state, we hypothesise that protonated Glu292 not only serves as a counterion of the photoproduct but also constitutes part of the Cl⁻ binding site in the dark state of Antho2a. In Cl⁻-pumping microbial rhodopsins, where Cl⁻ serves as a counterion, the Cl⁻ binding site near the Schiff base is typically formed by hydrogen-bonding between Cl⁻, the protonated Schiff base, and key amino acids such as serine, threonine and aspartic acid, and water molecules (Besaw et al., 2020; Hosaka et al., 2016; Kolbe et al., 2000). Additionally, the bovine rhodopsin double mutant E113Q/A292E exhibits a higher sensitivity to hydroxylamine than the wild type, indicating instability of the Schiff base and increased solvent accessibility (Tsutsui and Shichida, 2010). These observations suggest that Glu292 in Antho2a may also facilitate the accessibility of Cl⁻ ions to the Schiff base environment. To determine the precise nature of the Cl⁻ binding site and clarify the roles of Glu292 and other residues in the Cl⁻-dependent counterion system of Antho2a, detailed spectroscopic and structural experiments will be necessary.

We also found that the cells expressing the Antho2a E292A mutant show a lower Ca²⁺ elevation upon light stimulation compared to the wild type (Fig. 5). This effect could be due to Glu292 not only serving as a counterion in the photoproduct but also allosterically influencing G protein activation. It has been reported that the introduction of Glu or Asp at position 292 can affect various molecular properties of opsins. For instance, the A292E mutation in human rhodopsin has been shown to result in constitutively active apoproteins, leading to congenital night blindness (Dryja et al., 1993; Jin et al., 2003). These data suggest that Glu292 has the potential to influence functional properties, such as G protein activation. Several studies have reported that the counterion can affect diverse properties of opsins beyond their absorption spectra. For instance, in bovine rhodopsin, Glu113 is involved not only in visible light absorption in the dark state but also in the efficient activation of G proteins by the photoproduct (Terakita et al., 2004). It has been suggested that these pleiotropic functions of the counterion have been achieved through evolutionary modifications of the protein structure once the counterion was acquired (Terakita et al., 2004). This concept may similarly apply to Antho2a.

One of the notable features of the opsins in the ASO-II group that use chloride as a counterion is the relatively low pK_a of the Schiff base (e.g., ∼9.0 for Antho2a in the dark state, Fig. 3A) compared to other animal opsins with a “regular” amino acid counterion (e.g., >16 for bovine rhodopsin, with Glu113 as a counterion (Steinberg et al., 1993) or 11 for jumping spider Rh1, with Glu181 as a counterion (Nagata et al., 2019)). We hypothesise that opsins in the ASO-II group, with lower Schiff base pK_a, may change their spectral sensitivity and G protein activation profile within the physiological pH range of corals. For instance, these opsins could exhibit decreased sensitivity to visible and increased sensitivity to UV light under alkaline pH conditions. The extra- and intracellular pH environments in symbiotic cnidarians such as corals are spatially and temporally variable due to the photosynthesis of symbiotic algae (Barott et al., 2017). For example, pH values in the cytosol of algae-hosting cells (a group of endodermal cells with algal occupancy) in corals reportedly increase by approximately 0.5 pH units upon light treatment (Venn et al., 2009). The pH values in the gastrovascular cavity (coelenteron), which is in contact with endodermal cells, range from 6.6 to 8.5 (Agostini et al., 2012; Al-Horani et al., 2003; Cai et al., 2016) and increase in the presence of light, presumably due to the consumption of CO₂ by photosynthesis (Barott et al., 2017). Additionally, alkalinisation in extracellular region of calcifying cells reportedly increases ∼1.0 pH units from dark to light conditions reaching levels above pH 9.0 (Al-Horani et al., 2003). These pH changes, generally an increase due to photosynthetic activity, may result in variable light sensitivity of the “low pK_a” opsins in the ASO-II group. Recent studies have reported that members of the ASO-II group may be associated with the symbiotic relationship between host anthozoans and symbiotic algae. For example, using the symbiotic sea anemone Exaiptasia diaphana, Gornik et al. (2021) showed that the mRNA levels of several opsins in the ASO-II group were higher in symbiotic adults than in apo-symbiotic ones (Gornik et al., 2020). In sea anemone, it has also been reported that behavioural responses to light differ between symbiotic and apo-symbiotic individuals (Foo et al., 2020; Kishimoto et al., 2023), with these responses potentially being driven by the activation of ASO-II opsins that are up-regulated in the presence of symbiotic algae. Moreover, single-cell RNA-seq analysis in the reef-building coral Stylophora pistillata has revealed that an opsin belonging to the ASO-II group is specifically expressed in algae-hosting cells (a group of endodermal cells with 50% algal occupancy) (Levy et al., 2021). Although further studies are needed, we suggest that the unique use of Cl⁻ as the counterion in opsins of ASO-II group, rather than a negatively charged amino acid, may be associated to their pH-sensitive light response and, ultimately, to their role in photosynthesis-related functions in symbiotic cnidarians.

It is widely accepted that opsins have evolved from a non-opsin GPCR (Feuda et al., 2012). Our finding of the native chloride counterion in opsins shows a possibility that in the first stage of the evolutionary process, the “primitive” opsins having Lys296 but no counterion residue could absorb not only UV light but also a wide range of wavelengths that extends into the visible region by embracing chloride ions. Namely, the chloride counterion system might have been a preliminary step in the evolution of amino acid counterions in animal opsins. Further empirical and bioinformatic studies are required for disentangling the evolutionary trajectories of the Schiff base – counterion system, including the chloride counterion.

Materials and Methods

Experimental design

We first identified and cloned opsins from a reef-building coral, Acropora tenuis, and then expressed opsins belong to the ASO-II group in mammalian cultured cells. We performed spectroscopic measurements of purified pigments of the opsins in different pH and Cl⁻ conditions to identify their effects on the acid dissociation constant of the protonated Schiff base of the opsins, leading to the determination of the counterion. Computational modelling and QM/MM calculations were also conducted to elucidate the retinylidene Schiff base environment in the dark state of Antho2a. Light-evoked Ca²⁺ responses were assessed by aequorin-based bioluminescent reporter assay to evaluate the G protein activation of Antho2a.

Identification of Acropora tenuis opsins and phylogenetic tree inference

Acropora tenuis (Dana, 1846) is a common reef-building coral distributed throughout the Indo-Pacific Ocean. Candidate sequences of A. tenuis opsins were identified by homology search against public genome and transcriptome datasets (Shinzato et al., 2020; Voolstra et al., 2015) and their phylogenetic relationships to known opsins were inferred by subsequent phylogenetic tree reconstruction. We first conducted BLASTP and TBLASTN searches with an E-value cut-off of 10⁻¹⁰ using Acropora palmata Acropsin 1–3 (JQ966100 – JQ966102), two Nematostella vectensis opsins (BR000676 – BR000677), human rhodopsin (NM_000539), and squid rhodopsin (X70498) as queries. We aligned the collected opsin homologs and excluded sequences which did not contain a retinal-binding lysine residue (Lys296) in the seventh transmembrane helix. We modified the fragmented sequences by reference to the genome sequence of A. tenuis and opsin sequences of other Acropora species (A. palmata or A. millepora). The candidate sequences of A. tenuis opsins were combined with the representative opsin sequences. The final sequence set was aligned using MAFFT (Katoh and Standley, 2013) and trimmed by TrimAl (Capella-Gutiérrez et al., 2009) with the “gappyout” function. The ML tree was reconstructed using RAxML-NG v1.1.0 (Kozlov et al., 2019) assuming the LG + G4 model of protein evolution, which was selected by MoldelTest-NG v0.2.0 (Darriba et al., 2020). The ML branch supports were estimated with 1000 bootstrap replicates.

Sample collection, total RNA extraction, and cDNA synthesis

Colonies of A. tenuis were collected from < 3 m depth on the fringing reef on Sesoko Island, Okinawa (N26°37.58′, E127°52.01′) and were maintained in flow-through aquaria at Sesoko Station (Tropical Biosphere Research Center, University of Ryukyus, Okinawa, Japan). Four days after spawning, motile larvae and small branches of adult colonies were preserved in RNAlater® Stabilization Solution (Thermo Fisher Scientific, MA, USA). Total RNAs were extracted from the larval and adult samples using TRIzol reagent (Thermo Fisher Scientific) or Sepasol-RNA I Super G (nacalai tesque, Kyoto, Japan) and purified using Qiagen RNeasy mini kits (Qiagen, Hilden, Germany) following the manufacturer’s protocol. cDNAs were synthesised from the total RNA by reverse transcription using High-Capacity cDNA Reverse Transcription kits (Thermo Fisher Scientific).

Expression and purification of Acropora tenuis opsins

The coding regions of A. tenuis opsins were amplified by PCR with gene-specific primers and were tagged with the epitope sequence of the anti-bovine rhodopsin antibody Rho1D4 (ETSQVAPA) at their C-termini. Site-directed mutants were produced by overlap extension PCR using PrimeSTAR Max DNA Polymerase (TAKARA, Shiga, Japan) with site-specific primers and were also tagged with the 1D4 epitope sequence. The tagged cDNAs were inserted into the pUSRα vector (Kayada et al., 1995) digested with HindIII and EcoRI or the pMT vector (Ridge and Abdulaev, 2000) digested with EcoRI and NotI using In-Fusion HD cloning kit (TAKARA). The plasmids (15 µg per 100 mm culture dish) were transfected into COS-1 cells using the polyethyleneimine (PEI) transfection method. The transfected cells were maintained for 24 h after transfection at 37°C under 5% CO₂ and then 11-cis retinal was added to the medium (1 µL of 4 mM 11-cis retinal to 100 mm culture dish) following 25°C or 30°C incubation for another 24 h in the dark before collecting the cells. The reconstituted pigments were extracted from the cell membranes with 1% dodecyl β-D-maltoside (DDM, Dojindo, Kumamoto, Japan), 50 mM HEPES, and 140 mM NaCl (pH 6.5). The solubilised samples were mixed with 1D4-condjugated agarose beads overnight, and the mixture was transferred into Bio-Spin columns (Bio-Rad, Hercules, CA, USA) and washed in the buffer containing 0.02% DDM, 50 mM HEPES, and 140 mM NaCl (pH 6.5, buffer A). The purified pigments were eluted with buffer A containing 0.5 ∼ 1 mg/mL 1D4 peptide (custom peptide synthesis by GenScript Japan Inc., Tokyo, Japan). To obtain pigments in solutions of various anions (SO₄²⁻, Br⁻, and I⁻) other than Cl⁻, samples were prepared as described above and in the final step the mixture of solubilised samples and 1D4-agarose beads was washed with buffer A followed by the additional wash with buffers including different sodium salts of anions (0.02% DDM, each of 70 mM Na₂SO₄, 140 mM NaBr, or 140 mM NaI, and 50 mM HEPES). Then the pigments were eluted with the buffer including the appropriate sodium salt of anion containing 0.5 ∼ 1 mg/mL 1D4 peptide. Alternatively, for some pigments that were unstable under the absence of Cl⁻, we quickly removed Cl⁻ by gel-filtration chromatography on PD MiniTrap desalting columns with Sephadex G-25 resin (Cytiva, Marlborough, MA, USA). The columns were first equilibrated with the buffer including 0.02% DDM, 70 mM Na₂SO₄, and 50 mM HEPES, 500 µL of samples were loaded onto the columns and eluted with the buffer. We collected 800 µL fractions and used them for subsequent spectroscopic analyses.

UV-visible spectroscopy

Spectroscopic measurements were performed at 0°C using a V-750 UV-visible spectrophotometer (JASCO Corporation, Tokyo, Japan). The pH of the samples was adjusted with 100 mM CAPS including NaOH for alkaline conditions and 500 mM NaH₂PO₄ for acidic conditions. pH values were measured using a pH meter (B-211; HORIBA, Kyoto, Japan) immediately after each spectroscopic measurement. The concentration of Cl⁻ in the samples was adjusted by addition of different concentrations of NaCl solutions which were prepared in 70 mM Na₂SO₄ buffer (see above). A 100-W halogen lamp was equipped on the spectrophotometer and used to illuminate samples with a set of optical interference filters (420 nm, Toshiba) and cut-off filters (O-55 or O-56, AGC Techno Glass Co., Shizuoka, Japan). Absorption spectra of some UV-absorbing pigments were recorded using the V-750 UV-visible spectrophotometer, equipped with a 300-W xenon lamp (MAX-350; Asahi Spectra Co., Tokyo, Japan) that was used for illumination of samples in combination with a UV-transmitting filter (UTVAF-50S-36U, SIGMA KOKI, Tokyo, Japan).

HPLC analysis

An HPLC analysis was carried out to analyse the conformations of retinal present in the purified pigments. 100 µL of purified pigments were mixed with 210 µL of cold 90% methanol which was stored in −20°C and 30 µL of 1 M hydroxylamine to convert retinal chromophore in a sample into retinal oxime. The retinal oxime was extracted with 700 µL of n-hexane. 200 µL of the extract were injected into a YMC-Pack SIL column (particle size 3 μm, 150 × 6.0 mm) and eluted with n-hexane containing 15% (v/v) ethyl acetate and 0.15% (v/v) ethanol at a flow rate of 1 mL/min.

Bioluminescent reporter assays for Ca²⁺ measurements in cultured cells

Ca²⁺ levels in opsin-expressing cultured cells were assessed by an aequorin-based luminescent assay as described previously (Koyanagi et al., 2022). Briefly, the plasmid containing open reading frames of each opsin was transfected into HEK293S cells in 35 mm dishes by the PEI method with the aequorin plasmid obtained by introducing a reverse mutation A119D into the plasmid [pcDNA3.1+/mit-2mutAEQ] (Addgene no. 45539) (de la Fuente et al., 2012). The transfected HEK293S cells were incubated for ∼24 hours at 37°C under 5% CO₂ with the addition of 0.2 µM/dish of 11-cis retinal 4 ∼ 5 hours after the transfection. Before the luminescence measurements, the culture medium was replaced with a medium containing Coelenterazine h, and the cells were incubated to equilibrate with the media at 25°C for at least 2 hours. Dishes of cells were then stimulated with light and luminescence values were recorded using GloMax 20/20n Luminometer (Promega). A green (495 nm) LED light (colour: “Cyan”, Ex-DHC; BioTools Inc., Gunma, Japan) and arrayed LEDs on a board with spectral emission peaks at 390 nm and 510 nm (SPL-25-CC; REVOX Inc., Kanagawa, Japan) were used as light sources.

Computational modelling and QM/MM calculations

The three-dimensional structure of Antho2a was predicted from the primary amino acid sequence using AlphaFold2 (Jumper et al., 2021). The 11-cis retinal chromophore linked to the protonated Schiff base was incorporated into the AlphaFold model using as a template the high-resolution structure of bovine rhodopsin solved by time-resolved serial femtosecond X-ray crystallography (Gruhl et al., 2023). A Cl⁻ anion was initially placed in close proximity to the retinal protonated Schiff base, as observed in the microbial chloride-pump halorhodopsin (Mous et al., 2022). Water molecules were added using HomolWat (Mayol et al., 2020). We determined the pK_a values of the titratable amino acid residues at pH 6.5 using the PROPKA program (Olsson et al., 2011; Søndergaard et al., 2011) and subsequently, the protein was protonated using the tleap program in the AMBER software package (Case et al., 2016). The geometry of this initial model was first relaxed by molecular mechanics energy minimisation with the Amber ff14SB force field (Maier et al., 2015) using steepest descent for 10,000 steps before switching to a conjugate gradient minimiser for an additional 10,000 steps. During energy minimisation, a positional restraint of 10 kcal mol^-1Å^-2 was applied to all atoms, including hydrogens. The SHAKE algorithm (Ryckaert et al., 1977) was used to constrain the motion of bonds involving hydrogen. Finally, the geometry of the system was optimised using hybrid quantum mechanics/molecular mechanics (QM/MM) calculations without considering any external environment and with the backbone of the protein frozen (Mroginski et al., 2021; Senn and Thiel, 2009). The QM part consists of the retinal chromophore linked to the lysine sidechain cut between the Cδ and Cε atoms forming the protonated Schiff base, along with Cl⁻, Glu292 and Ser186. The hydrogen link atom (HLA) scheme was used at the QM/MM boundary. The QM part was treated using the BP86-D3 (BJ) functional (Becke, 1988; Grimme et al., 2011) in conjunction with the cc-pVDZ basis set (Dunning, 1989) and the def2/J auxiliary basis set for the resolution of identity (RI) (Weigend, 2008). The Chain of Spheres exchange (COSX) algorithm was utilised in combination with the resolution of identity for the Coulomb term (RI-J). The rest of the protein was treated with the Amber ff14SB force field. Water molecules were treated with the TIP3P model(Jorgensen et al., 1983). The QM/MM calculations were performed using the quantum chemistry program Orca 5.0.2 (Neese, 2022) interfaced with the DL_POLY module of the ChemShell 3.7.1 software package (Metz et al., 2014; Sherwood et al., 2003). The optimised ground state geometries and partial charges were used to calculate the vertical excitation energies using the simplified time-dependent density functional theory (sTD-DFT) (Bannwarth and Grimme, 2014; Runge and Gross, 1984) at the CAM-B3LYP/cc-pVTZ level of theory (Dunning, 1989; Yanai et al., 2004) using the Orca program. The excitation energies were also calculated using the RI-ADC(2) method (Hättig, 2005) with frozen core orbitals and the cc-pVTZ basis set in association with the corresponding auxiliary basis by utilising the Turbomole 7.5.1 program package (Furche et al., 2014). The three-dimensional models were visualised using the molecular graphics program PyMOL 2.5.5.

Supplementary table and figures

Table S1.

Fig. S1.

Fig. S2.

Fig. S3.

Fig. S4.

Fig. S5.

Fig. S6.

Fig. S7.

Fig. S8.

Fig. S9.

Fig. S10.

Data and materials availability

The cDNA sequences of Acropora tenuis opsins in this paper are available in GenBank: eight Cnidopsins (accession no. LC844924–LC844931), seven opsins in the ASO-II group (accession no. LC844932–LC844938), and one opsin in the ASO-I group (accession no. LC844939). All data needed to evaluate the conclusion in this paper are present in the paper and the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Acknowledgements

We thank Dr. Robert S. Molday (University of British Columbia) for kindly supplying rho 1D4-producing hybridoma. We thank Dr. Masayuki Hatta (Ochanomizu University) for coral sampling. This work was supported by the Japanese Ministry of Education, Culture, Sports, Science and Technology Grants-in-Aid for Scientific Research 23H02516 (to AT), 22H02663 (to MK) and JP20J01841 (YS); and Japan Science and Technology Agency (JST) Core Research for Evolutional Science and Technology (CREST) Grant JPMJCR1753 (to AT). YS was supported by Grant-in-Aid for JSPS Fellows. This work has also been supported by the Swiss National Science Foundation (project grant 192780 to XD) and by the European Union’s Horizon 2020 research and innovation programme (grant agreement 951644 to GFXS and Marie Skłodowska-Curie grant agreement 884104 (PSI-FELLOW-III-3i) to SS). We finally thank the members of the High-Performance Computing and Emerging Technologies (HPCE) group at the Paul Scherrer Institute for technical support and assistance with high-performance computing.

Additional information

Author contributions

YS, MK, and AT conceptualised the study; YS, MK, GFXS, XD, and AT designed the study; YS, TS, YK, and MI performed the experiments and analyses; SS and XD performed the computational modelling; YS and AT wrote the original draft with inputs from SS and XD; All authors reviewed the manuscript and gave final approval for the manuscript.