1 Introduction

Rhodopsins comprise a large family of retinal-binding proteins found in microbes and animals [1]. Microbial rhodopsins are categorized into different functional subgroups, including light-driven ion pumps, light-gated ion channels, light-sensors and light-activated enzymes [1,2,3,4,5,6], while animal rhodopsins are G protein-coupled receptors [1, 7, 8]. Rhodopsins are key tools in optogenetics that control activities of neurons or other cells by light [9,10,11,12,13,14]. The retinal chromophore of microbial rhodopsins is in an all-trans form (Fig. 1), while that of animal rhodopsin is mostly in an 11-cis form. The retinal molecule forms a Schiff base linkage with a lysine residue in the 7th transmembrane helix (TM7), and the Schiff base is protonated in many cases (Fig. 1) [1,2,3,4,5,6,7,8].

Fig. 1
figure 1

Photoisomerization pathway in microbial rhodopsins, where all-trans PRSB is dominant in the dark state. Upon light absorption, selective isomerization takes place into the 13-cis form of many microbial rhodopsins (green arrow). Exceptional cases can be seen in bestrhodopsin or NeoR, whose photoreaction intermediates contain 11-cis or 7-cis forms, respectively (red arrows)

Upon light absorption, photoisomerization of the protonated retinal Schiff base (PRSB) initiates each function of microbial and animal rhodopsins. The isomerization pathway of PRSB was known to be highly selective in protein; all-trans to 13-cis in microbial rhodopsins and 11-cis to all-trans in animal rhodopsins [1]. Selective isomerization in rhodopsins can take place even at cryogenic temperatures such as 77 K, where molecular motions are severely restricted [1, 15]. Although such specific isomerization was common knowledge in the rhodopsin field, unusual photoisomerization pathways have been recently found in microbial rhodopsins. In 2022, we reported an all-trans to 11-cis isomerization in a microbial rhodopsin, bestrhodopsin (Fig. 1) [16]. In the same year, we also reported an all-trans to 7-cis isomerization in an enzymerhodopsin NeoR (Fig. 1) [17]. Such unique photoisomerization was prohibited at < 150 K for bestrhodopsin, and at < 273 K for NeoR [16, 17]. Both bestrhodopsin and NeoR exhibit absorption maxima at > 650 nm [16, 18], which is also unusual. In rhodopsins, absorption is red-shifted if the counterion of the protonated Schiff base is lengthened and the electrostatic interaction between the Schiff base and counterion becomes weak [1]. All-trans PRSB absorbs maximally at 610 nm in gas phase [19], in the absence of counterion or at infinite distance of counterion, and the value was believed at the longest wavelength for rhodopsins.

These findings raised various questions about color tuning mechanism and photoisomerization pathways of PRSB. In particular, formation of the 7-cis product in NeoR was very unusual. It should be noted that 7-cis retinal is formed by illumination of retinal in solution [20, 21], but it is unstable because of a steric hindrance between the 9-methyl group and methyls in the β-ionone ring (Fig. 1). In contrast, the 7-cis product was stable in NeoR at room temperature [17]. Bovine rhodopsin can accommodate 7-cis forms [22,23,24], whereas none of microbial rhodopsins had the 7-cis form before the report [17]. To compare the photoisomerization pathways between protein and solution, all-trans PRSB formed with butylamine in solution is more important than all-trans retinal. According to the previous HPLC analysis, all-trans PRSB in solution is isomerized into the mixture of cis products (82 % 11-cis, 12 % 9-cis, and 6 % 13-cis in methanol), but 7-cis PRSB was never produced [25].

One possibility is that illumination of all-trans PRSB does not produce the 7-cis form, unlike retinal. Another possibility is that 7-cis PRSB cannot be isolated from all-trans PRSB in the previous HPLC analysis [25]. We suspect the latter being the case, as the illuminated PRSB was directly applied to HPLC in the previous study [25]. On the other hand, in the study of NeoR, we applied retinal oximes to HPLC analysis after the hydroxylamine-bleach, indicating the formation of both syn and anti-peaks for each isomer [17]. As the retention time of the 7-cis syn form was similar to that of the all-trans syn form, our analysis focused on the anti-peak. This paper aimed at analyzing the isomeric ratio of the photoproducts from all-trans PRSB in solution upon light absorption, particularly focusing if the 7-cis form is produced or not. We also studied photoreaction of unprotonated retinal Schiff base (URSB) to compare the results with those of PRSB. The HPLC results clearly demonstrated formation of the 7-cis form in URSB and PRSB, which was possible by the analysis of anti-peaks. Photoisomerization pathways of all-trans URSB and PRSB will be discussed based on the present HPLC analysis.

2 Materials and methods

Sample preparation and illumination. In this paper, we used the previously published HPLC data on truncated OmNeoR for detailed analysis [17]. In addition, all-trans URSB and PRSB were prepared as follows. All-trans retinal powder was first dissolved in acetonitrile, methanol, ethanol, or n-hexane. Then, a fivefold molar excess of n-butylamine was added at 5 °C, forming all-trans URSB. Conversion of URSB to PRSB was performed by adding HCl (final concentration: 0.17 M) into the URSB solution in acetonitrile, methanol, and ethanol. All these processes were carried out under red light conditions.

The URSB or PRSB samples were illuminated by use of 380 nm (KL38 interference filter, Toshiba) or 440 nm (KL38 interference filter, Toshiba) light for 4 min in each solvent, respectively. Absorption spectra were recorded by using a V-650 visible spectrophotometer (JASCO Corp.).

HPLC analysis of the retinal configuration. The configuration of retinal in URSB or PRSB was analyzed by high-performance liquid chromatograph (HPLC), after being converted into retinal oximes by hydroxylamine, according to the method normally applied to rhodopsins [17]. Namely, retinal oxime was prepared by adding hydroxylamine (final conc. 1 M) to 50 μ1 sample solution at 4 °C, followed by addition of 66% (v/v) methanol. The retinal oxime was extracted with hexane, and analyzed by HPLC (solvent composition: 12% (v/v) ethyl acetate and 0.12% (v/v) ethanol in hexane) with 1.0 ml min−1 flow rate. A silica column (6.0_150 mm; YMC-Pack SIL) was used for the analysis. The molar fraction of each retinal isomer was calculated from the ratio of the areas of corresponding peaks in the HPLC patterns, where we used molar extinction coefficient of each isomer at 360 nm in the previous report [26]. To estimate the experimental error, three identical measurements were performed for both dark- and light-adapted samples.

3 Results

Figure 2a–c reproduced the published HPLC data of the truncated OmNeoR protein, which reported the production of the 7-cis form upon light absorption of microbial rhodopsins for the first time [17]. When the chromophore was extracted by hydroxylamine-bleach method from dark-adapted OmNeoR, all-trans peak was predominant for both syn (ts) and anti (ta) forms of the retinal oxime (black line in Fig. 2a). In contrast, when the chromophore was extracted after conversion to P372, a stable photoproduct at room temperature, a peak appeared in the region of anti-forms, whose position was different from those of all-trans (ta), 9-cis (9a), 11-cis (11a), and 13-cis (13a) forms (Fig. 2a and c). By use of NMR spectroscopy, we identified the peak owing to the 7-cis form [17]. It should be noted that the syn peaks of the 7-cis (7s) and all-trans (ts) forms overlap with each other under the standard HPLC conditions (Fig. 2b), while the anti-peaks were well separated between 7-cis (7a) and all-trans (ta) forms (Fig. 2c). Therefore, we used only anti-peaks for the analysis of the previous paper; 92.8 % all-trans, 0.9 % 7-cis, 2.2 % 9-cis, 1.7 % 11-cis, and 2.4 % 13-cis forms in the dark, and 3.7 % all-trans, 90.7 % 7-cis, 1.0 % 9-cis, 3.3 % 11-cis, and 1.3 % 13-cis forms by light [17].

Fig. 2
figure 2

a Photoisomerization of retinal Schiff base in near-infrared light absorbing enzymerhodopsin (NeoR) derived from Obelidium mucronatum. Black, and red lines show HPLC profiles of retinal oxime form in dark and light irradiated (> 660 nm) conditions, respectively. Syn and anti-peaks in retinal oximes appear at the retention time of 6–10 minutes and 11–19 min, respectively. b Expanded figure of 7-cis syn (7s) and all-trans syn (ts) oxime region (retention time: 8.0–10.5 min) in (a). c Expanded figure of 7-cis anti (7a) and all-trans anti (ta) oxime region (retention time: 16.0–18.5 min) in (a)

We then applied HPLC analysis to the chromophore molecules in solution, which provides a good control of those in rhodopsins. In addition to the photoreactions of all-trans PRSB, we also studied those of all-trans unprotonated retinal Schiff base (URSB) in solution. Solvents studied in this paper were acetonitrile, methanol, ethanol, and n-hexane for URSB, and acetonitrile, methanol, and ethanol for PRSB. Black line in Fig. 3a shows absorption spectrum of URSB in acetonitrile, whose λmax is located at 357 nm. Illumination at 380 nm little changes the absorption spectrum (red line in Fig. 3a), but the HPLC analysis will show that photoisomerization takes place from the all-trans form as shown below. This indicates similar absorption spectra for the all-trans and various cis forms. Black line in Fig. 3b shows absorption spectrum of PSB in acetonitrile, whose λmax is located at 441 nm. Acidification of URSB dramatically alters the color from pale yellow (Fig. 3a) to dark orange (Fig. 3b). Illumination at 440 nm little changes the absorption spectrum (red line in Fig. 3b), though the HPLC analysis will show formation of cis products as shown below. This indicates similar absorption spectra for the all-trans and various cis forms.

Fig. 3
figure 3

(a) UV-visible spectra of unprotonated retinal Schiff base (URSB) in acetonitrile. Black and red lines are obtained in the dark and light-irradiation (380 nm) conditions, respectively. The inserted picture shows its color. (b) UV-visible spectra of protonated retinal schiff base (PRSB) in acetonitrile. Black and red lines are obtained in the dark and light-irradiation (440 nm) conditions, respectively. The inserted picture shows its color

We next analyzed HPLC profiles of URSB. Raw data of the HPLC profile for URSB are shown in Fig. S1, where three independent measurements provided similar results. Top panels of Fig. 4 show HPLC profiles for the anti-forms of all-trans URSB in acetonitrile (a), ethanol (b), methanol (c), and n-hexane (d), where black and red lines represent the measurements in the dark and light (380 nm), respectively. In addition to the labeled peaks, an unlabeled peak was observed at ~12 min, whose origin was unclear (Top panel of Fig. 4). As it appears both in the dark and light, we did not take into account the band. Similarly, there was a shoulder peak on the longer reetention time of the 11a peak (Top panel of Fig. 4), which was not used for calculation. Then, the areas of 13-cis, 11-cis, 9-cis, 7-cis and all-trans forms were calculated, whose ratios were determied by using molar extinction coefficient of each isomer at 360 nm. Table 1 shows such isomeric compositions in the dark and light (380 nm), and middle panels of Fig. 4 visualize the isomeric compositions in the dark (black) and light (red). The results indicate that illuminattion converts all-trans URSB into various cis products in all solvents.

Fig. 4
figure 4

HPLC analysis of URSB. (Top panel) HPLC profiles of the retinal oximes (anti-peaks), which are prepared from all-trans URSB in the dark (black line) or after illuminated by a 380 nm light (red line) in acetonitrile (a), ethanol (b), methanol (c), and n-hexane (d). (Middle panel) Isomeric compositions of URSB in the dark (black bar) or after illuminated by a 380 nm light (red bar) in acetonitrile (a), ethanol (b), methanol (c), and n-hexane (d). These values are calculated from peak area in the HPLC profile, where three independent experiments are averaged (standard deviations are shown in the panel). (Bottom panel) Isomeric ratio of the photoproducts of all-trans PRSB after illuminated by a 440 nm light (blue bar) in acetonitrile (a), ethanol (b), methanol (c), and n-hexane (d). These values are obtained by subtraction of the dark value (black bar) from the light value (red bar) in each cis peak of the middle panel

Table 1 Isomeric compositions of unprotonated retinal Schiff base (URSB) under dark and light at 380 nm in each solution. Averaged values of three independent measurements are shown with the standard deviations

We then estimated the isomeric ratio in the photoproducts by illumination at 380 nm (Table 2). Bottom panels of Fig. 4 represent the isomeric ratio in the products by photoisomerization, which shows clear solvent-dependence. In n-hexane, the 13-cis form was predominantly produced (Fig. 4d). Various isomers were produced in other solvents, while the most produced isomer was 9-cis in acetonitrile (Fig. 4a), 13-cis in ethanol (Fig. 4b), and 11-cis in methanol (Fig. 4c). Koyama et al. directly applied URSB or PRSB to HPLC analysis [25], while we applied retinal oximes to the HPLC analysis. Nevertheless, essentially similar results were obtained. It should be noted that 7-cis form was also produced in this study; 3.3 % in acetonitrile, 1.4 % in ethanol, 2.4 % in methanol, and -0.1 % in n-hexane among various isomers (Table 2). These values were small, but statistically significant except for that in n-hexane. Formation of the 7-cis form was never reported in the previous studies including Koyama et al. [25].

Table 2 Isomeric ratio in the products by photoisomerization of unprotonated retinal Schiff base (URSB) by 380 nm light, which are obtained by subtraction of the dark value from the light value in each cis peak of Table 1

We next studied the HPLC analysis of PRSB. Raw data of the HPLC profile for PRSB are shown in Fig. S2, where three independent measurements provided similar results. Top panels of Fig. 5 shows HPLC profiles for the anti-forms of all-trans PRSB in acetonitrile (a), ethanol (b), and methanol (c), where black and red lines represent the measurements in the dark and light (440 nm), respectively. Table 3 and middle panels of Fig. 5 qantified isomeric compositions in the dark (black) and light (red), indicating that illuminattion converts all-trans PRSB into cis products as well as the case for URSB (Fig. 4). Table 4 and Bottom panels of Fig. 5 reprsent the isomeric ratio in the products by photoisomerization. Solvent dependence was much smaller than in the case of URSB, and the major product was the 11-cis form (60–70 %) regardless of solvents used. These results reproduced the previous studies [25, 27, 28]. One noticeable observation was the formation of the 7-cis product. In fact, formation of the 7-cis form was 8.2 % in acetonitrile, 3.2 % in ethanol, and 4.5 % in methanol among various isomers (Table 4). These values are larger than those of URSB, and we conclude that all-trans PRSB produces the 7-cis product by light. Namely, unusual photochemical property of NeoR originates from those of the all-trans chromophore itself, though the 7-cis form is exclusively produced in the protein.

Fig. 5
figure 5

HPLC analysis of PRSB. (Top panel) HPLC profiles of the retinal oximes (anti-peaks), which are prepared from all-trans PRSB in the dark (black line) or after illuminated by a 440 nm light (red line) in acetonitrile (a), ethanol (b), and methanol (c). (Middle panel) Isomeric compositions of PRSB in the dark (black bar) or after illuminated by a 440 nm light (blue bar) in acetonitrile (a), ethanol (b), and methanol (c). These values are calculated from peak area in the HPLC profile, where three independent experiments are averaged (standard deviations are shown in the panel). (Bottom panel) Isomeric ratio of the photoproducts of all-trans PRSB after illuminated by a 440 nm light (blue bar) in acetonitrile (a), ethanol (b), and methanol (c). These values are obtained by subtraction of the dark value (black bar) from the light value (blue bar) in each cis peak of the middle panel. Because of decrease of the 13-cis form by light in acetonitrile (a), it represents the negative value

Table 3 Isomeric compositions of protonated retinal Schiff base (PRSB) under dark and light at 440 nm in each solution. Averaged values of three independent measurements are shown with the standard deviations
Table 4 Isomeric ratio in the products by photoisomerization of protonated retinal Schiff base (PRSB) by 440 nm light, which are obtained by subtraction of the dark value from the light value in each cis peak of Table 3

4 Discussion

It has been well established that photoisomerization of the retinal chromophore in rhodopsins is highly selective; all-trans to 13-cis in microbial rhodopsins, and 11-cis to all-trans in animal rhodopsins. These reactions can occur even at 77 K, though retinal changes its shape by isomerization. Therefore, molecular mechanism of efficient isomerization in rhodopsins has been of great interests among researchers. Recent discoveries of different isomerization pathways in bestrhodopsin [16] and NeoR [17] further raised questions about the mechanism, which have been investigated experimentally and theoretically [29,30,31]. To understand unique isomerization in rhodopsins, chromophore molecules in solution offer a good control. For the retinal chromophore, four kinds of mono-cis states are possible for C=C bonds: 7-cis, 9-cis, 11-cis, and 13-cis (Fig. 1). Previous HPLC analysis of all-trans PRSB in solution reported the photoproducts to be a mixture of cis products (82 % 11-cis, 12 % 9-cis, and 6 % 13-cis in methanol), but 7-cis PRSB was not observed [25, 27, 28].

The present study revealed that all-trans and 7-cis forms are only distinguishable for the anti-peaks under standard HPLC conditions, but not for the syn peaks. Therefore, if the 7-cis form may be involved in the photoproducts of rhodopsins, only the anti-peaks in retinal oximes have to be analyzed in the HPLC measurements. In this case, the data from the syn peaks should not be taken into account, though the syn peaks are generally larger in amounts than the anti-peaks [20, 21, 26].

The present study showed that the averaged formation upon illumination of all-trans PRSB was 63 % 11-cis, 25 % 9-cis, 7 % 13-cis, and 5 % 7-cis. This suggests isomerization at central portion being preferential for the photoreaction of all-trans PRSB in solution, whereas selective isomerizations into 13-cis and 7-cis forms are possible in microbial rhodopsins [1, 17]. It should be noted that these values do not represent the quantum yield of the product formation. To determine quantum yield of photoisomerization, sample should be illuminated by weak light, where product formation linearly increases with illumination time, and relative quantum yields can be gained between different cis forms by calibrating their molar extinction coefficients [25]. In this study, we illuminated the sample for 4 min, where photoequilibria were achieved. Although the amounts in this study do not show the quantum yields directly, these values (63 % 11-cis, 25 % 9-cis, 7 % 13-cis, and 5 % 7-cis) are related to preferential isomerization pathways after photoexcitation of all-trans URSB or PRSB.

After Franck–Condon excitation of all-trans PRSB in solution, reaction channels are open to all mono-cis states in the excited state. Ultrafast spectroscopy reported femtosecond and early picosecond relaxations in solution [32,33,34,35,36,37], suggesting that fast reaction dynamics originate from the molecular properties of the chromophore itself. In solution, formations of the 11-cis and 9-cis forms are major and second reaction pathway, respectively. This resembles bestrhodopsin, as 9-cis is also formed together with 11-cis [16]. On the other hand, most of microbial rhodopsins discovered the reaction channel to the 13-cis form, possibly by sterically restricted protein environments. Similar mechanism might take place in NeoR, where central and the Schiff base regions are hardly isomerized, and consequently 7-cis form was produced [17]. Unusual photoisomerization pathway in NeoR is determined by surrounding amino acid residues. Structure of bestrhodopsin may be a good reference because both NeoR and bestrhodopsin possess near-infrared light absorption, together with unusual photoisomerization pathways. Low light-energy absorption may influence potential surfaces of the excited and ground states of NeoR and bestrhodopsin. The near-infrared light absorption originates from three carboxylates near the Schiff base, which was called “counterion triad” [16,17,18, 29, 30]. Important role of the counterion triad will be examined by mutation in future.

5 Summary

By the HPLC analysis of retinal oximes, we report that only anti-peaks should be used for the study of isomeric compositions of the retinal chromophore if formation of the 7-cis state is possible. Here we found formation of the 7-cis state to be 3.3 % in acetonitrile, 1.4 % in ethanol, 2.4 % in methanol, and negligible in n-hexane by the photoisomerization of all-trans URSB, and 8.2 % in acetonitrile, 3.2 % in ethanol, and 4.5 % in methanol by the photoisomerization of all-trans PRSB. It is likely that the all-trans PRSB in NeoR finds a pathway after photoexcitation into the 7-cis form exclusively. In the present study, we also observed strong solvent effect to the products from all-trans URSB, but much less for the photoreaction of all-trans PRSB.