Existence of two O-like intermediates in the photocycle of Acetabularia rhodopsin II, a light-driven proton pump from a marine alga

Abstract

A spectrally silent change is often observed in the photocycle of microbial rhodopsins. Here, we suggest the presence of two O intermediates in the photocycle of Acetabularia rhodopsin II (ARII or also called Ace2), a light-driven algal proton pump from Acetabularia acetabulum. ARII exhibits a photocycle including a quasi-equilibrium state of M, N, and O (M⇄N⇄O→) at near neutral and above pH values. However, acidification of the medium below pH ~5.5 causes no accumulation of N, resulting in that the photocycle of ARII can be described as an irreversible scheme (M→O→). This may facilitate the investigation of the latter part of the photocycle, especially the rise and decay of O, during which molecular events have not been sufficiently understood. Thus we analyzed the photocycle under acidic conditions (pH ≤ 5.5). Analysis of the absorbance change at 610 nm, which mainly monitors the fractional concentration changes of K and O, was performed and revealed a photocycle scheme containing two sequential O-states with the different molar extinction coefficients. These photoproducts, termed O₁ and O₂, may be even produced at physiological pH, although they are not clearly observed under this condition due to the existence of a long M-N-O equilibrium.

Microbial rhodopsins undergo a cyclic photochemical reaction (photocycle) induced by photoexcitation of all-trans retinal (RET) as a chromophore. The photolyzed protein after all-trans to 13-cis isomerization of RET thermally returns to the initial unphotolyzed state. During their respective photocycles, several spectroscopically different photoproducts, referred to as K, L, M, N, and O, appear in a sequential manner [1]. The transitions between successive states are accompanied by several chemical and structural events, such as reisomerization of RET, proton (or other ion species such as Cl⁻ and Na⁺) movement, and conformational changes of the protein, playing a crucial role in exerting the respective functions of microbial rhodopsins [1]. Among four microbial rhodopsins in haloarchaea, the photocycles of bacteriorhodopsin (BR) as a light-driven proton pump and two sensory rhodopsins (SRI and SRII) commonly include a blue-shifted M with deprotonated retinal Schiff base (SB). However, the decay of M in BR is relatively fast, whereas those in SRI and SRII are slow due to a defect in efficient SB reprotonation. On the other hand, M is not found in the photocycle of halorhodopsin (HR), a light-driven inward chloride pump. Although the transitions between each photo-intermediate described above (K-O) occur with large spectral shift, transitions without apparent spectral changes are also often observed in their photocycles [2]. The most famous example of this is found in the first half of the photocycle of BR [3]. A spectrally silent transition between two M-like substates, M₁ and M₂, which works as the accessibility switching process for SB from the extracellular (EC) side to cytoplasmic (CP) side, leads to unidirectional outward proton transport by BR [3]. Another example is the L₁–L₂ transition that occurs in the photocycle of HR. In this transition, it has been deduced that the position of Cl⁻ changes inside the protein [1,4,5]. In addition, it is considered that the formation of multiple M species in SRI and SRII may participate in signal transduction to a transducer by F-helix tilting [6,7]. Therefore, spectrally unchanged processes during the photocycle may also happen on other intermediates and be worth further investigation if they exist.

In contrast with the first half of the photocycle, the second half of the photocycle has been incompletely understood. The presence of two N-states after M was reported in BR [3]. The N₁ (N) to N₂ (N′) conversion changes the connectivity of a proton donor to SB (D96^BR) from SB at the intracellular surface to take up a proton from the CP aqueous phase [3]. This process also serves for the regulation of the direction of proton translocation like the M₁–M₂ conversion. The N-O transition accompanies the isomerization. The isomerization-switch-transfer (IST) model by Haupts et al. assumes that the switch (orientation change of N-H of SBH⁺ between CP and EC) occurs after the isomerization [8]. On the other hand, Wang et al. carried out the molecular dynamic (MD) simulations during the N-O transition of BR, and the first switch is followed by the isomerization in the latter part of the photocycle, which is contrast to IST model [9]. Wang et al. thought the important role of the interaction of SBH⁺ with deprotonated D212^BR in EC [9].

In this article, we report the presence of two O states in the photocycle of Acetabularia rhodopsin II (ARII or also called Ace2), which is one of two eukaryotic light-driven proton pump homologues from the marine alga Acetabularia acetabulum [10,11]. As reported previously, the latter half of the photocycle of ARII at neutral pH is described as M⇄N⇄O→ARII′→ARII [11]. The existence of reversible reactions between M, N and O makes analysis of the photocycle complicated. However, the equilibrium between M and N in the above scheme shifts toward M due to a rapid back reaction at acidic pH (< ~5.5). In addition, the N→O reaction is fast under this condition because the medium pH is below the pK_a (5.9 or 6.3) of D92^ARII (corresponding to D96^BR) during H⁺-uptake at N-decay [11]. Therefore, N is not accumulated at pH < ~5.5, allowing the scheme to be simplified as M→O→ARII′→ARII. Consequently, the kinetic analysis becomes easy, and we found two O intermediates in the photocycle of ARII.

Materials and Methods

Sample preparation

The procedure for the synthesis of ARII protein by cell-free expression and its purification method were the same as previously described [10,11]. ARII solubilized with 0.05% n-dodecyl-β-D-maltoside (DDM) was used.

Flash photolysis

Measurements were performed by using the same apparatus and procedure described previously [12]. Absorbance changes generated by the excitation of proteins with a laser pulse (Nd:YAG 532 nm, 7 ns, 5 mJ/pulse) were collected at three characteristic wavelengths (400, 520, and 610 nm) at 20°C. The experimental medium was a solution containing 400 mM NaCl and 6-mixed buffer (citrate, MES, HEPES, MOPS, CHES, CAPS, at 2 mM each). Data analysis was conducted using the Microcal Origin software (OriginLab, Microcal Software).

Results and Discussion

To detect three characteristic photocycle intermediates, K, M, and O, flash-induced absorbance changes were measured at three selected wavelengths at three different pH values, as shown in Figure 1. Note that L cannot be clearly observed, likely due to kinetics. The obtained ΔA₄₀₀ and ΔA₅₂₀ signals mainly reflect concentration changes of M and the original pigment (ARII), respectively. On the other hand, two red-shifted photointermediates, K- and O-states, have an absorbance maximum wavelength (λ_max) at a similar red-shifted spectral range [11]. It is worth noting that the absorbance change at 610 nm over the early time range prior to ~0.02 ms represents the time-dependent fractional concentration change of K, whereas the rise and decay of O are detected at the latter time range.

Figure 1.

Simulation of the observed ΔA₆₁₀ signal in the photocycle of ARII under acidic condition (pH≤5.5) by three analytical models. The data at pH 4.1 are analyzed by (A) a model with simple O (see Eq. (4)), (B) a model with two O-states, O₁ and O₂ (ɛ_O1=ɛ_O2) (see Eq. (6)), and (C) a model with O₁ and O₂ (ɛ_O1≠ɛ_O2) (see Eq. (7)). The upper panels show the fitting results. In these figures, the observed and fitting curves are shown as black noisy and red smooth lines, respectively. The original state (ARII), M, and K plus O were monitored at 520, 400, and 610 nm, respectively. The red dotted, broken, and chain lines in these panels stand for the calculated fractional concentration changes of K, O₁ (or simple O), and O₂, respectively (see Eqs. (1), (3), and (5)). A small negative absorbance change in the ΔA₄₀₀ signal at the time range from ~0.1 to ~10 ms may originate from the contribution of the small absorbance of the original pigment at this wavelength (also see Fig. 3B), which may lead to a fitting error with Eq. (2). The lower thin panels represent the difference at 610 nm between the observed and regression curves. The panels (D)–(F) and (G)–(I) show the corresponding fitting results for the data at pH 4.8 and 5.5, respectively. The assumptions are: a single O-intermediate (Eq. (4)) is adopted in the left column, two O-intermediates with the same ɛ (Eq. (6)) are in the middle column, and two O-intermediates with different ɛs (Eq. (7)) are in the right column. The ratio of α_O1 and α_O2 were estimated to be 0.63 and 0.64 at pH 4.8 and 5.5, respectively. Measurements were performed using the ARII protein solubilized by 0.05% DDM in medium containing 400 mM NaCl buffered with 2 mM 6-mixed buffer at 20°C.

In agreement with the simplified scheme described above, the decay of M (ΔA₄₀₀) is single exponential (see the data below pH 4.8 in the inset of Fig. 2A), and its decaying phase seems to match the rising phase of O (ΔA₆₁₀) (also see Table 1). When assuming that the concentration of K at the initial moment (t = 0) is 1, the time-dependent fractional concentration changes of K, M, and O in the simplified scheme (K→M→O→) can be derived as the following equations, (1), (2) and (3), respectively.

Figure 2.

Coincidence of the pH-dependent appearance of M⇄N equilibrium and lack of detection of O₁-formation. Panels A and B show the pH-dependent changes of the ΔA₄₀₀ and ΔA₆₁₀ signals with an increase in pH from 4.1 to 7.3, respectively. The inset in panel A shows semilogarithmic plots of the ΔA₄₀₀ signals. These data were obtained under the same experimental condition as described in Figure 1. The ΔA₆₁₀ signals at pH≤5.5 are observed as a sum of K and two O substates (O₁, O₂), whereas those at pH>5.5 are approximated as a sum of K and simple O (O₂). The smooth curves show the fitting curves for the observed data (noisy lines) at each pH. The data regression at pH≤5.5 and above was done using Eq. (7) and (4). In near agreement with the pH-dependent change of the ΔA₆₁₀ signal, the decay of the ΔA₄₀₀ signal changes from mono- to bi-phasic, indicating the appearance of M⇄N equilibrium with increasing pH.

Table 1.

Decay rate constants of various photointermediates under acidic conditions^a

	k0 (K-decay)	k0′b (K-decay)	k1′b (M-decay)	k1 (M-decay)	k2 (O1-decay)	k3 (O2-decay)
pH 4.1	99.3	79.8	40.0	30.2	2.25	0.285
pH 4.8	86.4	66.8	17.7	13.1	1.53	0.324
pH 5.5	74.8	67.7	6.08	5.46	1.36	0.340

^aThe unit of all values is ms⁻¹.

^bk_0′ and k_1′ were estimated by the fitting for the ΔA₄₀₀ signals with Eq. (2) and distinguished from the corresponding values (k₀ and k₁) estimated from the fitting for the ΔA₆₁₀ signals. Two k₀ and k₁ values are similar each other, although there are small differences between them presumably due to fitting error for the ΔA₄₀₀ signals (also see the legend of Fig. 1).

$$\mathrm{\Delta }\text{K}=e^{-k_{0}t}$$ (1)

$$\mathrm{\Delta }\text{M}=\frac{k_0}{k_1-k_0}(e^{-k_{0}t}-e^{-k_{1}t})$$ (2)

$$\mathrm{\Delta }\text{O}=-\frac{k_0{k}_1}{(k_0-k_1)(k_1-k_2)(k_2-k_0)}[(k_1-k_2)e^{-k_{0}t}+(k_2-k_0)e^{-k_{1}t}+(k_0-k_1)e^{-k_{2}t}]$$ (3)

where k₀, k₁, and k₂ signify the rate constants of K-, M-, and O-decay, respectively. When scaling constants for the amplitude of the absorbance of K and O at 610 nm, which contain molar extinction coefficients (ɛ) for their respective intermediates, defined as α_K and α_O, respectively, ΔA₆₁₀ signals can be expressed as the following equation.

$$\mathrm{\Delta }{A}_{610}\approx {\alpha }_{\text{K}}\mathrm{\Delta }\text{K}+{\alpha }_{\text{O}}\mathrm{\Delta }\text{O}$$ (4)

Eq. (4) was applied to the analysis for the ΔA₆₁₀ signal. This analysis, however, did not give a good fitting result (see Fig. 1A, D, and G). As seen in the lower panels of these figures, the fitting deviation is especially prominent in the absorbance change originating from the rise and decay of O (the time region from ~0.02 to ~10 ms). At first glance, the rise of O is not monophasic, irrespective of the single decay of M (see the strain line in the inset of Fig. 2A). This discrepancy can be solved by assuming a scheme that contains another O-like state after the first O, and these two sequential O-states are referred to as O₁ and O₂.

In the alternative scheme designed above (K→M→O₁ →O₂→), the rise and decay rate constants of the two O substates were redefined as k₁ for O₁-rise, k₂ for O₁-decay (O₂-rise), and k₃ for O₂-decay. The time-dependent fractional concentration change of O₁ can be represented by Eq. (3), whereas that of O₂ can be derived as the following equation.

$$\mathrm{\Delta }{\text{O}}_2=-\frac{k_0{k}_1{k}_2}{(k_0-k_1)(k_1-k_2)(k_2-k_0)(k_3-k_0)(k_3-k_1)(k_3-k_2)}\left[\begin{array}{l}(k_1-k_2)(k_3-k_1)(k_3-k_2)e^{-k_{0}t}\hfill \\ +(k_2-k_0)(k_3-k_0)(k_3-k_2)e^{-k_{1}t}\hfill \\ +(k_0-k_1)(k_3-k_0)(k_3-k_1)e^{-k_{2}t}\hfill \\ +(k_0-k_1)(k_1-k_2)(k_2-k_0)e^{-k_{3}t}\hfill \end{array}\right]$$ (5)

We first performed fitting analysis using the following equation under the assumption that the ɛs for the two O-states are equal.

$$\mathrm{\Delta }{A}_{610}\approx {\alpha }_{\text{K}}\mathrm{\Delta }\text{K}+{\alpha }_{\text{O}}(\mathrm{\Delta }{\text{O}}_1+\mathrm{\Delta }{\text{O}}_2)$$ (6)

This fitting result is shown in Figure 1B, E, and H. The fit is largely improved, but a small difference is yet observed (see the lower thin panels of Fig. 1B, E, and H).

Careful inspection reveals a small second increasing phase in the ΔA₆₁₀ signal originating from O, which may imply that the ɛ of the second O (O₂) is different from that of the first one (O₁). Moreover, the results of the multiexponential global fitting analysis under acidic conditions (pH 5.0) also support this idea (see Fig. 3). This analysis reveals a photocycle scheme containing five photointermediates: P₀ (ARII)→P₁ (K)→P₂ (M)→P₃ (O₁)→P₄ (O₂)→P₅ (ARII′) →P₀ (ARII), which is almost consistent with the above described scheme, although a small contamination of K is observed in the P₂-state. The maximum absorbance of P₄ (O₂) is apparently larger than that of P₃ (O₁). Therefore, this together with the discrepancy of Eq. (6) with observed data led us to consider that ɛ_Oi (i.e. α_Oi) are different each other. The following combined function of Eqs. (1), (3), and (5) was used for the fitting.

Figure 3.

Global fitting analysis for the flash photolysis data in ARII at pH 5.0. (A) Flash-induced absorbance changes at three chosen wavelengths (400, 530, and 610 nm). The data were obtained in a solution containing 0.05% DDM, 400 mM NaCl and 10 mM 6-mixed buffer at 20°C. The observed data and multiexponential fitting curves are shown as black and red lines, respectively. These two curves are nearly completely overlapping. The vertical lines stand for the decay time constants of five kinetically defined states (P₁–P₅). (B) Relative absorbance spectra of P₀–P₅. Data analysis was performed for the obtained dataset at 320–700 nm with a 10 nm interval according to a sequential irreversible model including five P-states (P₀→P₁→ ... →P₅→P₀, where P₀ denotes the dark state). The detailed procedures of this analysis should be referred to our previous papers [12,24,25]. (C) Closeup view of O₁- and O₂-spectra around the absorption peak. As seen in the panel B, P₃ and P₄ are composed of only red-shifted absorbance band (λ_max ~560 nm), suggesting that these states can be identified as O. With the P₃ to P₄ conversion, the λ_max of the spectrum almost does not change (Exactly, as shown in the panel C, that of P₃ is shorter by small difference (~5 nm) than that of P₄), but its maximum value increases from approx. 0.7 to 0.9 despite the lack of contribution from other absorption bands. Therefore, this may imply that the ɛ of O in P₃ (O₁) is different from that in P₄ (O₂). The ratio of their amplitudes at λ_max (ɛ_{3, max} /ɛ_{4, max}) is 0.78. The reason for the discrepancy between this value and that described in the text (ca. 0.7) is unclear, but may be experimental due to error.

$$\mathrm{\Delta }{A}_{610}\approx {\alpha }_{\text{K}}\mathrm{\Delta }\text{K}+{\alpha }_{{\text{O}}_1}\mathrm{\Delta }{\text{O}}_1+{\alpha }_{{\text{O}}_2}\mathrm{\Delta }{\text{O}}_2$$ (7)

Eq. (7) simulates the observed ΔA₆₁₀ signal well (see Fig. 1C, F, and I). Therefore, we concluded that there were two O-intermediates with different α values. The estimated values of the k_is (i = 0–3) are shown in Table 1. The values of α_O1 and α_O2 at pH 4.1 were 30.3 and 45.3, respectively, and the ratio of these values (α_O1/α_O2) was 0.67. This approximately 0.7 value is reflected by the ratio of ɛ_O1 to ɛ_O2, which is almost the same as the ratio of the peak values (Fig. 3B and C).

As the medium pH increases, the rise of O becomes monophasic (see the lower panel B in Fig. 2). This may be attributed to the lack of accumulation of O₁ due to the prolonged decay of its precursor (N). Indeed, it appears that lack of contribution of O₁-formation to the ΔA₆₁₀ signal occurs simultaneously with the appearance of the biphasic M-decay in the ΔA₄₀₀ signal originating from the formation of M-N equilibrium due to the delay of N-decay, which was detectable at a pH above ~5.5 (Fig. 2). Thus, ΔA₆₁₀ signals are observed as a sum of K and simple O (O₂) at a physiological neutral or weak alkaline pH. However, we infer that O₁ exists even under these pH conditions, although it cannot be detected because of the appearance of the M-N-O quasi-equilibrium. It is worth noting that Chizhov et al. reported the existence of two Os in HR from Natronomonas pharaonis (NpHR) [13].

What molecular event does occur during the transition from O₁ to O₂? Wang et al. proposed the following two possible sequences at the N→O transition of BR [9]:

1. Sequence 1: N (13-cis RET, 15-anti SBH⁺)

   → (all-trans RET, 15-syn SBH⁺)

   → (all-trans RET, 15-anti SBH⁺),

2. Sequence 2: N (13-cis RET, 15-anti SBH⁺)

   → (13-cis RET, 15-syn SBH⁺)

   → (all-trans RET, 15-anti SBH⁺).

The MD simulation by Wang et al. showed the possibility of Sequence 2 and the interaction of N-H with deprotonated D212^BR was pointed out [9]. Furthemore, all-trans RET, 15-syn SBH⁺ in Sequence 1 is considered to be metastable [14]. Considering these facts, Sequence 2 may be more plausible. As a broadly received conception, the O-state in BR has a twisted all-trans RET [15] and the large spectral red-shift during the N-O transition is attributed to the isomerization of RET from 13-cis to all-trans. In contrast, Subramaniam and coworkers reported the long life-time of O with 13-cis RET in L93A^BR mutant which induces the slow reisomerization of RET due to abolishment of van der Waals interaction between the 13-methyl of RET and the terminal methyl groups of L93^BR [16]. In addition, Zhang et al. solved its X-ray crystal structure and concluded that this long-lived O took 13-cis RET, 15-syn SBH⁺ configuration [17]. These findings demonstrate at least the existence of a 13-cis O in L93A^BR. Tóth-Boconádi et al. presumed two substates of O in L93A^BR, and described that their distinction is the difference of the configuration of RET, 13-cis or all-trans [18,19]. Furthermore, Milder postulated the presence of two consecutive O-intermediates having 13-cis and all-trans RET in the wild-type BR [20]. Hence, the O₁-to-O₂ transition in this study may be also accompanied by the cis-to-trans isomerization of RET. In this respect, we can see that in Figure 3B, λ_max of P₃ (O₁) is a little bit smaller than P₄ (O₂) (also see Fig. 3C), and that ɛ of P₃ (O₁) is clearly smaller than the other. The 13-cis isomer in the dark-adapted BR has smaller λ_max and ɛ values compared with those of the all-trans isomer [21]. Other microbial rhodopsins such as HR from Halobacterium salinarum and Anabaena sensory rhodopsin showed similar characteristic [22,23]. This spectral property is consistent with the assumption that O₁ and O₂ take the RET configuration of 13-cis and all-trans, respectively. In the anion-pumping photocycle of NpHR proposed by Kouyama et al., two O-intermediates (O′ and O) with 13-cis and -trans form of RET were assumed [5], which is also consistent with our present assumption. Thus, together with the model by Wang et al. [9], we propose a conformational scheme of RET and SBH⁺ upon the transition from N to O, which is shown in Figure 4.

Figure 4.

Schematic of configuration change of RET and SBH⁺ in the latter half of the photocycle. A picture is depicted in the case of BR.

In the acidic condition adopted here, N-decay is very fast and the rate of the isomerization is relatively pH-independent, which may result in the observation of O₁. Under neutral or alkaline conditions, N-decay is slow and the O₁-to-O₂ conversion rate is presumably faster, which results in the lack of detection of O₁. Allthough O-intermediate in the crystal structure of L93A^BR takes 13-cis RET, 15-syn SBH⁺ in agreement with the expected configuration of O₁ in the present paper [17], the precise RET configuration of O₁ of the wild-type BR should be determined in further study. In addition, we should explore its existence for other microbial rhodopsins. However, there is a possibility that in many microbial rhodopsins, observation of O₁ (13-cis RET, 15-syn SBH⁺) is difficult (although its existence is in the case) maybe due to M-N-O equilibrium and/or its short life time. In this case, a problem arises why this intermediate can be observable in ARII. This is an intriguing subject in future.

Conclusion

In this study, we analyzed the photocycle of ARII under acidic conditions (pH ≤ 5.5) at which accumulation of N is almost negligible. The kinetic analysis with three models revealed that the scheme containing two sequential O-intermediates (O₁ and O₂) with different ɛ values is best. Considering the smaller λ_max and ɛ of O₁ than those of O₂, we expected that the O₁-to-O₂ transition is accompanied by the RET isomerization from 13-cis to all-trans, which agrees with the proposed sequence of two configurational changes in RET and SBH⁺ based on the MD simulations by Wang et al. [9]: thermal reisomerization of RET is preceded by switch of the accessibility of SBH⁺. Whether there are two O-intermediates in the photocycle of other microbial rhodopsins including BR should be investigated in future. However, the presence of the M-N-O equilibrium may make the detection of the first O (O₁ in the present study) difficult.

Significance.

The latter half of the photocycle of microbial rhodopsins remains incompletely understood. Although the transition from N to O intermediate is accompanied by thermal reisomerization of retinal chromophore and accessibility switch of the protonated Schiff base, it has been an unsolved problem which event occurs first, even in bacteriorhodopsin (BR), the most intensively-studied microbial rhodopsin. The present work revealed the formation of two O-states in the latter part of the photocycle of Acetabularia rhodopsin II, a BR-like light-driven proton pump from a marine alga. This finding would help us to understand the latter molecular event in the photocycle of microbial rhodopsins.