Chimeric Microbial Rhodopsins Containing the Third Cytoplasmic Loop of Bovine Rhodopsin

Abstract

G-protein-coupled receptors transmit stimuli (light, taste, hormone, neurotransmitter, etc.) to the intracellular signaling systems, and rhodopsin (Rh) is the most-studied G-protein-coupled receptor. Rh possesses an 11-cis retinal as the chromophore, and 11-cis to all-trans photoisomerization leads to the protein structural changes in the cytoplasmic loops to activate G-protein. Microbial rhodopsins are similar heptahelical membrane proteins that function as bacterial sensors, light-driven ion-pumps, or light-gated channels. Microbial rhodopsins possess an all-trans retinal, and all-trans to 13-cis photoisomerization triggers protein structural changes for each function. Despite these similarities, there is no sequence homology between visual and microbial rhodopsins, and microbial rhodopsins do not activate G-proteins. However, it was reported that bacteriorhodopsin (BR) chimeras containing the third cytoplasmic loop of bovine Rh are able to activate G-protein, suggesting a common mechanism of protein structural changes. Here we design chimeric proteins for Natronomonas pharaonis sensory rhodopsin II (SRII, also called pharaonis phoborhodopsin), which has a two-orders-of-magnitude slower photocycle than BR. Light-dependent transducin activation was observed for most of the nine SRII chimeras containing the third cytoplasmic loop of bovine Rh (from Y223, G224, Q225 to T251, R252, and M253), but the activation level was 30,000–140,000 times lower than that of bovine Rh. The BR chimera, BR/Rh223-253, activates a G-protein transducin, whereas the activation level was 37,000 times lower than that of bovine Rh. We interpret the low activation by the chimeric proteins as reasonable, because bovine Rh must have been optimized for activating a G-protein transducin during its evolution. On the other hand, similar activation level of the SRII and BR chimeras suggests that the lifetime of the M intermediates is not the simple determinant of activation, because SRII chimeras have two-orders-of-magnitude's slower photocycle than the BR chimera. Activation mechanism of visual and microbial rhodopsins is discussed on the basis of these results.

Introduction

Rhodopsin (Rh) is one of the G-protein-coupled receptors that has diverged into a photoreceptive protein in retinal visual cells (1–5). It is a membrane protein consisting of a single polypeptide opsin and a light-absorbing chromophore 11-cis-retinal. The opsin contains seven transmembrane α-helices, the structural motif typical of G-protein-coupled receptors. The 11-cis-retinal is bound to the Lys²⁹⁶ in the transmembrane helix 7 through a protonated Schiff base linkage. Absorption of a photon by the chromophore causes isomerization to the all-trans form, followed by conformational changes of protein (1,5). Several intermediate states in the bleaching process are identified as photorhodopsin, bathorhodopsin, lumirhodopsin, metarhodopsin-I (Meta-I), and metarhodopsin-II (Meta-II). Meta-II catalyzes the GDP-GTP exchange reaction in the trimeric G-protein transducin (1–4).

Some archaea and bacteria possess retinal binding proteins, archaeal rhodopsins that contain an all-trans retinal as a chromophore. Among them, the most studied is bacteriorhodopsin, which is found in Halobacterium salinarum (5–7). This archaea contains four retinal-bonding proteins: bacteriorhodopsin, halorhodopsin, sensoryrhodopsin I (SRI), and sensoryrhodopsin II (SRII; also called phoborhodopsin). Bacteriorhodopsin (BR) and halorhodopsin are light-driven ion pumps, which act as an outward proton-pump and an inward Cl⁻-pump, respectively (5–10). SRI and SRII are photoreceptors of this halophilic archaea, which act for attractant and repellent responses in phototaxis, respectively (9,10). Thus, functions of these four proteins are different, but it is known that they have similar protein architecture. Recently, genomic analysis revealed the presence of similar proteins in Eubacteria and Eukaryotes, and they are classified as microbial rhodopsins.

There are no sequence homologies between visual and microbial rhodopsins, though both possess similar chromophore (retinal) and protein (7-transmembrane helices) structures. Therefore, it is believed that both have been evolved independently. In fact, visual rhodopsins do not pump ions, while microbial rhodopsins do not activate G-proteins. However, it was reported that BR chimeras containing the third cytoplasmic loop of bovine rhodopsin are able to activate G-protein (11). This observation suggests the common structural feature for light-induced activations between BR and bovine Rh, where helix-opening motion at the cytoplasmic side probably causes exposure of the third loop to possible binding with G-protein. According to the previous results, the highest G-protein activation by the BR chimera (BR/Rh223-253) was ∼25% of that by bovine Rh (11). This was surprising, because bovine Rh must be optimized to activate G-protein during its evolution.

In this study, we have also studied chimeric proteins from SRII of Natronomonas pharaonis, which has a slower photocycle than BR (12,13). Because activation of G-protein presumably depends on the lifetimes of the active intermediate state, we expect that G-protein activation is higher for the SRII chimera than for the BR chimera. We have designed nine SRII chimeric proteins (Fig. 1), whose positions are identical to the case for the reported BR chimera (11). In the case of the BR chimera, four chimeras (BR/Rh223-252, 223-253, 225-251, and 225-252) showed G-protein activation, whereas no G-protein activation was reported for BR/Rh223-251, 224-251, 224-253, and 225-253 (no purple protein was gained for BR/Rh224-252) (11). This study shows very different results for the SRII chimera. G-protein activation is observed for most of the nine chimeras, but the activation is much lower (∼10⁴ times) than that by bovine Rh. In this study, we have also reproduced the G-protein activation by BR/Rh223-253, but the efficiency is ∼10⁴ times lower than that of bovine Rh, which is contrary to the previously reported value of 25% (11). On the basis of the molecular properties of these chimeras, activation mechanism of visual and microbial rhodopsins will be discussed.

Figure 1.

Design of chimeric proteins from Natronomonas pharaonis SRII. (a) We removed the sequence of ESASQRSSG in the E-F loop from SRII, and inserted nine constructs of the third loop of bovine Rh. This design of nine SRII/Rh chimeras is identical to those for nine BR/Rh chimeras by Geiser et al. (11). (b) Schematic drawing of the secondary structure of the chimera. The SRII sequence of E-helix (…PMT) is connected to the third loop of bovine Rh, which further continues to F-helix (… IKS) of SRII.

Materials and Methods

Sample preparation

The chimera constructions were designed based on the wild-type SRII of Natronomonas pharaonis, and the DNA template of bovine Rh loop was exchanged by the following three-step PCR. At first, three PCR products were constructed and purified:

• The SRII segment preceding the exchange region with the 15-mer Rh loop attached (i.e., fSRII).

• The segment of SRII trailing the exchange region (i.e., lSRII).

• The bovine Rh loop with a 15-mer of SRII attached near the exchange region (Rh loop).

The products of the first PCR were used to amplify a second-round PCR product. Then, fSRII was extended to the loop region by PCR with the Rh loop and lSRII treated the same. The final full-length chimera fragment was amplified from the former two products and they were cloned into pET21c vector, by inserting with NdeI/XhoI digestion. After ligation, the plasmids were transformed in Escherichia coli strain JM109. All of the chimeras were confirmed by DNA sequencing (Hokkaido System Science, Sapporo, Hokkaido, Japan). The wild-type and chimeric proteins of SRII possessing a six-histidine tag at the C-terminus were expressed in E. coli, solubilized with 1% n-dodecyl β-D-maltoside (DDM), and purified by a Ni²⁺-column chromatography as described previously (14). Reconstitution of the purified protein into L-α-phosphatidylcholine (PC) liposomes was performed using Bio-Beads (Bio-Rad, Hercules, CA) where the molar ratio of added PC was 30 times that of the proteins (15).

The BR chimera, BR/Rh223-253, was designed as described previously (11). BR/Rh223-253 and the wild-type BR proteins were purified from Halobacterium salinarum as purple membrane sheet according to a standard method (16). Wild-type bovine rhodopsin was expressed in HEK293S cells and was purified by using the monoclonal antibody rho1D4 against the C-terminal sequence of bovine rhodopsin (17). A G-protein transducin was purified from bovine rod outer segments according to the method previously described (18).

Spectroscopy

Absorption spectra of solubilized SRII (300 mM NaCl, 300 mM imidazole, 50 mM Tris-HCl, pH 7.0, and 0.1% DDM) or BR in purple membrane suspension were measured at 20°C using a model No. UV-2400PC UV-visible spectrophotometer (19) (Shimadzu, Kyoto, Japan). Photobleaching and acid denaturation experiments were carried out using a model No. V-550 UV-visible spectrophotometer (JASCO, Oklahoma City, OK). For the photobleaching experiment, the sample was illuminated with the >480 nm light from a 1 kW projector lamp at room temperature. Acid denaturation experiment was performed by adding 20 μL HCl (5 M) into the sample solution (600 μL), followed by measuring absorption spectra of the sample.

Photocycles of the wild-type and chimeric proteins were measured by use of a flash photolysis apparatus as described previously (20). Each purified SRII sample (chimera and wild-type) was resuspended in buffer (300 mM NaCl, 300 mM imidazole, 50 mM Tris-HCl, pH 7.0, and 0.1% DDM). The BR sample (chimera and wild-type) was resuspended in buffer (20 mM phosphate buffer, 200 mM NaCl, pH 7.0). Flash-induced absorption changes were acquired with 20 ms and 1 ms intervals for SRII and BR, respectively, by using a commercial flash photolysis system (Hamamatsu Photonics, Hamamatsu City, Japan), which consists of a charge-coupled device linear detector (Multichannel Spectral Analyzer PMA-11 C8808-01; Hamamatsu), a continuous-wave xenon lamp L8004 as a light source and a sample room (Hamamatsu). Excitation of each sample was done using 500-nm nanosecond laser pulses from a Nd³⁺-YAG laser apparatus (LS-2134UT-10; LOTIS TII, Minsk, Belarus; 355 nm, 7 ns, 60 mJ) through an optical parametric oscillator (model No. LT-2214-OPO/PM; LOTIS TII) for SRII, and 532-nm nanosecond laser pulse of the second harmonics generation of a Nd3⁺ YAG laser (INDI-40-10, Spectraphysics, Santa Clara, CA) for BR. The energy of one laser pulse was 0.13 mJ. For signal/noise improvement, 20 and 50 photoreactions were averaged for each sample solution of SRII and BR, respectively. Absorbance was adjusted to be 0.5 at the λ_max. The temperature of each sample was kept at 25°C.

Photocycles of the wild-type and chimeric SRII were also measured using a model No. V-550 UV-visible spectrophotometer (JASCO). The sample, whose temperature was kept at 4°C in the sample cell, was illuminated at >480 nm from a 1 kW projector lamp for 4 s, and absorbance at 360 nm or 500 nm was monitored at every 0.1 s after the illumination. Because the obtained M-decay time constants (τ_1/e) were in the range of 1–2 s, this measurement could not detect the fast components of the M-decay.

G-protein activation assays

A radionucleotide filter-binding assay, which measures a light-dependent GDP/GTPγS exchange by transducin, was carried out as described previously (21). All procedures were carried out at 20°C. The assay mixture consisted of 50 mM HEPES (pH 7.0), 140 mM NaCl, 5 mM MgCl₂, 1 mM DTT, 0.05% DDM, 1 μM [³⁵S]GTPγS, and 2 μM GDP. The mixture of rhodopsin (final concentration; 2.4 μM SRII, 1.6 μM BR, and 5 nM bovine Rh) and transducin (final concentration; 600 nM) was constantly irradiated with white light or was kept in the dark. After incubation for the selected time, an aliquot (20 μL) was removed from the sample into 200 μL of stop solution (20 mM Tris/Cl (pH 7.4), 100 mM NaCl, 25 mM MgCl₂, 1 μM GTPγS, and 2 μM GDP), and it was immediately filtered through a nitrocellulose membrane to trap [³⁵S]GTPγS bound to transducin. The amount of bound [³⁵S]GTPγS was quantified by assaying the membrane with a liquid scintillation counter (Tri-Carb 2910 TR; Perkin Elmer, Waltham, MA).

Results

Absorption properties of the SRII/Rh and BR/Rh chimera

In this study, we have expressed the SRII/Rh chimera in E. coli, followed by solubilization with DDM and purification through Ni:NTA column. The purified protein possesses the identical λ_max to that of the wild-type SRII (500 nm), as shown for the case of SRII/Rh225-252 in Fig. 2 a. Identical λ_max are obtained for all nine chimeras, as shown in Fig. S1 in the Supporting Material. This is also the case for BR/Rh223-253 (Fig. 2 b), which shows the identical λ_max with the wild-type BR (562 nm) under the dark-adapted conditions. This is reasonable, because the introduced loop is located at the cytoplasmic surface and distant from the retinal binding region. In other words, no structural perturbation of the retinal binding region by the replacement of the loop is implicated by the identical absorption spectra.

Figure 2.

(a) Absorption spectra of the wild-type SRII (broken line) and SRII/Rh225-252 (solid line) in 0.1% DDM solution. (b) Absorption spectra of the wild-type BR (broken line) and BR/Rh223-253 (solid line) in purple membrane suspension. One division of the y axis corresponds to 0.05 absorbance unit.

While λ_max were identical, absorptions of chimeras were larger at 300–450 nm for both SRII (Fig. 2 a) and BR (Fig. 2 b). In the case of BR, larger absorption originates from contamination of impurity (Fig. 2 b), where less expression of chimeric proteins yields less-purified BR chimera. The reason was entirely different for the SRII chimera (Fig. 2 a). At the early stage of this study, the peak at ∼370 nm has been more prominent, and we have found that the formation of the species (P370) is dependent on the exposure to light. Therefore, we have tried to keep the sample under dark to avoid photobleaching.

Fig. S2 shows the photobleaching behavior of the SRII/Rh chimera. Unlike visual rhodopsins, microbial rhodopsins such as BR and SRII exhibit photocycle. In fact, little bleaching was observed for the wild-type SRII due to continuous illumination at >500 nm. In contrast, illumination caused bleaching of SRII chimeras with half-decay of 2 min. Continued illumination with the same light source converted nearly all of the remaining 500-nm state into the 370-nm product with a nonexponential time course.

How is P370 produced, and what are its molecular properties? One important question is whether the retinal molecule is bound to the Lys residue or not. Although both free retinal and deprotonated Schiff base absorb at ∼370 nm, acid denaturation causes color change of the latter to ∼450 nm (protonated Schiff base), but the former does not change color. Fig. S3 a shows the absorption spectra of P370 before and after acidification by addition of HCl. By lowering pH, the absorption at 370 nm diminishes, while that at 400–550 nm newly appears. The difference spectrum in Fig. S3 b clearly shows the conversion of P370 (negative peak at 366 nm) to a product absorbing at 460 nm, which is characteristic of a protonated Schiff base. Thus, P370 possesses the deprotonated Schiff base like the M intermediate of SRII, though the M state returns to SRII for the wild-type.

Photochemical properties of the SRII/Rh and BR/Rh chimera

SRII and BR have unique cyclic reactions that are comprised of a series of intermediates, such as blue-shifted M and red-shifted O intermediates. Next, we have tested how photocycles are modified by the introduction of the Rh loop. It should be noted that BR/Rh223-253 exhibits the photocycle like the wild-type BR, but all SRII/Rh chimera bleach. This fact raises another question about the bleaching yield. In the case of Rh, photoisomerization causes 100% bleaching (no photocycle). In contrast, bleaching yields of SRII and BR are 0%. Flash photolysis experiment under single-photon event can determine the bleaching yield of SRII/Rh chimera. Because flash photolysis experiment requires numerous amount of sample because of bleaching, here we have applied the laser photolysis experiment to SRII/Rh225-252 and SRII/Rh224-252 for which expression levels were high. Then, we have compared photoreaction kinetics of all SRII/Rh chimeras by lowering temperatures (4°C).

Fig. 3 a shows room-temperature laser-induced absorbance changes of the wild-type SRII at 390 nm (blue line), 500 nm (green line), and 560 nm (red line), which monitor the M intermediate, the depletion of SRII, and the O intermediate, respectively. The decay time constants (τ_1/e) of the M and O intermediates were 165 ± 3 ms, and 380 ± 30 ms, respectively, and the time constant (τ_1/e) of the recovery of SRII at 500 nm was 250 ± 20 ms (mean ± SD). Fig. 3 b shows the corresponding data of SRII/Rh225-252, where time constants (τ_1/e) of the M and O decays were 173 ± 5 ms and 1650 ± 470 ms, respectively, and time constants (τ_1/e) of the recovery were 229 ± 30 ms (69%) and 1670 ± 470 ms (31%). This indicates that the M decay is not changed between SRII and SRII/Rh225-252, but the O decay and the recovery are 4–6 times slower in SRII/Rh225-252 than in SRII.

Figure 3.

Laser flash photolysis results of SRII (a and b) and BR (c and d) at room temperature. Light-induced absorbance changes of the wild-type SRII (a) and SRII/Rh225-252 (b) are monitored at 390 nm, 500 nm, and 560 nm, which monitor the M intermediate, the depletion of SRII, and the O intermediate, respectively. (Dotted lines) Data points, which were averaged for 20 experiments. (Solid lines) Fitting curves (390 nm; single-exponential, 500 nm; double-exponential, 560 nm; double-exponential of rise and decay). Note that the green line in panel b does not recover to the zero level. One division of the y axis corresponds to 0.02 (a) and 0.01 (b) absorbance units. Light-induced absorbance changes of the wild-type BR (c) and BR/Rh223-253 (d) are monitored at 412 nm, 570 nm, and 643–680 nm, which monitor the M intermediate, the depletion of BR, and the O intermediate, respectively. (Dotted lines) Data points, which were averaged for 50 experiments. (Solid lines) Fitting curves (412 nm; single-exponential, 570 nm; double-exponential, 643–680 nm; double-exponential of rise and decay). One division of the y axis corresponds to 0.02 absorbance unit.

Another important piece of information is gained from the green line in Fig. 3 b, where the transient depletion at 500 nm of SRII/Rh225-252 mostly recovers. We have estimated the bleaching yield of SRII/Rh225-252 to be <10%, indicating that SRII/Rh225-252 exhibits the similar photocycle to the wild-type SRII, but some portion (<10%) cannot return to the original state. Similar kinetics and bleaching yield were obtained for SRII/Rh224-252 (data not shown).

Photoreaction kinetics among all SRII/Rh chimeras were compared at 4°C, where absorbance change was monitored after the 4-s illumination of the >480-nm light in spectrophotometer. Table 1 compares the time constants (τ_1/e) of the decay of the M intermediate, and the recovery of the original state. The decay time constants (τ_1/e) of the M intermediate of nine chimeras were similar, ranging between 1.1 and 2.1 s. There are no clear correlations between the M-decay and their sequences, and the decay is ∼2-times slower than the wild-type. The ground-state recovery was fitted by a single exponential for the wild-type SRII, but two components were needed for nine chimeras. The fast major (>80%) and slow minor time constants were in the range between 2 and 4 s, and between 8 and 35 s, respectively. Photoreaction kinetics of the nine chimeras were similar with each other, and as described below, there are no correlations between the photoreaction kinetics and G-protein activity.

Table 1.

Time constants (τ_1/e; s) of the M decay, and original-state recovery monitored at 360 nm, and 500 nm, respectively, at 4°C

	M Decay	Recovery
Wild-type SRII	0.7	1.6
SRII/Rh223-251	1.7	3.5 (81%), 22.6 (19%)
SRII/Rh223-252	1.8	3.9 (80%), 27.9 (20%)
SRII/Rh223-253	2.1	3.3 (82%), 22.2 (18%)
SRII/Rh224-251	1.5	3.0 (85%), 23.0 (15%)
SRII/Rh224-252	1.5	2.6 (94%), 31.0 (6%)
SRII/Rh224-253	1.4	2.6 (90%), 23.0 (10%)
SRII/Rh225-251	1.1	2.6 (91%), 9.8 (9%)
SRII/Rh225-252	1.1	2.5 (96%), 34.5 (4%)
SRII/Rh225-253	1.1	2.4 (95%), 7.9 (5%)

We have also examined the photochemical properties of the BR chimera, which was not reported in the previous study (11). Fig. 3, c and d, shows the results of BR and BR/Rh223-253, respectively, where photocycles are approximately two-orders-of-magnitude faster than in SRII (Fig. 3, a and b). The M decays were almost identical between BR (τ_1/e = 1.97 ± 0.05 ms) and BR/Rh223-253 (τ_1/e = 1.91 ± 0.10 ms). On the other hand, the O decay and the recovery are 4–6 times slower in BR/Rh223-253 than in BR. Thus, the effects of chimera on the photoreactions were very similar between SRII and BR.

G-protein activation properties of the SRII/Rh and BR/Rh chimera

Finally, we have measured G-protein activation by these chimeras. Fig. 4 shows the time-course of the binding of GTPγS to transducin, where light-dependent GDP/GTPγS exchange is monitored using [³⁵S]GTPγS. In the case of bovine Rh, light-dependent G-protein activation was clearly observed by the significant binding of GTPγS (Fig. 4 a, dotted lines). Fig. 4 a also shows that G-protein activation by SRII/Rh225-252 is little, regardless of light and dark conditions, which are comparable to that of bovine Rh in the dark. It should be noted that the final concentration of SRII/Rh225-252 is 2.4 μM, which is much more numerous than that of bovine Rh (5 nM), but the GTPγS binding is negligible (Fig. 4 a). Nevertheless, light-dependent G-protein activation by SRII/Rh225-252 is observable if the y axis of Fig. 4 a is expanded. Fig. 4 b shows the results of the wild-type SRII and SRII/Rh225-252. In the case of the wild-type SRII, there are no changes in the GTPγS binding between light and dark (dotted lines), indicating no G-protein activation by SRII. In contrast, clear light-dependent GTPγS binding is observed for SRII/Rh225-252, although it is tiny if compared to the case of bovine Rh.

Figure 4.

G-protein activation by SRII/Rh (a and b) and BR/Rh (c and d) chimeras. Time-dependent GTPγS-binding to transducin is monitored under light (open circle or square) and dark (solid circle or square) conditions. (Solid lines, a and b) Fitting results of SRII/Rh225-252. (Dotted lines, a and b) Fitting results of bovine Rh and wild-type SRII, respectively. (Solid lines, c and d) Fitting results of BR/Rh223-253. (Dotted lines, c and d) Fitting results of bovine Rh and wild-type BR, respectively. Concentrations of bovine Rh, SRII (SRII chimera), and BR (BR chimera) are 5 nM, 2.4 μM, and 1.6 μM, respectively.

It is now possible to compare, quantitatively, the activation level between bovine Rh and SRII/Rh225-252. For bovine Rh, the GTPγS binding is calculated to be 3.75 mol/min/mol, using the molar extinction coefficient of bovine Rh (40,600). For SRII/Rh225-252, we have calculated the GTPγS binding to be 99 × 10⁻⁶ mol/min/mol (Table 2), assuming that the molar extinction coefficient of SRII/Rh225-252 is identical to that of the wild-type SRII (42,000) (13,22). Thus, SRII/Rh225-252 can activate G-protein, but the activation efficiency is ∼38,000 times lower than that of bovine Rh.

Table 2.

Light-dependent G-protein activation of the wild-type and chimeric microbial rhodopsins

	Activity(μmol/min/mol pigment)
Wild-type SRII	15 ± 12
SRII/Rh223-251	36 ± 19
SRII/Rh223-252	32 ± 22
SRII/Rh223-253	27 ± 17
SRII/Rh224-251	42 ± 20
SRII/Rh224-252	38 ± 21
SRII/Rh224-253	44 ± 22
SRII/Rh225-251	95 ± 25
SRII/Rh225-252	99 ± 26
SRII/Rh225-253	63 ± 20
Wild-type SRII (PC liposomes)	13 ± 10
SRII/Rh225-252 (PC liposomes)	126 ± 30
Wild-type BR	22 ± 18
BR/Rh223-253	102 ± 30

Light-dependent activity in the microbial rhodopsin (48 pmol SRII or 32 pmol BR) was calculated by the difference of the activities in the light and dark conditions as shown in Fig. 5.

We have also tested G-protein activation by BR/Rh chimera. Fig. 4 c compares the results between bovine Rh and BR/Rh223-253 (final concentration; 1.6 μM BR and 5 nM bovine Rh). As is the case for SRII (Fig. 4, a and b), G-protein activation by the BR/Rh chimera is not clear in Fig. 4 c. However, the expanded Fig. 4 d shows that there is no activation for the wild-type BR (dotted lines), but clear light-dependent GTPγS binding is observed for BR/Rh223-253. We have calculated the GTPγS binding to be 102 × 10⁻⁶ mol/min/mol, assuming that the molar extinction coefficient of BR/Rh223-253 is identical to that of BR (63,000). This study shows that BR/Rh223-253 can activate G-protein like bovine Rh, but the activation is ∼37,000 times lower than in bovine Rh.

Fig. 5 summarizes the G-protein activation by the wild-type and chimeric proteins. For the wild-type SRII and BR, GTPγS bindings are slightly higher under light than dark, but the difference is within the error. Therefore, we conclude that the wild-type SRII and BR cannot activate G-protein. In contrast, nine SRII/Rh chimeras appear to exhibit light-dependent G-protein activation, though the activation levels of SRII/Rh223-251, 223-252, and 223-253 are statistically insignificant. The activation values are shown in Table 2. As is shown, most of the SRII/Rh chimeras can activate G-protein, but the activation level is much smaller than in bovine Rh. The activation level ranges from 27 × 10⁻⁶ mol/min/mol (SRII/Rh223-253) to 99 × 10⁻⁶ mol/min/mol (SRII/Rh225-252). If the M intermediate is the active state, one may expect the positive correlation between the lifetime of the M state and G-protein activation. However, the M-decay and ground-state recovery kinetics are similar among chimeras, and there is no clear correlation. Instead, three chimeras whose insertion starts at position 225 show higher G-protein activation than others, suggesting the important role of the sequence alignment.

Figure 5.

Comparison of G-protein activation ability by SRII/Rh and BR/Rh chimeras. GTPγS-binding to transducin is monitored at 10 min under light (open bar) and dark (solid bar) conditions. Data are presented as the means ± SD of more than three independent experiments, and the marked chimeras (^∗) exhibit a significant difference between light-dependent and dark activations (p < 0.05; Student's t-test, one-tailed).

It should be noted that SRII/Rh chimeras are solubilized by detergent, and the G-protein activation may be different for the membrane-embedded proteins. Therefore, we have also tested the G-protein activation for the wild-type SRII and SRII/Rh225-252 reconstituted into PC liposomes. Fig. 5 shows that the light-dependent activity is higher in PC liposomes (126 × 10⁻⁶ mol/min/mol) than in detergent (99 × 10⁻⁶ mol/min/mol), but the activity is not significantly enlarged. We have also tested the G-protein activity of P370 (Fig. S3). If P370 mimics the active state, it may be constitutively active. The activation level of P370 is obtained to be (46 ± 18) × 10⁻⁶ mol/min/mol, which is comparable to the activation level of the dark state of SRII chimeras. We thus conclude that P370 does not activate G-protein.

Discussion

Properties of chimeras

In this study, we have prepared the chimeric proteins of SRII and BR containing the third loop of bovine Rh. All nine constructs of the SRII/Rh chimeras show characteristic absorption spectra (Fig. 2 and Fig. S1), indicating no structural perturbation of the retinal binding region by the replacement of the loop. Photoreaction cycles of the SRII chimeras are not much altered from the wild-type (Fig. 3), but we have found, during the photoreaction cycles, some portion is bleached like bovine Rh (Fig. S2). This fact suggests that introduction of the Rh loop does not affect the protein structure in the dark, but destabilizes the intermediate state during photocycle, resulting in the formation of P370. P370 contains the Schiff base linkage with Lys²⁰⁵, as shown by the acid denaturation experiment (Fig. S3) and the HPLC analysis has shown that P370 possesses 13-cis retinal (data not shown).

Thus, an M-like conformation is formed that is irreversible to the original state for all SRII chimeras. In contrast, BR/Rh223-253 does not form such a product, and fully returns to the original state after photoexcitation. It should be noted that the photoreactions of the SRII and BR chimeras are tested in detergent and purple membrane, respectively, suggesting that membrane environment is more stable for protein. However, P370 is also formed for the SRII chimera reconstituted into PC liposomes (data not shown). Thus, it is likely that BR is more stable than SRII in terms of photocycling reactions of chimera possessing the third loop of Rh. Organization of BR into a two-dimensional lattice, purple membrane, might be important for such stability.

G-protein activation by chimeras

This study shows that most of the chimeras of SRII and BR activate a G-protein transducin, but the activation is tiny compared to that of bovine Rh, the native receptor. Table 2 shows the activity to be 27–126 × 10⁻⁶ mol/min/mol, which is ∼30,000–140,000 times lower than that of bovine Rh. It should be noted that a strict quantitative comparison can only be made if Rh and the SRII and BR chimeras are fully bleached with light once and remain in the active conformation during the assay period. This holds for bovine Rh, which is rapidly converted to Meta-II under these experimental conditions. In contrast, SRII and BR chimeras are activated repeatedly by the continuous illumination. In this sense, the relative efficiency of transducin activation by SRII and BR chimeras may be lower than 1:30,000–1:140,000 of bovine Rh.

The observed large difference in activity is reasonable, we believe, because bovine Rh must be optimized to activate bovine transducin during its evolution. The activation process includes

• 1.Exposure of the interaction site in Rh.
• 2.Complex formation with transducin.
• 3.GDP release from the complex.
• 4.GTP binding to the complex.
• 5.Dissociation of transducin from Rh, for which the cytoplasmic surface structure of Meta-II, the active state of Rh, is important.

Crucial role of the third and second loops as well as the eighth helix in the transducin activation has been reported (21,23,24). In contrast, these SRII and BR chimera contain only the third loop of bovine Rh. In addition, because of photocycling reactions of microbial rhodopsins, lifetime of the active state is much shorter in the chimera, which must also contribute to their lower activity compared to that of bovine Rh. Therefore, we conclude that the obtained less-efficient G-protein activation (10⁴–10⁵ times) is reasonable for the chimeras, and greater activity will be gained by the further improvement of chimeras such as additional introduction of the second loop and/or long-lived active states.

A puzzling issue out of this study is the discrepancy to the previous results. Geiser et al. (11) had reported that the G-protein activation by BR/Rh223-253 was ∼25% of that by bovine Rh. According to our measurements, the G-protein activation by BR/Rh223-253 is 0.0003 of that by bovine Rh, not 0.25, which was reported by Geiser et al. (11). The huge quantitative discrepancy (1000 times) between the two reports is unclear, but it may originate from

• 1.Our underestimate of the value of BR/Rh223-253, or
• 2.An underestimate by Geiser et al. (11) of the value of bovine Rh.

We have carefully checked the G-protein activation conditions of chimeras from nM concentration, which is the standard condition of bovine Rh (5 nM in this study). Nevertheless, we could not observe the activation, so that we have increased the concentration of chimeras and finally we have established the concentration of 1–3 μM for chimera, where small but reproducible light-dependent activity is observed. The experimental conditions are similar between these and previous experiments except for the concentration of receptors. In this study, we have used 5 nM bovine Rh and 1.6 μM chimera, where GTPγS binding increases linearly with illumination time (Fig. 4). Geiser et al. (11) had used 0.5–10 μM receptors, indicating that the concentration of bovine Rh was >100 times higher than in this study. Possibly, the high Rh concentration in their experiment led to saturation of GTPγS binding. For example, using this transducin activity by bovine Rh (3.75 mol/min/mol), transducin would be fully bound with GTPγS within the first minute for a 2.3:1 Rh/transducin ratio in Geiser et al. (11), but the data reported appears to be a single 10-min point. This would lead to a gross underestimate of the value of bovine Rh. We thus emphasize the importance of the linear increase of the G-protein activation with the stimuli (light in this case), which is crucial for the quantitative comparison.

G-protein activation mechanism by chimeras

There are no sequence homologies between visual and microbial rhodopsins, but extensive studies have suggested similar structural changes at the cytoplasmic surface. For instance, spin-labeling studies reported transient enlarged distance between helices 3 and 6 for all bovine Rh (25–27), BR (28–30), and SRII (30–32), which is suggested to originate from outward motion of the sixth helix. The same conclusion was obtained for bovine Rh from engineered metal-ion-binding study (33) and x-ray diffraction (34), for BR from cryo-electron microscopy (35), x-ray diffraction (36,37), and neutron diffraction (38). In addition, recent high-speed atomic force microscopy directly visualized real-time and real-space imaging of the opening motion of the E-F loop (third loop) by 7 Å in D96N BR (39). Such motion exposes the inserted third loop, enabling formation of a complex with transducin. This study indeed supports the common structural changes, because most of the SRII and BR chimeras activate G-protein.

Lower activation of the chimera may be correlated with the extent of the outward motion of the helix 6 (F-helix). According to the x-ray crystallographic analysis of bovine Rh, helix 6 relatively moves by 5–7 Å between the dark and peptide-bound opsin states, the latter of which mimics the active state (34). Compared to such large motion in bovine Rh, outward motion of the F-helix is smaller in microbial rhodopsins. It should be noted that the 7 Å outward motion observed by high-speed atomic force microscopy is at the cytoplasmic surface (top) of the E-F loop of D96N BR (39), and cryo-electron microscopy study of D96G/F171C/F219L BR estimates 3.5 Å outward motion at the top of F-helix (35). Such smaller structural alteration in microbial rhodopsins may be also correlated with the fact that microbial rhodopsins exhibit photocycles, which is not the case for visual Rh. Using a site-directed fluorescence labeling approach, Tsukamoto et al. (40) showed that the relative movement of helices 5 and 6 upon photoactivation is much greater in bovine Rh than in parapinopsin (a nonvisual rhodopsin), which explains the 20-fold-more-effective G-protein activation of bovine Rh than that of parapinopsin (41). They concluded that the different amplitude of the helix movement is responsible for functional diversity of G-protein-coupled receptors, and this conclusion may be applicable to the chimeric microbial rhodopsins.

In the crystal structure of bovine opsin complexed with the C-terminal peptide of the α-subunit of transducin, Tyr²²³ forms a hydrogen bond with Arg¹³⁵ of the conserved ERY motif in bovine Rh, which further interacts with the main-chain carbonyl of Cys³⁴⁷ of the transducin peptide (34). Because the hydrogen bond between Tyr²²³ and Arg¹³⁵ is absent in the unphotolyzed Rh, structural rearrangement of Tyr²²³ is important for G-protein activation. In contrast, these results report that the three SRII chimeras possessing Tyr²²³ only exhibit statistically insignificant activation (Fig. 5). Apparently negative effect of tyrosine at this position may originate from the absence of the ERY motif (Arg¹³⁵) in chimeras. We observed that the starting sequence of the inserted loop (Fig. 1) is more sensitive to transducin activation than the ending sequence, where the activation level of transducin is greatest for SRII/Rh225-25x, middle for SRII/Rh224-25x and statistically insignificant for SRII/Rh223-25x (Fig. 5). This may suggest that the orientation of the E-helix side is more important than that of the F-helix side in the transducin activation by SRII chimeras.

The active state of bovine Rh is Meta-II that has a deprotonated Schiff base. Similarly, it is suggested that the largest structural changes occur in the M intermediate of BR (28–30,35–38) and SRII (30–32), which has a deprotonated Schiff base. If it is the case, however, a question arises why G-protein activation is similar between SRII and BR chimeras (Fig. 5). As shown in Fig. 3, photocycle of the SRII chimera is much slower than that of the BR chimera. In fact, the time-constants of the M decay are 1.91 ± 0.10 ms and 173 ± 5 ms for BR/Rh223-253 and SRII/Rh225-252, respectively (Fig. 3), while the G-protein activity is similar (Fig. 5 and Table 2). This indicates that the lifetime of the M intermediate is not simply correlated to G-protein activation.

The determinant of the G-protein activation for chimeras is unclear at present, but structural changes of SRII may be smaller than those of BR. Another possibility is that the structural changes at the loop region occur on a faster timescale than the lifetime of the M intermediate in the SRII chimera. In that case, transient absorption technique is not sufficient, and we need different spectroscopic methods to monitor protein structural changes such as infrared spectroscopy (42,43) and transient grating (44).

The effects of transducin (or its peptides) on the photocycles of chimeras is also interesting. Further experimental study including other chimeras and/or chimeras of other microbial rhodopsins will reveal a more-detailed mechanism of the G-protein activation.