The structures of the active center in dark-adapted bacteriorhodopsin by solution-state NMR spectroscopy

Abstract

The two forms of bacteriorhodopsin present in the dark-adapted state, containing either all-trans or 13-cis,15-syn retinal, were examined by using solution state NMR, and their structures were determined. Comparison of the all-trans and the 13-cis,15-syn forms shows a shift in position of about 0.25 Å within the pocket of the protein. Comparing this to the 13-cis,15-anti chromophore of the catalytic cycle M-intermediate structure, the 13-cis,15-syn form demonstrates a less pronounced up-tilt of the retinal C12—C14 region, while leaving W182 and T178 essentially unchanged. The N—H dipole of the Schiff base orients toward the extracellular side in both forms, however, it reorients toward the intracellular side in the 13-cis,15-anti configuration to form the catalytic M-intermediate. Thus, the change of the N—H dipole is considered primarily responsible for energy storage, conformation changes of the protein, and the deprotonation of the Schiff base. The structural similarity of the all-trans and 13-cis,15-syn forms is taken as strong evidence for the ion dipole dragging model by which proton (hydroxide ion) translocation follows the change of the dipole.

A number of important biological processes, including the reception of hormon- or light-encoded signals, depend on the function of hepta-helical transmembrane proteins (1, 2). Certain members of this family, like the visual pigment rhodopsin and the archaeal proton pump bacteriorhodopsin (BR), contain a light reactive prosthetic retinal group that is attached via a protonated Schiff base linkage (−C = NH+) to the ɛ-amino group of a lysine residue in the seventh helix. Their light sensitivity is based on photoisomerization of this retinal moiety. The specific amino acid composition in the active center determines the wavelength of chromophore absorption, the energy barriers for bond rotation in the retinylidene moiety, and the regioselectivity of isomerization. Pioneering structural investigations were done with electron microscopy (3). Recently, high-resolution three-dimensional structures of BR (4–13) (initial state and M-, K-intermediates), the related halorhodopsin (14), and rhodopsin (15) have been made available.

Bacteriorhodopsin converts light energy into that of a proton gradient that is subsequently used by the transmembrane protein ATP-synthase to produce chemical energy in the form of ATP. At the beginning of the photocycle, the catalytically active form of BR absorbs a photon by the all-trans,15-anti retinal cofactor that is linked to K216 forming a Schiff base. During the photoexcitation, the retinal undergoes a rearrangement of the electronic structure of its extended conjugated π system, resulting in the trans → cis isomerization to a 13-cis,15-anti form, and causing reduced proton affinity on the charged Schiff base. The M intermediate is formed by the transfer of a proton from the Schiff base to D85 in the extracellular half-channel. Reprotonation of the Schiff base is mediated by D96 in the cytoplasmic half-channel of the protein in response to the large conformational change (16, 17). In contrast to most retinal proteins, the ion pumps BR and halorhodopsin have the unique capacity to thermally re-isomerize the retinal moiety from the 13-cis,15-anti to the all-trans,15-anti state in milliseconds, a reaction that is completely prevented for a free protonated Schiff base retinal in solution by a high-energy barrier (Fig. 1).

Figure 1.

(a) Configurations and atom numbering of the chromophore in the binding pocket of bacteriorhodopsin during the initial state and the photocatalytic cycle. (b) Regions of NOESY spectra of uniformly deuterated BR samples containing ¹H-retinal (Left and Center), and ¹H-retinal, threonine, and lysine (Right). Cross peaks denoted with c or t and an atom number are caused by interactions involving the 13-cis,15-syn or the all-trans form of the chromophore, respectively. A capital letter indicates the respective amino acid type. All spectra were recorded with mixing times of 40 ms. The starting point for the assignment were short intraresidue distances between retinal protons in the two isomers of the dark-adapted form, causing characteristic cross peak patterns (Left and Center). H15 in the case of the all-trans form, and H14 in case of the 13-cis,15-syn-form show only one strong NOE to a methyl group, which at the same time is involved in NOE with H11 (Center). H₁₅ of the 13-cis,15-syn form shows NOE involving H₁₂ and H₁₄ (Left).

The active center in its initial state can accommodate, in addition to the all-trans form, a second 13-cis form that, however, is in a 15-syn configuration. This chromophore does not release a proton, and therefore does not mediate proton translocation but equilibrates thermally with the active all-trans form in the minute range. The 40:60 mixture of all-trans/13-cis,15-syn is called dark-adapted BR (Fig. 1a). The reaction is catalyzed by nucleophiles in the surrounding protein scaffold that are also involved in the thermal re-isomerization of the 13-cis,15-anti retinylidene group during the light-induced catalytic cycle. A detailed description of the structural differences of the two 13-cis forms is of key importance for the understanding of the functional isomerization reaction in the proton pump. High-resolution structural information on the 13-cis,15-syn form of the chromophore present in the dark-adapted state is not yet available (4–13), and, for principal reasons, is very difficult to obtain by x-ray crystallography from the mixture present in the wild-type protein.

Here, solution-state NMR is applied to derive this structure by using uniformly ²H-labeled but amino acid-type specifically ¹H-labeled, detergent-solubilized samples. This approach was chosen because carbon line widths up to 200 Hz were observed in selectively carbon labeled samples (18), preventing the application of heteronuclear three-dimensional nuclear Overhauser effect spectroscopy (NOESY) techniques to resolve structure-relevant side chain-retinal nuclear Overhauser effects (NOEs). Similar experiences were made with a 75% ²H/100% ¹³C,¹⁵N-labeled sample. Chemical shift data were used to calculate quality parameters for selecting structures (19). The present work is seen as supplementary to transverse relaxation-optimized spectroscopy (TROSY)-based procedures (20) that were applied to porins (21, 22).

Materials and Methods

Samples.

Residue-specifically labeled purple membranes (PM) were prepared as described (18). BR was then solubilized in dodecyl maltoside with a deuterated dodecyl moiety (dDM), either by detergent exchange of Triton X-100 micelles or by direct, ultrasound-assisted solubilization of PM in dDM solution: PM (12–20 mg BR) was suspended in a solution of 1% dDM in 10 mM potassium phosphate buffer in D₂O at a measured pH of 5.6. The slurry was sonicated for 15 min (Branson Sonifier, Cell Disruptor B15, Microtip, pulsed 50%, output control 5, T < 30°C) before a spectrum was recorded. Sonication was repeated until no further decrease of λ_max was observed (2–4 times). The solution was diluted to 10 ml with buffer in D₂O and centrifuged (80,000 × g, 15°C, 15 min). The supernatant was concentrated to ≈0.5 ml (Filtron; 50-kDa exclusion size, 5,000 × g, 10°C, 1 h). Excessive detergent was removed by a further cycle of dilution with the buffer and concentration. All samples contained approximately 120 molar equivalents of dDM. In total, six samples were used for the assignment of NOEs entered into structure calculations (Table 1).

Table 1.

Samples used in this study

Sample no.	1H-residue
I	Ret
II	Trp, Thr
III	Ret, Trp, Thr
IV	Ret, Thr
V	Ret, Trp
VI	Ret, Thr, Lys

Sample stability was monitored by UV-visible spectroscopy before and after measurements, and proved to be stable over weeks, including several days of measurements at elevated temperatures.

NMR-Spectroscopy.

NOESY spectra were recorded at 308, 313, and 318 K at 800.13 or 750.13 MHz with mixing times in the range of 5–160 ms for each sample. The ¹H²HO resonance was saturated by mild irradiation during the relaxation delay. The spectra were processed with appropriate weighting functions and baseline corrections after Fourier transformation.

NOE cross peak intensities were translated into distance constraints by either integration and comparison to a reference interaction, or estimated and used as a distance range in the calculations. Nonoverlapped cross peaks appearing in spectra with mixing times of 10 ms or less were integrated, assuming direct proportionality between the NOE intensity and the distance between the two contributing proton types. This assumption was validated by monitoring a linear buildup rate of the NOEs in this regime (data not shown). The isolated intraresidue NOEs of W86 were used as a reference.

Distances were then calculated by

where I is the integrated intensity, r_ref and I_ref are the reference distance and integral, respectively, and n is the number of involved protons (23). Overlapping peaks where assigned to a distance range (see below) by comparing the isolated part of the peak volume to reference peaks.

Cross peaks in spectra with mixing times of 20, 40, and 80 ms were assumed to represent a distance belonging to one of three distance range classes, taking into account that spin diffusion affects the peak intensity considerably. The distance range of class I was set to 3.0–4.5 Å, of class II was set to 4.0–6.5 Å, and of class III was set to 4.5–8.0 Å. Cross peaks were assigned to a specific class after comparison of their intensity with those of integrated cross peaks.

Structure Calculations.

The NMR structures were calculated by using the program amber (version 4.1) (24), and the x-ray structure 1BRR as a template. A simulated annealing protocol (25) was applied in that only atoms within a distance of 10 Å from the retinal were allowed to move (elevating temperature to 1500 K at 0–5 ps, cooling down to 0 K at 5–10 ps with slowly increasing van der Waals repulsion terms from 0.001 to 1.0 between 0 and 5 ps). Point charges for a protonated chromophore have been calculated with gamess u.s. (26) and the resp programs following the method of Cieplak et al. (27, 28). Torsion barriers for the alternating single and double bonds of protonated retinal were set to 10 kcal/mol, according to quantum chemical calculations (29). D96 and D115 were included as protonated (neutral) amino acids. Four hydrogen-bonded water molecules observed in the crystal structure 1C3W were included in the calculation (including the water molecule between D85 and the Schiff base proton). Hydrogen bonds in helices were restrained to their original values at the end of the calculation. Forces representing distance restraints were applied by way of a square-well function (rk = 50.0 kcal⋅mol⁻¹⋅Å⁻²). A distance of 0.4 Å was added to a determined value to allow for methyl pseudo atoms, 1.5 Å was added to take account of the C₁₆/C₁₇ pseudo atom (23). From both, the all-trans and the 13-cis,15-syn calculations, ensembles of 12 structures with low chemical shift penalties were selected from 100 simulated annealing runs.

Chemical Shift Calculations.

Chemical shifts were calculated from the coordinates by using the model in the program shifts with the parameters of Ösapay and Case (30). The application of an alternative parameter set using larger ring current intensity factors based on quantum chemical calculations (31) led to a significant overestimation of the ring current contributions to the chemical shifts as judged by comparison to the experimental results.

Structure Comparisons.

All distance and chemical shift deviations are specified as the average standard deviation of the corresponding values in the 12 members of a structure family. The average displacement of the polyene chain atoms C₁ to C₁₁ in the 13-cis, 15-syn family of structures with respect to the all-trans family of structures was calculated by (i) rotating first each member of the all-trans family into a coordinate system with the axis along the principle components of inertia of the C₅-C₁₅ fragment of the retinal to generate a reference position. In this system, the x- and y-axis corresponding to the largest and second largest moments of inertia are in the plane of an ideal polyene chain, with x pointing along its main axis. (ii) Overlaying each member of the 13-cis,15-syn family of structures individually to each member of the all-trans family of structures by minimising the backbone rmsd of each pair of structures; and (iii) determining the components of the translation vector by which each member of the 13-cis,15-syn family has to be moved to superimpose the retinal atoms C₁—C₁₁ of each pair of structures (atoms C₁—C₁₁ have been chosen because the conformation of this part of the molecule is unchanged in all-trans and 13-cis,15-anti conformations).

Results

Appropriately labeled BR samples were prepared by growing retinal-deficient cells in a fully deuterated medium complemented with 1H-retinal, 1H-threonine, and 1H-tryptophan in appropriate combinations and investigated by NOESY for resonance assignment and collection of distance constraints (Fig. 1b). One sample contained also 1H-lysine. A combined evaluation of six samples (see Table 1) according to the exclusion principle yielded the classification of cross peaks as intraresidue tryptophan, threonine, or retinal, and interresidue Trp–Thr, Trp–Ret, or Thr–Ret signals (Table 2).

Table 2.

Sets of spectra used for assignment of a certain type of inter-residue NOE

NOEs	Retinal	Trp	Thr
Retinal	I	III, V	III, IV, VI
Trp	–	II, III, V	II, III
Thr	–	–	II, III, IV

A comparison of 6 different spectra allows a unique identification of intra- and inter-residue peaks for the residues threonine, tryptophan, lysine, and the retinal. Cross peaks between tryptophan and threonine, for example, show up in spectra II and III but must not occur in I, IV, or V. 

The structure of the all-trans, 15-anti form was determined first to estimate the precision of the procedure by comparing the result to the available x-ray structures (4–13). The structure was calculated by using 32 distance constraints (see Table 3), and the x-ray structure 1BRR (6) as a template. The three water molecules that were identified in cavities around the Schiff base by Luecke et al. (7) were included in the calculations. A simulated annealing protocol (25) was applied, allowing only atoms in a sphere of 10 Å around the retinal to move. To determine the influence of the NOE-derived potential and of the force field on the final result, calculations without NOE constraints and without including water molecules were also performed. When NOE constraints were used, nearly all calculations converged to globally similar but locally divergent structures with only insignificant energy differences. We have therefore used the chemical shifts of the assigned protons as additional selection criteria. Final structures of the active center of bacteriorhodopsin were then selected by using the rms distance (RMSD) between observed and calculated chemical shifts for 14 proton signals as a criterion (Table 4). The chemical shifts were calculated by using the program shifts (V.3.062) with the original parameter set from Ösapay and Case (30).

Table 3.

Classification of NOE restraints

No. of NOE restraints used	cis	trans
Total	39	32
Integrated peaks	6	1
2.90 Å (±10%) (class I)	7	7
3.50 Å (±10%) (class I)	4	4
3.0–4.5 Å (class II)	3	5
4.0–6.5 Å (class III)	14	8
4.5–8.0 Å (class IV)	5	7

Table 4.

Comparison of observed and calculated chemical shifts for selected proton signals

Proton	13-cis,15-syn		all-trans				M
	Exp.	Calc.	Exp.	Calc.			Calc.
		NMR		1c3w	1qhj	NMR	1c8s
W86 Hδ1	6.64	6.89  ± 0.02	6.88	6.82	6.85	6.90  ± 0.03	6.86
W86 Hζ2	7.06	6.87  ± 0.09	6.91	6.85	6.93	6.83  ± 0.07	7.27
W86 Hη2	5.97	5.89  ± 0.26	6.15	6.26	6.22	6.00  ± 0.22	5.90
W86 Hζ3	6.64	6.33  ± 0.12	6.73	6.85	6.74	6.46  ± 0.07	6.39
W86 Hɛ3	6.99	6.85  ± 0.11	7.03	7.18	7.15	6.90  ± 0.06	7.23
T90 Hα	4.24	4.10  ± 0.09	4.26	4.09	4.13	4.06  ± 0.09	3.76
T90 Hβ	4.56	4.18  ± 0.12	4.60	3.99	4.01	4.18  ± 0.11	4.18
T90 Hγ2	0.28	0.64  ± 0.13	0.22	0.36	0.35	0.60  ± 0.08	0.64
T142 Hα	3.63	3.78  ± 0.02	3.63	3.68	3.77	3.78  ± 0.03	3.70
T142 Hβ	3.43	3.83  ± 0.20	3.43	3.81	3.93	3.96  ± 0.06	3.81
T142 Hγ2	0.83	0.71  ± 0.08	0.83	0.76	0.79	0.76  ± 0.03	0.82
T178 Hα	3.87	4.08  ± 0.04	3.77	3.93	4.04	4.05  ± 0.04	4.65
T178 Hβ	4.08	4.25  ± 0.03	4.00	4.26	4.27	4.24  ± 0.03	4.41
T178 Hγ2	−0.05	0.35  ± 0.06	−0.13	−0.37	−0.38	0.31  ± 0.09	0.34

Twelve structures of the calculated all-trans form were selected in this manner and compared with the best-resolved x-ray structures as shown in Fig. 2a, the helices are indicated with cartoons and the side chains of the amino acids in the environment that contributed NOE constraints are included. Defined positions for the retinal, for the tryptophan side chains 86, 138, 182, and 189, for threonines 90, 142, and 178, and for the Schiff base moiety are observed, but strong variations of the lysine side chain positions occurred (Table 5). The positions of the tryptophans 86, 138, and 189 and of the retinal match 1C3W very well, whereas the aromatic ring system of W182 in our structure is slightly parallel shifted with respect to 1C3W and 1BRR, expressed in a somewhat larger RMSD when comparing to the two x-ray structures (Table 6). This parallel shift is also the reason for the larger deviation between the chemical shifts of T90 and T178 H_γ2 calculated for the NMR ensemble and the observed ones (Table 4). On the whole, both 1C3W and 1BRR fit well into the spread of calculated structures of the all-trans form.

Figure 2.

(a) Comparison of the calculated structures of the all-trans form with the x-ray structures 1BRR (green) and 1C3W (magenta). (b) Superposition of 12 structures of the active center of the 13-cis,15-syn form of bacteriorhodopsin as derived by solution NMR, and comparison to the calculated structures of the all-trans form. (c) Comparison of the 13-cis,15-syn form (yellow) with the x-ray structures of the all-trans ground state 1C3W (magenta) and M (1C8S, orange). (d) Positional change of the N—H dipole during thermal and photoisomerization. Only photoisomerization positions the dipole in an assumed energetically unfavorable situation, therefore storing the energy for ion dragging and providing the driving force for the catalytic cycle.

Table 5.

RMSD of the cis- and trans-ensemble relative to the corresponding average structures

RMSD relative to the average structure	cis	trans
Cα atoms (TM regions)	0.07  ± 0.01	0.07  ± 0.01
Retinal	0.26  ± 0.09	0.28  ± 0.09
K216	0.48  ± 0.22	0.59  ± 0.17
Bound water molecule	0.67  ± 0.55	0.44  ± 0.31
W86	0.24  ± 0.08	0.23  ± 0.10
W138	0.11  ± 0.04	0.13  ± 0.05
W182	0.33  ± 0.10	0.25  ± 0.13
W189	0.18  ± 0.06	0.20  ± 0.08

Table 6.

RMSD of the trans ensemble relative to other BR structures

RMSD relative to other BR structures (TRANS)	AT/WT	AT/WT	AT/WT	AT/WT	AT/WT	AT/WT	M/D96N	AT/D96N	AT/WT	M/E204Q	AT/E204Q	AT/WT	AT/WT
	1C3W	1BRX	1BRR	1AT9	1AP9	2BRD	1C8S	1C8R	1BM1	1F4Z	1F50	2AT9	1QHJ
Cα (TM regions)	0.39  ± 0.01	0.40  ± 0.01	0.14  ± 0.01	0.59  ± 0.01	0.63  ± 0.01	1.00  ± 0.01	0.52  ± 0.01	0.40  ± 0.01	0.50  ± 0.01	0.42  ± 0.01	0.42  ± 0.01	0.60  ± 0.01	0.30  ± 0.01
Retinal	0.54  ± 0.07	0.62  ± 0.10	0.50  ± 0.09	0.93  ± 0.07	0.69  ± 0.08	0.75  ± 0.10	0.82  ± 0.08	0.59  ± 0.05	0.73  ± 0.11	0.57  ± 0.08	0.53  ± 0.06	0.58  ± 0.07	0.46  ± 0.06
K216	0.94  ± 0.14	0.90  ± 0.16	0.91  ± 0.14	1.00  ± 0.21	0.97  ± 0.20	0.95  ± 0.23	1.08  ± 0.21	0.95  ± 0.16	0.98  ± 0.26	1.12  ± 0.16	0.96  ± 0.15	1.00  ± 0.13	1.01  ± 0.17
W86	0.43  ± 0.09	0.36  ± 0.11	0.35  ± 0.08	0.85  ± 0.12	0.48  ± 0.12	0.84  ± 0.09	0.60  ± 0.09	0.43  ± 0.10	0.37  ± 0.08	0.40  ± 0.10	0.50  ± 0.09	0.48  ± 0.09	0.38  ± 0.09
W138	0.47  ± 0.08	0.54  ± 0.06	0.27  ± 0.06	1.12  ± 0.07	0.57  ± 0.09	0.34  ± 0.05	0.32  ± 0.08	0.50  ± 0.09	0.34  ± 0.03	0.48  ± 0.09	0.47  ± 0.09	0.73  ± 0.07	0.36  ± 0.09
W182	1.17  ± 0.22	1.50  ± 0.21	0.80  ± 0.20	2.54  ± 0.08	0.80  ± 0.07	1.26  ± 0.15	1.14  ± 0.11	1.25  ± 0.22	0.81  ± 0.15	0.71  ± 0.20	1.17  ± 0.22	1.37  ± 0.16	1.02  ± 0.22
W189	0.67  ± 0.18	0.68  ± 0.16	0.30  ± 0.14	1.87  ± 0.08	1.36  ± 0.13	0.86  ± 0.15	0.43  ± 0.06	0.61  ± 0.18	0.82  ± 0.17	0.48  ± 0.14	0.75  ± 0.17	0.94  ± 0.17	0.45  ± 0.17

For all structure comparisons, the ensemble of structures was fitted on either the average structure or the reference structure using all Cα atoms. In case of 1BRR.pdb (L. O. Essen and D.O.), monomer B was used for structure comparisons. AT, all-trans form; WT, wild type; M, M form. 

The structures of the 13-cis,15-syn form were then calculated by using 39 NOE signals. The 12 structures with lowest chemical shift RMSD (Table 5) are shown in superposition in Fig. 2b (yellow), including a comparison of both sets of NMR structures (in cyan for all-trans,15-anti and yellow for 13-cis,15-syn). The two sets are on the whole very similar, but there is a small displacement of the retinal along its main axis by about 0.25 ± 0.25 Å. This effect is caused by subtle differences in the experimentally observed NOE patterns. No significant differences in the positions of the tryptophan or threonine side chains of both solution NMR structures could be observed. Both sets of structures reproduce reasonably well earlier measurements of the distance between retinal C₁₄ and W86 N_ɛ by solid state NMR (32). We observed distances of 4.6 ± 0.11 and 3.8 ± 0.15 Å for the 13-cis,15-syn and all-trans,15-anti forms, respectively, in comparison to 4.2 and 3.9 Å observed by solid state NMR. In both sets, the position of retinal C15 is well defined by NOE's involving W86 and T90, thus defining also the position of the Schiff-base nitrogen because of the double bond to C15 in the respective anti and syn orientations.

The ensemble of selected 13-cis,15-syn structures (yellow) is compared with the x-ray structures 1C3W (magenta) of the all-trans state and of the photointermediate M¹¹ (1C8S, orange) in Fig. 2c. Major structural differences between all three structures occur in the direct environment of the Schiff base, and in the area of W182 and T178. The latter residues experience a change in the M intermediate in comparison to the other two forms to give space for the methyl group at C₁₃ of the retinal, which involves also a slight displacement of the retinal moiety C₁₂—C₁₄ toward W182. This behavior is the only characteristic structural difference in the environment of the C₁—C₁₃ retinal portion between the all-trans form and the M intermediate, apart from a small longitudinal displacement of the retinal, and including a change in the position of the side chain of T178 in M. Interestingly, W182 shows the same position in both NMR structures, and its position in the 13-cis,15-syn form is much closer to the one in the x-ray structure of the all-trans form than to the one in M. Although closer to C₁₃, T90 is much less affected because the movement of the retinal is toward W182, and not T90.

The retinal chain in its 13-cis,15-syn form is thus harbored in an unchanged pocket in the protein, and a potentially occurring structural strain is most likely compensated by torsional angle changes within the Lys-216 side chain. However, the structure has to give space to the unavoidable changes in shape on trans-to-cis isomerization observed in the positioning of the C₁₂—C₁₅ moiety, resulting in a slight up-tilt of the C₁₂—C₁₄ area in the 13-cis,15-syn form toward the cytoplasmic face, and a tilting downwards of C₁₅, including methyl groups. The similarity of the protein structure around the retinal C₁—C₁₁ moiety suggests considerable rigidity, which minimizes the differences in retinal positions of the all-trans,15-anti and 13-cis,15-syn forms. This finding confirms earlier elastic incoherent neutron scattering experiments, which showed that the retinal environment is considerably more rigid than the rest of the protein (33).

Discussion

These observations lead to interesting conclusions concerning the mechanism of proton transfer. The photoisomerization of the all-trans to the 13-cis,15-anti form and the thermal isomerization of the all-trans to the 13-cis,15-syn form show functionally important differences. The direction of the N—H dipole of the Schiff base is changed on photoisomerization from down (extracellular) to up (cytoplasmic), but not on thermal isomerization. Therefore, only the photoisomerization places the proton into a potentially energetically unfavorable environment. Ion dipole dragging of a proton down to the aspartic acid 85 or a hydroxide ion (chloride in halorhodopsin; ref. 14) up leads to relaxation by deprotonation of the Schiff base in BR on photo isomerization, but not on thermal isomerization, of the all-trans chromophore (Fig. 2d). Therefore, the change of the N-H dipole is the primary cause for energy storage and its release in all subsequent steps in the catalytic cycle. The consequence of the deprotonation is a change in charge distribution in M that could cause the stronger up-tilt of the C₁₂—C₁₅ region and the changes around W182/T178, which are not observed for the protonated 13-cis,15-anti form.

A rationale for the ease of the thermal isomerization process and for the re-isomerization of the protonated Schiff base that is relevant in the photocycle can be deduced from a comparison of the structures of the two forms occurring in the dark and the M intermediate by analyzing the positions of two nucleophilic side chain atoms in the retinal binding pocket. The oxygen and sulfur atoms of S141 and M118 are located in the vicinity of electrophilic carbon atoms of the protonated chromophore, C₅ and C₇, respectively. It is, therefore, conceivable that a nucleophilic attack takes place that shifts the bond orders in the retinylidene moiety and thereby reduces the torsion energy of the formal double bond between C₁₃ and C₁₄. For several reasons, M118 appears to be the most favorable candidate for a nucleophilic attack: (i) sulfur has a softer nucleophilic character and larger nonbonding orbitals than an alcohol oxygen and should interact more readily with the soft electrophilic centers of a polarized polyene; (ii) the ethylene group of methionine would provide sufficient conformational flexibility; and (iii) the secondary sulfur would guarantee for the necessary reversibility of the interaction. Because of the structural similarity of the retinal C₁—C₁₂ moiety in the two investigated forms and the M state, M118 can contribute to both, the thermal re-isomerization during the catalytic cycle and the thermal equilibration of the two isomers in the initial state in the dark.

This hypothesis is supported by mutation studies on BR (34, 35). The mutant M118A shows a different equilibrium in the dark (78:22 all-trans/13-cis,15-syn, respectively), and a strongly reduced thermal decay of M. Milder effects are observed for the mutant M118E supposedly because of the electron-donating capacity of the glutamic acid head group. According to the data obtained on the mutant S141A, there is a smaller influence of S141 on the isomerization reactions. The thermal decay of M is only 6-fold slower, and the equilibrium in the dark is closer to the wild-type situation. In line with our expections, a cysteine in place of S141 accelerates the thermal decay. Also, 95% all-trans isomer is observed in the dark for the S141C mutant. A recent investigation of Shimono et al. (36) on sensory rhodopsin II (phoborhodopsin) has also demonstrated that these residues have some influence on the absorption spectrum. Sensory rhodopsin II lacks nucleophilic groups at these positions (V108, G130) and shows a blue-shifted absorption spectrum (≈500 nm). The spectrum of sensory rhodopsin II mutant V108M/G130S is considerably red-shifted. Similarly, the BR mutants M118A and S141A and the double mutant show blue-shifts. The fact that the local structure around the retinal C₁—C₁₃ moiety is unperturbed on thermal cis–trans isomerization ensures that M118 and S141 can still serve as nucleophiles, which explains the similar absorption maxima of both forms occurring in the dark.

In summary, the two NMR structures of the BR chromophore in its initial state described here bear out strong significance on the mechanism of active proton transfer in the molecule and the catalysis of thermal C⩵C double bond isomerization in bacteriorhodopsin.

Abbreviations

BR

bacteriorhodopsin

NOESY

nuclear Overhauser effect spectroscopy

NOE

nuclear Overhauser effect

RMSD

root mean square distance