2023 Volume 20 Issue Supplemental Article ID: e201006
It marked half a century since the discovery of bacteriorhodopsin two years ago. On this occasion, I have revisited historically important diffraction studies of this membrane protein, based on my recollections. X-ray diffraction and electron diffraction, and electron microscopy, described the low-resolution structure of bacteriorhodopsin within the purple membrane. Neutron diffraction was effective to assign the helical regions in the primary structure with 7 rods revealed by low-resolution structure as well as to describe the retinal position. Substantial conformational changes upon light illumination were clarified by the structures of various photointermediates. Early trials of time-resolved studies were also introduced. Models for the mechanism of light-driven proton pump based on the low-resolution structural studies are also described. Significantly, they are not far from the today’s understanding. I believe that the spirit of the early research scientists in this field and the essence of their studies, which constitute the foundations of the field, still actively fertilizes current membrane protein research.
In an era when sophisticated methodologies of membrane proteins had not yet been developed, research grappled with essential questions about the molecular mechanism of light-driven proton pumping in bacteriorhodopsin. Current membrane protein researchers need not have a deep knowledge of the early works. It is nevertheless important to understand how the problems were tackled in the first 30 years following the discovery of bacteriorhodopsin. In particular, how ground-breaking new diffraction methods were developed to obtain the low-resolution structure, and how important questions were addressed from this limited knowledge. Through these old studies, we should learn spirit as a scientist.
In world history, we were taught the concept of Before Christ (BC) and Anno Domini (AD). This concept is especially useful to understand Western history, because the Western civilization based on Christianity began at the change in era from BC to AD. In biology, we have encountered many critical and essential changes that transformed the science dramatically. I would like to call them, symbolically, transitions from BC to AD. A typical example is “Before Channelrhodopsin (ChR, BC)” to “After Deisseroth (AD)”. By this BC to AD transition, a new research field, optogenetics, was born. A few other examples are: (1) Before Cloning (BC) and After DNA (AD) for molecular biology; (2) Before Crystallization (BC) and After 3-Dimensional structure (AD) for structural biology; (3) Before Computer (BC) and After Desk-top (AD) for structural biology, theoretical and computational biology; (4) Before Computer programing (BC) and After Deep learning (AD) for theoretical and computational biology. Nobody will disagree that these changes from BC to AD have been brought about the revolutions in biology.
Two years ago, 2021 was the 50th anniversary of the discovery of bacteriorhodopsin (bR) [1]. Research in bR has been also influenced by these BC to AD changes. Several bacterial rhodopsins including ChR have been isolated and purified through persistent efforts of biochemical, physiological, and bacteriological researches for certain types of bacteria [2–6]. By the above-mentioned BC to AD change (1), exhaustive genome searches identified thousands of new bacterial rhodopsins [7,8], which developed and enlarged the retinal protein research. By the changes (2) combined with the change (3), structural studies of bR progressed significantly and are now completely different from the research in the BC era.
I am a structural biophysicist who worked on bR in the BC era. I left from the field when crystal structure analysis became a major stream of bR structural research. On the occasion when half a century passed since the discovery of bR [1], I have revisited historically important diffraction studies of this membrane protein, based on my recollections. This review is not intended a comprehensive review of bR structural studies, but is limited mainly to the studies until 2000. It may be biased by my interest. However, it is useful to understand how researchers tackled problems at the time. These ‘ancient’ studies are the fossils of this field. Similarly to Prof. Pǎǎbo, the Nobel laureate in physiology or medicine last year, in his analysis of DNA from fossil humans, I believe that the spirit of the early research scientists in this field (research DNA), including myself, constitutes a foundation to fertilize ongoing studies.
Following the discovery of the purple membrane (PM) of Halobacterium halobium [9], bR was discovered as the sole polypeptide in the PM and identified as the origin of the purple color [1]. At the same time of the discovery of bR [1], an X-ray diffraction pattern of PM suspension was also reported [10] (Fig. 1). A beautiful X-ray diffraction pattern of the PM suspension is composed of a series of concentric sharp rings and diffuse scatter (Fig. 1). By stacking the membrane through drying or centrifugation, the origin of the sharp rings was found to arise from the in-plane structure, a two-dimensional hexagonal array with a unit cell size of 63 Å [10]. 75% protein and 25% lipid content were estimated for the PM [1,10]. In the ref. [10], the authors focused on the structure perpendicular to the membrane rather than the in-plane structure to suggest that the PM is made up of a lipid bilayer of the thickness less than 49 Å [10]. There were serious discussions about the structure of biological membranes at that time [11, for example]. This was previous to the fluid-mosaic membrane model [12] and the interest focused on how bR was located within the PM with respect to the lipids, rather than on the protein structure itself. The function of bR was also unknown and the possibility of an alternating up-and-down orientation of bR was considered. In fact, it was proposed that the structure had the protein molecules in two layers, and it was symmetric in projection onto the profile-axis [10]. After the finding that bR was a light-driven proton pump [13], it was considered that all bR molecules in the PM were oriented in the same direction.
When I read paper [10] in 1974, as the first year of graduate student, I noticed two characteristic points: 1) sharp rings are observed up to high angles, although their intensities drastically drop down around 7 Å–1 (Fig. 2); 2) the diffuse scattering is observed around 10 Å–1. Although the broad 10 Å–1 ring was interpreted as α-helical packing for fibrous proteins [14], a detailed analysis in terms of membrane proteins had not been given. If the atoms of bR and lipids were distributed evenly in the PM, such peculiar diffraction characteristics would not have been observed. My curiosity was aroused and then completely satisfied by Henderson [15].
X-ray diffraction profile and its change upon light absorption of PM with D96N mutant bR taken by our group using BL15A, Photon Factory. Reflections up to 7 Å-1 are indexed according to P3. Solid line, dark state; dot, under continuous illumination. The figure is provided by Dr. Hironari Kamikubo.
The X-ray diffraction patterns of the PM (Fig. 1 and Fig. 2) comprise inherent difficulties for the further crystallographic analysis. They are powder patterns. The concentric diffraction rings arise from the circular average of diffraction spots from the in-plain crystalline arrangement, because each PM sheet in the specimen is randomly oriented around the normal to its surface. Structural analysis from a powder pattern is not trivial. Proteins do not have a center of symmetry so that the intensities of (h, k) and (k, h) reflections are different but their intensities combine in the observed powder pattern. Phasing is also the problem. Henderson very carefully examined and analyzed X-ray diffraction pattern as well as carrying out the trial of heavy atom labeling for the phasing [15].
He indexed the diffraction rings up to 7 Å–1 to conclude that the protein and lipid molecules packed in P3 hexagonal lattice with one bR molecule per an asymmetric unit [15] (Fig. 2). It led to a conclusion that bR forms a trimer around the three-fold axis. He ruled out the other possibilities of space group such as P31 to conclude bR molecules are in a single layer in PM and that each bR is oriented vectorially in the same direction across the membrane [15]. The most important conclusion was that a bR molecule is composed of α-helices that are 25 to 35 Å long, roughly arranged perpendicular to the plane of the membrane [15]. He also suggested that the α-helices form superhelical groupings of the “coiled-coil” type within the bR molecule [15]. Since the modification of the membrane with heavy atoms such as uranyl acetate did not affect the X-ray diffraction pattern significantly, he claimed that the surface of the membrane was flat, with no bumps or dimples, and that the arrangement of phospholipids was rather random in the lattice compared to the protein [15]. Similar conclusions were derived in the subsequent paper by Blaurock [16]. Blarurock rejected his earlier model on the orientation of bR [10] based on the function as a light driven proton pump [16]. These stand as the most thorough analyses of a powder X-ray diffraction pattern that can be performed at that time. This may be a reason why Henderson shifted his field from X-ray diffraction to electron microscopy (EM), although he had started his carrier as a protein X-ray crystallographer [17].
The three-dimensional structure of bR as well as a projection density map was obtained by EM and electron diffraction from an unstained single layer of PM [18,19]. EM has been a useful technique to investigate membrane structure. Although EM provided images in real space, in contrast to the X-ray diffraction pattern in reciprocal space, the observation was limited to either the morphology of the membrane or the distributions of protein molecules in the membrane, because the contrast of transmitted electron is rather weak. Furthermore, the specimen is placed in a high vacuum environment within the microscope. Therefore, special techniques for EM observation were required for biological specimens such as negative staining, freeze-fracture and freeze-replica methods. The actual molecular structures were either deteriorated or lost by these procedures. To overcome these disadvantages, Unwin and Henderson brought revolutionary developments in EM observation, such as unstaining, low dose bright-field observation with defocusing, and glucose-embedded specimen [18]. They examined the error propagation due to electron dose [18]. Consequently, they succeeded to obtain the projected structure of the PM to 7 Å resolution (Fig. 3) [18]. They also applied their methodology to the tilted specimen, which enables to cover good fractions of both real and reciprocal space [19]. Finally, they obtained the three-dimensional structure of bR to 7 Å resolution to show that the molecule is composed of seven trans-membrane α-helices (Fig. 3) [19]. These results brought light to X-ray diffraction studies again, because the phasing and the separation of intensities are now possible with the aid of the structure. I personally profited from the spirit expressed in these studies bR structure [18,19], which prepared me intellectually as an experimental biophysicist. I wrote an essay about the experience, published in 2020 in the bulletin of the Biophysical Society of Japan [20]. Henderson went further in the development of cryo-EM as a powerful structural method, with which he solved the bR structure at atomic resolution [21,22]. He was awarded the Nobel Prize in Chemistry of 2017.
After its three-dimensional structure of bR was established [18,19], the primary structure of bR was reported by two groups (Fig. 4) [23,24]. There are 7 differences in amino acid residues between two sequences. Subsequently, the DNA sequence of the gene [25] verified the correctness of the result in paper [24]. This is a consequence of the BC to AD change in molecular biology. However, the DNA sequencing at the time was still a very time-consuming and requiring hard labor because neither the polymerase-chain-reaction technique nor an automated DNA sequencer existed. The next essential task towards the three-dimensional atomic structure in the BC era was the mapping of amino acid residues (Fig. 4) onto the 7 α-helical rod structures of the tertiary structure and projection map (Fig. 3). Two approaches were initially taken for the purpose.
In the first, α-helical segments in the amino acid sequence (Fig. 4) were mapped into the 7 rods [19] or the 7 peaks in the two-dimensional projection map (Fig. 3) [18] by intensive examination of enzyme cleavage sites used in the sequencing. Such sites should be located at the turn segments exposed to the aqueous milieu rather than in the transmembrane parts. Ovchinikov et al. proposed a set of 7 α-helical in their sequence [23], and a similar conclusion was reported by Khorana et al. [24]. Engelman et al. refined the approach to identify the 7 α-helical regions [26]. There are 7! (5040) possibilities for the assignment of 7 regions to the 7 density peaks of the map. They examined each model carefully in terms of connectivity of the helical regions, charge neutralization, and total scattering density per helix to propose one plausible chain-trace model [26]. The model was proven to be correct by the high-resolution structure [21,22], although some postulated ion pairs were incorrect.
The second approach consisted in using diffraction data to map the amino acid residues on the rod structures or the 2-dimensional density projection. For the purpose, heavy atom labeling of the specific amino acid residues was expected to be effective, although a failure had already been reported [15]. Henderson tried uranyl acetate treatment of the PM for phasing, but he could not obtain the favorable results on the location of the uranium [15], which led him to conclude that lipid molecules in the hexagonal lattice were disordered [15]. He also suggested that the Pt labeling might be useful, because Pt binds to protein through Met or Lys residues while U is coordinated to the lipid phosphate groups [15]. The Pt position was determined by EM [27], however the labeled site in the primary structure could not be identified [27].
An impressive result was published by Engelman and Zaccai using neutron diffraction and deuterated amino acid residues [28]. Note that neutron can discriminate deuterium from hydrogen, providing a powerful labeling method. They biosynthesized the PM with either the deuterated valine or deuterated phenylalanine and determined the distribution of these amino acids in the projection [28]. Valine was found to be distributed toward the periphery of each bR molecule while phenylalanine was distributed toward its center [28]. Combined with the model proposed by Engelman et al. [26], they carefully examined the distribution of all amino acids in the helical region. Consequently, they concluded that the charged and polar amino acids were predominantly located inside of bR, and that the nonpolar residues were predominantly located on the surface in contact with the lipid hydrocarbon chains [28]. In the case of soluble globular proteins, both the charged and the polar residues are predominantly located on the protein surface, whereas the non-polar residues tend to be buried to ensure solubility. Therefore, they called bR an “inside-out” protein [28]. The result contained a further important implication that the proton channel for the function is located within the bR molecule, and not in the space around the three-fold axis.
The neutron diffraction study was extended to deuterated leucine and deuterated isoleucine [29]. Combined with the earlier data [28], 11 plausible models for chain tracing were obtained [29]. These models [29] were, however, inconsistent with the model proposed earlier [26,28], and were later proved to be wrong. What caused the discrepancy? In the former model [28], they generated the difference map from the diffraction changes with the phase and the separation ratios of the intensities of nonequivalent reflections based on the low-resolution structure [18]. In the latter model, however, they calculated neutron diffraction intensities from the models and constructed the difference map from the calculated intensities for deuterated and non-deuterated models [29]. My impression at present is that the model used for calculation was too simple to reflect the true structure. It assumed simply that the helices were rod-shaped and erected vertical to the membrane plane [29], while some helices are tilted and/or kinked in the real structure [21,22]. Later, Engelman and Zaccai confirmed the validity of the first model [26] through neutron diffraction of the deuterated reconstituted PM [30].
Another important message from the work [28] is that the interpretation of the difference map based on the low-resolution structure [18,19], using the phase and separation ratios given by Henderson et al. [31], is effective and appropriate. This approach was successfully applied to the structural analysis of bR photointermediates.
In my lab, we attempted to determine the distribution of tyrosines through light-dependent iodination but failed because iodination caused serious disorder in the crystalline lattice [32]. We also applied Hg labeling for 5 site-directed cysteine substituted mutant bRs. We could determine each mercury position within 1 Å accuracy [33]. The observed positions [33] were in agreement with the expected sites from the high-resolution structure [22]. It was too late to extend the mercury labeling to whole molecule, because the high-resolution structure of bR was already available. To our regret, this approach turned out to be nothing more than a feasibility study at the beginning of the AD era.
As was introduced above, neutron diffraction in combination with specific deuteration produced unique and valuable information to understand the structure of bR. The identification of the location and orientation of the chromophore, retinal, in the low-resolution structure was another essential question. The trans-membrane location of retinal in PM was determined by neutron diffraction as its β-ionone ring was situated centrally in the membrane [34], which is close to the high-resolution structure [21,22]. Authors incorporated deuterated retinal into PM by the reconstitution from the bleached membrane [34]. They applied the technique to determine the in-plane location of retinal. They concluded that Schiff base was located on the surface of helix B facing to helix D of a neighboring bR, and β-ionone ring occupied the space surrounded by helices B, C, and G [35]. It is obvious that the conclusion is incorrect [21,22]. It was considered that retinal was linked to K41 at that time [36]. With this wrong information, they might interpret one peak near helix B among three peaks as Schiff base in their noisy difference map between the deuterated and the hydrogenated retinal [35]. If one accepted the result, proton channel might be located in the space around three-fold axis. This possibility was eliminated as shown above [28]. Zaccai et al. carried out neutron diffraction of the PM with deuterated retinal by biosynthesis [37]. They indicated that the retinal was extended from helix G toward helix D [37], which is almost identical with the high-resolution structure [21,22]. The difference map of Zaccai et al. was convincing and much clearer than the previously reported map [35,37]. If the former group had a correct information that retinal is linked to K216 [38], they might deduce a correct answer for the position of Schiff base, because they observed a difference peak at the position [35].
Since bR is a light driven proton pump, the information on the distribution of water molecules in a protein is an important information. It is now understood that three water molecules coordinate near the Schiff base and the proton accepting group to form hydrogen bond network [39], and that a special water, named water 402 [39], plays an essential role in the proton pump mechanism [39,40]. A pioneering work was reported by Zaccai et al. [41]. They applied H2O/D2O exchange to neutron diffraction and obtained difference map of water distribution. The map showed that exchangeable water molecules were located more in the lipid areas than in the protein areas. They also indicated that the protein had neither pockets nor pits with 12 or more water molecules, in the other words, bR does not have bulk water channels extending into the membrane by more than 5 Å [41]. This was an important conclusion because it ruled out the possibility of proton pathway through hydrogen bond network comprised of a chain of water molecules [41]. Neutron diffraction with H2O/D2O exchange was extended to identify the differences between wet and dried PM. The difference map thus obtained showed that 4 water molecules were tightly bound close to the chromophore binding site [42]. This was the first direct evidence for the bound water which are revealed by crystal structure analysis [39,40].
I joined the bR research field in 1981. My target was the structural analysis of photointermediates by X-ray diffraction. We considered following two distinct paths: time-resolved measurements after light excitation and static measurements of stabilized intermediates. Since a time-resolved measurement system was not available in Japan at the time, I joined the construction team of the small-angle diffractometer with synchrotron radiation at the Photon Factory [43]. For the second path, because it was almost impossible to obtain mutants to stabilize the photointermediates at that time, we tried both cryo-trapping and the stabilization of M intermediate with chemical treatments. We designed a Dewar vessel for the diffractometer at the Photon Factory [44] based on the one used for low-temperature spectroscopy [45]. It turned out not to be suitable, however, because the thin windows required for sufficient X-ray propagation were too fragile or too soft to contain a wet sample in vacuo. Previously reported conditions to stabilize the M intermediate turned out to distort the crystalline lattice. Other conditions required high salt concentration, inappropriate for X-ray diffraction because of strong absorption. After 7 years of effort, we finally found out the arginine treatment which was suitable for our purpose [46]. We had always worried that another group might scoop us—then, our nightmare became a reality.
The first report on the structural changes in the M intermediate was brought by neutron diffraction with guanidium hydrochloride treated PM under cryogenic temperature [47]. Both conditions were impossible to use in X-ray diffraction. The neutron data showed positive peaks in the difference nuclear density map [47]. In contrast, electron diffraction studies using dried, glucose-embedded PM demonstrated no substantial structure changes under continuous illumination of a slowly cooled specimen [48]. Therefore, the further structural studies on the M intermediate were required to resolve the discrepancy. Substantial structural changes at the M intermediate were shown by X-ray diffraction from two groups using different conditions [49,50] (Fig. 2, Fig. 5(a), (b)): the time-resolved X-ray diffraction with synchrotron radiation for the D96N mutant bR [49], and our X-ray diffraction measurement [50] for the wild type bR soaking in arginine solution [46] (Fig. 5 (a), (b)). Both studies showed the similar difference peaks in the projection map each other [49,50] and to the earlier report [47]. The changes were found around helices B, F and G [47,49,50]. The substantial structural changes were further confirmed by electron microscope and electron diffraction [51]. The correlations among these 4 different and independent works [47,49–51] were examined carefully and concluded that they observed essentially the same structural changes [51]. The earlier conclusion of no substantial structural changes [48] was attributed to the early M or M1 intermediate [51,52].
Difference density maps between various intermediates and the original state obtained by our group [57]. Thin solid lines and thin dashed lines represent positive and negative changes, respectively. (a) The difference map for M intermediate obtained for D96N (pH 11.0, 5°C) [57]. This difference map is essentially identical with the map obtained by Koch et al. [49]. (b) The difference map for M intermediate obtained for wild type with arginine treatment [46] (pH 10.0, room temperature) [57]. The map is essentially identical with the map reported previously [50]. (c) The difference map for N intermediate obtained for F171C at pH 7.0, 5°C [57]. No significant differences are observed between (a) and (c). The figure is provided by Dr. Hironari Kamikubo.
Possible structural changes observable in the low-resolution projection map are winding/unwinding, tilt or bend, and shift of α-helices [50]. The latter two cases were plausible because a pair of positive and negative difference peaks was found in helices B, F, and G. Such movements are really observed by fast atomic force microscopy [53], although the observed motions are larger than we imagined. The motion of helix G was further examined using Hg labeled I222C mutant bR [33]. We observed the shift of Hg position by 2 Å at M intermediate [54]. However, we considered that the motion was not a simple shift or tilt of helix G but rather due to the motion of flexible regions [54]. This large movement was confirmed by high resolution electron diffraction [55].
We paid attention to the small but noticeable differences among 4 structures [47,49–51] (Fig. 5). We claimed that the characteristic changes were around helices B and G, because the difference peak height around helix F was much smaller [50]. Koch et al. indicated that the structural changes were more remarkable for helices F and G than for helix B, because the peak height around helix B was considerably smaller [49]. Subramaniam et al. [51] supported the changes around helix F and G for D96G. However, a significant positive peak around helix B was observed for wild type [51]. Therefore, the changes around helix F and G are characteristic of the D96 mutant bR, while the change around helix B is observed for wild type only. Subramaniam et al. suggested the possibility that the changes in B and F reflected the M to N transition [51]. Fourier transform infrared study on the D96N mutant bR indicated that it took up the structure of the N intermediate with the deprotonated Schiff base (M-like) under the condition of Koch et al. [56]. To clarify these small but distinct differences, we analyzed the structure of the N intermediate with the F171C mutant [57] (Fig. 5(c)). We succeeded to accumulate the N intermediate exclusively and to obtain an X-ray diffraction pattern. We also analyzed the intermediate of D96N under the same conditions as Koch et al. [49] (Fig. 2). The difference map of the N intermediate showed the changes in helix F and helix G, which were identical to those of the photointermediate of D96N [57] (Fig. 5(a), (c)). Consequently, we confirmed that the changes in helix F and G were characteristic of the N intermediate, and that the changes in helix B and G were attributed to the M intermediate [57]. We named these structures the N-type and the M-type, respectively [57,58]. The structure of the N intermediate was analyzed by electron diffraction for glucose-embedded F219L mutant bR, which led the same conclusion as ours [59]. The authors also concluded that the changes around helix F and G were characteristic of the N (or MN) intermediate, although they misunderstood that we had shown the changes in helices F and G in the arginine treated M intermediate [59]. Since the differences between the M- and N-type structures were subtle and small, some considered them trivial [60,61]. However, we clearly showed the hydration dependent transition between the M- and N-type structures [62]. Hydration dependent conformational changes after the formation of M intermediate were described independently by Zaccai and coworkers [42]. They also indicated that hydration water is required for the completion of the photocycle [63].
After entering the AD era, several groups carried out the synchrotron radiation crystallographic analyses of the M and N intermediates [64–70]. There were some differences and contradictions among the published structures [64–70]. We evaluated the low-resolution projection maps of these structures to discriminate between them [71]. The structures were clearly classified into three types of structures, the M-type (M2 type), the N-type and a type displaying no substantial structural changes (M1 type). Some crystal structures were fell into a class different from the intermediate intended by the authors [71]. The origins of the differences may be attributed to the differences in the methods and the conditions for the crystallization as well as the systems for illumination and trapping to stabilize the intermediates of the interest. In spite of the low resolution, X-ray diffraction of PM under physiological conditions can give useful information to interpret the high-resolution crystal structures.
Based on the technological progress, it is now possible to perform the time-resolved crystal structure analysis of bR with femto-second time resolution [40,72,73]. Even in the BC era, however, researchers attempted the challenge set by time-resolved diffraction measurement. To the best of my knowledge, the first trial was carried out with laser plasma X-rays [74]. The diffraction pattern was recorded 1 msec after photoexcitation. The authors observed the disordering of crystalline structure of PM induced by light with little structural change [74]. Since the diffraction pattern was too broad to separate overlapping reflections quantitatively, it was difficult to observe the intensity changes, if any, caused by the formation of intermediates. They applied the radial autocorrelation function to analyze the disordered profiles [74], a method that had been developed by myself for the analysis of non-crystalline membrane diffraction [75].
Koch et al. carried out a time-resolved X-ray diffraction with 15 msec time resolution for the decay process of M intermediate in the D96N mutant by synchrotron radiation [49]. They observed the time course of the intensity changes of a few reflections. The kinetics of the structural changes were similar to those of the absorbance changes at neutral pH, but lagged behind the absorbance change at alkaline pH [49]. They considered that the formation of N intermediate contributed to the structural decay at alkaline pH [49]. The structure of M intermediate (MN intermediate as mentioned above) was revealed in the initial state of the decay process [49]. This work was the first report on structural kinetics in bR.
Subramaniam and Henderson performed the time-resolved electron diffraction at liquid nitrogen temperature [61]. They trapped the generated photointermediate states in the range from 1 to 35 msec after illumination [61]. The structural changes at 1 msec and at 35 msec were essentially identical, although the peak heights in the difference map around helices F and G increased at the 35 msec stage [61], suggesting that both M and N intermediates had similar structures. They claimed that the snap shot method was effective to investigate the structure of the reaction intermediates [61].
We also carried out the time-resolved X-ray diffraction of both wild type and D96N bR with synchrotron radiation [76–78] (Fig. 6). We used the snap-shot method for before and after light excitation (Fig. 6), while Koch et al. used the continuous measurements during the decay of M intermediate [49]. The time resolution of our work was improved from 244 msec [76] and 122 msec [77] to 6 μsec [78]. We applied the singular value decomposition (SVD) method to the analysis of the diffraction patterns [76–78]. We observed two distinct components other than the un-photolyzed structure appearing during the decay process after illumination for both wild type [76] and D96N [77]. The difference maps of two components were identical to the M-type and the N-type structures as explained above [76,77], confirming that the change around helix B was real and characteristic of the M intermediate, while the change of helix F was characteristic of the N intermediate, and the change of helix G was common to both intermediates. The time course of the rise of the M intermediate was evaluated by SVD analysis and compared with the absorbance change, which led to the conclusion that the deprotonation of the Schiff base triggered the structural change [78]. The conclusion supported the charge-controlled conformation change model [52,79] and the reprotonation switch model [79–81].
Time-resolved diffraction pattern of wild type bR (pH 9.0, 10°C) before and after light excitation. Time-resolution was 244 msec. Time frames were only shown up to the 50th frame, although the patterns were recorded up to the 200th frame. BR was excited at the 11st frame with a Xenon flash lamp [75]. The figure is provided by Dr. Toshihiko Oka.
The final goal of structural studies of bR, even at low resolution, is the understanding of the molecular mechanism of the light driven proton pump. Excellent comprehensive reviews of the field are available [82–85]. The proton pathway was identified both spectroscopically and by X-ray crystallography [40,72,73,82–85]. The proton transfer is understood with five steps [82]: i) deprotonation of the Schiff base and protonation of D85 (L-to-M1 transition); ii) proton release from D85 to the proton release group (E204 and/or E194) (M1-to-M2 transition); iii) reprotonation of the Schiff base from protonated D96 (M-to-N transition); iv) reprotonation of D96 from aqueous milieu (N-to-O transition); v) proton release from the proton release group to aqueous milie (O-to-bR transition) [82]. Global conformational changes ensure the vectorial proton transfer, which occur at the step ii) and the step iv) [52,58,62,79–81,86]. The structure of the step iii) would be different from the structure of the step ii) [82], which we had shown [57]. We had proposed a conformation-controlled conformation change model [58,62,86]. Its essence was that the local conformations (the local electrostatic interactions, isomer form of the retinal and protonation/deprotonation of the residues of interest) and the global conformation correlate closely with each other [58,62]. We had also incorporated, into the model of reference [86], the consistency principle of protein folding [87].
In contrast, some have considered that a conformational change is not a prerequisite [55,40,88]. Henderson and Subramaniam claimed that the changes in retinal curvature would be more important than the conformational change [55]. Such a curvature change largely affects the orientation of the Schiff base nitrogen to switch its exposure from the extracellular to the cytoplasmic side [55,84]. The conclusion was derived from the study of the triple mutant, D96G/F171C/F219L, that the mutant pumps protons without the remarkable conformational changes [55,79]. Conformational changes observed in the wild type would thus reinforce the retinal curvature change. Another model without conformational change was the isomerization/switch/transfer (IST) model [88]. In the model, the isomerization state of the retinylidene moiety rather than protein conformation governed the accessibility of the Schiff base in the protein to establish the vectorial transport [88]. The causality between the conformational changes and the vectorial transport is still under debate [40,53,84].
Even though the current understanding of the molecular mechanism of the bR proton pump is greatly improved, the previous models based on low resolution structure [52,55,58,62,79–81,86] were not far from today’s understandings [82–85]. However, it is still some ways from our objectives. The sophisticated spatio-temporally ultrahigh-resolution structural analyses of bR have revealed atomic motions step by step of the early stages of the photoreaction [40,72,73,82–85]. Further, improved molecular information should be required to reach the final goal. Based on my interest, three significant questions to be unraveled are the clarification of the transient formation/breaking of the hydrogen bond network upon light absorption, the direct observation of the proton movement, and the detailed analysis of energy balance. We strongly anticipate the technological and theoretical progress that will enable us to answer these questions. My dream is that we shall reach our final goal in the next half century.
The scope of this review is not an updated comprehensive survey of bR researches, but is an introduction to the old structural studies. We can learn from the spirit of the first scientists in the bR field. Their experiments with the limited technology of the time were carefully carried out to derive the best results. Their original, specific methodological developments and analyses to tackle the problems are carefully described in the papers. Their conclusions were derived from profound interpretation and consideration of the experimental results based on sound scientific logic. Even where their conclusions were incorrect in some cases, the papers do not lose the scientific soundness or validity for us to learn from the mistakes. More sophisticated and improved techniques and methodologies are applied to current studies, with derived conclusions that are beautiful and convincing. However, sophisticated methodologies often hide the personality of the researcher. The old studies remind us of an important lesson that science is made by human beings.
During preparation of this manuscript, I received the sad news of the passing of Prof. Dieter Oesterhelt, one of the discoverers of bR. Although some of his important works were cited [1,13,38,49,51,52,60,79,88], I could not survey his tremendous contribution to this field, because they were beyond the scope of this review. His great contribution to this field is obvious and will be introduced elsewhere [89]. I pray he rests in peace.
None.
MK designed and wrote the manuscript.
All the information should be referred to the original references, which are properly cited.
I thank F. Tokunaga, D. M. Engelman, T. Iwasa, H. Kamikubo, Y. Imamoto, Y. Yamazaki, M. Nakasako, T. Oka, J. K. Lanyi, R. Needleman and my students for their continuous collaboration and discussion. I also thank G. Zaccai for his critical readings of the manuscript as well as his warm support, and R. Henderson for his warm support.
