Surface and Dynamic Structures of Bacteriorhodopsin in a 2D Crystal, a Distorted or Disrupted Lattice, as Revealed by Site-directed Solid-state 13C NMR


  • This paper is part of the Proceedings of the 12th International Conference on Retinal Proteins held at Awaji Island, Hyogo, Japan on 4–8 June 2006.

*Corresponding author email: (Hazime Saitô)


The 3D structure of bacteriorhodopsin (bR) obtained by X-ray diffraction or cryo-electron microscope studies is not always sufficient for a picture at ambient temperature where dynamic behavior is exhibited. For this reason, a site-directed solid-state 13C NMR study of fully hydrated bR from purple membrane (PM), or a distorted or disrupted lattice, is very valuable in order to gain insight into the dynamic picture. This includes the surface structure, at the physiologically important ambient temperature. Almost all of the 13C NMR signals are available from [3-13C]Ala or [1-13C]Val-labeled bR from PM, although the 13C NMR signals from the surface areas, including loops and transmembrane α-helices near the surface (8.7 Å depth), are suppressed for preparations labeled with [1-13C]Gly, Ala, Leu, Phe, Tyr, etc. due to a failure of the attempted peak-narrowing by making use of the interfered frequency of the frequency of fluctuation motions with the frequency of magic angle spinning. In particular, the C-terminal residues, 226–235, are present as the C-terminal α-helix which is held together with the nearby loops to form a surface complex, although the remaining C-terminal residues undergo isotropic motion even in a 2D crystalline lattice (PM) under physiological conditions. Surprisingly, the 13C NMR signals could be further suppressed even from [3-13C]Ala- or [1-13C]Val-bR, due to the acquired fluctuation motions with correlation times in the order of 10−4 to 10−5 s, when the 2D lattice structure is instantaneously distorted or completely disrupted, either in photo-intermediate, removed retinal or when embedded in the lipid bilayers.


Bacteriorhodopsin (bR) is a light-driven proton pump from Halobacterium salinarum. It assembles into naturally occurring 2D crystalline patches known as purple membrane (PM) in which its trimeric unit is hexagonally packed (1–7) to form PM under physiological conditions. Many integral membrane proteins in the membrane environment are also known to assemble into oligomeric complexes rather than monomers to lead to tertiary and quarternary structures as revealed by X-ray diffraction or cryo-electron microscope studies. This has been demonstrated for chloride pump halorhodopsin (8,9), phototaxis receptor sensory rhodopsin II (10–12), a photosynthetic reaction center (13), a light-harvesting complex (14,15), cytochrome c oxidase (16,17), potassium and mechanosensitive channels (18,19), bovine rhodopsin (20), the calcium pump of sarcoplasmic reticulum (21), etc. This is because the conformation and dynamics of such membrane proteins as determined by X-ray diffraction or cryo-electron microscope studies are mainly regulated by specific lipid–protein and protein–protein interactions with the surrounding lipids and neighboring proteins as structural determinants, especially those leading to a 2D assembly in bR (22).

Nevertheless, the surface structure data of many such proteins, including bR, are still missing, either obscured or inconsistent, among a variety of 3D structures as revealed by cryo-electron microscopy and X-ray diffraction studies (3–7). These could be biologically very important, providing information on the transport and signal transduction of these proteins. In addition, it appears that such biological activities are retained also in a monomeric form present as solubilized in detergents or reconstituted in lipid bilayers, rather than in a crystalline or oligomerizied form. Indeed, the functional unit of bR, as a naturally occurring 2D crystal as PM which is responsible for the photocycle, is the monomer itself (23–25). However, the lifetimes of the L and N intermediates are significantly shorter than in the monomer (26,27). Therefore, a plausible conformation and dynamic change during conversion from the 2D crystal to an instantaneous monomeric structure may also play an important role for its biological activity.

The site-directed 13C NMR approach on selectively13C-labeled membrane proteins from either the 2D crystalline state or the monomer turn out to be a very valuable, alternative means revealing conformational features as well as the dynamics of whole areas of fully hydrated proteins at ambient temperature. This includes the surface residues arising from both the N- or C-terminal residues as well as the interhelical loops. We briefly describe here the surface and dynamic structures of fully hydrated bR in the 2D crystal. The distorted or disrupted lattice can be evaluated with respect to its relation to its biological function, as revealed by site-directed 13C NMR. Understanding of such pictures is crucially important, as a reference, for interpretation of the 13C NMR data on a variety of membrane proteins reconstituted in lipid bilayers.


X-ray and cryo-electron microscopic studies have yielded considerable structural information about the intramembrane portion of bR but little is known about the disposition of the loops and the N- or C-terminus (2–7). The transmembrane α-helices thus obtained are shown in the boxes as demonstrated in the schematic representation of the primary structure of bR taking into account its secondary structure (Fig. 1) (3,5). The remaining portions are flexible enough to allow a variety of fluctuation motions with various ranges of correlation time from 10−4 to 10−9 s (28–32). This is because some 3D crystal packing of bR may limit the conformational flexibility of the loops, while the surface in the 2D crystals is not hindered by 3D contacts (33,34). Therefore, a number of fully hydrated membrane proteins of biological relevance are far from rigid bodies even in a 2D crystal or monomeric state, at least at ambient temperature.

Figure 1.

 Schematic representation of the secondary structure of bR taking into account of its secondary structure revealed by X-ray diffraction of 3D structure. It consists of the C-terminal α-helix (helix G′ protruding from the membrane surface) held together by the C–D and E–F loops to form the surface structure or cytoplasmic complex, together with the seven transmembrane α-helices (A–G) (31). The transmembrane α-helix identified by X-ray diffraction and the C-terminal α-helix revealed by 13C NMR are shown by the boxes. The loops and N- or C-terminus residues, however, are not well defined either by lack of defined electron density or different among published data depending upon the crystallization conditions. The Ala and Val residues are highlighted by the circles and rectangular boxes, respectively.

Appropriate isotopic enrichment (labeling), as high as possible approaching to 100%, is essential for solid-state 13C NMR studies of membrane proteins, in order to enhance both the sensitivity and the selectivity of their labeled signals from the background signals arising from residues of natural 13C abundance. However, it is cautioned that 13C NMR signals from portions undergoing fast isotropic, fluctuation motions such as the N- or C-terminal residues of bR (with a correlation time shorter than 10−8 s) are preferentially suppressed by cross polarization-magic angle spinning (CP-MAS) NMR. This is because the 13C magnetizations from such regions, generated by 13C-1H dipolar interactions, are completely time-averaged by isotropic fluctuation motions. In such cases, the corresponding 13C NMR spectra can be conveniently recorded by an alternative, dipolar decoupled magic angle spinning (DD-MAS) NMR experiments (28,29). Here we compare the 13C DD-MAS and CP-MAS NMR spectra of fully hydrated [3-13C]Ala-labeled bR from PM (centrifuged pellet) as a 2D crystal (Fig. 2a,b, respectively) (35). Up to twelve 13C NMR signals, including contributions from the five single carbon signals among 29 Ala residues in bR (see Fig. 1 for the location of the individual Ala residues), are well-resolved in the two types of spectra. The three intense 13C NMR signals marked by gray in the DD-MAS NMR spectrum (Fig. 2a) are significantly suppressed in the corresponding CP-MAS NMR spectrum (Fig. 2b). These suppressed peaks in the latter are obviously ascribed to the N- or C-terminal regions protruding from the membrane surface which undergo fast isotropic fluctuation motions with correlation times shorter than 10−8 s (28,29). Only three peaks could be resolved, however, when lyophilized preparations were used instead of the fully hydrated pelleted samples currently used (36).

Figure 2.

 Comparison of the 13C DD-MAS (a) and CP-MAS (b) NMR spectra of [3-13C]Ala-labeled bR from the purple membrane (2D crystal). The three peaks marked by gray on the upper trace are from the Ala residues located at the N- and C-terminal moieties (37).

The presence of the picosecond internal motions of the fully hydrated bR/lipid complex was also revealed by neutron scattering studies (37). It appears that the above-mentioned N- or C-terminal residues, rather than the major transmembrane α-helices, are responsible for such fast thermal fluctuations detected by the neutron diffraction. These are from small-amplitude atomic and molecular vibrations up to large-amplitude stochastic re-orientational motions of the molecular subunit, which are not hindered by any crystal contacts. Indeed, water molecules “lubricate” such picosecond motions and the light-triggered micro to millisecond tertiary structural changes of the protein (38). In addition, the lipids and their ability to attract solvent molecules play an important role in producing such hydration-induced flexibility. Thus, well-resolved 13C NMR spectra are available from the fully hydrated, intact proteins under physiological conditions.

Spectral resolution of uniformly or extensively 13C-labeled proteins, however, might be desperately deteriorated, if a directly bonded 13C-13C sequence is present in the dense 13C spin networks. If fluctuation motions with correlation times from 10−4 to 10−5 s are present, as described later, these are manifested by the extremely broadened 13C CP-MAS NMR spectra of [1,2,3-13C3]Ala-labeled bR from PM (39). This is the reason why site-directed or amino-acid-selective isotope labeling is favorable for membrane proteins to avoid such dense 13C spinnetworks.


A specific peak, with reduced 13C peak-intensity, from a site-directed mutant, in which an amino-acid residue is replaced by another residue, can be assigned to the residue of interest with reference to that of the wild-type. Here, we assume that any additional spectral change due to an accompanying conformational change can be neglected. For instance, the Ala Cβ13C NMR peaks of Ala 196 (F-G loop) and Ala 126 (at the corner of the helix D) of [3-13C]Ala-labeled bR are straightforwardly assigned to the peaks suppressed in the site-directed mutants, A196G and A126G, respectively (see Fig. 2 for the assigned peaks). Because no additional spectral change is noted upon the introduction of such site-directed mutagenesis (35), a variety of selectively 13C labeled amino-acid residues can be utilized for this purpose. It is cautioned, however, that several 13C NMR signals could be missing from the site(s) of 13C-labeling, depending upon their locations and samples (2D crystal or monomer). Therefore, a prior knowledge is required as to whether or not the incorporated 13C NMR signals are fully visible from the site(s). It is shown that the 13C NMR signals for Ala Cβ are almost fully visible from the fully hydrated [3-13C]Ala-labeled bR from PM at ambient temperatures (40).

In many instances, however, more complicated spectral changes could be induced in the difference spectrum between the wild-type and the site-directed mutant. These are due to local conformational changes accompanied by site-directed mutagenesis, yielding several sets of dispersion peaks (41). To remove such undesirable dispersion peaks, Tuzi et al. have proposed a powerful means to eliminate such dispersion peaks from the surface areas by accelerated Mn2+-induced 13C transverse relaxation (42,43). This is applicable to those residues giving disturbing signals which are mainly located near the surface areas, or where a residue of interest is located within the transmembrane helices as in A215G, A81G, A184G, A53G, etc. (44). The assigned peaks thus obtained can be used as a very convenient probe to examine the local conformation of the residues under consideration, with reference to the data base of the conformation-dependent displacement of 13C chemical shifts (28,29,45,46). As to the 13C NMR signals of [3-13C]Ala-bR, two kinds of α-helices, αΙ- and αΙΙ-helices, are conveniently distinguished as viewed from their peaks with reference to those of (Ala)n in the solid and in a hexafluoroisopropanol (HFIP) solution, respectively. The concept of the αII helix was initially proposed by Krimm and Dwivedi based on their infrared study (47) and later extended to the interpretation of the 13C chemical shifts utilizing their sample system (40).

This approach is not always successful for the assignment of peaks from several [3-13C]Ala-labeled proteins. However, it is applicable when several peaks are simultaneously displaced or suppressed, due to induced global conformational changes of the proteins by a site-directed mutagenesis, as encountered for the D85N mutant of bR (48). In such cases, use of [1-13C]Val-labeled proteins rather than [3-13C]Ala-D85N turned out to be more useful (48,49), because this label is in some instances less sensitive to such conformational changes.


Surprisingly, the 13C NMR signals turn out to be not always fully visible, especially from the surface areas of membrane proteins undergoing local fluctuation motions, in spite of the 2D crystalline preparations of bR from PM (40,41). To clarify this problem, a relative contribution to the 13C NMR signals (f) from residues located near the surface areas was evaluated by comparing their peak-intensities in the presence (I) with those in the absence of 40 μM Mn2+ (I0),


The relative contributions of the 13C NMR signals from the surface areas (I0I) can be defined as residues located within 8.7 Å from the membrane surface (44), because these 13C NMR signals from the residues within these areas can be completely suppressed (giving rise to a line width larger than 100 Hz) by the accelerated transverse relaxation process caused by the Mn2+ ion (42,43) (see Table 1). Alternatively, the proportion of such residues near the surface (8.7 Å from the membrane surface; see the gray area in Fig. 1), g, can be easily counted by examination of the primary sequence of bR taking into account its secondary structure (see Fig. 1),


where n and n0 are the number of residues concerned located within such areas (gray areas in Fig. 1) and the total number of residues, respectively. If g, there are no further suppressed peaks caused by fluctuation motions in the transmembrane α-helices near the surface areas, as seen for [1-13C]Val- and Ile-bR (Table 1). If g , on the other hand, more 13C NMR signals than those of the surface areas could be suppressed by interference of the incoherent fluctuation frequencies (104 Hz) with the coherent frequency of the magic angle spinning (50), as demonstrated for [1-13C]Gly-, Ala-, Leu-, Phe-, Trp-labeled bR.

Table 1.   Comparison of relative proportion of the 13C NMR signals from the surface areas of [1-13C]amino-acid labeled bR from purple membrane as estimated from the 13C NMR intensity ratio with (I) and without (I0) Mn2+ ion.
 Estimated from 13C signals (1-I/I0), fPredicted amounts of residues at the surface area (8.7 Å from the membrane surface), gSuppressed 13C NMR peaks from the surface areas caused by slow motions
  1. *Reference 41

Alaca. 0*0.62Almost completely suppressed

The completely or partially suppressed peaks (in the absence of Mn2+) are related to the presence of such low frequency, residue-specific dynamics, in relation to the possibility of conformational fluctuations caused by the time-dependent deviation from the torsion angles corresponding to the lowest energy minimum of a particular conformation. Naturally, such conformational space which allows fluctuation motions may be limited to a very narrow area for the Val or Ile residues with bulky side-chains at Cα, together with limited χ1 rotation around the Cα–Cβ bond in the Cα–CβH(X)(Y) moiety in the peptide unit where X and Y are substituents at Cβ. This minimum may be very shallow, however, for Gly, Ala, or Leu residues in view of the expected, widely allowed conformational space. Therefore, it is plausible that the above-mentioned low frequency, residue-specific backbone dynamic with a fluctuation frequency of 104 Hz interferes with the frequency of the magic angle spinning for the [1-13C]Ala, Leu, Phe and Trp residues. The backbone dynamics in these systems could be coupled with a possible rotational motion of the χ1 angle around the Cα–Cβ bond, as schematically represented by the Cα–CβH2–Z system where Z is H, isopropyl, phenyl, or the indole group.

Accordingly, the recommended probes to yield correctly the 13C NMR signals of incorporated 13C-labeled nuclei are [3-13C]Ala, [1-13C]Val or Ile (51).


Surface structures of bR, if any, could be easily modified by a variety of intrinsic or environmental factors during the course of crystallization leading to 2D or 3D crystals. Such factors are temperature, pH, ionic strength, crystallographic contact, etc. (28,33). The C-terminal residues, 226–235, participate in the formation of the C-terminal α-helix (39) protruding from the cytoplasmic surface (as manifested by the peak-position of 15.91 ppm for [3-13C]Ala-bR and also for the Ala Cα and C = O 13C NMR peaks from the helix G′) (see Fig. 1). This applies to the above-mentioned conformation-dependent 13C chemical shifts (28,29,45,46). Only part of this α-helix, however, is visible by X-ray diffraction (52) due to the presence of the fluctuation motions with correlation times of the order of 10−6 s at ambient temperature, as judged from the carbon spin–lattice relaxation times, T1C and spin–spin relaxation times, T2C, under CP-MAS conditions (39). The involvement of Ala 228 in the C-terminal α-helix was also shown by the obviously reduced peak-intensities of A228G at 15.91 ppm, both in the CP-MAS and the DD-MAS NMR spectra (53).

Therefore, it is shown that the C-terminal α-helix is held together by the loops at the cytoplasmic surface to form the cytoplasmic surface complex or surface structure, stabilized by formation of salt bridges and metal-mediated linkages, as schematically illustrated by the dotted lines in Fig. 3. This structure could be destabilized when environmental factors are changed to such as high ionic strength, low pH or high temperature (53). The surface structures are naturally altered either by site-directed mutations at the C-terminal α-helix (R227Q) or at the loop (A160G, E166G and A168G), and by the manner of altered cation binding, as viewed from the 13C chemical shifts of Ala 228 and 233 (C-terminal α-helix), Ala 103 (C–D loop) and Ala 160 (E–F loop). In contrast, the cytoplasmic ends of the B and F helices are found to undergo fluctuation motions of the order of 10−5 s when such a surface structure is disrupted (54). To make proton uptake efficient during the photocycles, the following surface structure is proposed: the C-terminal α-helix of the wild-type at ambient temperature is tilted toward the direction of the B and F helices. This facilitates efficient proton uptake by preventing unnecessary fluctuations of the helices. Such a surface structure, however, is destabilized in that the C-terminal α-helix is straightforwardly extended from the helix G at low temperatures or in an M-like state (54). This view is consistent with the previously published data for the “proton binding cluster” consisting of Asp 104, Glu 166 and Glu 224 (55–57).

Figure 3.

 Schematic representation of the secondary structure and dynamics for bR, consisting of the C-terminal α-helix (helix G′ protruding from the membrane surface) held together by the C–D and E–F loops to form the surface structure or cytoplasmic complex, together with the seven transmembrane α-helices (A–G) (31). The protein dynamics differs substantially between the 2D crystal and the monomer in the lipid bilayer.

It is interesting to note that the role of such a C-terminal α-helix is more important for pharaonis phoborhodopsin (ppR) than bR for the sake of stabilization of the 2:2 complex formation with its cognate transducer (pHtrII) through interaction with cytoplasmic α-helices protruding from the cytoplasmic surface, besides the mutual interaction at the transmembrane α-helices (58,59). As a result, such a C-terminal helix is made visible by X-ray diffraction as a result of complex formation with pHtrII (60).


To our dismay, we found out that several 13C NMR peaks of certain amino-acid residues could be suppressed depending upon their sites, and the manner of sample preparation when the 13C NMR spectra of fully hydrated bR were recorded at ambient temperature. Thus structural information from such sites could be obscured. Therefore, closer examination of such suppressed signals obviously provides an unrivaled means to evaluate the presence and location of millisecond or microsecond motions for such sites.

The expected 13C NMR line width 1/πT2C under CP-MAS or DD-MAS NMR conditions depends strongly on the incoherent frequency of any fluctuation motion, if it is interfered with either by the coherent frequency of the proton decoupling or by the magic angle spinning. (50,61). In such cases, the overall relaxation rate 1/T2C can be dominated by the second or third terms given in Eq. (3) instead of the first term which is applicable to the static component,


where (1/T2C)S is the transverse component due to static C–H dipolar interactions, and (1/T2C)MDD and (1/T2C)MCS are the transverse components due to the fluctuation of the dipolar and chemical shift interactions in the presence of internal fluctuation motions, respectively. The latter two terms are given as a function of the correlation time τc by


Here ωI and ωS are the gyromagnetic ratios of the I and S nuclei, respectively, and r is the internuclear distance between spins I and S. ω0 and ωI are the carbon resonance frequency and the amplitude of the proton decoupling RF field, respectively. ωr is the rate of spinner rotation. δ is the chemical shift anisotropy and η is the asymmetric parameter of the chemical shift tensor.

As far as carbonyl groups with large chemical shift anisotropies are concerned, a decoupling field of 50 kHz is sufficient to reduce the static component at a magnetic field of 9 T and the (1/T2C)MCS term will be dominant in the overall value of 1/T2C, while CH2 or CH3 signals are determined by the (1/T2C)MDD term alone. It is expected that the Cα signal could be affected by both the (1/T2C)MDD and (1/T2C)MCS terms, depending upon the frequency range of either 50 kHz (ωI) or 4 kHz (ωr), respectively. Of course, it is possible to avoid the above-mentioned interference with a frequency of the order of 10−4 s by increasing the spinning rate up to 20 kHz.

These considerations provide unique means for a model-free evaluation of such fluctuation frequencies, if the fully visible 13C NMR spectra are available as reference spectra for further evaluation. It is noticed that there exist local fluctuation motions in certain locations even for bR from a 2D crystalline lattice (PM): correlation times for such fluctuation motions for the loops, C-terminal α-helix, and N- and C-terminal ends are 10−4 s, 10−6 s and 10−8 s. These are obtained from the suppressed peaks interfered by the MAS frequency, relaxation parameters and from the suppressed peaks in the CP-MAS NMR spectra, respectively. However, the frequency for the transmembrane α-helices is estimated as 10−2 s for the 2D crystal as found by a chemical exchange process reflecting the presence of the crystalline lattice as summarized in Fig. 3 (39). Indeed, the correlation time for the loops is estimated to be of the order of 10−4 s, because the 13C NMR signals of [1-13C]Ala-bR from such region are suppressed by interference of the fluctuation frequency by the magic angle spinning, as already described (39). Such backbone dynamics of the loops is also evaluated by examination of site-specific 13C-1H dipolar couplings in [3-13C]Ala-bR, yielding slow or intermediate frequency motions (62). Naturally, the above-mentioned surface structure of the cytoplasmic complex is far from a static picture. The structure undergoes fluctuations in the order of 10−4 to 10−6 s as found by taking into account the estimated correlation times of its constituent components.

A particular portion of the transmembrane α-helices acquires fluctuation motions with a correlation time of up to 10−5 s, leading to suppressed peaks for [3-13C]Ala-bR due to a failure of the peak-narrowing by proton decoupling (61). This occurs when a 2D crystalline lattice is either instantaneously or permanently distorted or is disrupted by removing retinal (41), reconstitution in a lipid bilayer (63), a modified lipid–protein (64), retinal-protein interactions (41) or the M-like state of D85N mutant of bR (65). This kind of dynamic picture, however, has never been seriously taken into account for the data available at cryo-temperatures.


It is noteworthy that the 13C NMR signals of the loops and some transmembrane α-helices of [3-13C]Ala-bR which resonate at 16.1–16.4 ppm are either completely or substantially suppressed, respectively, when the 13C NMR signals are recorded in a lipid bilayer in which bR is present as a monomer (63) (Figs. 4a,b), as compared with those from bR in a 2D crystalline lattice (Figs. 4c,d). This is obviously caused by acquired low-frequency fluctuation motions (105 Hz) in the absence of the helix–helix contact due to a disrupted 2D crystalline lattice (see Fig. 3) interfering with the proton decoupling frequency (61). Therefore, the backbone dynamics could be modified when the 2D lattice assembly is similarly distorted or disrupted, as in bacterio-opsin (bO) prepared from either hydroxylamine-treated bR or the retinal-deficient E1001 strain in which the helix–helix interactions are substantially modified due to lack of retinal (64). The same applies to W80L or W12L mutants in which the side chain of one of two Trp residues, which is oriented outward from the transmembrane α-helices at the interface of the lipid–protein interaction, is absent (64). Therefore, very similar spectral changes are induced for the 13C NMR spectra of [3-13C]Ala-bO, W80L and W12L, except for the presence of one of the loop signals for [3-13C]Ala-bO (41). The resulting characteristic spectral changes are summarized in Table 2, showing the presence or absence of the loop signals. As a measure of the suppressed 13C NMR of the transmembrane α-helices, the intensity ratio of the transmembrane α-helical peak at 16.4 ppm (I16.4) vs. the C-terminal α-helix at 15.9 ppm (I15.9) varies from 1.2 to 0.4, depending upon the type of 13C-labeled preparation. Further, it was shown that the 13C NMR spectra of some of the [1-13C]Val-labeled proteins are almost completely suppressed in the distorted or disrupted lattice (Table 2). These changes are interpreted in terms of the presence of such a distorted or disrupted 2D lattice, which leads to slow fluctuation motions with a correlation time of 10−5 s for the cytoplasmic ends of the helix B and F as deduced from the reduced peak at 16.4 ppm due to Ala 39 and 168, as compared with a time of 10−2 s in the 2D crystalline sample. The acquisition of this kind of motional freedom can be well recognized by the fact that the helices B and F are located either at the interior of the trimer or the outer boundary of the trimeric structure (22) which plays an important role in the helix–helix interaction. Here, it is mentioned that the trimeric structure is preserved for W12L but is disrupted for W80L as judged from the observed CD spectra (64).

Figure 4.

13C CP-MAS (left) and DD-MAS (right) NMR spectra of [3-13C]Ala-labeled bR reconstituted in an egg PC bilayer (a and b; 1:50 mol ratio) and from PM (c and d). The intense peaks at 19.7 and 14.1 ppm are ascribed to lipid methyl groups from Halobacteria and egg PC, respectively (65).

Table 2.   Characteristic loop 13C NMR signals* and intensity ratio of the transmembrane α-helix at 16.3 ppm (I16.3) vs. C-terminal α-helix at 15. 9 ppm (I15.9).
 bR 2D crystalW12LD85NW80LbObR deionizedbR monomer†D85N monomer†
pH 7pH 10
  1. *++, fully visible; +, partially suppressed; –, suppressed. †In lipid bilayer. ‡Displaced also.

Loop C-D (Ala 103)+++‡++
E-F (Ala 160)+++‡
F-G (Ala 196)+++‡
13C NMR of [1-13C]Val-labeled proteins++++++

This finding suggests that the fluctuation motion of bR with a correlation time of the order of 10−2 s in the 2D crystal (30,31) could be accelerated for the rest of the connecting transmembrane α-helices in a distorted or disrupted lattice (Fig. 3). Indeed, it is possible to visualize such an accelerated backbone motion with a correlation time of the order of 10−4 s (or fluctuation frequency of 104 Hz) for the transmembrane α-helices from the suppressed 13C NMR spectra of the [1-13C]Val-labeled proteins (64,65). The loop 13C NMR signals of [3-13C]Ala-D85N are completely suppressed at pH 10. This is concomitant with the taking of the M-like state. In fact, a significant dynamic change is induced together with deprotonation of the Schiff base (SB) at this pH, leading to the deletion of the salt bridge with Asp 85 as a structural constraint as encountered in bO (48,49). However, it is cautioned that the resulting spectral features of the [3-13C]Ala proteins could not be distinguished even between bR and the D85N mutant as far as they were examined in the lipid bilayer (65).

In particular, D85N acquires local fluctuation motions with a frequency of 104 Hz in the transmembrane B α-helix in the M-like state as manifested from the suppressed 13C NMR signal of the [1-13C]Val-labeled Val 49 residue (49). The local dynamics of D85N at Pro 50 with Val 49 as its neighbor turn out to be unchanged, irrespective of the charged state of SB as viewed from the 13C NMR spectra of Pro 50 in [1-13C]Pro-labeled D85N (49,66). This means that the transmembrane B α-helix is able to acquire a fluctuation motion with a frequency of 104 Hz beyond the kink at Pro 50 in the cytoplasmic side, as schematically illustrated in Fig. 5. Concomitantly, a fluctuation motion at the C-helix with a frequency of the order of 104 Hz is found to be prominent, due to deprotonation of SB at pH 10, as viewed from the suppressed 13C NMR signal of Pro 91. Accordingly, a novel mechanism as to proton uptake and transport is proposed on the basis of a dynamic aspect such that a transient environmental change from a hydrophobic to a hydrophilic nature at Asp 96 and SB is responsible for the reduced pKa which makes a proton uptake efficient as a result of the fluctuation motion at the cytoplasmic side of the transmembrane B and C α-helices.

Figure 5.

 Schematic representation of the dynamic behavior of the B and C α-helices of the D85N mutant, accompanied by protonation of the Schiff base, as viewed from the 13C NMR spectral behavior of Val 49 and Pro 91. (a) ground state (pH 7); (b) M-like state at pH 10 (52).

It is also interesting to note that the 13C NMR peak of the C-terminal α-helix was displaced to low frequencies due to the disrupted cytoplasmic complex due to the removal of salt bridges and metal ions. Consequently, such a surface structure is stabilized when blue membranes are prepared by either complete removal of surface-bound cations (deionized blue) or the neutralization of surface charge by a lowered pH to 1.2 (acid blue) (35). As a result, the accompanying low frequency motions lead to suppressed 13C NMR signals from the loop. Partial neutralization of the Glu and Asp residues at the extracellular side such as E194Q/E204Q (2Glu), E9Q/ E194Q/E204Q (3Glu), and E9Q/E74Q/E194Q/E204Q (4Glu) also causes global fluctuation motions at these loop regions as well as in the disorganized trimeric form (67). It turns out that these changes at the extracellular side do not strongly affect the protein dynamics as described above.


We have found that several degrees of fluctuation motion, fast and slow motions, are present in bR in the 2D crystalline state, depending upon the particular sites of bR. Well-resolved 13C NMR spectra are fully visible at ambient temperature as far as [3-13C]Ala- or [1-13C]Val-labeled bR are concerned, although the 13C NMR signals from the surface areas including loops and transmembrane α-helices near the surface (8.7 Å depth) are suppressed for the preparations labeled with [1-13C]Gly, Ala, Leu, Phe, Tyr, etc. due to interference between the frequency of the fluctuation motions and the frequency of the magic angle spinning. Even in the case of [3-13C]Ala or [1-13C]Val-labeled bR, 13C NMR signals from the loops and transmembrane α-helices are also suppressed when bR is reconstituted in the lipid bilayer due to the acquired fluctuation frequency which interferes with the frequency of the proton decoupling. This possibility should always be taken into account when the NMR spectra of any kind of membrane proteins are examined as monomers or oligomers in lipid bilayers. In order to achieve maximum peak-intensity and spectral resolution, a low temperature study leading to a 2D array is very helpful if specific lipids arising from expression from Halobacteria are present with bR (65).

A dynamic picture of the surface structure or cytoplasmic complex of bR from PM, consisting of the C-terminal α-helix and nearby loops, is well characterized by careful analysis of the 13C NMR spectra, especially those obtained by DD-MAS NMR. The C-terminal α-helix is also present in ppR which is directly involved in the complex formation with a cognate transducer (58,59,68,69). In particular, the C-terminal residues of ppR, protruding from the cytoplasmic surface, consist of the C-terminal α-helix stem and tip regions, in addition to the randomly coiled terminal. The latter tip region in ppR is in direct contact with the cytoplasmic α-helix of pHtrII(1-159) which provides stabilization, as well as the mutual helix–helix interaction which is stabilized by hydrogen bonds between the transmembrane α-helices (Kawamura et al., unpublished). This interaction, however, is destabilized in D75N with a pHtrII(1-159) complex as a photo-activated state. Therefore, Kawamura et al. proposed that a switching of pairs in the helix–helix interaction occurs from pHtrII-ppR (59) to pHtrII itself when ppR is converted to an activated state (D75N). These conformational and dynamical changes of ppR and pHtrII are related to the transmission of a signal to pHtrII (68). Further, it has been demonstrated that the surface structure of ppR, consisting of the C-terminal α-helix and E–F loop, is directly involved in the stabilization of the complex (68).

E. coli diacylglycerol kinase (DGK) is a membrane enzyme which catalyzes the conversion of diacylglycerol to phosphatic acid in membranes. Yamaguchi et al. have showed that the 13C NMR spectra of [3-13C]Ala-, [1-13C]Val-labeled DGK are broadened to yield rather featureless peaks at physiological temperatures, both in DM solution and in lipid bilayers in the liquid crystalline phase. This is due to interference of the motional frequencies of DGK with the frequencies of magic angle spinning or proton decoupling (104 or 105 Hz, respectively) (69). Distinct 13C NMR signals for up to six peaks, however, are well resolved due to the absence of such fluctuation motions in the gel-phase lipids, for the domains for the transmembrane and amphipathic α-helices and loops. This finding again indicates that acquisition of such low frequency motions with time scales of microseconds to milliseconds is essential for the protein backbone to facilitate the efficient enzymatic activity of DGK. The individual DGK monomers at physiological temperature are not always tightly packed as anticipated from the trimeric form but are loosely held together in order to allow the backbone fluctuations, even though they could be tightly packed reflecting the dynamic state of the surrounding lipids in the gel phase lipid.

These findings clearly indicate that the presence of slow fluctuation motions with correlation times of 10−4 to 10−5 s plays an essential role in their respective biological activities, although such motions could be a serious obstacle as viewed from the detailed analysis of the 3D structure at ambient temperature.


Acknowledgements— The authors are grateful to Professors J. K. Lanyi, Richard Needleman, Esteve Padrós, N. Kamo and James U. Bowie for cooperative works and stimulating discussions.