¹H and ¹³C MAS NMR evidence for pronounced ligand–protein interactions involving the ionone ring of the retinylidene chromophore in rhodopsin

Abstract

Rhodopsin is a member of the superfamily of G-protein-coupled receptors. This seven α-helix transmembrane protein is the visual pigment of the vertebrate rod photoreceptor cells that mediate dim light vision. In the active binding site of this protein the ligand or chromophore, 11-cis-retinal, is covalently bound via a protonated Schiff base to lysine residue 296. Here we present the complete ¹H and ¹³C assignments of the 11-cis-retinylidene chromophore in its ligand-binding site determined with ultra high field magic angle spinning NMR. Native bovine opsin was regenerated with 99% enriched uniformly ¹³C-labeled 11-cis-retinal. From the labeled pigment, ¹³C carbon chemical shifts could be obtained by using two-dimensional radio frequency-driven dipolar recoupling in a solid-state magic angle spinning homonuclear correlation experiment. The ¹H chemical shifts were assigned by two-dimensional heteronuclear (¹H-¹³C) dipolar correlation spectroscopy with phase-modulated Lee–Goldburg homonuclear ¹H decoupling applied during the t₁ period. The data indicate nonbonding interactions between the protons of the methyl groups of the retinylidene ionone ring and the protein. These nonbonding interactions are attributed to nearby aromatic acid residues Phe-208, Phe-212, and Trp-265 that are in close contact with, respectively, H-16/H-17 and H-18. Furthermore, binding of the chromophore involves a chiral selection of the ring conformation, resulting in equatorial and axial positions for CH₃-16 and CH₃-17.

Rhodopsin is the photosensitive protein of the rod photoreceptor in the vertebrate retina that mediates dim light vision. Rhodopsin represents a paradigm for the large and diverse family of the G protein-coupled membrane receptors (GPCRs) (1). The GPCRs play an essential role in the transduction of signals from the extracellular environment across the plasma membrane to the interior of every cell type and thus represent an important target for pharmacological intervention (1). Rhodopsin consists of 348-aa residues arranged in seven transmembrane α-helices that span the disk membranes of the rod outer segment (2–4). The chromophore of rhodopsin is an 11-cis-retinylidene prosthetic group that is bound to the protein via a protonated Schiff base (pSB) linkage to amino acid residue Lys-296 (Fig. 1A) (5). The trigger for photoreceptor activation is the light-induced isomerization of the 11-cis-retinylidene ligand to the all-trans configuration.

Figure 1.

Molecular structure and International Union of Pure and Applied Chemistry numbering of the 11-cis-12-s-trans-retinylidene chromophore in rhodopsin (A) and the 11-cis-12-s-cis-retinylidene pSB model (B).

The chromophore of rhodopsin has been studied extensively during the past decades by a variety of techniques. Solid-state ¹³C magic angle spinning (MAS) NMR spectroscopy has been used in the past to resolve essential details of the spatial and electronic structure of the chromophore. These studies have focused on the retinylidene chain by using ¹³C-labeled retinals and have assigned the chemical shifts for the polyene carbon atoms (6–12).

Recently, a 10-fold ¹³C-labeled 11-cis-retinal was incorporated in the active site of rhodopsin (12). With solid-state two-dimensional (2D) correlation spectroscopy, detailed information on the electronic structure of the end part of the polyene near the pSB was obtained. Analysis of the NMR data showed that the excess positive charge from the pSB is partially delocalized into the polyene chain, yielding a polaronic conjugation defect close to the nitrogen of the Schiff base (12, 14). This finding demonstrates the utility of multispin labeling in combination with solid-state 2D correlation spectroscopy for the study of ligand–protein interactions for GPCRs.

In the present study uniformly ¹³C-labeled 11-cis-retinal is reconstituted into native opsin in the natural membrane environment, enabling a comprehensive ¹H and ¹³C NMR assay of the electronic structure of the chromophore in the active site of the protein. The ¹H and ¹³C shifts of the ionone ring methyl groups are strongly perturbed by close contacts with the protein environment. The results are discussed in the context of recently established structural models of the chromophore binding pocket.

Materials and Methods

Uniformly ¹³C-labeled retinal was prepared by total synthesis, starting from commercially available 99% enriched ¹³C₂-acetonitrile, ¹³C₃-acetone, ¹³C₂-acetic acid, [2-¹³C]-acetic acid, and [1,2,3,4-¹³C₄]-ethyl acetoacetate (15). The 11-cis-retinal was obtained by a standard illumination and HPLC purification procedure (9). All subsequent manipulations with the purified 11-cis isomer and rhodopsin were performed in the dark or under dim red light conditions with λ > 620 nm. Opsin was isolated from fresh cattle eyes and reconstituted with the uniformly ¹³C-labeled 11-cis-retinal according to standard procedures (16). The resulting A₂₈₀/A₅₀₀ ratio of 1.9 ± 0.1 showed that the efficiency of the regeneration was better than 95% (17).

NMR spectra were acquired at 750-MHz ¹H frequency (Bruker, Karlsruhe, Germany). The sample was cooled to 223 K, and the MAS spin rate was 12 kHz for all experiments. The spectra were recorded by using 2.0-ms ramped cross polarization and two-pulse phase modulation decoupling during acquisition (18, 19). Radio frequency-driven dipolar recoupling correlation spectra were acquired by using a pathway-selective phase cycling method (20). The 2D heteronuclear correlation spectra were obtained with phase-modulated Lee–Goldburg decoupling during the t₁ period (21, 22).

¹H and ¹³C chemical shifts of the ligand are reported relative to tetramethylsilane, using the ¹H and ¹³C chemical shift assignments of the phospholipids that are present in the rhodopsin sample as an internal reference (23, 24). In the 2D MAS heteronuclear correlation spectrum, the C⩵C carbons of the phospholipid acyl chain are easily resolved from the C-10 chromophore response. Based on a global comparison of solid- and solution-state NMR data for lipids, we assign this lipid peak to isotropic proton and carbon shifts of 5.3 ppm and 128.7 ppm, respectively. The lipid methylenic ¹³C is clearly resolved at 28.7 ppm (7, 12). The associated phospholipid methylenic proton response at 2.7 ppm was used to calculate a Lee Goldburg scaling factor of 0.610, which is in line with the theoretical value of 1/√3 = 0.577 (25).

To calculate the Δσ (carbon NMR ligation shift) and Δσ (proton NMR ligation shift) for the polyene region, shifts reported for the pSB model N-(11-cis-12-s-cis-retinylidene)-n-propyliminium trifluoro acetate dissolved in CDCl₃ were used (26). Both σ (isotropic proton shift of pSB model compound) and σ (isotropic carbon shift of pSB model compound) of this model compound were published, whereas only the σ are reported for the N-(11-cis-12-s-cis-retinylidene)-n-propyliminium chloride that was used as a reference in our prior investigations (12). The σ of the ring methylene protons of the 11-cis-retinylidene pSB model has not been reported. Because there is ample experimental evidence that in free retinylidene compounds the chemical shifts of these protons are insensitive to the configuration of the C11⩵C12 bond, an all-trans-retinylidene pSB model, N-(all-trans-12-s-cis-retinylidene)-n-butyliminium triflate, was used to estimate the Δσ of the ring methylene protons (27–29).

Normalization of the Δσ and Δσ is based on the difference of the proton and carbon chemical shift scale. The protons have a chemical shift dispersion of ≈15 ppm whereas the carbon response has a much larger range of ≈200 ppm. Scaling according to Δσ̃ (normalized proton NMR ligation shift) = Δσ/15 and Δσ̃ (normalized carbon NMR ligation shift) = Δσ/200 provides a normalized image that is useful to compare proton and carbon NMR ligation shifts.

Results

The one-dimensional ¹³C cross polarization/MAS spectra from rhodopsin membranes containing either the uniformly ¹³C-labeled or the unlabeled 11-cis-retinylidene chromophore are shown in Fig. 4, which is published as supporting information on the PNAS web site, www.pnas.org. Several narrow signals, from methyl and methylene ¹³C in the ionone ring of the chromophore, are detected in the aliphatic region (5–40 ppm) of the spectrum. In the vinylic region (120–150 ppm) the resonances of the carbon labels in the polyene chain can be identified. The response at σ (isotropic carbon shift of chromophore) = 127.9 ppm is superimposed on the broad natural abundance signal around ≈130 ppm of the unsaturated carbon atoms of the phospholipid acyl chains and the aromatic side chains of the protein. Two signals with σ = 168.3 ppm and σ = 165.8 ppm with corresponding spinning side bands at σ = 103.9 ppm and σ = 101.4 ppm are from vinylic carbon atoms that are shifted downfield because of a relatively high positive atomic charge density. Finally, the broad signal at σ = 175 ppm is caused by the carbonyl moieties of peptide bonds and lipid ester groups.

In the composite Fig. 2, contour regions from 2D homonuclear (¹³C-¹³C) and 2D heteronuclear (¹H-¹³C) dipolar correlation spectra of the rhodopsin containing the uniformly ¹³C-labeled chromophore are shown. The network of nearest-neighbor correlation signals is indicated in the plot with pairs of regular numbers. The correlations between carbon nuclei of the molecular frame C₁–C₁₅ are indicated with solid lines in the corresponding panels of the 2D radio frequency-driven dipolar recoupling spectrum. Furthermore, correlations between the carbon atoms of the molecular frame and the directly attached methyl groups are shown with dashed lines.

Figure 2.

Contour regions of the 2D homonuclear (¹³C-¹³C) and 2D heteronuclear (¹H-¹³C) dipolar correlation spectra of the uniformly ¹³C-labeled retinylidene in rhodopsin. The regions shown in the Top Left (C^AC^V) and Middle Right (C^VC^A) display correlations between vinylic and aliphatic carbon nuclei. Correlations between vinylic ¹³C are in the Middle Left (C^VC^V), couplings between aliphatic ¹³C are revealed in the Top Right (C^AC^A). (Bottom) The correlations between protons and vinylic (C^VH) or aliphatic (C^AH) ¹³C are shown.

Even for the relatively short mixing time of 1.23 ms that was used for the radio frequency-driven dipolar recoupling experiment, relayed transfer along the ¹³C-labeled network gives rise to additional correlations, which are indicated in Fig. 2 with italic numbers. In particular, the methyl groups of the chromophore show relayed correlations with vinylic carbon atoms. The longer-range correlations are generally weaker than the nearest-neighbor correlations. An exception is the relatively strong long-range correlation between carbon C-4 of the ionone ring and CH₃-18. Its signal is superimposed on a broad and weak cross peak of C-3 and C-4.

The 2D ¹³C dataset leads to a complete assignment of the ¹³C responses of the chromophore (Table 1). The σ for the polyene chain correspond with the assignments reported by Smith et al. (7) and Verhoeven et al. (12) within the experimental errors of 0.4 and 0.6 ppm, respectively. This finding validates the procedure of calibrating the ¹³C shift scale by using the lipid response. The ¹³C assignment can be used for an assignment of the proton signals from the chromophore bound to its receptor target in the natural membrane environment. High-field ¹H-¹³C heteronuclear correlation data from the rhodopsin containing the uniformly ¹³C-labeled chromophore are shown in the bottom of Fig. 2. These data were collected with a short cross polarization contact time of 50 μs to excite predominantly heteronuclear correlations between ¹³C nuclei and directly bound protons. For most heteronuclear correlation signals, the high resolution in the carbon dimension is sufficient to assign the corresponding isotropic proton shifts of the chromophore (σ_lig^H, Table 2). The aliphatic region of the 2D ¹H-¹³C spectrum (C^AH) is complicated because of strong overlap with correlations involving protons attached to carbons of phospholipid and protein. A 2D ¹H-¹³C heteronuclear correlation spectrum of natural abundance rhodopsin was therefore collected to identify the “background” signals (data not shown).

Table 1.

The complete carbon assignment σ_lig^C of the 11-cis-retinylidene chromophore of rhodopsin compared with σ_pSB^C for N-(11-cis-retinylidene)-n-propyl-iminium trifluoro acetate in solution to obtain the Δσ_lig^C

Position	σligC, ppm	σpSBC, ppm*	ΔσligC, ppm
C-1	34.0	34.1	−0.1
C-2	40.3	38.9	1.4
C-3	20.3	18.8	1.5
C-4	34.0	33.0	1.0
C-5	130.9	132.1	−1.2
C-6	137.0	137.2	−0.2
C-7	132.8	132.3	0.5
C-8	139.1	137.2	1.9
C-9	149.0	147.8	1.2
C-10	127.9	126.4	1.5
C-11	141.4	138.7	2.7
C-12	132.2	128.7	3.5
C-13	168.3	165.8	2.5
C-14	122.3	120.5	1.8
C-15	165.8	163.3	2.5
C-16	30.6	28.9	1.7
C-17	26.1	28.9	−2.8
C-18	21.7	22.1	−0.4
C-19	14.4	12.6	1.8
C-20	16.3	18.8	−2.5

^*Data from Shriver et al. (26). 

Table 2.

The complete proton assignment σ_lig^H of the 11-cis-retinylidene chromophore in rhodopsin

Position	σligH, ppm	σpSBH, ppm	ΔσligH, ppm
H-2	1.0	1.49*	−0.5
H-3	1.6	1.63*	0.0
H-4	1.0	2.06*	−1.1
H-7	6.4	6.55†	−0.2
H-8	6.2	6.36†	−0.2
H-10	6.4	6.98†	−0.6
H-11	7.2	7.12†	0.1
H-12	7.2	6.31†	0.9
H-14	7.0	6.71†	0.3
H-15	8.1	9.19†	−1.1
H-16	0.8	1.05†	−0.3
H-17	0.6	1.05†	−0.5
H-18	0.5	1.73†	−1.2
H-19	2.2	2.14†	0.1
H-20	1.7	2.57†	−0.9

The Δσ_lig^H are calculated from σ_lig^H minus σ_pSB^H as described in the text. 

^*Data from Elia et al. (29). 

^† Data from Shriver et al. (26). 

Discussion

By comparing the shifts for the chromophore in rhodopsin with the responses collected from model compounds, an NMR assay of the spatial and electronic structure of the chromophore can be obtained (12). The Δσ_lig = σ_lig − σ_pSB are listed in Tables 1 and 2 for the ¹³C and the ¹H responses, respectively. The Δσ_lig reflect differences in the electronic and spatial molecular structure between the pSB model compound and the chromophore bound in the active site of the protein. They can be used as a probe for ligand–protein interactions between the chromophore and the protein binding pocket, including protein-induced conformational restraints in the ligand or electronic effects on the chromophore exerted by the protein.

It has been shown that Δσ can be quite informative when considered at the molecular level and reveal systematic variations in the bound ligand (12). In comparison, Δσ is even more sensitive in providing information about nonbonding interactions between the chromophore and the protein. A Δσ ≥ 0.5 ppm can be considered highly significant on the proton shift scale of ≈15 ppm. Thus the combination of Δσ and Δσ can give comprehensive information about localized nonbonding interactions between ligand and receptor.

Fig. 3A provides a visual representation of the Δσ and Δσ of the chromophore in rhodopsin. The color-encoded NMR ligation shift patterns can be used for a discussion of local protein–ligand interactions. In addition, the scaled Δσ̃ and Δσ̃ provide a view of the relative magnitude of the ¹³C and ¹H effects and an overview of significant ligand–protein interactions acting on the chromophore bound in the active site of the protein (Fig. 3B).

Figure 3.

Visual representation of (A) the Δσ and Δσ NMR ligation shifts in Tables 1 and 2 and (B) normalized Δσ̃ and Δσ̃. The model corresponds with the image of the ground state structure calculated with Carr–Parrinello molecular dynamics (10, 14). Blue and red colors reflect the upfield and downfield Δσ_lig, respectively, and the larger the shift the darker the color.

Charge Delocalization in the Polyene Chain of the 11-cis-Retinylidene Chromophore.

Previous experiments already revealed downfield Δσ for the C10–C15 region in the tail end of the polyene chain of the retinylidene in rhodopsin in comparison with N-(11-cis-retinylidene)-n-propyliminium chloride (12). The Δσ was ascribed to three synergistic contributions leading to an excess of positive charge in the polyene: (i) the electronegative nitrogen, (ii) the protonation, and (iii) the counterion strength. The color-encoded data in Fig. 3A show a similar pattern for the Δσ of the vinylic carbon nuclei in the tail end of the polyene chain of the chromophore, with the largest downfield shift for C-12, Δσ = 3.5 ppm (Table 1).

The stronger positive charge delocalization into the rhodopsin polyene compared with the model compounds is thought to be caused by electrostatic interactions with polar or negatively charged side chains of amino acid residues that are in close contact with the tail end of the polyene (30–32). Based on earlier solid-state NMR studies on ¹⁵N-Lys-labeled rhodopsin, an effective center-to-center distance between the counterion and the pSB nitrogen of 0.43 ± 0.01 nm was estimated (33, 34). This effective radius is larger than the 0.34 nm derived from the recently published crystal structure and implies the presence of a complex counterion (3, 4). Such a complex counterion could be organized around a hydrogen-bonded network positioning a water molecule between the positively charged nitrogen of the pSB and the negatively charged carboxylate of Glu-113 (34). Several studies showed that upon addition of D₂O the proton of the pSB exhibits rapid H-D exchange already in the dark, indicating that the Schiff base region is easily accessible for bulk water molecules (35, 36).

The Palczewski model shows another glutamic acid residue, Glu-181, in the second extracellular loop, and a tyrosine residue, Tyr-268, in transmembrane helix VI, to be in close contact with the tail end of the polyene (3, 4). Site-directed mutagenesis studies of either residue leads to significant shifts in the spectral properties of the pigment (37, 38). Therefore, similar to Glu-113, Glu-181 and Tyr-268 may be involved in modulating the positive charge delocalization in the polyene, resulting in the observed large downfield Δσ of the carbon resonances in the tail end of the polyene chain. In addition, these two amino acid residues might contribute to a hydrogen-bonded network. Ab initio molecular dynamics simulations provided evidence that the ¹³C shifts of the carbons of the tail end of the polyene are sensitive to the position of the counterion (30). These simulations also suggest that other charged residues that are located in the vicinity of the C10-C13 region, like Glu-181 and Tyr-268, may contribute to the observed downfield Δσ.

Conformation of the 11-cis-Retinylidene Chromophore.

By using a uniformly ¹³C-labeled chromophore, the assay of the charge delocalization is extended to the ring end of the polyene. In the past, NMR studies with 5-¹³C-labeled all-trans-retinal reconstituted into bacteriorhodopsin have demonstrated that the counterion strength affects the shift of carbon C-5 in the ionone ring of the chromophore (39, 40). This finding contrasts with the data for rhodopsin, because the charge delocalization appears to be restricted to the region between carbon C-7 and C-15 (Fig. 3A). The difference between the two retinal proteins may be related to the conformation of the C6–C7 single bond. At an early stage it was established with MAS NMR shift tensor measurements using rhodopsin reconstituted with ¹³C-5-labeled retinal and retinoic acid model compounds that the 6–7 single bond has a 6-s-cis conformation in rhodopsin, which is different from the 6-s-trans-retinylidene chromophore in bacteriorhodopsin (7).

Recent deuterium MAS experiments on oriented rhodopsin preparations have been interpreted in terms of a 6-s-trans conformation of the chromophore (41). In contrast, the x-ray data point to a 6-s-cis structure, but the resolution is insufficient to rule out the possibility of a mixture of 6-s-cis and 6-s-trans (4). MAS NMR chemical shifts have been shown to be a sensitive indicator of this bond conformation, and our shifts match those of the previous MAS NMR study, thus reconfirming the 6-s-cis conformation (7, 39). Additionally, as can be seen in Fig. 2, there is no evidence of signal doubling or splitting and therefore the ring conformation must be unique.

In the past, NMR studies have provided convincing evidence for a conformational change leading to a nonplanar C10–C13 segment for the polyene chain in the chromophore relative to the models in solution (Fig. 1) (9, 10). Because the 12-s-cis conformation is energetically favorable over the 12-s-trans, the equilibrium will be shifted in solution toward 12-s-cis (Fig. 1B) (42). In the protein, however, there is a nonplanar 12-s-trans conformation (Fig. 1A) (4, 10, 43). Also the most recent refined x-ray model for rhodopsin includes an out-of-plane distortion in the C₁₀–C₁₃ segment with a dihedral angle for C₁₀–C₁₁–C₁₂–C₁₃ of ϕ = 7.9° (4). This distortion is in line with the torsional angle of C₁₀–C₁₁–C₁₂–C₁₃ of ϕ = 17°, which was calculated from the NMR distance constraints with Carr–Parrinello molecular dynamics (44). The out-of-plane distortion is distributed over several bonds (10).

The ¹H shift data provide additional qualitative support for a conformational change involving the C10–C13 segment of the chromophore by binding to the protein, because the σ of the 12-s-trans chromophore in rhodopsin is different from the σ of the pSB model in the 12-s-cis conformation in solution.

The downfield Δσ̃ of H-12 and H-14 may be caused by a reduced intramolecular steric interaction between CH₃-20 and the nearby protons H-12 and H-14, relative to the pSB model. In addition, the upfield Δσ̃ of H-10 and H-20 may reflect the steric interaction between H-10 and the methyl group C-20 in the 11-cis-12-s-trans-retinylidene chromophore bound to the protein, which is released in the free 12-s-cis conformation. Furthermore, the observed upfield Δσ = −2.5 ppm of CH₃-20 is in line with a distorted s-trans conformation of the 12-s bond. Addition of a methyl group at C-10, giving 10-methylrhodopsin, yields a larger out-of-plane distortion in the isomerization region compared with rhodopsin and an additional upfield Δσ = −1.6 ppm compared with the C-20 response in rhodopsin (10, 45, 46). This finding suggests a correlation between torsion and the ¹³C-20 shift. Finally, a relatively large Δσ = −1.1 ppm is detected for the H-15. This Δσ may be produced by a difference in the conformation around the C14–C15 bond between the 11-cis-retinylidene chromophore and the pSB model compound or by an interaction with a nearby protein residue (3, 4, 33, 34).

The absence of conformational shifts for C-7, C-8, and C-19 provides strong evidence for a highly similar electronic and spatial conformation of the rhodopsin chromophore and the pSB model. In particular, Fourier transform IR and resonance Raman spectroscopic studies have been interpreted in terms of a close contact between the C-19 methyl group and the surrounding protein (47, 48). Although these studies indicate strong ligand–protein interactions for CH₃-19, it is clear from the NMR that the protein has no significant effect on the conformation of the chromophore around C-19.

Finally, both methyl groups C-16 and C-17 of the pSB model compound in solution resonate with σ = 28.9 ppm (26). These methyl groups have significantly different shifts in rhodopsin (54). One methyl group is detected with σ = 26.1 ppm, corresponding with a Δσ of −2.8 ppm, whereas the other resonates with σ = 30.6 ppm, a Δσ of +1.7 ppm. In contrast, the Δσ = −0.5 ppm and −0.3 ppm for the protons of these two methyl groups are both upfield and of comparable magnitude.

The single resonance of CH₃-16 and CH₃-17 of the pSB model compound in solution is caused by motional averaging between the equatorial and axial conformation for the two methyl groups attached to the sp³-hybridized C-1 of the six-membered ring (49). Evidently the protein prevents the ring flip that occurs in solution. In this way, our data contribute to converging evidence that the chromophore conformation is unique with an equatorial position for CH₃-16 and an axial CH₃-17 methyl caused by the steric constraints from the protein binding pocket (54). The relatively large upfield Δσ of C-17 can be explained in terms of a γ effect caused by steric hindrance between the protons of C-17 and H-3 (50, 54). Because the CH₃-16 occupies an equatorial position, it experiences weaker steric interactions than in the pSB model in solution, which can explain the observed downfield NMR ligation shift.

NMR Ligation Shifts in the Ionone Moiety of the Chromophore Indicate Pronounced Interactions with the Protein Environment.

Binding studies on rhodopsin using acyclic retinal analogs have shown that the ionone moiety plays a central role in the formation of a stable photoactive pigment. In particular, the methyl groups attached to the ring appear to be essential for binding (51). Reconstitution studies have provided evidence for specific protein–ligand interactions, which are essential for correct positioning of the ionone ring and in particular the methyl groups. The complete proton and carbon assignments of the ionone moiety of the chromophore provide a view on the molecular details of these interactions with atomic selectivity.

In particular, the proton shifts of the ligand can reveal the presence of aromatic side chains in the binding pocket. Calculations of ring current intensities demonstrate that an extension of the delocalization of electrons in an aromatic system produces an increase of the magnitude of the ring current effect that leads to upfield proton shifts (52). These ring current effects can become highly significant on the proton shift scale of ≈15 ppm. The Δσ̃ reveal relatively strong ligand–receptor interactions for the ionone ring of the chromophore. The CH₃-18 protons show the largest upfield Δσ̃ observed in the chromophore. The protons of the adjacent carbon C-4 also demonstrate a large upfield Δσ̃. These relatively large Δσ̃ indicate the presence of a tryptophan amino acid residue (52). Earlier studies already suggested that a tryptophan residue should be located in the vicinity of the ionone ring (13, 53). Based on these studies and the recently published x-ray model, a tryptophan residue at a distance of ≈3.8 Å to C-18, Trp-265 in transmembrane helix VI, is most likely responsible for these interactions (3, 4). The Δσ̃ for the CH₃-16 and CH₃-17 can be explained in terms of ring current effects from two nearby phenylalanine side chains, Phe-208 and Phe-212 of transmembrane helix V (3, 44, 52). According to the x-ray model, Phe-212 is also located close to the methylene protons attached to C-2 (≈4 Å), which is in line with the observed upfield Δσ̃ for the C-2 protons. In this way the NMR data contribute to converging evidence that nonbonding interactions between the methyl groups and the nearby phenylalanines contribute to the “scaffolding” of the protein structure and the retinylidene binding pocket.

Other parts of the ionone moiety appear less important for binding to the protein. The carbon responses of C-2 and C-3 are relatively broad, which indicates disorder of this part of the ring. This finding correlates with ligand analog experiments, which demonstrated that acyclic retinal analogs could bind as well. It also suggests that C-2 and C-3 play only a modest part in rhodopsin–ligand recognition in the dark inactive state of the receptor (51).

It is difficult to attribute the small Δσ̃ to a specific mechanism. For instance, C-5 shows a small upfield Δσ = −1.2 ppm, with Δσ̃ = −0.4 ppm for the adjacent CH₃-18. The x-ray model and rotational resonance solid-state MAS NMR studies demonstrate that the C5–C8 segment has a twisted out-of-plane conformation (3, 4, 54). Ab initio calculations on pSB models suggested that the C5⩵C6 bond and the adjacent methyl group C-18 lie in one plane (42). Conformational distortions might be the origin of the observed small Δσ̃ for C-5 and C18. In addition, the x-ray model of rhodopsin indicates that the carboxylic group of a glutamic acid residue, Glu-122, is in close contact to C-5 (≈3.8 Å) (3, 4). Interaction with the side chain of such a polar amino acid could also result in a small upfield Δσ.

Conclusions

High-field solid-state 2D homonuclear and 2D heteronuclear MAS NMR dipolar correlation spectroscopy on rhodopsin reconstituted with a uniformly ¹³C-labeled 11-cis-retinylidene chromophore provided a complete ¹³C and ¹H chemical shift assignment for the chromophore. The Δσ_lig and Δσ̃_lig reflect the spatial and electronic structure of the chromophore in the active site of rhodopsin relative to the pSB model in solution and thus provide a detailed view on the mechanisms of ligand–protein interactions with atomic selectivity.

Pronounced Δσ̃ are observed for the methyl protons of the ring moiety. The data provide converging evidence for nonbonding interactions between the chromophore and the protein binding pocket involving the aromatic amino acid residues Phe-208, Phe-212, and Trp-265 that are in close contact with H-16/H-17 and H-18, respectively. The Δσ̃ in the polyene chain and the relatively large downfield Δσ observed for CH₃-20 are consistent with a nonplanar conformation of the 12-s bond. Finally, binding of the chromophore involves chiral selection by fixation of the ring puckering, which results in well-defined axial and equatorial positions for CH₃-17 and CH₃-16.

This study shows that the combination of a uniformly ¹³C-labeled ligand incorporated into the active binding site of a GPCR with 2D solid-state NMR correlation spectroscopy can provide detailed information about ligand–protein interaction patterns. In this way a solid-state NMR “snapshot” of the spatial and electronic ground state structure of a ligand bound to its GPCR target in the natural membrane is obtained.

Abbreviations

GPCR

G protein-coupled receptor

pSB

protonated Schiff base

MAS

magic angle spinning

2D

two-dimensional

Δσ

proton NMR ligation shift

Δσ

carbon NMR ligation shift

Δσ̃

normalized proton NMR ligation shift

Δσ̃

normalized carbon NMR ligation shift

σ

isotropic proton shift of chromophore

σ

isotropic carbon shift of chromophore

σ

isotropic proton shift of pSB model compound

σ

isotropic carbon shift of pSB model compound