Characterisation of Schiff base and chromophore in green proteorhodopsin by solid-state NMR

The proteorhodopsin family consists of hun-dreds of homologous retinal containing membrane proteins found in bacteria in the photic zone of the oceans. They are colour tuned to their environment and act as light-driven proton pumps with a potential energetic and regulatory function. Precise structural details are still unknown. Here, the green proteorhodopsin variant has been selected for a chemical shift analysis of retinal and Schiff base by solid-state NMR. Our data show that the chromophore exists in mainly all- trans conﬁguration in the proteorhodopsin ground state. The optical absorption maximum together with retinal and Schiff base chemical shifts indicate a strong interaction network between chromophore and opsin.


Introduction
Proteorhodopsin (PR) is a retinylidene membrane protein found in the genomes of many species of marine c-proteobacteria in the oceans (Beja et al. 2000;Sabehi et al. 2005). Approximately 800 variants have been found so far which share 80% sequence identity but show different absorption maxima shifted by *35 nm between green and blue PR (Wang et al. 2003). The green absorbing form (27 kDa, primary accession number Q9F7P4, k max *520 nm) has been the focus of a number of studies. Based on comparison with bR, its 249 residues are predicted to be arranged in seven transmembrane helices. Green proteorhodopsin (PR) shares a 30% sequence identity with bacteriorhodopsin (bR). The primary proton acceptor D85, the proton donor D96, the Schiff base linkage K216 and the counterions of the Schiff base R82 and D212 in bR correspond to D97, E108, K231, R94 and D227 in PR, respectively. In contrast to bR, PR has no ionisable extracellular groups for a fast release of protons to the extracellular environment. The pK a of the primary proton acceptor in PR is five pH units higher compared to bR (PR: pK a,D97 *7.2, bR: pK a,D85 *2.5). Outward driven proton pumping has been found at pH [ pK a,D97 but experiments Mark Lorch and Andreas C. Woerner contributed equally to this work.
Electronic supplementary material The online version of this article (doi:10.1007/s10858-007-9203-5) contains supplementary material, which is available to authorized users. by Friedrich et al. have also shown an inversion at pH \ pK a,D97 (Friedrich et al. 2002). These findings suggest that PR could also have a regulatory function, but the proton acceptor is usually deprotonated under sea water conditions allowing the build up of an electrochemical potential. Considering the high abundance of PR, a very important role in the conversion of light energy to chemical energy in the photic zone of the oceans can be envisaged (Walter et al. 2007). PR undergoes a photocycle with a short turnover time of about 20 ms which contains K, M, N and O intermediates (Beja et al. 2000;Dioumaev et al. 2002). While the photocycle has been studied spectroscopically in some detail (Bergo et al. 2004;Dioumaev et al. 2002;Imasheva et al. 2005;Lenz et al. 2006Lenz et al. , 2007Varo et al. 2003;Wang et al. 2003), no structural data are available yet.
Here, we present a first solid-state NMR characterisation of chromophore and Schiff base of PR reconstituted into lipid bilayers. Solid-state NMR has been very extensively used to study its archaeal homologue bR, the best characterised membrane protein in terms of structure and function (for reviews see for example Lanyi and Schobert 2002;Luecke and Lanyi 2003). In fact it was the first membrane protein to be studied by solid-state NMR (Engelhard et al. 1990;Harbison et al. 1984;Kinsey et al. 1981). Therefore, an excellent knowledge base is available to which our data can be compared. Here, we report the 13 C chemical shifts of 10,11-13 C 2 retinal in PR and for comparison in bR. The number of resonance lines allows a direct assessment of the number of retinal populations within the protein. Their chemical shifts are valuable indicators for interactions within the binding pocket and could be used for quantum chemical calculations (Dreuw 2006). Our data show unambiguously an all-trans retinal conformation in dark adapted PR. We have also measured 15 N chemical shifts of the protonated Schiff base 15 N f-K231 in the wild type and upon a D97N single site mutation. This mutation emulates a photocycle intermediate state by keeping the primary proton acceptor in a ''protonated'' state. Comparing the relationship of the 15 N chemical shift and chromophore absorption maximum with those found in bR provides new insight into the PR retinal environment.
Preparation of [ 15 N f-Lys]-PR: Expression of PR (wt and D97N) was essentially done as described (Friedrich et al. 2002) but cells were grown in defined medium (Muchmore et al. 1989) with 50 lg/mL kanamycine until an OD 578 of 0.8 was reached. After spin-down, cells were resuspended in defined medium containing 15 N f-Lys. After 15 min of incubation, 1 mM IPTG and 0.7 mM of all-trans retinal were added. Cells were harvested after three hours of incubation.
Preparation of [10,11-13 C 2 Ret]-PR: LB medium was used instead of defined medium. Cells were induced by 1 mM IPTG when an OD 578 of 0.8 was reached. Simultaneously, 0.7 mM 10,11-13 C 2 all-trans retinal was added.
Purification: His-tagged PR was essentially purified as described by others (Hohenfeld et al. 1999) with only minor modifications (supporting information). An average protein yield of DDM solubilised PR of 10 mg/L culture was obtained (A 280 /A 518 = 1.9).
Reconstitution: To prepare proteoliposomes, 1 mg of DDM solubilised protein (1 mg/ml) was added to 5 mg of lipids (molar ratio POPC:POPG = 5:1) dissolved in 1 ml buffer containing 300 mM NaCl, 50 mM MES, 2% OG, pH 6. After incubation for 1 h at room temperature, polystyrene beads were added for detergent removal for two hours at room temperature followed by incubation over night at 4°C. Detergent removal by dialysis and the use of total E. coli lipid mixture worked equally well. After centrifugation, the pellet was transferred to a 4 mm MAS rotor. Light adaption was performed by spreading the sample as a thin film and illuminating the sample with white light for 20 min at 150 W and subsequent freezing into liquid nitrogen.
Optical spectroscopy: UV/VIS spectra of detergent solubilised PR were acquired on a Jasco V-550 spectrophotometer. Spectra at various pH values were taken in buffers containing 300 mM NaCl, 0.5% DDM including 20 mM Na-acetate (pH 4-5), 50 mM MES (pH 6), 50 mM HEPES (pH 7), 50 mM Tris (pH 8-9) or 50 mM phosphate buffer (pH 10-13). 13 C and 15 N MAS-NMR: Experiments were carried out on a Bruker Avance 600 equipped with a 4 mm MAS DVT probe head. Cross polarization experiments were performed at 60.88 MHz or 150.92 MHz for 15 N and 13 C, respectively. An 80-100% ramped proton pulse of 0.5 ms for 15 N and 1.0 ms for 13 C was used during the Hartmann-Hahn match. Proton decoupling (Fung et al. 2000) of typically 90 kHz was applied during a 50 ms acquisition time. The recycle delay time was 2 s. Spectra were referenced externally to the 15 N resonance of solid 15 N ammonium sulphate at 27 ppm and to the 13 C signal of TMS at 0 ppm. All measurements were carried out at sample spinning of 10 kHz at 210 K. 13 C double-quantum filtering was achieved by using a POST-C7 experiment with 110 kHz decoupling during 457 ls double-quantum excitation and reconversion (Lee et al. 1995). An amount of 4-6 mg of protein was used per sample and 30-40 k scans were acquired per experiment.

Results and discussion
13 C chemical shifts of 10,11-13 C 2 retinal in PR 13 C MAS NMR spectra of PR with 10,11-13 C 2 retinal reconstituted in lipid bilayers are shown in Fig. 1. For dark adapted PR, two signals are detected at 138.5 and 130 ppm (Fig. 1a). For comparison, we have carried out the same experiment on dark adapted bR (Fig. 1c). Here, due to the cis-trans isomerisation equilibrium of the chromophore, four resonances are found, which can be assigned to all-trans-C 11 , 13-cis-C 11 , all-trans-C 10 and 13-cis-C 10 (Harbison et al. 1985).
The chemical shifts of both peaks in PR are within the range expected for C 11 and C 10 , based on the comparison with model compounds, with bR and with rhodopsin (Han et al. 1993;Harbison et al. 1985). In retinal, all oddnumbered carbon positions are always less shielded than even ones (Bremser and Paust 1974). Therefore, we assign the resonances at 138.5 and 130 ppm to C 11 and C 10 , respectively. The line widths of the C 11 and C 10 peaks are 3 and 2 ppm. Their differences could be caused by a small retinal heterogeneity or stronger coupling to adjacent protons of C 11 . The fact that only two peaks are observed instead of four shows unambiguously that ground state PR contains only all-trans retinal. In previous extraction experiments, where the retinal is studied outside the protein, all-trans mixed with 5% (Imasheva et al. 2005) or 20% (Friedrich et al. 2002) 13-cis retinal had been found. Our results are also supported by FTIR and Raman data which had been interpreted in favour of a predominantly all-trans retinal population (Bergo et al. 2004;Dioumaev et al. 2002;Imasheva et al. 2005).
Extensive studies on retinal protonated Schiff base model compounds have been carried out in order to identify factors which influence 13 C chemical shifts and absorption maxima of the chromophore (Albeck et al. 1992;Hu et al. 1994). It has been found that k max increases with the chemical shifts of odd numbered carbons (Albeck et al. 1992). Comparing the C 11 chemical shifts of all-trans retinal in bR and PR (bR: 139.1, PR: 138.5 ppm) with their absorption maxima (bR: 568 nm, PR: ca. 520 nm) shows that they follow this trend but also deviate significantly from the model compounds. These differences are most likely caused by the steric interactions with residues in the binding pocket and their effects onto the retinal structure, the retinal ring orientation (6s-cis or 6s-trans), the effects Fig. 1 Double quantum filtered 13 C MAS spectra of PR (a, b) and bR (c) with 10,11-13 C 2 retinal. Dark adapted PR shows two well resolved resonances at 130 and 138.5 ppm which can be assigned to C 10 and C 11 (a). Two small shoulders appear upon light adaptation, but chemical shifts and intensities of both main resonances are not significantly altered (b). In contrast, dark adapted bR shows four resonances at 139.1, 135.1, 133 and 129.7 ppm which correspond to all-trans-C 11 , 13-cis-C 11 , all-trans-C 10 and 13-cis-C 10 , respectively of charges, or differences in the effective electricity constant within the retinal environment. However, the individual contributions of these factors must be addressed by further solid-state NMR experiments.
C 10 chemical shifts are less dependent on the parameters mentioned above, but the difference between bR and PR is significantly larger (3 ppm) than for C 11 (0.6 ppm). This could be explained by a different chemical environment such as a tighter binding pocket: Although PR shares a general 30% sequence identity (70% in binding pocket) with bR, important differences exist such as the high pK a of D97 and the presence of a highly conserved His-75 close to the retinal binding pocket.
Upon light adaption, a slight broadening of the baseline with two small shoulders at 139.9 and 125.8 ppm is observed (Fig. 1b). This could be due to conformational changes of a small fraction of the retinal. Resonance Raman spectra did also provide evidence for an almost entirely all-trans retinal conformation in light adapted PR (Dioumaev et al. 2002;Krebs et al. 2003). 15 N chemical shifts of the protonated Schiff base 15 N f-K231 in PR As shown for bR, 15 N f chemical shifts of the Schiff base also report on the retinal conformation (Harbison et al. 1983;Hu et al. 1997b). We have therefore inspected the 15 N chemical shift of the protonated Schiff base at K231. The 15 N CP MAS spectrum of dark adapted 15 N f-lys PR is shown in Fig. 2a. A single resonance of the protonated Schiff base is found at 186 ppm while all other 15 N f-lys resonate at 37 ppm. The spectrum does not change upon light adaption (Suppl. Material). In dark adapted bR, the protonated Schiff base is found at 171 ppm for all-trans and at 178 ppm for 13-cis retinal, which changes to a single resonance at 171 ppm upon illumination (re-referenced to solid ammonium sulphate) (Harbison et al. 1983). It is intriguing to note that the chemical shift in PR is similar to bovine rhodopsin (Creemers et al. 1999), but this seems merely a coincidence considering the differences in retinal conformations, protein sequence and evolutionary origin. The spectral resolution obtainable in our samples is comparable to purple membrane preparations and should be sufficient to resolve different species in PR, if they did exist. As this is not the case, our 15 N measurement supports the conclusion of a predominantly all-trans conformation of the retinal, which was drawn from our 13 C data. Compared to bR, the protonated Schiff base in PR is less shielded. Based on a large number of Schiff base chemical shifts (d) from all-trans retinylidene model compounds with both 6s-cis and 6s-trans ring orientation, an empirical relationship with the distance from the Schiff base (d) to the counterions and the absorption maximum (k max ) in form of d * 1/d 2 * 1/k max had been found which agrees very well with bR (Hu et al. 1997a) but differs for PR (Fig. 3). Based on the larger chemical shift in PR compared to bR, a stronger counterion interaction (i.e., a shorter effective distance) should be expected. For example, an outward movement of the arginine (R94) side chain in PR compared to bR (R82) would weaken the interaction between R94 and D97 and thus strengthen the coupling of D97 and the protonated Schiff base (PSB) (Krebs et al. 2003). When comparing the correlation between chemical shifts and k max (Fig. 3), an absorption maximum of 480 nm (20833 cm -1 ) is predicted while *520 nm (19231 cm -1 ) in detergent and 545 nm (18349) in the reconstituted samples were actually observed. Additional factors must contribute to this red shift, such as twisting around double bonds within the retinal (Hu et al. 1997a) or negative charges close to the chromophore (Hu et al. 1994). Indeed, homology modelling of PR onto bR (1C3W) (Friedrich et al. 2002) reveals the location of a negative charged D142 within a distance of 8 Å to the retinal. The corresponding residue in bR, A126, is neutral.
A monotonic relationship had been described for the principal values r 33 and r 22 of the 15 N chemical shift anisotropy tensor of the PSB in model compounds and bR 568 (de Groot et al. 1989). Using low speed MAS, we were also able to determine the 15 N chemical shift anisotropy to -132 ppm with an asymmetry parameter of  Fig. 4). This is similar to bR but r 33 /r 22 deviates from the empirical rule most likely due to a stronger network of couplings around the Schiff base as discussed above. Furthermore, FTIR and Raman studies suggest a strong hydrogen coupling network between chromophore and adjacent residues or water molecules, which could also contribute to the chemical shift and k max differences between PR and bR (Ikeda et al. 2007) (Fig. 3).
A comparison of UV/VIS absorption spectra with 15 N MAS NMR data is shown in Fig. 5 for different pHs. For wild type PR, k max shifts from 545 nm at pH 4 to 516 nm at pH 9, which is thought to be mainly due to the deprotonation of D97 (Friedrich et al. 2002). At the same time, the Schiff base chemical shifts and signal intensities remain almost constant between pH 6 and 11 (Fig. 5a). Below pH6 and above pH12 resonance line broadening has been observed. Signals for the deprotonated Schiff base could only been detected for pH [ 12 where the protein starts to denature. Schiff base deprotonation is seen in the UV/VIS spectra where an additional absorption maximum at 366 nm above pH12 occurs (Imasheva et al. 2004). The high pH stability of K231 could be explained with a high dielectric constant potentially resulting from a water cluster which could stabilize the PSB. Indeed, recent FTIR data suggests a strongly hydrogen bonded water cluster in alkaline PR (Ikeda et al. 2007).
The 15 N CP MAS spectrum of the 15 N f-lys labelled D97N mutant differs from the wild type as the Schiff base resonance is found at 179 ppm and a general broadening is observed (Fig. 5b) which is most likely caused by an altered counterion interaction (Hatcher et al. 2002). The absorption maximum k max is found at 560 nm between pH 4 and 9 and shifts to 400 nm above pH 9 at which also the 15 N resonance starts to decay (Fig. 5b). The formation of such an M-like state is also supported by the observation of a downfield shifted peak at 306 ppm (data not shown) corresponding to the deprotonated Schiff base. Therefore, the D97N mutant could potentially offer the opportunity to study functional and structural properties of the M-like state in the PR photocycle by solid-state NMR.

Summary
The 13 C chemical shifts of 10,11-13 C 2 retinal and the 15 N chemical shift of the protonated Schiff base K231 in green proteorhodopsin show unambiguously that ground state PR only contains all-trans retinal. The C 10 and C 11 are more shielded in PR compared to bR. The large 15 N chemical shift of 186 ppm of the protonated Schiff base indicates a strong counterion interaction, while additional charges or constraints on the retinal conformation are needed to explain the absorption maximum. In contrast to bR, residue D142 adjacent to the chromophore could cause an additional charge effect. The existence of a strong coupling network in the Schiff base vicinity is also supported by the high pH stability of the PSB chemical shift, which is The spectrum decays into spinning sideband families of the Schiff base, lysine side chains and amide backbone. Their isotropic signals are labelled with (*). Indices describe the PSB spinning sidebands. The spectrum was deconvoluted and fitted against a calculated sideband pattern using Topspin 2.0 (Bruker, Karlsruhe). For the protonated Schiff base (black), a chemical shift anisotropy of -132 ppm with g = 0.73 was found (r 11 = 54 ppm, r 22 = 203 ppm, r 33 = 300 ppm, fitting error 10%) weakened in the D97N mutant. The pK a of the Schiff base in the D97N mutant is decreased compared to the wild type. Considering the fact that the pK a of the proton acceptor in PR is 5 pH units higher than in bR, different interaction strength with the Schiff base is expected. Unfortunately, our data do not allow a direct conclusion about the retinal ring orientation in PR. This question can be answered by further solid-state NMR experiments, but it is interesting to note that in other prokaryotic rhodopsins such as bacteriorhodopsin (Luecke et al. 1999), halorhodopsin (Kolbe et al. 2000) and sensory rhodopsin II (Luecke et al. 2001), the all-trans retinal polyene chain is always coplanar with the beta-ionone ring, which is in the 6s-trans conformation. The interactions of Schiff base and retinal with opsin will be further studied by solid-state NMR using multiple amino acid labelling. Fig. 5 Correlation of UV/VIS with 15 N f-Lys 231 CP spectra of wt (a) and PR D97N (b) at different pH values. In wt PR, the 15 N signal of the PSB remains relatively constant over a pH range from 6 to 11 suggesting that the Schiff base remains mainly in its protonated form. The deprotonated form at 300 ppm could not be observed below pH 12. However, a peak at 366 nm starts occurring at pH 11 indicting deprotonation of the Schiff base. The acidic form and the basic form of the ground state appear at 545 and 516 nm respectively. For the D97N mutant, a decrease of the 15 N signal of the PSB occurs at high pH. This reveals the formation of an M-like state with a deprotonated SB which is also seen in the UV-VIS spectrum where an M-state signal at 400 nm occurs. Between pH 5 and 9, the mutant seems to be trapped in an N-like state as 15 N chemical shift and k max correspond to a protonated Schiff base. The pK a for D97N PR was determined to be 9.7