The role of water molecules in spectral tuning of proteins has been left largely unexplored. This topic is important because changing hydrogen bond patterns during the activation process may lead to spectral shifts which can be of diagnostic value for the underlying structures. Arguments put forward in this article are based on spectral shift calculations of the rhodopsin and bathorhodopsin chromophore due to wat2a and 2b in the presence and absence of the counterion and of the amino acids lining the rhodopsin binding pocket. They show, among others, that a single water molecule can shift the absorbance by up to 0.1 eV or 34 nm depending on the environment of the chromophore.
Water molecules in protein cavities continue to stimulate the curiosity of researchers due to their profound role in the evolution of key biological networks where they are often linked to highly conserved residues (1,2). In rhodopsin, the membrane protein which is responsible for dim-light vision in vertebrates, seven water molecules have been located in the most recent 2.6 Å (3) and 2.2 Å (4) structures, and all are probably instrumental for the light-induced activation process. Five of them mediate interactions between the bundle of seven transmembrane helices whereas two interact with the chromophore of rhodopsin, 11-cis-retinal protonated Schiff base, pSb11: wat2b binds to the primary counterion (E113) forming a complex which stabilizes the twisted C11=C12 bond (5) and the protonation state of the chromophore (6), whereas the other, wat2a, connects to polar residues of the extracellular loop (EII) and may be involved in the postulated counterion switch mechanism from E113 to E181 during the activation process (7). These water molecules attest to the emerging paradigm that “intraprotein water molecules can be as essential for biological functions as amino acids” (8).
For the study we used the quantum-mechanically optimized structure of the chromophore binding pockets based on chain B of the 2.2 Å rhodopsin (PDB:1U19) (4) and 2.6 Å (PDB:2G87) (9) bathorhodopsin crystal structures. As the topic of interest concerns the chromophore and its immediate environment we chose chain B over chain A due to its less distorted structure near the Schiff base region. The optimized structures are obtained by embedded quantum chemistry (QM/MM) using the SCC-DFTB (10) and CHARMM codes. No geometry optimization was performed on any substructure obtained from this model. The MM point charges for the rhodopsin and bathorhodopsin environment consisting of 28 amino acids and two water molecules have been calculated using the sophisticated natural population analysis method (11) with 6-31G** basis. Ground- and excited-state energies were calculated by the CASSCF/CASPT2 method as provided by the MOLCAS set of routines (12). Six-root state-averaged wavefunctions were expanded in an atomic natural orbital basis set (13). We have used the contraction of the present work, C,N,O[4s3p1d]/H[2s] before in the studies of retinal model chromophores (14–17). It yielded highly accurate excited-state energies, the average deviation being 0.05 eV. The active space is (12,12), i.e. all pseudo π-electrons and valence pseudo π-orbitals were considered. Second-order corrections to the CASSCF energies were obtained with CASPT2. All core orbitals were kept frozen during the calculations, and the level shift to avoid the effect of intruder state was set uniformly to 0.3 au. For the oscillator strengths, CASPT2-corrected state energies were combined with transition dipole moments calculated by the CAS state interaction method.
In each system, the effect of water molecules on the chromophore was studied in three different scenarios: the retinal chromophore in its protein-induced distorted conformation (pSb11dist and pSbtdist), the ion-paired chromophore-E113 complex (pSb11ip and pSbtip) and the protein-embedded chromophore (pSb11pe and pSbtpe). In the last case the water molecules treated as point charges while they are part of the quantum-mechanical supermolecule in the first two.
Results and discussion
The matrix shown as Table 1 gives the calculated energies of the ground and the two lowest excited states of the chromophore without water and in the presence of the three different water combinations. Turning to rhodopsin (Scheme 1A), the ground state (S0) energies first we note that the chromophore itself forms a more stable complex with wat2a (pSb11dist+wat2a) than with wat2b (pSb11dist+wat2b) by 3.20 kcal mol−1, a consequence of the former being significantly closer to the unsaturated chromophore chain (smallest distance is 3.701 Å from C14) than wat2b, which is 4.283 Å from N16. The stability is reversed in the chromophore–counterion complex (Scheme 1B) with pSb11ip+wat2b being more stable than pSb11ip+wat2a by 4.89 kcal mol−1 because of the strong hydrogen bond formed between wat2b and the charged counterion.
Table 1. CASPT2 calculated energies* of pSb11, pSbt chromophore complexes and (in bold) wavelengths of optical transitions in rhodopsin and bathorhodopsin.
*S0 energies in au, S1 and S2 energies in eV relative to the corresponding S0 state (in parentheses: wavelengths in nm). †Water molecules treated as point charges. ‡From Schreiber et al. (17).
In the strongly allowed S1 state (the oscillator strength of this state in any of the calculated complexes never drops below 0.8; for a complete listing of these and other spectroscopic data, see Supporting Information) there is a shift in the electron density toward the positive azomethine nitrogen (18). Negatively charged groups in the vicinity of N16 will destabilize this state, i.e. they will cause a strong shift in the absorbance toward higher energy (19). From pSb11dist to pSb11ip the destabilization due to the negative counterion amounts to 0.62 eV, corresponding to a shift from 643 to 486 nm. The effect of polar substituents should be smaller; also, it should depend on the orientation of their molecular dipoles. Perusal of Table 1 (bold entries) reveals that the shifts effected by one water molecule range from 0.1 eV or 34 nm (wat2a or wat2b in pSb11dist) to 0.02 eV or 4 nm (wat2b in pSb11pe). This is in good agreement with the findings of Kusnetzow et al. (20) where the water residue directly associated with the Schiff-base proton was found to induce a blueshift of 0.16 eV, an aspect also observed by Schreiber et al. (21) who calculated a shift greater than 0.11 eV using very similar methods but with a truncated five double bond chromophore model. Interestingly, the calculated shifts predicted for the water molecules associated with the Schiff-base proton in bacteriorhodopsin (bR) and sensory rhodopsin are 0.26 and 0.27 eV, respectively (22). Thus our findings are consistent with previous theoretical studies, but smaller in magnitude compared to the calculated spectral shift. Fortuitously, in the case of pSb11ip, pSbtip and pSb11pe, the effect of the two water molecules is strictly additive presumably due to the presence of a neutral environment envisaged in the presence of the counterion and also a consequence of the point-charge approximation.
In the case of bathorhodopsin, in the absence of counterion and in the presence of polar and/or nonpolar groups the absorption maximum and stability of the pSbtdist complexes are more or less similar to that of the pSb11dist counterpart. However, in the presence of the counterion for the strongly allowed S1 state from pSbtdist to pSbtip, the destabilization due to the negative counterion amounts to 0.47 eV, corresponding to a shift from 649 to 521 nm. As wat2b directly assists in maintaining the stability of the ion-pair salt-bridge network, the chromophore-wat2b (pSbtip+wat2b) complex emerges as a more stable isomer compared to that of its wat2a counterpart (pSbtip+wat2a) by 5.33 kcal mol−1 in bathorhodopsin. However, we would like to emphasize that resolving the mechanism of energy storage of bathorhodopsin is not an issue to be examined here as a single water molecule contributes anywhere in between 1 and 2.0 kcal mol−1 when added to pSb11ip, pSbtip models.
In this context, it may be interesting to correlate our results with that of the 42 nm spectral shift observed in the dehydration studies on bR (23) where when the content of water in the purple membrane is decreased, the absorption of bR shifts to the blue. This effect could be due to the absence of retinal–water interaction in the binding pocket which can further inhibit the formation of purple chromophore from bacterio-opsin and all-trans retinal (24). It is also known that three water molecules (wat401, wat402, wat406, and wat22, wat24, wat50) are present in the active centers of bR (25,26) and halorhodopsin (hR) (27,28). QM/MM calculations (29) and FTIR probes (30) have shown that the hydrogen bonding interaction of wat402 is stronger with Asp85 than with Asp212 in bR, while in the case of hR, the hydration of chloride ion was found to be stabilized by weak hydrogen bonds of water (31). In conclusion, our calculations on visual rhodopsin highlight the role of water molecules in color regulation and reiterate the need for the presence of water channels and the accessibility of the binding site to them.
Acknowledgements— This work was supported by grants from the Deutsche Forschungsgemeinschaft (FOR480). This work is dedicated to Prof. Dr. Volker Buss on the occasion of his 66th birthday.