dynamic structure of retinylidene ligand of rhodopsin...

doi:10.1016/j.jmb.2007.06.047 J. Mol. Biol. (2007) 372, 906–917

Dynamic Structure of Retinylidene Ligand of RhodopsinProbed by Molecular Simulations

Pick-Wei Lau1, Alan Grossfield2, Scott E. Feller3

Michael C. Pitman2 and Michael F. Brown1,4,5⁎

1Department of Biochemistryand Molecular Biophysics,University of Arizona,Tucson, AZ 85721, USA2IBM TJ Watson ResearchCenter, Yorktown Heights,NY 10598, USA3Department of Chemistry,Wabash College, Crawfordsville,IN 47933, USA4Department of Chemistry,University of Arizona,Tucson, AZ 85721, USA5Department of Physics,University of Arizona,Tucson, AZ 85721, USA

Present address: P.-W. Lau, DeparBiology, The Scripps Research Instit92037, USA.Abbreviations used: FTIR, Fourie

GPCR, G protein-coupled receptor;hydrogen-out-of-plane; MD, molecumetarhodopsin I; MII, metarhodopsData Bank; QM/MM, quantum mecmechanics; SDPC, 1-stearoyl-2-docoglycero-3-phosphocholine; SDPE, 1-docosahexaenoyl-sn-glycero-3-phospE-mail address of the correspondi

[email protected]

0022-2836/$ - see front matter © 2007 E

Rhodopsin is currently the only available atomic-resolution template forunderstanding biological functions of the G protein-coupled receptor(GPCR) family. The structural basis for the phenomenal dark state stabilityof 11-cis-retinal bound to rhodopsin and its ultrafast photoreaction are activetopics of research. In particular, the β-ionone ring of the retinylidene inverseagonist is crucial for the activation mechanism. We analyzed a total of 23independent, 100 ns all-atom molecular dynamics simulations of rhodopsinembedded in a lipid bilayer in the microcanonical (N,V,E) ensemble.Analysis of intramolecular fluctuations predicts hydrogen-out-of-plane(HOOP) wagging modes of retinal consistent with those found in Ramanvibrational spectroscopy. We show that sampling and ergodicity of theensemble of simulations are crucial for determining the distribution ofconformers of retinal bound to rhodopsin. The polyene chain is rigidlylocked into a single, twisted conformation, consistent with the function ofretinal as an inverse agonist in the dark state.Most surprisingly, theβ-iononering is mobile within its binding pocket; interactions are non-specific and thecavity is sufficiently large to enable structural heterogeneity. We find thatretinal occupies two distinct conformations in the dark state, contrary tomost previous assumptions. The β-ionone ring can rotate relative to thepolyene chain, thereby populating both positively and negatively twisted6-s-cis enantiomers. This result, while unexpected, strongly agrees withexperimental solid-state 2HNMR spectra. Correlation analysis identifies theresidues most critical to controlling mobility of retinal; we find that Trp265moves away from the ionone ring prior to any conformational transition.Our findings reinforce how molecular dynamics simulations can challengeconventional assumptions for interpreting experimental data, especiallywhere existing models neglect conformational fluctuations.

© 2007 Elsevier Ltd. All rights reserved.

Keywords: G protein-coupled receptor; molecular dynamics; retinal;rhodopsin; vision
*Corresponding author
tment of Molecularute, La Jolla, CA

r transform infrared;HOOP,lar dynamics; MI,in II; PDB, Proteinhanics/molecularsahexaenoyl-sn-stearoyl-2-hoethanolamine.ng author:

lsevier Ltd. All rights reserve

Introduction

G-protein coupled receptors (GPCRs) constitutethe largest superfamily of proteins in the humangenome, and they initiate many signal transductionpathways.1–3 This class of proteins is of enormouspharmaceutical interest,4,5 because over 50% ofrecently launched drugs are targeted to GPCRswith annual worldwide sales exceeding $30 billion.6

As with many membrane proteins it is typicallyquite difficult to express and characterize GPCRs.The exception is rhodopsin, the primary dim lightreceptor, which is the only GPCR whose atomic-level structure has been published.7–10 A great deal

d.

mailto:[email protected]

http://dx.doi.org/10.1016/j.jmb.2007.06.047

907Dynamic Structure of Retinal Ligand of Rhodopsin

of effort has been put into developing our under-standing of rhodopsin function,11–20 both for its ownsake and as a template to explain activation of otherGPCRs.3 Here we address the general problem ofprotein–ligand interactions in GPCRs in relation totheir agonist and inverse agonist or antagonist roles.Specifically, we investigated the basis for thephenomenal dark state stability and activation of11-cis retinal bound to rhodopsin at the single mole-cule level using an ensemble of molecular dynamics(MD) simulations. Our findings yield a deeperunderstanding of the structural basis for dark stateretinal bound to rhodopsin, with potential insightsinto GPCR activation in general.3

The light-absorbing moiety 11-cis retinal of rho-dopsin is bound by a protonated Schiff base linkageto the side-chain of Lys296. In its 11-cis conformation,retinal functions as a strong inverse agonist,21

virtually locking the protein in the inactive state.Remarkably, upon absorbing a photon it instantlyisomerizes (within 200 fs)17,22 to the all-trans state,and subsequently becomes an agonist.13,14 Theprotein relaxes through a series of spectroscopicallydefined intermediates23 until it reaches the metarho-dopsin I (MI) state, which exists in equilibrium withthe active metarhodopsin II (MII) state.13,14,16,18 Thecentral role of retinal in directing this relaxation andpromoting MII formation has been extensivelyinvestigated by a number of approaches. Studies ofrhodopsin regenerated with chemically modifiedretinals as well as rhodopsin mutants show thatspecific interactions between retinal and the bindingpocket of the receptor are implicated both in rho-dopsin pigment formation and receptor activa-tion.13,16,24 Most of the altered ligands are lesseffective agonists in the all-trans state, shifting theMI–MII equilibrium back toward MI.25–27 The C9-methyl group distal to the retinylidene Schiff base ofLys296 is known to play an essential role.25,27,28

Moreover, the β-ionone ring of retinal is cruciallyimportant for rhodopsin activation.26,29,30 While theflexibility of dark state retinal has not been addressedspecifically in the literature, some studies havepostulated that the β-ionone ring is the moredynamic region of the retinal molecule.31 Forexample, 2H NMR results have been publishedsuggesting both a 6-s-trans32 and 6-s-cis33 conforma-tion. Recent experiments, including a 2.2 Å crystalstructure of rhodopsin9 and solid state 13C NMRmeasurements,31 suggest the β-ionone ring is in anegatively twisted 6-s-cis conformation. Nonetheless,there is some ambiguity in interpreting these results,since in the crystal structure retinal has severe stericclashes, while the 13C NMR experiments cannotdifferentiate between negatively and positivelytwisted enantiomers.In addition to experimental studies, computational

approaches, including molecular mechanics andquantummechanical calculations, have been appliedto understand the behavior of rhodopsin.34–37

Although several calculations have focused on thestructure of the retinal,38 none considered the pos-sibility of multiple retinal conformers. In part, this is

because such transitions occur very rarely on thetime scale of typical molecular dynamics calcula-tions. Here we show that by utilizing an ensemble ofMD simulations, we are able to relate the temporaldynamics of retinal in dark state rhodopsin to solid-state 2H NMR and resonance Raman data.17,33 Byconducting multiple MD simulations, with indepen-dent starting membrane configurations, we moreeffectively sample the conformational space of theretinylidene ligand of rhodopsin. The results ofindividual simulations differ significantly, indicatingthey are not long enough to be compared directly toexperimental results on an individual basis. Yetwhen taken as an ensemble, the simulations producetheoretical 2H NMR spectra in excellent agreementwith the experiment.33 The data clearly show thatretinal undergoes fluctuations on timescales com-parable to the simulation trajectories. A model in-cluding multiple retinal conformations in the darkstate is most consistent with experimental measure-ments. The picture that emerges suggests specificinteractions with the polyene chain and the proto-nated Schiff base; butmore surprisingly, non-specificinteractions with the ionone ring allow the bindingpocket to accommodate conformational heterogene-ity that may optimize receptor activation.

Results

A total of 23 independently equilibrated MDsimulations of 100 ns duration were carried out fora single rhodopsin molecule embedded in a fullyhydrated bilayer of native-like phospholipids com-prising 1-stearoyl-2-docosahexaenoyl-sn-glycero-3-phosphoethanolamine (SDPE)/ 1-stearoyl-2-docosa-hexaenoyl-sn-glycero-3-phosphocholine (SDPC)/cholesterol (50:49:24). Simulations were conductedin the microcanonical (N,V,E) ensemble using theatomic-level structure of rhodopsin (PDB code1U19)9 shown in Figure 1(a). The 11-cis retinal chro-mophore of rhodopsin is bound through a proto-nated Schiff base linkage to Lys296, cf. Figure 1(b).Gas phase quantum chemical39 and molecularmechanics calculations indicate three stable rotamersfor the C6–C7 torsion angle which defines the orien-tation of the β-ionone ring relative to the polyenebackbone. These include a skewed 6-s-trans con-formation (χ≈160°), a negatively twisted 6-s-cis con-former (χ≈−60°), and a positively twisted 6-s-cisconformer (χ≈55°). Themost recent crystal structureof rhodopsin9 reports a C6–C7 torsion angle of −34°,quite far from any minima in the potential energysurface. However, interpreting this structure isproblematic at best because of steric clashes betweenthe C8 and C5-methyl (C18) hydrogen atoms. Thediscrepancy could in part be due to the fact thatcrystallography averages away thermal motion toproduce only a single structure. By contrast, MDsimulations directly reveal conformational hetero-geneity that would be hidden experimentally. Wewere able to interpret experimental results, includinghydrogen-out-of-plane (HOOP) wagging modes

Figure 1. Structure of dark-state rhodopsin and retinal in its binding cavity is illuminated by X-ray crystallography,resonance Raman, and solid-state 2H NMR spectroscopy. (a) Structural model of rhodopsin (PDB code 1U19)9 showingretinal (green) within its binding pocket. (b) Extended view of the retinal binding pocket showing the C6–C7 bond(yellow) and the key C5, C9, and C13-methyl groups. (c) HOOP modes for H11 and H1217 are identified at 910 cm−1 andare independent of both negative (cyan) and positive (green) 6-s-cis conformers. (d) Comparison of experimental solid-state 2H NMR spectra33 to theoretical spectra for C5, C9, and C13-methyl groups with membrane tilt angles of 0, 45, and90° and bond orientation distributions from MD simulations.

908 Dynamic Structure of Retinal Ligand of Rhodopsin

from Raman vibrational spectroscopy17 and solidstate 2H NMR spectra33 (cf. Figure 1(c)–(d)) in termsof multiple conformers of the retinal ligand bound torhodopsin. Comparison was made to detailed expe-rimental data in all cases.17,33,40,41

Hydrogen-out-of-plane wagging modes areidentified in simulations

Unlike NMR and X-ray methods, Raman spectro-scopy probes the fluctuations arising from molecu-lar vibrations. These fluctuations are extremelysensitive to conformational changes in retinal, andcan serve as an indicator of the different photo-intermediates.42 In dark-state rhodopsin, a peak at969 cm−1 is assigned to concerted HOOP waggingmotions of the C11 and C12 hydrogen atoms.17Upon photoactivation, the frequency of this modeshifts significantly. When we characterized theretinal fluctuations in our simulations using princi-

pal component analysis, we observed an analogousvibrational mode with a frequency of 910 cm−1 thatis depicted in Figure 1(c) (see Methods for details).Given the difficulties in reproducing vibrationalspectra using classical molecular mechanics forcefields, a 10% discrepancy between simulation andexperiment is impressive. We performed this calcu-lation separately for retinals in both the negativelyand positively twisted conformations, and foundthat they produced similar modes at the samefrequency, suggesting that this vibrational mode isindependent of the conformation of β-ionone ring.As a result, the Raman experiments do notdifferentiate between the two conformations.

Multiple β-ionone ring conformations areconsistent with 2H NMR data

In solid state NMR spectroscopy, molecular in-teractions are measured through angular and dis-

Figure 2. Multiple conformations of the β-ionone ringof the retinylidene inverse agonist are shown bymolecularsimulations. (a) Distribution of the C6–C7 torsion angle inall 23 simulations of 100 ns duration. The two mostprevalent conformations are shown, representing nega-tively and positively twisted 6-s-cis conformations of theβ-ionone ring. (b) Plot of C8-to-C18 internuclear distance as afunction of C6–C7 torsion angle showing the symmetrywith respect to 0°. (c) Three representative C6–C7 torsionangle distributions from the 23 individual 100 ns trajec-tories comprising the distribution in (a).


tance restraints, giving both structure and dynamics.Salgado et al.33 reported 2H NMR spectra of alignedrecombinant membranes containing rhodopsinregenerated with retinal deuterated at the C5, C9,or C13-methyl positions. By varying the membranetilt angle with respect to the magnetic field, theyobtained detailed information about the orientationθB for the individual retinylidene C–C2H3 bonds.Datawere analyzed by assuming a single orientationfor each methyl group. In the present work, weadopted a different strategy and used the simulationresults to predict the 2HNMR spectra.We did this byusing the probability distribution from the MDsimulations to compile a weighted-average spec-trum in terms of theoretical spectra for each bondorientation.43 The results are shown in Figure 1(d).The C5–C2H3 bond axis distribution is particularlyimportant, because changes in the C6–C7 torsionangle report on the β-ionone ring conformation. Asshown in Figure 1(d), the spectrum computed fromthe simulation matches the experiment extremelywell. Interestingly, the probability distribution con-tains significant contributions from distinct confor-mations whose 2H NMR spectra differ greatly fromexperiment (see below). The simulated spectra aremost similar to experiment when computed from theprobability distribution derived directly from thecombined MD simulations. For completeness, weperformed the same analysis for the C9 and C13-methyl groups, as also shown in Figure 1(d), wherethe simulations predict a single dominant conformerconsistent with the crystal structure. The fact that thespectra agree well in all cases lends support to theidea that the β-ionone ring has more than oneconformation in the dark-state rhodopsin bindingpocket.

Flexibility of retinal ligand is illuminated bylarge-scale simulations

Turning next to the issue of the β-ionone ringconformation, Figure 2(a) shows the distribution ofthe C6–C7 torsion angles compiled from all 23simulations. Two 6-s-cis conformers are populated,with the negatively twisted (χ≈−60°) conformerpredominating, but with a significant probability forthe positively twisted (χ≈+70°) conformation. Gasphase computations indicate equal probability for thetwo 6-s-cis conformers, suggesting that the asymme-try in the distribution for the β-ionone ring of retinalwhen bound to rhodopsin is most likely due tointeractionswith the surroundingprotein side-chains.Previous experiments using chemically modifiedligands have demonstrated that the binding cavitydifferentiates between retinal enantiomers.16,40 OurMD simulations imply that the cavity has a moderateselectivity for the negatively twisted enantiomer,yielding positive ellipticity in circular dichroism(CD) spectra consistent with experimental data.40

The conformation of the β-ionone ring has alsobeen investigated through measurements of the C8-to-C18 distance using solid-state rotational reso-nance 13C NMR spectroscopy.31 In Figure 2(b) the

probability distribution for the C8-to-C18 carbondistance is plotted as a function of the C6–C7 torsionangle obtained in the simulations. Figure 2(b) clearlydemonstrates that the distance is correlated with theβ-ionone ring orientation; it also shows that thedistance is governed solely by the magnitude of the6-s-cis skew and not the sign, due to the mirrorimage symmetry of the two enantiomers. As a directresult, the rotational resonance experiment cannotdistinguish between the positively and negativelytwisted conformers. Moreover, the C8-to-C18 dis-tance varies over a large range, from approximately3.0 Å to 4.5 Å, with the smaller distances corre-sponding to a conformation in the 6-s-cis region, andthe larger distances representing the 6-s-trans region.For a particular value of the C6–C7 torsion, there aresubstantial variations in the observed C8-to-C18internuclear distances, arising from fluctuations inbond angles and bond lengths. This is importantwhen using the distance from NMR as a restraint forthe conformation of retinal, because the distance fora single conformer will in general differ from thatcomputed by averaging over the conformationalfluctuations. Figure 2(c) shows representative C6–C7 torsion angle distributions from individual simu-lations, which are further discussed below.We now shift our focus to the conformation of the

polyene moiety of the retinylidene chromophorewithin the binding cavity of the photoreceptor,showing in Figure 3 the probability distributions forthe torsion angles along the entire conjugated chain.In contrast to the C6–C7 dihedral angle, the torsionsin the polyene chain all have narrow unimodal

Figure 3. The polyene chain of retinylidene ligand islocked in a twisted conformation in the inverse agonistform. (a)–(h) Torsion angle distributions along the conju-gated polyene chain from C7 to C15. Twisting of thepolyene chain from planarity (torsion angle χ=180°) ismainly confined to the C10–C11 and C12–C13 torsions toeither side of the C11=C12 double bond.


distributions, indicating that it exists in a single well-defined state. All the torsion angles (C7 to C15) arein the trans conformation, except the C11=C12 bondwhich is in the cis conformation. Any concerteddeviation from planarity is an indication of straininduced by the protein environment, and maystrongly alter the photochemistry of the polyenechain. For retinal bound to rhodopsin, the twotorsions adjacent to the cis-double bond are sig-nificantly perturbed from their planar state; theC10–C11 torsion angle is 164(±12)° and the C12–C13torsion angle is 161(±9)°, in accord with 2H NMRdata for rhodopsin in the dark state.33 The torsionaldeformation alleviates the steric clashes between theC13-methyl group and the H10 hydrogen.38 Resultsof the MD simulations are generally consistent withthree planes of unsaturation for retinal within thebinding cavity of rhodopsin,16,33,41 a model pre-viously used to interpret 2H NMR results.33

Sampling and ergodicity are establishedthrough multiple trajectories

Although the polyene chain of the retinylidenechromophore occupies a single well-defined con-formation, the ensemble of MD simulations reveals

that the β-ionone ring populates two distinctenantiomers (cf. Figure 2(a)). Previous MD andquantum mechanics/molecular mechanics (QM/MM) simulations did not report multiple retinalconformations,9,34,35,44 most likely due to limita-tions in time scale. Based on the number of transi-tions in our simulations, the lifetimes of the twoconformers are≈50 ns, comparable to the longest ofthe other trajectories. Because transitions are rareevents, the probability distributions for independenttrajectories differ significantly, indicating that the100 ns timescale is not long enough to allowstatistical convergence. This point is best seen inFigure 2(c), which shows probability distributionsfrom three of the 23 trajectories. These differenceshighlight the risks involved in interpreting a singletrajectory, no matter how long, when the systemmay contain long-lived states. For any given trajec-tory, the C6–C7 torsion angle undergoes only a smallnumber of transitions. Indeed, in several simulationsthe torsion remains in its initial conformationthroughout the trajectory, with no transitions at all.In others, there is a single transition leading to apositively twisted conformation, and still othersshow a number of transitions back and forth.As noted above, trajectories where the retinal

primarily samples the positive-twisted conforma-tion produce 2H NMR spectra that bear littleresemblance to the experimental ones; interpretedby themselves they would lead us to conclude thatthe simulations and experiments were fundamen-tally inconsistent. On the other hand, combining thepositive 6-s-cis conformers with the negativelytwisted state, using the ensemble of simulations toprovide the weighting, improves the fit to experi-mental data33 versus the negatively twisted statealone. It is only by interpreting the ensemble averageof the simulations in accord with the ergodichypothesis that we are able to conclude that thereare multiple important conformations within theactual distribution. To illustrate this aspect further,Figure 4 shows 2H NMR spectra calculated byrestricting the angular range of the probabilitydistribution from all 23 simulations. Part (a) con-siders only values of the C6–C7 torsion angle b0°;whereas part (b) only includes angles N0°, with theentire distribution indicated in (c). As can be seen,considering only a part of the distribution leads tovery poor agreement of the synthetic spectra withthe experimental 2H NMR data; yet when the entiredistribution is included the agreement is more grati-fying (cf. Figure 1).

Contact residues of β-ionone ring depend on itsconformation

To assess the change in protein environment due toisomerization about theC6–C7bond,wemapped theinteractions of the three retinal methyl carbons (C5,C9, and C13) with the neighboring protein residues.Specifically, we identified the amino acid residuesfound within a 4 Å sphere about each methyl carbonin each of the 23 simulations. For each simulation, we

Figure 4. Retinal ligand of rhodopsin has conforma-tions of β-ionone ring whose 2H NMR spectra differ fromexperiment. (a) Predicted 2H NMR spectra consideringonly distribution of C6–C7 torsion angles b0° (cyan) and(b) torsion angles N0° (purple). Synthetic spectra in eachcase are superimposed on experimental 2H NMR data33

for membrane tilt angles of 0, 45, and 90°. (c) Probabilitydistribution for all 23 simulations (cf. Figure 2) showingrange of C6–C7 torsional angles used to calculate thesynthetic spectra.


constructed a vector containing the probability that aselected region of retinal formed a contact with eachresidue, and combined them to form a contactprobabilitymatrixC. The average contact probabilityfor each residue was computed over all 23 simula-tions and subtracted from each column of the Cmatrix, producing a fluctuation matrix. The covar-iancematrixV=CTCwas thenconstructed,whereCT

is the transposed contactmatrix. Covariancematricesfor each of the methyl groups are represented inFigure 5(a)–(c), and allow one to compare thetrajectories and identify patterns in ligand–proteininteraction. Figure 5(a) shows V for the C5-methylgroup, themostmobile of the three due to its locationon the β-ionone ring. This map can be divided intothree different blocks, giving three characteristicdistributions of the C6–C7 torsion: (i) simulations 1–11,whichmostly populate the negatively twisted 6-s-cis, (ii) simulations 12–15, where both positively andnegatively twisted states are populated, and (iii)simulations 16–23, where the positively twisted statepredominates. The simulations in each group arestrongly correlated with each other, and anti-corre-lated with those in the other groups, indicating thatthe torsional state directly modulates the residueswhich interact with the ligand. In contrast, the

covariancematrix for the C9-methyl and C13-methylgroups in Figures 5(b) and 5(c) are featureless,meaning that fluctuations of these bonds do notalter interactions of retinal with its environment.To further enumerate the important residues in

contact with the C5-methyl group, we present itscontact matrix in Figure 5(d). Note that Thr118,Gly121, and Glu122 appear frequently in simula-tions in the first category, while residues Phe208,Tyr268, and Ala269 are often in close proximity tothe C5-methyl group for simulations in the thirdcategory. Residues in contact with the C9 and C13-methyl groups are shown in Figures 5(e) and 5(f).Streaks of the same color are observed across all 23simulations, indicating that the presence or absenceof a particular residue is consistent from onesimulation to another. Most striking is the closeproximity of Ile189 to the C9-methyl group acrossall simulations. The closest residues to the C13-methyl group are Ala117, Trp265, Tyr268, Ala292,and Ala295. It follows that the polyene chainsegments occupy well-defined binding sites,which may account for the phenomenal darkstate stability of retinal acting as an inverse agonist.By contrast, the binding site of the β-ionone ring isnot as well defined, presumably due to relativelynon-selective interactions consistent with a pocketof moderate to low specificity.16

Trp265 inhibits 6-s-trans retinal formation withinthe binding pocket

The residue Trp265 is highly conserved across theGPCR family A, and is thought to be involved inligand binding. Interactions between the Trp265side-chain and the β–ionone ring in rhodopsin areimplicated in the inverse agonist role of retinal.20,24

Consistent with this interpretation, Figure 6(a)indicates that Trp265 is usually packed againstretinal, specifically the ionone ring. With this inmind, we attempted to relate motions of Trp265 toconformational transitions of the retinal C6–C7torsion angle. Examination of the probability dis-tribution for the C6–C7 torsion, Figure 2(a), showsthat the primary path between the two twisted 6-s-cisstates is via the 6-s-trans state. Hence the trans statecan be used as an indicator of retinal conformationaltransitions. By examining the time correlation func-tion between the presence of the trans state and thedistance between the β-ionone ring and Trp265(Figure 6(b)), we can interpret the functionalsignificance of these interactions. Specifically, wecomputed C(t)= ⟨[d(t+τ)– ⟨d⟩] [ϑ(t)– ⟨ϑ⟩]⟩/σdσϑwhere d(t) is the instantaneous distance, and ϑ(t) isan indicator function that is unity when the C6–C7torsion is greater than 120° or less than −120°, andzero otherwise. Here ⟨d⟩ and ⟨ϑ⟩ denote averagingover all times in all simulations, and σd and σϑ arecorresponding standard deviations.We computed this correlation function indepen-

dently for each of the 19 trajectories in which wefound the trans state. Figure 6(b) shows the averageof these correlation functions (the error bars are the

Figure 5. Covariance analysis of individual 100 nssimulations shows variability of protein environment ofretinylidene C5, C9, and C13-methyl groups. Elements ofthe covariance matrix are depicted by colors and rangefrom −1 (completely anti-correlated; blue) to 1 (completelycorrelated; red). (a)–(c) C5, C9, and C13-methyl positions,respectively. Note that the matrix in (a) has three differentblocks: (i) simulations 1– 11; (ii) simulations 12– 15; and(iii) simulations 16–23. Each of the blocks has a distinct C5-methyl group environment. By contrast, in (b) and (c) thecolor map for the C9 and C13-methyl groups has aconsistent green color, indicating all 23 simulations yieldsimilar methyl group environments. (d)–(f) Contactmatrices for the C5, C9, and C13-methyl groups, respec-tively. Columns list the simulation number and rowscorrespond to residues within a 4 Å radius sphere centeredon the methyl carbon. The color scale ranges from blue(low fraction value; indicating residue is infrequent within4 Å radius sphere) to red (high fraction value within 4 Åsphere). For each block in (a), distinct residues appearconsistently around the C5-methyl group showing sam-pling of different environments. For the C9 and C13-methyl groups in (b) and (c), bands of similar colors acrossall 23 simulations evince similar environments.


standard error, computed as r=ffiffiffiffiffiffi

N;p

where σ is thestandard deviation and N is the number of simula-tions). The correlation function has a statisticallysignificant positive peak at τ=−1 ns (the resolutionof the time series used to construct the correlations),indicating that on average the formation of the transstate is preceded by an increase in the distancebetween Trp265 and the β-ionone ring. It appearsthat conformational transitions between the twotwisted 6-s-cis conformers are inhibited by thepresence of Trp265 in its normal configuration. Tofurther test this idea, we ran a short simulation inwhich we mutated Trp265 to glycine in silico. In thistrajectory, we saw a significant population of 6-s-trans retinal, in contrast to the b1% population in theprimary simulations. This may have functionalimplications, as previous work24 suggests thatrhodopsin activation moves the β-ionone groupaway from Trp265 toward Ala169.

Discussion

The present work shows that all-atom MDsimulations provide an important framework for

relating the temporal fluctuations of retinal to itsbiological activity. Upon photon absorption byrhodopsin, the retinal cofactor isomerizes to thetrans form of the polyene, which drives non-equilibrium relaxation of the protein from the darkstate toward the inactive MI state. In MI, trans-retinal acts as an agonist, shifting the MI-MIIequilibrium toward the active MII state, whichbinds the G protein transducin and initiates visualperception.13,14,16 Besides the polyene chain ofretinal, the β-ionone ring in particular is crucial tothe reaction mechanism20,26,29,30,45–47 The β-iononering is connected with protein movements that yieldreceptor activation,14,20,48–50 following release of thespring-loaded conformational switch of the polyene.The torsion about the C6–C7 bond defines theorientation of the β-ionone ring relative to thepolyene chain. Free retinal has been shown withboth experimental51 and computational39 methodsto exist as multiple isomers represented by 6-s-cisand 6-s-trans states, which is consistent with CDdata.40,41,52 By contrast, the retinal conformationbecomes more restricted upon binding to rhodopsin,where it functions as an inverse agonist. It isinteresting that in both bacteriorhodopsin51,53 andsensory rhodopsin,54 retinal is 6-s-trans; while indark-adapted rhodopsin it is widely thought toadopt the 6-s-cis conformation.33,40,55

Activation mechanism of the GPCR prototyperhodopsin

An understanding of why nature has selecteddistinct conformations of retinal for different receptorsmay provide additional insights into the activationmechanism of rhodopsin.33,56 As a step in thisdirection, we have used MD simulations to exploreprotein–ligand interactions and flexibility of retinalwithin the binding pocket of dark state rhodopsin. Oursimulations are complementary to bioorganic16,40,41

and biophysical17,32,33 approaches, and aid the inter-pretation of experimental data. Binding of retinal toopsin leads to significant rotary strength in CDspectroscopy due to twisting of the chromophore.57A sign reversal is evident in bathorhodopsin, followedby a diminution in the subsequent intermediates.52

Investigations of rhodopsin with chemically modifiedligands further indicate how specific interactions ofretinal within its binding cavity are connected with itsdark state stability, aswell as activation of the receptor.Removal of the C9-methyl group appears to increaseligand flexibility in the binding pocket. The absence ofHOOP modes in Fourier transform infrared (FTIR)spectra for MI indicates the C10-to-C13 region hasrelaxed to a more planar conformation, which onlyoccurs inMII for natural retinal.25 Aback shiftingof theMI–MII equilibrium is evident leading to reduced Gprotein activation.25 Opening the retinal ionone ring inacyclic analogs similarly reduces the stability of MIIrelative to MI,26,29,45 although in this case the MI stateappears to differ structurally from that formed withunmodified retinal.26 Such acyclic modifications alsoeliminate the MI HOOP modes, thus implicating the

Figure 6. Conformation of the β-ionone ring iscorrelated with its distance to Trp265 side-chain. (a)Distance between center of the β-ionone ring and theindole 6-carbon ring of Trp265 side-chain used in thecross-correlation calculation is shown by the white line. (b)Average cross-correlation profile of the 19 simulationscontaining the 6-s-trans state shows a non-zero correlationvalue above the noise level at τ=−1 ns. Thus a correlationbetween the C6–C7 torsion angle and the distancebetween the β-ionone ring and Trp265 is revealed.


ring structure in transmitting force from the all-transretinal ligand to the protein.The current simulations show that the polyene

chain of retinal is rigidly locked into a single, twistedconformation when bound to rhodopsin, consistentwith its role as an inverse agonist in the dark state.However, interactions of the β-ionone ring within itsbinding pocket are non-specific, and the cavity islarge enough to allow conformational heterogeneity.It follows that activation of rhodopsin can beconsidered in terms of three (sub)sites for bindingof the ligand to the receptor.16,25,33,38 These corre-spond to (i) the protonated Schiff base and itsassociated counterion,14,58–60 (ii) the mid-portion ofthe polyene chain,25,28 and lastly (iii) the β-iononering within its hydrophobic binding pocket.31,33 Thethree sites are specifically probed by 2H NMR of theC13, C9, and C5-methyl groups, respectively.32,33 Inthe dark state, the retinal polyene chain is underconsiderable strain, as indicated by the HOOPmodes measured by stimulated Raman17 or FTIR26

spectroscopy, and by 2H NMR of the C9 and C13-

methyl groups.33 The HOOP modes shift theirfrequencies in the MI state of native rhodopsin,but disappear in MII. Likewise, solid-state 2H NMRdemonstrates a reduction of the twist of retinal inMI.61 This finding suggests that the polyene strainis used to drive the large-scale protein rearrange-ments that occur in the MII state.14 It is logical toconsider the ring as necessary to anchor retinal tothe protein, guaranteeing that retinal relaxationinduces protein reorganization and gives activephotoreceptor. Our work predicts that, contrary toprevious assumptions, retinal populates more thanone stable conformation under physiological con-ditions. The stron-gest evidence for the existence ofmultiple conformers is the excellent agreementbetween the experimental 2H NMR spectra andthose computed directly from the distribution ofbond orientations observed in the MD simulations.There are additional indications of such conforma-tional heterogeneity from rotational resonance 13CNMR31 and FTIR62 spectroscopy.

Conformational flexibility of retinal in the inverseagonist state

Multiple retinal conformations31 could be impor-tant for understanding conflicting reports in theliterature that address isomerization of the β-iononering.32–34,38,40,55,63 Up until now, most investigatorshave assumed that the β-ionone ring of retinaladopts a single conformer. Biophysical studiesusing 2H NMR,32,33 13C NMR,64 and CD40 spectro-scopy have been interpreted in terms of only oneisomer of the β-ionone ring. By contrast, MDsimulations suggest that retinal may populatedifferent conformers while bound to rhodopsin.Our results imply that the net positive ellipticity ofthe dark state of rhodopsin40,41,52 originates from anenantiomeric excess of the negative isomer. Thepresence of symmetric conformers with both nega-tive and positive C6–C7 torsional angles is notdetectable by 13C NMR studies.31,64 As a result,previous interpretations of structural data indicatinga single negatively twisted 6-s-cis conformation,while natural in light of the X-ray crystal structure,may be incomplete. Although the C9 and C13-methyl groups undergo only small librations, andconsistently interact with the same parts of theprotein, the flexibility of the C6–C7 torsion enablesthe C5-methyl group to interact with a more diverseset of residues. Interestingly, these internal motionsdo not significantly alter the overall shape of thebinding pocket. Figure 7 shows the two primaryretinal conformations found in this work superposedin the context of the protein cavity. Only minor localdisplacements of individual amino acid residues arenecessary to accommodate the retinal conforma-tional transitions. More importantly, specific inter-actions selecting only one of the enantiomers areapparently absent, despite the highly tuned functionof rhodopsin.Our findings highlight the importance of per-

forming multiple MD simulations to effectively


explore conformational space with fewer assump-tions and to assess convergence. In particular, thecorrelation between the motions of Trp265 and theretinal could not be assessed from a singletrajectory; it was only by examining an ensemblewith a significant number of independent transi-tions that the pattern became apparent. The dangerof analyzing a single trajectory is that the systemmay be undersampled due to the limited duration,the large number of accessible states, and therelatively long lifetimes of individual conforma-tions. The situation is exactly analogous to an issuethat arises when analyzing single molecule experi-ments. Any single measurement may appearinconsistent with a bulk measurement, becausean individual molecule may be found in states farfrom the mean of the population.65,66 Conse-quently, the experiments must be repeated manytimes in order to obtain dependable statisticalinformation, which directly parallels the computa-tional approach taken here. We further note thereare important long-timescale fluctuations of retinalso that the simulations may not necessarily beergodic. (The ergodic hypothesis states that thetime average for any one member of an ensembleis equivalent to the ensemble average at any oneinstant.) These statistical fluctuations are mani-fested as significant variations in the distributionof retinal conformers. Sampling from a number ofindependently equilibrated systems allows us tomore closely estimate the actual molecular proper-ties, while accurately assessing the degree ofstatistical convergence. However, we cannot dis-count the possibility that even the ensemble ofsimulations taken together does not represent afully converged probability distribution for theconformation of retinal. In each simulation, theretinal was initially placed in the negativelytwisted form, as suggest by the crystal structure.While this is clearly the most tenable choice, theresulting simulations may as a result overestimatethe population of the negatively twisted form.Unfortunately, proving ergodicity from a finite setof data is extremely difficult, especially when thelifetime of some states is comparable to the lengthof the trajectories.To conclude we propose the following mechan-

istic hypothesis: in the dark state, retinal acts as aninverse agonist by forming specific interactionsthrough the polyene chain and protonated Schiffbase. However, the ionone ring interactions are non-specific, and the binding pocket is large enough toallow interconversion of positively and negativelytwisted 6-s-cis conformers. This stands in contrast toprevious analyses of crystallographic and NMRdata, which implicitly assume a single representa-tive conformation for retinal, yielding (perhaps forthis reason) a structure with significant internalsteric clashes. Our suggestion that retinal insteadpopulates different twisted 6-s-cis enantiomersenables us to resolve this difficulty. The resultingprobability distribution correctly reproduces experi-mental 2H NMR spectra, as well as resonance

Raman data. In its 11-cis form, retinal is a stronginverse agonist and locks the dark state into aninactive conformation that accounts for its phenom-enal stability. Since modification of the ionone ringgenerally diminishes activity, it may seem counter-intuitive that interactions between the ionone ringand the receptor are non-specific. Yet the excep-tional agreement of the experimental spectra calcu-lated from the simulation-derived conformationaldistributions supports multiple retinal conformersas our best model of dark state rhodopsin. Finally,the present results demonstrate that computersimulations can offer deep insights into the inter-pretation of experimental data, thus facilitatingmechanistic models that correspondingly motivatefuture experiments.

Methods

Molecular dynamics simulations

System construction in the microcanonical (N,V,E)ensemble is described by Grossfield et al. and itsSupplementary Material.67 Briefly, each of the systemscontained one rhodopsin molecule, 50 SDPE molecules,49 SDPC molecules, 24 cholesterol molecules, 14sodium ions, 16 chloride ions, and 7400 watermolecules (43 222 atoms total). Individual systemswere embedded in a 56.5 Å×79.2 Å×95.5434 Å peri-odic box, with the membrane normal aligned along thez-axis. The CHARMM version 27 force field was usedto represent the protein68 with parameters appropriatefor retinal,69 while the recently refined CHARMMsaturated chain,70 polyunsaturated chain,71 andcholesterol72 parameters were used to describe thelipids. Construction and equilibration were performedusing CHARMM27,68,73 while production calculationswere performed using Blue Matter,74 a moleculardynamics package specially written to take advantageof the Blue Gene/L hardware.74 The simulationsdescribed here were generally run on 512, 1024, or2048 Blue Gene nodes, generating 4, 6, or 9 ns/day,respectively. However, due to improvements in thecode, we are now able to generate roughly 40 ns/dayusing 4 Blue Gene racks (4096 nodes). It is worthnoting that this performance significantly exceeds thatof the most rigorous implicit membrane models,75,76

albeit with a dramatically larger computational invest-ment. Each of the simulations was constructed inde-pendently; we used the 1U19 PDB file for the proteininitial coordinates, and regenerated new lipid confor-mations from a library of structures. As a result, thedistribution of lipid species and cholesterol around theprotein was fully randomized at the beginning of eachtrajectory. The initial locations of sodium and chlorideions in the aqueous region were also determinedrandomly at the beginning of each simulation. Inde-pendently regenerating the membrane starting struc-ture is crucial to effective sampling, given themicrosecond timescale for lateral reorganization of thelipid bilayer. For further details see Grossfield et al.67

The vibrational modes of the retinal were computedusing the quasi-harmonic analysis facility withinCHARMM. We computed the time-averaged structurefrom the trajectory, and then obtained the atomic fluctua-tions in rectangular Cartesian coordinates from the

Figure 7. Dynamical structure of the retinal inverseagonist in its binding pocket is obtained for dark-staterhodopsin. Both positive (red) and negative (blue) 6-s-cisconformers are placed into the binding pocket ofrhodopsin. The three planes used to describe the structureof retinal are designated by A, B, and C (cf. the text).


difference between the instantaneous structures and theaverage. The cross-correlation matrix of the fluctuationswas computed and diagonalized to obtain the normalmodes of the system. The normalmode associatedwith theHOOP motion was found by taking the scalar product ofeach normal mode vector with a test vector constructed bydisplacing H11 and H12 out of the plane of the cis-doublebond. In each case, this procedure identified a singlenormalmode,whose overlapwith the displacement vectorwas an order of magnitude greater than any other normalmode. Details of the computation of normal modes byquasi-harmonic analysis in CHARMM are describedelsewhere.77

Calculation of 2H NMR spectral lineshapes

Theoretical 2H NMR lineshapes for specifically deut-erated retinal bound to rhodopsin in aligned membranebilayers were calculated as described.33,43 Spectra weresimulated for bond angles on a 1° grid; the final spec-trum was then computed as the average of the individualspectra, weighted by their probability as computed fromthe combined molecular dynamics simulations. Themosaic spread of the membrane stack was treated as afree parameter. Further details are provided by Salgadoet al.33

Acknowledgements

We thank W. Hubbell, K. Martínez-Mayorga, O.Miyashita, K. Nakanishi, A. Struts, F. Tama, and R.Vogel for valuable discussions. Members of the BlueMatter team including B. Fitch, R. Germain, A.Rayshubskiy, T. J. C. Ward, M. Eleftheriou, F. Suits,Y. Zhestkov, R. Zhou, J. Pitera, and W. Swopeprovided valuable support. S. E. F. acknowledgesthe NSF for funding through grant MCB-0091508and is a Henry Dreyfus Teacher–Scholar. M. F. B. isgrateful to the NIH for support from grant EY12049and the NSF from grant CHE-607917.

References

1. Bockaert, J. & Pin, J. P. (1999). Molecular tinkering of Gprotein-coupled receptors: an evolutionary success.EMBO J. 18, 1723–1729.

2. Pierce, K. L., Premont, R. T. & Lefkowitz, R. J. (2002).Seven-transmembrane receptors. Nature Rev. Mol. CellBiol. 3, 639–650.

3. Fanelli, F. & De Benedetti, P. G. (2005). Computa-tional modeling approaches to structure-functionanalysis of G protein-coupled receptors. Chem. Rev.105, 3297–3351.

4. George, S. R., O'Dowd, B. F. & Lee, S. R. (2002). G-protein-coupled receptor oligomerization and itspotential for drug discovery. Nature Rev. Drug Disc.1, 808–820.

5. Watts, A. (2005). Solid-state NMR in drug design anddiscovery for membrane-embedded targets. NatureRev. Drug Disc. 4, 555–568.

6. Drews, J. (2000). Drug discovery: a historical perspec-tive. Science, 287, 1960–1964.

7. Palczewski, K., Kumasaka, T., Hori, T., Behnke, C. A.,Motoshima, H., Fox, B. A. et al. (2000). Crystalstructure of rhodopsin: a G protein-coupled receptor.Science, 289, 739–745.

8. Okada, T., Fujiyoshi, Y., Silow, M., Navarro, J.,Landau, E. M. & Shichida, Y. (2002). Functional roleof internal water molecules in rhodopsin revealed byX-ray crystallography. Proc. Natl Acad. Sci. USA, 99,5982–5987.

9. Okada, T., Sugihara, M., Bondar, A.-N., Elstner, M.,Entel, P. & Buss, V. (2004). The retinal conformationand its environment in rhodopsin in light of a new2.2 Å crystal structure. J. Mol. Biol. 342, 571–583.

10. Li, J., Edwards, P. C., Burghammer, M., Villa, C. &Schertler, G. F. X. (2004). Structure of bovinerhodopsin in a trigonal crystal form. J. Mol. Biol.343, 1409–1438.

11. Brown, M. F. (1994). Modulation of rhodopsin func-tion by properties of the membrane bilayer. Chem.Phys. Lipids, 73, 159–180.

12. Szundi, I., Mah, T. L., Lewis, J. W., Jäger, S., Ernst,O. P., Hofmann, K. P. et al. (1998). Proton transferreactions linked to rhodopsin activation. Biochemistry,37, 14237–14244.

13. Sakmar, T. P., Menon, S. T., Marin, E. P. & Awad,E. S. (2002). Rhodopsin: insights from recentstructural studies. Annu. Rev. Biophys. Biomol. Struct.31, 443–484.

14. Hubbell, W. L., Altenbach, C., Hubbell, C. M. &Khorana, H. G. (2003). Rhodopsin structure, dyna-mics, and activation: a perspective from crystallogra-phy, site-directed spin labeling, sulfhydryl reactivity,and disulfide cross-linking. Adv. Protein Chem. 63,243–290.

15. Botelho, A. V., Gibson, N. J., Wang, Y., Thurmond,R. L. & Brown,M. F. (2002). Conformational energeticsof rhodopsin modulated by nonlamellar forminglipids. Biochemistry, 41, 6354–6368.

16. Fishkin, N., Berova, N. & Nakanishi, K. (2004).Primary events in dim light vision: a chemical andspectroscopic approach toward understanding pro-tein/chromophore interactions in rhodopsin. Chem.Rec. 4, 120–135.

17. Kukura, P., McCamant, D. W., Yoon, S., Wandsch-neider, D. B. & Mathies, R. A. (2005). Structuralobservation of the primary isomerization in visionwith femtosecond-stimulated Raman. Science, 310,1006–1009.


18. Palczewski, K. (2006). G protein-coupled receptorrhodopsin. Annu. Rev. Biochem. 75, 743–767.

19. Botelho, A. V., Huber, T., Sakmar, T. P. & Brown, M. F.(2006). Curvature and hydrophobic forces driveoligomerization and modulate activity of rhodopsinin membranes. Biophys. J. 91, 4464–4477.

20. Crocker, E., Eilers, M., Ahuja, S., Hornak, V.,Hirshfeld, A., Sheves, M. et al. (2006). Location ofTrp265 in metarhodopsin II: implications for theactivation mechanism of the visual receptor rhodop-sin. J. Mol. Biol. 357, 163–172.

21. Han, M., Lou, J. H., Nakanishi, K., Sakmar, T. P. &Smith, S. O. (1997). Partial agonist activity of 11-cis-retinal in rhodopsin mutants. J. Biol. Chem. 272,23081–23085.

22. Wang, Q., Schoenlein, R. W., Peteanu, L. A., Mathies,R. A. & Shank, C. V. (1994). Vibrationally coherentphotochemistry in the femtosecond primary event ofvision. Science, 266, 422–424.

23. Jäger, S., Lewis, J. W., Zvyaga, T. A., Szundi, I.,Sakmar, T. P. & Kliger, D. S. (1997). Chromophorestructural changes in rhodopsin from nanoseconds tomicroseconds following pigment photolysis. Proc.Natl Acad. Sci. USA, 94, 8557–8562.

24. Borhan, B., Souto, M. L., Imai, H., Shichida, Y. &Nakanishi, K. (2000). Movement of retinal along thevisual transduction path. Science, 288, 2209–2212.

25. Vogel, R., Fan, G.-B., Sheves, M. & Siebert, F. (2000).The molecular origin of the inhibition of transducinactivation in rhodopsin lacking the 9-methyl group ofthe retinal chromophore: a UV-Vis and FTIR spectro-scopic study. Biochemistry, 39, 8895–8908.

26. Vogel, R., Siebert, F., Lüdeke, S., Hirshfeld, A. &Sheves, M. (2005). Agonists and partial agonists ofrhodopsin: retinals with ring modifications. Biochem-istry, 44, 11684–11699.

27. Vogel, R., Lüdeke, S., Siebert, F., Sakmar, T. P.,Hirshfeld, A. & Sheves, M. (2006). Agonists andpartial agonists of rhodopsin: retinal polyene methy-lation affects receptor activation. Biochemistry, 45,1640–1652.

28. Han, M., Groesbeek, M., Sakmar, T. P. & Smith, S. O.(1997). The C9 methyl group of retinal interacts withglycine-121 in rhodopsin. Proc. Natl Acad. Sci. USA, 94,13442–13447.

29. Bartl, F. J., Fritze, O., Ritter, E., Herrmann, R., Kuksa,V., Palczewski, K. et al. (2005). Partial agonism in a Gprotein-coupled receptor. Role of the retinal ringstructure in rhodopsin activation. J. Biol. Chem. 280,34259–34267.

30. Domínguez, M., Álvarez, R., Pérez, M., Palczewski, K.& de Lera, A. R. (2006). The role of the 11-cis-retinalring methyl substituents in visual pigment formation.ChemBioChem. 7, 1815–1825.

31. Spooner, P. J. R., Sharples, J. M., Verhoeven, M. A.,Lugtenberg, J., Glaubitz, C. & Watts, A. (2002).Relative orientation between the β-ionone ring andthe polyene chain for the chromophore of rhodop-sin in native membranes. Biochemistry, 41,7549–7555.

32. Gröbner, G., Burnett, I. J., Glaubitz, C., Choi, G.,Mason, A. J. & Watts, A. (2000). Observations of light-induced structural changes of retinal within rhodop-sin. Nature, 405, 810–813.

33. Salgado, G. F. J., Struts, A. V., Tanaka, K., Fujioka, N.,Nakanishi, K. & Brown, M. F. (2004). Deuterium NMRstructure of retinal in the ground state of rhodopsin.Biochemistry, 43, 12819–12828.

34. Gascon, J. A. & Batista, V. S. (2004). QM/MM study of

energy storage and molecular rearrangments due tothe primary event in vision. Biophys. J. 87, 2931–2941.

35. Röhrig, U. F., Guidoni, L., Laio, A., Frank, I. &Rothlisberger, U. (2004). A molecular spring forvision. J. Am. Chem. Soc. 126, 15328–15329.

36. Huber, T., Botelho, A. V., Beyer, K. & Brown, M. F.(2004). Membrane model for the GPCR prototyperhodopsin: hydrophobic interface and dynamicalstructure. Biophys. J. 86, 2078–2100.

37. Lau, P.-W., Pitman, M. C., Feller, S. E. & Brown, M. F.(2005). Conformational sensitivity of retinylidenecofactor in rhodopsin studied by molecular dynamics.Biophys. J. 90, 442a.

38. Sugihara, M., Hufen, J. & Buss, V. (2006). Origin andconsequences of steric strain in the rhodopsin bindingpocket. Biochemistry, 45, 801–810.

39. Terstegen, F. & Buss, V. (1996). Ab initio study of theC6–C7 conformation of retinal model systems. Chem.Letters, 25, 449–450.

40. Fujimoto, Y., Ishihara, J., Maki, S., Fujioka, N., Wang,T., Furuta, T. et al. (2001). On the bioactive conforma-tion of the rhodopsin chromophore: absolute sense oftwist around the 6-s-cis bond. Chem. Eur. J. 7,4198–4204.

41. Fujimoto, Y., Fishkin, N., Pescitelli, G., Decatur, J.,Berova, N. & Nakanishi, K. (2002). Solution andbiologically relevant conformations of enantiomeric11-cis-locked cyclopropyl retinals. J. Am. Chem. Soc.124, 7294–7302.

42. Pan, D., Ganim, Z., Kim, J. E., Verhoeven, M. A.,Lugtenburg, J. & Mathies, R. A. (2002). Time-resolvedresonance Raman analysis of chromophore structuralchanges in the formation and decay of rhodopsin's BSIintermediate. J. Am. Chem. Soc. 124, 4857–4864.

43. Nevzorov, A. A., Moltke, S., Heyn, M. P. & Brown,M. F. (1999). Solid-state NMR line shapes of uniaxiallyoriented immobile systems. J. Am. Chem. Soc. 121,7636–7643.

44. Crozier, P. S., Stevens, M. J., Forrest, L. R. & Woolf,T. B. (2003). Molecular dynamics simulation of dark-adapted rhodopsin in an explicit membrane bilayer:coupling between local retinal and larger scaleconformational change. J. Mol. Biol. 333, 493–514.

45. Jäger, F., Jäger, S., Krautle, O., Friedman, N., Sheves,M., Hofmann, K. P. et al. (1994). Interactions of theβ-ionone ring with the protein in the visual pigmentrhodopsin control the activation mechanism. An FTIRand fluorescence study on artificial vertebrate rho-dopsins. Biochemistry, 33, 7389–7397.

46. Nakanishi, K. & Crouch, R. (1995). Application ofartificial pigments to structure determination andstudy of photoinduced transformations of retinalproteins. Isr. J. Chem. 35, 253–272.

47. Spooner, P. J. R., Sharples, J. M., Goodall, S. C., Bovee-Geurts, P. H. M., Verhoeven, M. A., Lugtenburg, J. et al.(2004). The ring of the rhodopsin chromophore in ahydrophobic activation switch within the bindingpocket. J. Mol. Biol. 343, 719–730.

48. Farrens, D. L., Altenbach, C., Yang, K., Hubbell, W. L.& Khorana, H. G. (1996). Requirement of rigid-bodymotion of transmembrane helices for light activationof rhodopsin. Science, 274, 768–770.

49. Patel, A. B., Crocker, E., Eilers, M., Hirshfeld, A.,Sheves, M. & Smith, S. O. (2004). Coupling of retinalisomerization to the activation of rhodopsin. Proc. NatlAcad. Sci. USA, 101, 10048–10053.

50. Kim, J.-M., Altenbach, C., Kono, M., Oprian, D. D.,Hubbell, W. L. & Khorana, H. G. (2004). Structuralorigins of constitutive activation in rhodopsin: role of


the K296/E113 salt bridge. Proc. Natl Acad. Sci. USA,101, 12508–12513.

51. Harbison, G. S., Smith, S. O., Pardoen, J. A., Courtin,J. M. L., Lugtenburg, J., Herzfeld, J. et al. (1985). Solid-state 13C NMR detection of a perturbed 6-s-transchromophore in bacteriorhodopsin. Biochemistry, 24,6955–6962.

52. Horiuchi, S., Tokunaga, F. & Yoshizawa, T. (1980).Circular dichroism of cattle rhodopsin and bathorho-dopsin at liquid nitrogen temperatures. Biochim.Biophys. Acta, 591, 445–457.

53. Luecke, H. & Lanyi, J. K. (2003). Structural clues to themechanism of ion pumping in bacteriorhodopsin.Adv. Protein Chem. 63, 111–130.

54. Vogeley, L., Sineshchekov, O. A., Trivedi, V. D., Sasaki,J., Spudich, J. L. & Luecke, H. (2004).Anabaena sensoryrhodopsin: a photochromic color sensor at 2.0 Å.Science, 306, 1390–1393.

55. Smith, S. O., Palings, I., Copié, V., Raleigh, D. P.,Courtin, J., Pardoen, J. A. et al. (1987). Low-temperature solid-state 13C NMR studies of theretinal chromophore in rhodopsin. Biochemistry, 26,1606–1611.

56. Moltke, S., Nevzorov, A. A., Sakai, N., Wallat, I., Job,C., Nakanishi, K. et al. (1998). Chromophore orienta-tion in bacteriorhodopsin determined from the angu-lar dependence of deuteriumNMR spectra of orientedpurple membranes. Biochemistry, 37, 11821–11835.

57. Yoshizawa, T. & Shichida, Y. (1982). Low-temperaturecircular dichroism of intermediates of rhodopsin.Methods Enzymol. 81, 634–641.

58. Yan,E.C.Y.,Kazmi,M.A.,Ganim,Z.,Hou, J.-M., Pan,D.,Chang, B. S. W. et al. (2003). Retinal counterion switch inthe photoactivation of the G protein-coupled receptorrhodopsin. Proc. Natl Acad. Sci. USA, 100, 9262–9267.

59. Lüdeke, S., Beck, R., Yan, E. C. Y., Sakmar, T. P., Siebert, F.& Vogel, R. (2005). The role of Glu181 in the photoactiva-tion of rhodopsin. J. Mol. Biol. 353, 345–356.

60. Martínez-Mayorga, K., Pitman, M. C., Grossfield, A.,Feller, S. E. & Brown, M. F. (2006). Retinal counterionswitch mechanism in vision evaluated by molecularsimulations. J. Am. Chem. Soc. 128, 16502–16503.

61. Salgado, G. F. J., Struts, A. V., Tanaka, K., Krane, S.,Nakanishi, K. & Brown, M. F. (2006). Solid-state 2HNMR structure of retinal in metarhodopsin I. J. Am.Chem. Soc. 128, 11067–11071.

62. Fahmy, K., Jäger, F., Beck, M., Zvyaga, T. A., Sakmar,T. P. & Siebert, F. (1993). Protonation states ofmembrane-embedded carboxylic acid groups in rho-dopsin and metarhodopsin II: a Fourier-transforminfrared spectroscopy study of site-directed mutants.Proc. Natl Acad. Sci. USA, 90, 10206–10210.

63. Singh, D., Hudson, B. S., Middleton, C. & Birge, R. R.(2001). Conformation and orientation of the retinylchromophore in rhodopsin: a critical evaluation ofrecent NMR data on the basis of theoretical calcula-tions results in a minimum energy structure consistentwith all experimental data.Biochemistry, 40, 4201–4204.

64. Creemers, A. F. L., Kiihne, S., Bovee-Geurts, P. H. M.,DeGrip, W. J., Lugtenburg, J. & de Groot, H. J. M.(2002). 1H and 13C MAS NMR evidence for pro-nounced ligand-protein interactions involving theionone ring of the retinylidene chromophore inrhodopsin. Proc. Natl Acad. Sci. USA, 99, 9101–9106.

65. Seol, Y., Skinner, G. M. & Visscher, K. (2004). Elasticproperties of a single-stranded charged homopoly-meric ribonucleotide. Phys. Rev. Letters, 93, 118102-1–118102-4.

66. Keller, D., Swigon, D. & Bustamante, C. (2003).Relating single-molecule measurements to thermo-dynamics. Biophys. J. 84, 733–738.

67. Grossfield, A., Feller, S. E. & Pitman, M. C. (2006). Arole for direct interactions in the modulation ofrhodopsin by ω-3 polyunsaturated lipids. Proc. NatlAcad. Sci. USA, 103, 4888–4893.

68. MacKerell, A. D., Jr, Bashford, D., Bellott, M.,Dunbrack, R. L., Evanseck, J. D., Field, M. J. et al.(1998). All-atom empirical potential for molecularmodeling and dynamics studies of proteins. J. Phys.Chem. B, 102, 3586–3616.

69. Saam, J., Tajkhorshid, E., Hayashi, S. & Schulten, K.(2002). Molecular dynamics investigation of primaryphotoinduced events in the activation of rhodopsin.Biophys. J. 83, 3097–3112.

70. Klauda, J. B., Brooks, B. R., MacKerell, A. D., Venable,R. M. & Pastor, R. W. (2005). An ab initio study on thetorsional surface of alkanes and its effect on molecularsimulations of alkanes and a DPPC bilayer. J. Phys.Chem. B, 109, 5300–5311.

71. Pastor, R. W., Venable, R. M. & Feller, S. E. (2002).Lipid bilayers, NMR relaxation, and computer simu-lations. Acc. Chem. Res. 35, 438–446.

72. Pitman, M. C., Suits, F., MacKerell, A. D. & Feller,S. E. (2004). Molecular-level organization of saturatedand polyunsaturated fatty acids in a phosphatidyl-choline bilayer containing cholesterol. Biochemistry,43, 15318–15328.

73. Brooks, B. R., Bruccoleri, R. E., Olafson, B. D.,Swaminathan, S. & Karplus, M. (1983). CHARMM: aprogram for macromolecular energy, mininization,and dynamics calculations. J. Comp. Chem. 4,187–217.

74. Fitch, B. G., Germain, R. S., Mendell, M., Pitera, J.,Pitman, M., Rayshubskiy, A. et al. (2003). Blue Matter,an application framework for molecular simulation onBlue Gene. J. Para. Distrib. Comp. 63, 759–773.

75. Tanizaki, S. & Feig, M. (2005). A generalized Bornformalism for heterogeneous dielectric environments:application to the implicit modeling of biologicalmembranes. J. Chem. Phys. 122, 124706-1–124706-13.

76. Tanizaki, S. & Feig, M. (2006). Molecular dynamicssimulations of large integral membrane proteins with animplicit membranemodel. J. Phys. Chem. B, 110, 548–556.

77. Karplus, M. & Kushick, J. N. (1981). Method forestimating the configurational entropy of macromo-lecules. Macromolecules, 14, 325–332.

Edited by B. Honig

(Received 13 March 2007; received in revised form 13 June 2007; accepted 18 June 2007)Available online 26 June 2007

dynamic structure of retinylidene ligand of rhodopsin...

Documents