16622
J. Phys. Chem. B 2009, 113, 16622–16631
QM/MM Study of the Monomeric Red Fluorescent Protein DsRed.M1 Elsa Sanchez-Garcia, Markus Doerr, Ya-Wen Hsiao, and Walter Thiel* Max-Planck-Institut fu¨r Kohlenforschung, Mu¨lheim an der Ruhr ReceiVed: July 21, 2009; ReVised Manuscript ReceiVed: October 21, 2009
We report a combined quantum mechanical/molecular mechanical (QM/MM) study of the DsRed.M1 protein using as QM component the self-consistent charge density functional tight-binding (SCC-DFTB) method in molecular dynamics (MD) simulations and hybrid density functional theory (DFT, B3LYP functional) in QM/MM geometry optimizations. We consider different variants of the chromophore (including the cis- and trans-acylimine and peptide forms) as well as different protonation states of environmental residues. The QM/MM calculations provide insight into the role of nearby residues concerning their interactions with the chromophore and their influence on structural and spectroscopic properties. QM/MM optimizations yield a single conformer for the anionic acylimine chromophore, whereas there are distinct cis- and trans-conformers in the anionic peptide chromophore, the latter being more stable. The calculated vertical excitation energies (DFT/MRCI) for the anionic chromophores agree well with experiment. The published crystal structure of DsRed.M1 with an anionic acylimine chromophore indicates a quinoid structure, while the QM/MM calculations predict the phenolate form to be more stable. 1. Introduction DsRed is a red fluorescent protein1,2 from the Discosoma coral which is much used in molecular biology and biomedical research.3-6 Like green fluorescent protein (GFP),7 the DsRed monomers consist of 11-stranded β barrels,2 and the chromophore is formed autocatalytically from the Gln66-Tyr67Gly68 peptide in the presence of molecular oxygen. This reaction is commonly assumed to pass through a GFP-like green chromophore with a trans-peptide bond between Phe65 and Gln66, followed by a so-called maturation step that yields the DsRed chromophore with a cis-acylimine group between Phe65 and Gln66 (see Figure 1), which extends the conjugation of the system and thus causes a red shift compared to GFP.2,7-9 In the DsRed protein, the latter conversion is not complete so that a mixture of green and red chromophores is formed.2,8,10 DsRed has shortcomings as a biological marker because it suffers from slow chromophore maturation, oligomerization into a stable tetramer, and association to even higher aggregates.2 Therefore, much effort has been spent on engineering DsRed to overcome these shortcomings and also to enhance its photostability and pH stability.11,12 This has led to the development of fast maturing tetrameric variants such as DsRed.T410 and of a variety of monomeric red fluorescent proteins with the desired properties, including mRFP1,13 mFruits,14,15 TagRFP,16 and mKate.17 In this Article, we focus on the fast maturing monomer DsRed.M1 which has recently been characterized biochemically, spectroscopically, and by high-resolution crystallography (together with DsRed.T4).10 The spectroscopic properties of DsRed.M1 are very similar to those of DsRed.T4, but its fluorescence intensity is significantly lower due to its lower quantum yield.10 The chromophore environment is known to play an important role in the maturation chemistry of the DsRed proteins, and it strongly influences their photophysical properties such as fluorescence wavelength and brightness.2,10,18 One of the key environmental residues is glutamic acid 215 (Glu215) which is the DsRed equivalent of the Glu222 residue in the green
Figure 1. Simplified scheme representing the formation and maturation of the chromophore in the DsRed protein. Shown are the neutral transpeptide chromophore and the anionic cis-acylimine chromophore in its quinoid and phenolate forms.
fluorescent protein from Aequorea Victoria (avGFP).2,7 In the WT DsRed and DsRed.T4 proteins, Glu215 is deprotonated (anionic) and interacts with Lys70 via a salt bridge so that it cannot mediate a reversible proton transfer as in avGFP.2 In DsRed.M1, Glu215 rearranges and no longer interacts with Lys70, and the position of its OE1 oxygen atom in the crystal structure suggests an NCRQ · · · HOE1Glu215 hydrogen bond between the available nitrogen atom of the imidazolinone ring and protonated (neutral) Glu215.10 The second-generation monomeric red fluorescent proteins mFruits confirm that Glu215 can adopt different protonation states: in mCherry and mStrawberry, Glu215 is protonated at pH < 10, while, in mOrange, it is more likely deprotonated.14 There have been many theoretical studies of GFP and RFP chromophores.9,19-38 The necessity of treating the chromophore within the native protein environment has been emphasized
10.1021/jp9069042 2009 American Chemical Society Published on Web 12/08/2009
QM/MM Study of DsRed.M1 repeatedly.20,21,39,40 Recently, Krylov et al.41 studied the cis-trans isomerization of the GFP chromophore in water and found that CASSCF calculations including solvent effects and perturbative corrections for dynamical correlation gave results close to experiment. Grubmu¨ller et al.42 investigated the reversible photoswitching mechanism of the fluoroprotein asFP595 using CASSCF calculations and QM/MM excited state molecular dynamics simulations with explicit surface hopping. They found that changes in the protonation state of the chromophore and of some proximal amino acids lead to different photochemical states that are essential for the photoswitching mechanism. In the present work, we apply combined quantum mechanical/ molecular mechanical (QM/MM) methods43,44 to study DsRed.M1 considering different protonation states of the chromophore and of environmental residues. DsRed proteins normally contain an anionic chromophore with a cis-acylimine group (see Figure 1); i.e., the phenolic oxygen is deprotonated and engaged in a salt bridge, e.g., in WT DsRed and DsRed.T4 with lysine163 (Lys163).2,9,45 In DsRed.M1, one of the most important internal substitutions is the replacement of Lys163 by histidine (His163), and therefore, we also investigate different protonation states of His163. Experimentally, the electron density of DsRed.M1 is best fit by an approximately equal mixture of the red (cisacylimine) and the green (trans-peptide) chromophore (see Figure 1). Therefore, we also consider the anionic and neutral forms of the trans-peptide chromophore, the latter with different orientations of the phenolic hydrogen atom. Structural and stability issues are addressed through QM/MM molecular dynamics (MD) and QM/MM geometry optimization techniques using as QM component the self-consistent charge density functional tight-binding (SCC-DFTB) method and density functional theory (DFT, B3LYP functional), respectively. Vertical excitation energies are determined by a DFT-based multireference configuration interaction treatment (DFT/MRCI). 2. Computational Details 2.1. QM/MM Molecular Dynamics. All MD calculations were performed using the program CHARMM in version 33b1.46,47 The initial set of coordinates (including crystal water) used for the MD simulations was taken from the X-ray structure available in the Protein Data Bank48 under the reference 2VAD.10 The orientations of the side chains of asparagine, glutamine, and histidine, which are not reliably determined from the X-ray data, were checked using the program Reduce49 and by visual inspection. Protonation states of titratable side chains were determined using the web interface of the PROPKA 2.0 software.50,51 For residues in the vicinity of the chromophore, the protonation states were assigned by visual inspection of the environment of these residues taking into account the information available from the crystal structure.10 Since His163 interacts with the phenolic oxygen of the chromophore, the calculations with the neutral chromophore employed a histidine residue protonated in the δ position (HSD). In the calculations with the anionic chromophore, His163 was protonated in the ε position (HSE) so it can form a hydrogen bond with the phenolic oxygen atom of the chromophore; the doubly protonated form of His163 (HSP) was not considered, since the crystal structure indicates the δ N atom of His163 to be engaged in a strong hydrogen bond with the N-H peptide hydrogen atom of Ala164 (N-N distance of 2.91 Å). The other histidine residues (all far from the chromophore) were protonated in the δ position, with the only exception of His22 which was doubly protonated. The protonation states of the remaining residues were determined with CHARMM using the information in the force field topology
J. Phys. Chem. B, Vol. 113, No. 52, 2009 16623 files and the HBUILD facility.52 Those for the residues in the vicinity of the chromophore are listed in Table S1 of the Supporting Information. In the QM/MM MD simulations, the protein was described by the CHARMM22 force field,53,54 and the water molecules were represented by the TIP3P model. SHAKE constraints55 were applied to all bonds involving hydrogen atoms. Due to the lack of suitable CHARMM force field parameters, the chromophore was treated using the semiempirical SCC-DFTB method56,57 as implemented in CHARMM.58 The QM region included the chromophore CRQ66, residue Phe65, atoms C and O of Gln61, and all atoms except C and O of Ser69. The QM region was connected to the MM region using generalized hybrid orbital (GHO) boundary atoms.44,59 Atoms CR of Gln64 and C of residue Ser69 served as boundary atoms. The latter is an sp2 carbon atom, while boundary atoms are generally assumed to be sp3 hybridized in the SCC-DFTB/CHARMM implementation. As a consequence, the average bond length to this boundary atom is 0.03 Å shorter than the experimental value, which is tolerable for our purposes. The bond involving the other boundary atom is only 0.01 Å too short. An important issue in a QM/MM approach is the treatment of long-range electrostatic interactions. While it is recommended to use no cutoff for the interaction between the QM and MM part of the system,43 it is usually necessary to truncate electrostatic interactions within the MM part to make the molecular dynamics feasible. To avoid an unbalanced treatment of the system, we used no cutoff for the QM-MM electrostatic interaction and applied the group-based extended electrostatics approach within the MM part.60 In this approach, the electrostatic interactions between particles closer than a cutoff distance (14 Å in our case) are treated by the conventional pairwise additive scheme, while the interactions at larger distance are approximated by a computationally cheaper multipole approach. After the initial setup, the system was solvated in a preequilibrated sphere of TIP3P water with radius 30 Å, located at the origin of the coordinate system which was chosen to be at the position of atom CA2 of the chromophore. In all MD calculations, a spherical quartic boundary potential acting on the O atoms of the water molecules was used in order to retain the shape of the water droplet and to avoid evaporation of outer water molecules. All protein atoms more than 20 Å away from the center of the coordinate system were kept frozen in all subsequent calculations. After initial placement of the water sphere, overlapping water molecules (distance between Ow and any heavy atom
