Photophysics of the LOV-Based Fluorescent Protein Variant iLOV

Article pubs.acs.org/JPCB

Photophysics of the LOV-Based Fluorescent Protein Variant iLOVQ489K Determined by Simulation and Experiment Mehdi D. Davari,*,† Benita Kopka,‡ Marcus Wingen,‡ Marco Bocola,† Thomas Drepper,‡ Karl-Erich Jaeger,‡,§ Ulrich Schwaneberg,†,∥ and Ulrich Krauss*,‡ †

Lehrstuhl für Biotechnologie, RWTH Aachen University, 52056 Aachen, Germany Institut für Molekulare Enzymtechnologie, Heinrich Heine University Düsseldorf, Forschungszentrum Jülich, 52426 Jülich, Germany § Institut für Bio- und Geowissenschaften, IBG-1, Biotechnologie, Forschungszentrum Jülich, 52426 Jülich, Germany ∥ DWI-Leibniz Institute for Interactive Materials, Forckenbeckstraße 50, 52056, Aachen, Germany ‡

S Supporting Information *

ABSTRACT: Light, oxygen, voltage (LOV) based fluorescent proteins (FPs) represent a promising alternative to fluorescent reporters of the green fluorescent protein family. For certain applications like multicolor imaging or the design of FRET-based biosensors, the generation of spectrally shifted LOV-based FPs would be required. In a recent theoretical study (Khrenova et al. J. Phys. Chem. B 2015, 119 (16), pp 5176−5183), the photophysical properties of a variant of the LOV-based fluorescent protein iLOV were predicted using quantum mechanics/ molecular mechanics (QM/MM) approaches. The variant contained a lysine residue at the position of a highly conserved glutamine residue (Q489K), which directly interacts with the O4 and N5 atom of the flavin mononucleotide (FMN) chromophore. On the basis of QM/MM calculations, iLOV-Q489K was suggested to possess substantially red-shifted absorption and fluorescenceemission maxima with respect to parental iLOV. Here, we describe the experimental characterization of this variant, which, surprisingly contrary to the theoretical prediction, shows blue-shifted absorption and fluorescence-emission maxima. Using molecular dynamics (MD) simulations and QM/MM calculations, the molecular basis for the contradictory theoretical and experimental results is presented. Essentially, our computational analysis suggests that, in the Q489K variant, two possible sidechain conformers exist: (i) a least populated conformer K489in forming a hydrogen bond with the O4 atom of FMN chromophore and (ii) a most populated conformer K489out with the side-chain amino group flipped away from the FMN chromophore forming a new hydrogen bond with the backbone oxygen of G487. QM/MM calculated spectra of the K489out conformer are blue-shifted compared to the calculated spectra of parental iLOV, which is in accordance with experimental data. This suggests that the change in the conformation of K489 from K498in to K489out accounts for the change in the direction of the spectral shift from red to blue, thus reconciling theory and experiment. All LOV-based FPs owe their fluorescence properties to the flavin mononucleotide (FMN) molecule, which represents the light-absorbing chromophore of all LOV proteins.10 To generate a LOV-based FP, usually, the strictly conserved cysteine residue of the photosensory LOV domain is exchanged with an alanine.1,2,11 The corresponding mutation abolishes the LOV photocycle, which in the native LOV domain involves the transient formation of a covalent FMN−cysteinyl−thiol adduct between the Sγ atom of the mentioned cysteine and the C4a atom of the FMN chromophore.10 In the corresponding alanine variant, no covalent FMN−protein adduct can be formed, which results in continuous cyan−green flavin fluorescence upon blue-light (450 nm) excitation. The LOV-based FP iLOV is based on the C426A mutant of the Arabidopsis thaliana phototropin (phot) 2 LOV2 domain.2 Although the protein is

1. INTRODUCTION Recently, fluorescent reporter proteins have been engineered from bacterial and plant light, oxygen, voltage (LOV) photoreceptors and were designated as FbFP (flavin mononucleotide (FMN)-binding fluorescent proteins1), iLOV2 or miniSOG,3 respectively. Because of their unique properties, they represent a promising new class of cofactor-dependent fluorescent proteins (FPs).1,2,4 In particular, their small size and oxygen independent fluorescence render them superior to FPs of the Green Fluorescent Protein (GFP) family for several applications.1,2,4−6 To this end, FbFPs have been used to analyze anaerobic pathogens like Bacteroides fragilis7 and Porphyromonas gingivalis8 during the infection process. The tolerance of LOV-based FPs toward oxygen limitation also led to the development of the first, genetically encoded FRETbased biosensor for oxygen that consists of an oxygen independent FbFP donor domain and an oxygen sensitive EYFP acceptor domain.9 © 2016 American Chemical Society

Received: February 13, 2016 Published: March 10, 2016 3344

DOI: 10.1021/acs.jpcb.6b01512 J. Phys. Chem. B 2016, 120, 3344−3352

Article

The Journal of Physical Chemistry B

BL21(DE3) as previously described.12 All cultures were grown in a volume of 1 L autoinduction TB medium supplemented with 50 μg mL−1 kanamycin in 5 L shake flasks at 37 °C for 24 h. Proteins were purified using Ni-NTA superflow-columns (Qiagen, Hilden, Germany) under standard operating conditions, as described by the manufacturer. Purified FPs were stored in protein storage buffer (10 mM NaCl, 10 mM NaH2PO4, pH 8.0) at 4 °C. 2-3. Spectroscopic Characterization. All spectroscopic measurements were carried out as described by Wingen and coworkers.12 In brief, all samples were adjusted to a maximum absorption of 0.1 at 450 nm and measured at 20 ± 2 °C, if not stated otherwise. Absorption spectra were recorded using a Cary 60 UV−vis spectrophotometer (Agilent Technologies, Santa Clara, USA). Fluorescence spectra and fluorescence quantum yields were determined using a Quanta-Master 40 fluorescence spectrophotometer (Photon Technology International, Birmingham, USA). The extinction coefficients were determined by comparison of the absorption of the flavin in the bound and unbound state (i.e., before and after thermal denaturation). Photobleaching kinetics were measured using a Cary Eclipse fluorescence spectrophotometer (Agilent Technologies, Santa Clara, CA) with a high-power blue-light LED (LUXEON Rebel LXML PR01 0425 royal blue, Philips Lumileds, San Jose, CA; center wavelength 448 nm; optical power 180 mW cm−2; operating current 350 mA; angle of radiation = 12°) that was placed on top of the cuvette. Data points were collected at an interval of 1 s, and the first data point with a fluorescence intensity below 50% of the initial value was taken as the bleaching halftime. 2-4. FoldX Prediction. A structural model of the iLOVQ489K variant was constructed in YASARA Structure version 13.9.814 using the YASARA-FoldX plugin15 and employing the FoldX method.16 The starting coordinates for the FoldX16 in silico mutagenesis experiment were taken from the X-ray structure of the parent iLOV protein (PDB-code: 4EES17). Since FMN is not recognized by FoldX, we constructed the FMN molecule using fragments of recognized molecules (amino acids and DNA base). Testing our fragment-based method gives rise to reliable results for self-mutations. A FoldX mutation run including rotamer search, exploring alternative conformations (10 independent runs) of Q489K and the surrounding side chains below 6 Å distance from residue Q489 was performed during the FoldX energy minimization employing a probability-based rotamer library. Stabilization energy calculation and interaction energies were computed with FoldX version 3.0 Beta16 using standard settings. Details of the FoldX calculations can be found in the Supporting Information. 2-5. Classical Molecular Dynamics (MD) Simulations. The initial coordinates were taken from the X-ray structure of parent iLOV (PDB-code 4EES, resolution: 1.8 Å17). 4EES contains one iLOV molecule per asymmetric unit and thus represents a monomeric structure. In the work by Khrenova et al.,13 another crystal structure of iLOV, i.e. one subunit from 4EET structure,17 which contains two molecules per asymmetric unit, was used. The all-atom RMSD for the residues in the binding pocket between 4EES and 4EET is 0.23 Å. Moreover, iLOV was shown to be a monomer in solution.17 We therefore decided to use the monomeric 4EES iLOV structure as starting coordinates for our simulations. Because of the small RMSD of the two structures a comparison between the 4EESand 4EET-derived calculated spectra should be possible. The protonation states of titratable residues were assigned on the

only 109 amino acids long, all studies of iLOV use the amino acid numbers of the full-length phot2 to indicate positions and mutations (phot2 amino acids 387−496). To further increase the fluorescence of the LOV2 domain harboring the C426A mutation, two rounds of DNA shuffling with three other LOV coding sequences of Arabidopsis phot1 and phot2 were carried out. The resulting variant, which possesses multiple amino acid exchanges compared to the initial Arabidopsis phot2 LOV2C426A protein, was termed iLOV due to its improved fluorescence properties.2 For certain applications, e.g. multicolor imaging or the design of FRET-based biosensors, the generation of spectrally shifted LOV-based FPs would be required. So far, only variants with a blue-shifted absorption and emission spectrum have been reported.12 In those variants a strictly conserved glutamine in the LOV domain (Q489 in iLOV) is exchanged by an apolar amino acid.12 Recently, based on QM/MM calculations, Khrenova and co-workers13 suggested that the introduction of a charged lysine residue at the respective position (iLOVQ489K) would result in a significant 52 and 97 nm red-shift of the absorption and fluorescence-emission bands of the FMN chromophore in the iLOV protein, respectively. To test this prediction, we generated the Q489K variant of iLOV and assessed the photophysical properties of the variant compared to the parent iLOV protein. Contrary to the previously predicted red-shift,13 iLOV-Q489K possesses a 10 nm blue-shifted excitation and 8 nm blue-shifted fluorescence emission maximum. In order to address this discrepancy between theory and experiment, we performed molecular dynamics simulations of the iLOV-Q489K variant, which revealed two probable side-chain conformers of K489 (least populated K489in and most populated K489out). QM/MM calculations for both conformers suggest that the most populated conformer (K489out), in which the lysine side chain does not form a hydrogen bond to the FMN molecule, is spectrally blue-shifted, thus essentially reconciling theory and experiment.

2. EXPERIMENTAL AND COMPUTATIONAL METHODS 2-1. Bacterial Strains and Plasmids. All bacterial strains used in this study were grown either in Luria−Bertani (LB) broth or in autoinduction (AI) terrific broth (TB) media as previously described.12 The genes coding for the parent iLOV protein (DNA sequence was derived from Addgene plasmid pGEX-iLOV, Addgene plasmid no. 26587) as well as for the iLOV-Q489K variant were obtained as synthetic genes from Life Technologies (Thermo Fisher Scientific Inc., Waltham, MA, USA) and were flanked with a 5′- NdeI and a 3′- SalI restriction endonuclease recognition site for subcloning. To generate the Q489K variant, a lysine codon (nucleotide sequence: AAA) was included instead of the Q489 codon (nucleotide sequence: CAG) in the respective synthetic gene. No codon-usage optimization was carried out during gene synthesis. Using the restriction endonucleases NdeI, and SalI, the respective iLOV genes were subcloned into the pET28a expression vector (Novagen/Merck, Darmstadt, Germany) to facilitate heterologous expression in E. coli and subsequent purification using an N-terminal His6-tag fused to the target protein for affinity chromatography (see below). All final expression vectors were verified by sequencing (SeqLab GmbH, Göttingen, Germany). 2-2. Protein Production and Purification. Expression of iLOV and the Q489K variant was carried out in E. coli 3345


Article

The Journal of Physical Chemistry B basis of pKa values obtained from the PROPKA 3.1 program18 and visual inspection. His471 and His495 were treated as HIE (NE2 protonated) and HIP (both ND1 and NE2 protonated), respectively. Side chains of Asn and Gln residues were checked for possible flipping. The Amber ff99SB force-field parameters19,20 for the protein and general Amber force field (GAFF)21 for flavin mononucleotide (FMN) were used. Geometry optimization of FMN (depicted in Figure S5 in Supporting Information) was performed applying density functional theory (DFT) by using the B3LYP functional22−24 and 6-31G(d) basis set.25,26 In order to assign partial charges to the FMN molecule, electrostatic potential (ESP) charges were calculated using the Hartree−Fock (HF) method and 631G(d) basis set.25,26 Amber ff99 compatible RESP charges for FMN (shown in Table S2 in Supporting Information) were calculated using the antechamber module27 of AmberTools14.28 Electronic structure calculations were carried out by using Gaussian09.29 The phosphate group of flavin was deprotonated, carrying a charge of −2, which leads to a total charge of −4e/-3e for parental iLOV and the iLOV-Q489K variant, respectively. To neutralize this charge, solvent water molecules that were at least 5.5 Å away from any protein atoms were replaced by Na+ ions. Hydrogen atoms were added employing the tleap module of AmberTools14.28 Crystal water molecules were kept. The protein was solvated in an octahedral TIP3P30 water box centered at the center of mass to ensure a water layer of 12 Å around the protein. The systems contained ∼27 000 atoms in total, including ∼8500 TIP3P30 water molecules. Initially, the solvent and the ions followed by the whole system were subjected to minimization using 10 000 steps of steepest descent followed by 3000 steps of conjugate-gradient minimization. The system was then slowly heated from 0 to 300 K for 50 ps. In all MD simulations, constant pressure periodic boundary conditions using the particle mesh Ewald (PME)31 method were employed. To calculate the electrostatic interactions a cutoff of 10 Å was used. After heating, the systems were equilibrated for 1000 ps at 300 K. Finally, three independent production runs (for 50 ns) were performed for both parental iLOV and iLOV-Q489K resulting in a total of 150 ns simulation time. All classical molecular dynamics (MD) simulations were performed with the Amber14 program.28 Pymol,32 VMD,33 and AmberTools 1428 were used for molecular visualizations and analysis of MD simulations. A geometric cutoff of 3.2 Å and 150° for hydrogen-bond distance and angle was used, respectively. The GROMACS 4.534 g_cluster tool and linkage method was used to perform cluster analysis of the structure of parent iLOV and iLOV-Q489K along MD trajectories based on the side-chain conformation of Q489 or K489 and the residues in FMN binding pocket with 1.35 Å RMSD as cutoff. The nearest structures (cluster representative) to the center of the most populated clusters from the production runs were chosen as starting structures for further QM/MM calculations. The Supporting Information provides further details on the MD analysis procedure. 2-6. QM/MM Calculations. ChemShell v3.535 was employed as QM/MM interface. The Turbomole 6.336 and DL-POLY37 force-field engine integrated in ChemShell were used to calculate the energies and gradients of the QM and MM parts, respectively. The same set of Amber force-field parameters (ff99SB and GAFF) implemented for the classical MD simulations was used for the MM part of the QM/MM calculations. We have applied the additive QM/MM scheme

with electrostatic embedding. The charge-shift scheme38 and hydrogen link atoms were employed to handle the QM-MM boundary region. During the QM/MM structure optimizations which were carried out using the DL-FIND39 optimizer module of ChemShell, we relaxed all residues (amino acids, FMN, water, counterions) that lie within 10 Å of the FMN chromophore. Everything beyond this selection is kept frozen during QM/MM geometry optimizations. The QM region consists of the lumiflavin moiety of the FMN chromophore while the covalent bond (between C13 and C14 for FMN; see Figure S5 in the Supporting Information) across the QM-MM boundary was capped with hydrogen link atoms. First, geometries of the ground state (S0) and first singlet excited state (S1) were optimized at QM/MM DFT/amber level of theory. To describe the excited state structures and excitation energies the time-dependent (TD) DFT method within QM/ MM approach was used.40,41 At the TD-DFT level, we applied B3LYP 22−24 functional and SVP basis set.42 Previous calculations of electronic states of the FMN chromophore justifies choosing this combination of method and basis sets as long as geometry and the order of electronic states is the main concern.43−53 2-7. Calculation of Vibronic Spectra. Vibronic absorption and fluorescence-emission spectra of FMN (modeled by lumiflavin, LF) were computed at the DFT level of theory with B3LYP22−24 functional and 6-31G(d) basis set.25,26 At the DFT level, we used the adiabatic Hessian (AH)54 to compute the vibronic spectra. First, geometries of the ground state (S0) and first singlet excited state (S1) of LF were optimized at B3LYP level of theory. To describe the excited state in the DFT calculations, we used the time-dependent formulation (TDDFT), for which analytical gradients are available. The optimizations at the B3LYP level were carried out with Cs symmetry, using analytical gradients. In (TD) DFT calculations, the Hessians of the ground and excited states were obtained by numerical differentiation. In addition, also the first derivatives of the transition dipole, required for Herzberg− Teller (HT) simulations, were obtained numerically. HT terms were only included at the DFT level. The spectra were computed within the Franck−Condon approximation (FC) at 0 K. All vibronic spectra were computed with the FCclasses code.54,55 For simulating spectra at 0 K, the time-independent (TI) approach was used. A maximum number of 25 overtones for each mode and 20 combination bands on each pair of modes were included in the calculations. The maximum number of integrals to be computed for each class was set to 109. The convergence in the total intensity is above 99%. Spectra are reported as normalized absorption and fluorescence-emission line shapes, where line shape is absorptivity divided respectively by radiation frequency or the frequency cube. We used Gaussian function with a full width at halfmaximum (fwhm) of 0.03 eV to model line broadening in our spectra. All DFT computations were performed with Gaussian09 package.29 Vibronic stick spectra can be found in Figure S11 in the Supporting Information.

3. RESULTS AND DISCUSSION 3-1. iLOV-Q489K Shows Blue-Shifted Absorption and Fluorescence Emission Band Maxima. In order to assess the recent prediction presented by Khrenova and co-workers13 we generated the Q489K mutant, which, according to theory, should possess a substantial red-shift in the absorption and 3346


Article

The Journal of Physical Chemistry B fluorescence-emission maxima of 52 and 97 nm, respectively (see Table S4 in the Supporting Information). The parent iLOV protein as well as iLOV-Q489K were heterologously produced in Escherichia coli and purified as His6-tagged fusion proteins by immobilized metal-ion affinity chromatography. Compared to the parent iLOV protein, iLOV-Q489K possesses a 10 nm blue-shifted excitation and 8 nm blue-shifted fluorescence-emission maximum (Figure 1, Table 1).

Figure 1. Normalized fluorescence excitation and fluorescence emission spectra of purified iLOV (black line) and iLOV-Q489K (red line). Excitation spectra and fluorescence-emission spectra are shown as solid line and dashed lines, respectively.

Table 1. Photophysical Properties of Parental iLOV and iLOV-Q489Ka ‑1

‑1

ε (M cm ) λmax excitation (nm) λmax emission (nm) ΦF τBL 50% (min)

iLOV

iLOV-Q489K

14 800 ± 300 450 497 0.33 ± 0.01 3.76 ± 0.26

16 100 ± 200 440 489 0.35 ± 0.01 1.55 ± 0.20

Figure 2. (a) Representative structure of the FMN chromophore binding pocket of parental iLOV taken from the MD trajectory. The snapshot was selected based on cluster analysis of the MD trajectories. FMN chromophore, Q489 and other neighboring residues (N468, N458, Q430) shown as ball-and-stick with carbon (green), oxygen (red), nitrogen (blue). Q489-NE2 keeps a hydrogen bond with FMNO4 during the simulations. (b) Distance distribution curve of the interatomic distances for the indicated residues (in Å) calculated over the representative MD trajectory of parental iLOV; the FMN-N5 and Q489-NE2 distance and the FMN-O4 and Q489-NE2 distance are shown. Nonpolar hydrogen atoms for amino acids and the ribityl phosphate part of FMN are not shown for clarity.

a ε: extinction coefficient. ΦF: fluorescence quantum yield. τBL 50%: photobleaching half-time. Errors represent the standard deviation of the mean, derived from three independent measurements.

To further characterize both proteins, we compared their extinction coefficients, fluorescence quantum yield, and photobleaching behavior (Table 1). The extinction coefficients at the respective excitation maximum as well as the fluorescence quantum yields of the Q489K variant and the parent protein are very similar. Only the photostability of the Q489K mutant shows a significant decrease, compared to the parent protein. Thus, in contrast to the theoretical proposal of a red shift by Khrenova and co-workers,13 iLOV-Q489K possesses blueshifted excitation and fluorescence emission maxima. This behavior is similar to that FbFP variants that possess an apolar substitution (e.g., Pp2FbFP-Q116 V) at the corresponding position.12 In order to explain this discrepancy, we theoretically characterized iLOV-Q489K by molecular dynamics simulations and hybrid quantum mechanics/molecular mechanics calculations. 3-2. MD Simulations of iLOV and iLOV-Q489K. Before analyzing and discussing the results of MD simulation of the iLOV-Q489K variant it is appropriate first to focus on the FMN binding pocket structure and on the side-chain orientations of Q489 in parental iLOV during the MD simulations. The representative structure of the FMN binding site in parental iLOV, obtained by cluster analysis of the respective MD trajectories, is shown in Figure 2a, which depicts the FMN chromophore and residues in the pocket. As in other LOV

proteins, the pteridine part of FMN participates in hydrogen bonding to the side chains of four conserved amino acids (Q489, N468, N458 and Q430), while the dimethylbenzene part is surrounded by nonpolar residues (not shown). Clustering analysis of the iLOV MD trajectories identifies only one highly populated cluster for the conformation of Q489 in parent iLOV, which is populated about 99% of the simulation time. In this dominant conformer, Q489 forms a hydrogen bond with FMN moiety (Q489-NE2 ··· FMN-O4; denoted as iLOV-Q489in). The distribution curve of the distance between FMN-O4 or FMN-N5 and Q489-NE2 is shown in Figure 2b. Analysis of the distances (FMN-O4··· Q489-NE2 and FMN-N5···Q489-NE2) shows that Q489 preserves the hydrogen bond with FMN-O4 during the simulations. Hydrogen bond analysis shows that the hydrogen bond network around the chromophore is persistent and there is a constant hydrogen bond between FMN-O4 and Q489-NE2 (with >70% occurrence) along all trajectories (Table S3 in Supporting Information). By contrast, in the iLOV simulation performed by Khrenova et al.,13 two almost equally populated conformations of Q489 along the trajectories were observed, one, which is stabilized by hydrogen bonding with the FMN-O4 atom (denoted as iLOV3347


Article


experimental iLOV spectrum can be assigned to a vibronic transition of the same electronic excited state with high confidence. Thus, invoking two Q489 conformers to account for the observed absorption and fluorescence-emission bands of FMN in iLOV is not needed. Moreover, the available X-ray structures of parent iLOV (PDB-IDs: 4EES and 4EET) show no evidence for the presence of increased conformational flexibility of the Q489 side chain, as evidenced by well-defined electron density in the corresponding 2Fo-Fc maps and low side-chain B-factors (see Figure S1 in Supporting Information). To investigate how the Q489K substitution affected the structure of iLOV, starting from the parental iLOV crystal structure (PDB-code 4EES), the iLOV-Q489K variant was constructed and a rotamer search was conducted with FoldX exploring alternative conformations of the surrounding side chains. Figures S2−S4 in Supporting Information show the side-chain orientations and the calculated difference in stabilization energies, as well as the FoldX derived differences in interaction energies upon mutation with respect to parent iLOV for 10 runs of the iLOV-Q489K variant. The side-chain orientations and distribution of the stabilization energy and interaction energies of 10 runs are shown in Figure S4 (Supporting Information for detailed discussion). From FoldX analysis of iLOV-Q489K, two distinct orientations of K489 are observed (denoted as K489in and K489out). In conformers occupying the so-called K489in position, there is a hydrogen bond between NZ of K489 and FMN-O4, whereas in the socalled K489out position there is a hydrogen bond formed between backbone oxygen of G487 (G487-O) and K489-NZ. FoldX favors the K489in because of steric hindrance of the backbone of G487, since backbone movements are not allowed using FoldX. To assess possible conformational changes due to the introduced Q489K mutation, we performed three 50 ns MD simulations of the iLOV-Q489K variant starting from the most stable FoldX predicted structure (K489in conformer, shown in Figure 4a). Analysis of the K489 side chain highlights the structural changes surrounding residues K489, near to FMN chromophore during MD simulations. In all three trajectories, when starting from the K489in conformer (Figure 4a), the side chain of K489 starts to move toward G487 after a few nanoseconds, thus populating the K489out conformer (shown in Figure 4b), which strongly suggests that the flipping of K489 toward G487 together with a slight backbone movement observed during MD provides sufficient space for the large K489 next to the FMN chromophore. We observed the formation of a hydrogen bond between the K489-NZ atom and G487-O while disrupting the hydrogen bond between the FMN-O4 atom and K489-NZ in a very early stage of the simulations (see the representative structures in Figure 4). The flip of the K489 side chain together with the main chain rearrangement of G487 to form an H-bond from K489-NH3 moiety toward the G487backbone oxygen leads to a structure that remains stable in all three independent MD simulations. The results of these simulations demonstrate that the K489out conformation is stable during the 50 ns trajectories and is the preferred conformer in iLOV-Q489K. Independent cluster analysis of the trajectories identifies two main clusters, with cluster 1 being the least populated conformation at 0.018% (denoted as K489in). Cluster 2 represents the most highly populated conformation at 98.2% (denoted as K489out). Hydrogen bonding analysis suggests that the K489 side chain forms a stable hydrogen bond between K489-NZ and G487-O (with ∼80% population)

Q489in) and one by forming a hydrogen bond between Q489NE2 and the N390 side chain (iLOV-Q489out). The existence of two conformers was justified by calculating the S0,min → S1 transition energies for both conformations. The resulting values appeared in good agreement with experimentally observed iLOV absorption maxima at 447 nm (Q489in) and 479 nm (Q489out). The authors thus assigned two electronic transitions to the two conformers. In contrast, the observed fine structure, which is less well resolved in isolated flavins, but observed for flavins in an aprotic solvent,56 can likewise be assigned to the vibronic structure of the same electronic transition.43 Figure 3

Figure 3. Vibrationally resolved (a) absorption and (b) fluorescenceemission spectra of the FMN model (lumiflavin) calculated at 0 K using (TD) DFT B3LYP functional and the 6-31G(d) basis set by using the adiabatic Hessian (AH) approach; Gaussian convolution with fwhm = 0.03 eV. The theoretical lineshapes (red lines) are compared to the experimental iLOV absorption and fluorescenceemission spectra (black lines). To make the comparison easier, the maxima of the theoretical spectra is shifted to the maximum of the experimental spectra. The stick vibronic spectra can be found in Figure S7 in the Supporting Information.

shows the vibrationally resolved absorption (a) and fluorescence-emission spectra (b) of the FMN chromophore (modeled by lumiflavin) calculated at B3LYP/6-31G(d) level of theory. Our simulated vibronic lineshapes are is in excellent agreement with the experimental spectra of iLOV. Our computations (Figure 3) as well as previous theoretical studies,43,45,46 of the absorption and fluorescence-emission vibronic lineshapes suggest that the two different bands in the visible region of the iLOV spectrum result from different vibronic transitions of the same electronic transition. Therefore, the shoulder at 479 nm of the main excitation peak in the 3348


Article


and K489 in ). For parental iLOV, QM/MM geometry optimizations were carried out for a representative snapshot obtained from the cluster analysis of MD trajectories (see further discussion in the Supporting Information). 3-3. QM/MM Calculation of Parent iLOV and iLOVQ489K Variant. QM/MM geometry optimization of the representative structure of parental iLOV was performed at the S0 and S1 state. In order to explore the influence of the K489out and K489in orientations on the spectral properties of the iLOVQ489K variant, we carried out QM/MM geometry optimizations on the two representative structures (K489out and K489in) obtained from MD simulations. Theoretical excitation and fluorescence emission wavelengths of parental iLOV and iLOV-Q489 variant in both K489in and K489out states were generated from excitation energies calculated from representative MD snapshots by using the QM/MM method. Calculations at the TD-B3LYP level of theory predict that the first strongly dipole allowed transition corresponds to the excitation of one electron from the HOMO to LUMO orbital (a π−π* transition), which is depicted in Figure S10. Table S4 in Supporting Information summarizes the calculated excitation and fluorescence-emission wavelengths for parental iLOV and considering the two above-described conformers of iLOV-Q489K, shown in Figures 3 and 5,

Figure 4. Representative structures of the FMN chromophore binding pocket of the iLOV-Q489K variant for the (a) K489out and (b) K489in conformations. FMN chromophore, K489 (K489in, K489out), and other neighboring residues (N468, N458, Q430) are shown as ball− stick with carbon (green), oxygen (red), and nitrogen (blue). K4989NZ forms a hydrogen bond with FMN-O4 in the K489in conformation while in the K489out conformation K489-NZ forms a hydrogen bond with G487-O. K489 out conformation is the most populated conformation. (c) Distance distribution curve of the interatomic distances for the indicated residues (in Å) calculated over the representative MD trajectory of iLOV-Q489K variant; The FMN-N5··· K489-NZ distance is shown in green, the FMN-O4···K489-NZ distance is shown in blue, and the K489-NZ···G487-O distance is depicted in red. Nonpolar hydrogen atoms for amino acids and the ribityl phosphate part of FMN are not shown for clarity.

Figure 5. Comparison of predicted and experimentally determined excitation and fluorescence-emission wavelength shifts of the iLOVQ489K variant relative to the experimentally determined values of the parental iLOV protein. In this work, excitation and fluorescenceemission shifts were computed for the two K489 conformations (K489in and K489out).

calculated at the QM/MM TD-DFT level. The excitation and fluorescence-emission wavelength maxima previously experimentally determined by Christie et al.17 as well as calculated by Khrenova et al.,13 are given for comparison. For parental iLOV (only showing the Q489in conformer in MD simulations; see Figure 2), the comparison of the computationally derived and experimentally determined excitation and fluorescence-emission maxima show that B3LYP/SVP systematically underestimates the excitation and fluorescence-emission wavelength maximum, while the overall shape is well reproduced by taking into account vibrational effects (shown in Figure 3). Please note that the direct comparison between QM/MM calculated excitation and fluorescence-emission wavelengths and the corresponding experimental maxima is difficult, because QM/MM methods calculate vertical transition energies of a given electronic

(see Figures S8−S9 and Table S3 in Supporting Information). The same trend can be observed from the corresponding distance distribution curves (Figure 4c), which depict the interatomic distances for the FMN-N5···K489-NZ, FMN-O4··· K489-NZ, and K489-NZ···G487-O interactions, respectively. To calculate excitation and fluorescence-emission wavelengths, we performed the QM/MM geometry optimization for the representative structures of the two clusters (K489out 3349


Article


wavelength maxima for both conformers by using QM/MM methods. For the K489out conformer of iLOV-Q489K, the calculated FMN chromophore excitation and fluorescence emission maxima are blue shifted with respect to parental iLOV as compared with those of the K489in conformer which are red shifted. Our results therefore strongly suggest that the flipping of K489 side chain accounts for the discrepancy between the theoretical prediction by Kheronova et al. and our experimentally determined photophysical properties of iLOVQ489K.

transition, which however do not directly compare to the experimentally observed excitation and fluorescence emission maxima due to vibronic effects.57 As shown in calculated vibrationally resolved spectra of the FMN model lumiflavin (Figure S11 in Supporting Information) and previously reported,43 by taking into account vibrational effects a systematic blue/red shift in vertical excitation and fluorescence-emission with respect to corresponding maxima is expected, respectively (see also Figure S11 and Table S4 in Supporting Information). Figure 5, depicts the direct comparison between previously predicted excitation and fluorescence-emission band maxima by Kheronova et al.13 and the corresponding predictions and experimental findings of this study. When compared with the vertical excitation energies of parental iLOV at the optimized (TD) DFT (B3LYP/SVP) geometry, our computed wavelengths of iLOV-Q489Kin and Q489Kout are shifted in opposite directions (red for Q489Kin and blue shift for 489Kout) with respect to parental iLOV. The excitation wavelength computed for Q489Kin at the TD-DFT level shows the same direction of shift (red-shift) as reported by Kheronova et al.,13 although the absolute excitation energies for the two methods are different. Both predictions for the iLOV-Q489Kin conformation (by Khrenova et al. and this study) are thus in direct opposition to the here presented experimental data (Figure 1, Table 1). In contrast, the excitation and fluorescence-emission wavelengths predicted for iLOV-Q489Kout are blue-shifted with respect to parental iLOV, which is in excellent agreement with the our experimentally determined blue-shifted values (Figure 1, Table S4). Taken together, our experimental and theoretical characterization of iLOV-Q489K suggests that in the iLOV-Q489K variant K489 does not occupy the previously suggested K489in conformation,13 which would result in red-shifted excitation and fluorescence-emission maxima, but instead its side chain flips away from the FMN molecule to form a hydrogen bond with G487-O, resulting in blue-shifted excitation and emission bands maxima.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.6b01512. Details of the FoldX prediction and MD and QM/MM simulations analysis (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*(M.D.D.) E-mail: [email protected]. *(U.K.) E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS M.D.D thanks the DFG Training Group 1166 “BioNoCo” (Biocatalysis in Nonconventional Media”) for financial support and Dr. Keyarash Sadeghian, Dr. Francisco Jose Avila Ferrer, and Gaurao V. Dhoke for fruitful discussions. Simulations were performed with computing resources granted by JARA-HPC from RWTH Aachen University under Project JARA0065. M.W., U.K., T.D., and K.-E.J. also acknowledge funding by the Federal Ministry of Education and Research (BMBF) in the framework of the collaborative research project “OptoSys” (FKZ 031A16).

■

4. CONCLUSIONS Recently, Kheronova et al.13 predicted a considerable red-shift (by 52 and 97 nm) in the absorption and fluorescence-emission maxima of the iLOV-Q489K variant, using QM/MM methods. According to our experimental data, the corresponding iLOVQ489K variant shows in fact a 10 nm blue-shifted excitation maximum and an 8 nm blue-shifted fluorescence-emission maximum compared to parental iLOV. We explored this observation by carrying out FoldX calculations and MD simulations of the respective variant, based on the iLOV Xray crystal structure (PDB ID: 4EES). FoldX predictions showed that iLOV-Q489K variant favors the K489in conformer due to steric hindrance with fixed backbone of G487. MD simulations suggest the existence of a most populated (K489out) and a least populated conformer of K489 (K489in), occurring by flipping of the side chain of K489 away from the FMN-O4 atom (K489in) toward G487 (K489out) after a slight backbone rearrangement to form a stable H-bond to the backbone oxygen. This obviously results in breaking of the hydrogen bond between the side-chain amino group of K489 and the FMN-O4 atom and the formation of a new hydrogen bond with the backbone O atom of G487. To reveal the effect of the flipping of the K489 side chain on the spectral characteristics of iLOV-Q489K, we calculated the excitation and emission

REFERENCES

(1) Drepper, T.; Eggert, T.; Circolone, F.; Heck, A.; Krauss, U.; Guterl, J.-K.; Wendorff, M.; Losi, A.; Gärtner, W.; Jaeger, K.-E. Reporter proteins for in vivo fluorescence without oxygen. Nat. Biotechnol. 2007, 25 (4), 443−445. (2) Chapman, S.; Faulkner, C.; Kaiserli, E.; Garcia-Mata, C.; Savenkov, E. I.; Roberts, A. G.; Oparka, K. J.; Christie, J. M. The photoreversible fluorescent protein iLOV outperforms GFP as a reporter of plant virus infection. Proc. Natl. Acad. Sci. U. S. A. 2008, 105 (50), 20038−20043. (3) Shu, X.; Lev-Ram, V.; Deerinck, T. J.; Qi, Y.; Ramko, E. B.; Davidson, M. W.; Jin, Y.; Ellisman, M. H.; Tsien, R. Y. A genetically encoded tag for correlated light and electron microscopy of intact cells, tissues, and organisms. PLoS Biol. 2011, 9 (4), e1001041. (4) Drepper, T.; Huber, R.; Heck, A.; Circolone, F.; Hillmer, A.-K.; Büchs, J.; Jaeger, K.-E. Flavin mononucleotide-based fluorescent reporter proteins outperform green fluorescent protein-like proteins as quantitative in vivo real-time reporters. Applied and environmental microbiology 2010, 76 (17), 5990−5994. (5) Zhang, L.-C.; Risoul, V.; Latifi, A.; Christie, J. M.; Zhang, C.-C. Exploring the size limit of protein diffusion through the periplasm in cyanobacterium Anabaena sp. PCC 7120 using the 13 kDa iLOV fluorescent protein. Res. Microbiol. 2013, 164 (7), 710−717. (6) Buckley, A. M.; Petersen, J.; Roe, A. J.; Douce, G. R.; Christie, J. M. LOV-based reporters for fluorescence imaging. Curr. Opin. Chem. Biol. 2015, 27, 39−45.

3350


Article

The Journal of Physical Chemistry B (7) Lobo, L. A.; Smith, C. J.; Rocha, E. R. Flavin mononucleotide (FMN)-based fluorescent protein (FbFP) as reporter for gene expression in the anaerobe Bacteroides fragilis. FEMS Microbiol. Lett. 2011, 317 (1), 67−74. (8) Choi, C. H.; DeGuzman, J. V.; Lamont, R. J.; Yilmaz, Ö . Genetic transformation of an obligate anaerobe, P. gingivalis for FMN-green fluorescent protein expression in studying host-microbe interaction. PLoS One 2011, 6 (4), e18499. (9) Potzkei, J.; Kunze, M.; Drepper, T.; Gensch, T.; Jaeger, K.-E.; Büchs, J. Real-time determination of intracellular oxygen in bacteria using a genetically encoded FRET-based biosensor. BMC Biol. 2012, 10 (1), 28. (10) Swartz, T. E.; Corchnoy, S. B.; Christie, J. M.; Lewis, J. W.; Szundi, I.; Briggs, W. R.; Bogomolni, R. A. The photocycle of a flavinbinding domain of the blue light photoreceptor phototropin. J. Biol. Chem. 2001, 276 (39), 36493−36500. (11) Mukherjee, A.; Weyant, K. B.; Agrawal, U.; Walker, J.; Cann, I. K.; Schroeder, C. M. Engineering and characterization of new LOVbased fluorescent proteins from Chlamydomonas reinhardtii and Vaucheria frigida. ACS Synth. Biol. 2015, 4 (4), 371−377. (12) Wingen, M.; Potzkei, J.; Endres, S.; Casini, G.; Rupprecht, C.; Fahlke, C.; Krauss, U.; Jaeger, K.-E.; Drepper, T.; Gensch, T. The photophysics of LOV-based fluorescent proteins−new tools for cell biology. Photochemical & photobiological sciences 2014, 13 (6), 875− 883. (13) Khrenova, M. G.; Nemukhin, A. V.; Domratcheva, T. Theoretical characterization of the flavin-based fluorescent protein iLOV and its Q489K mutant. J. Phys. Chem. B 2015, 119 (16), 5176− 5183. (14) Krieger, E.; Koraimann, G.; Vriend, G. Increasing the precision of comparative models with YASARA NOVAa self-parameterizing force field. Proteins: Struct., Funct., Genet. 2002, 47 (3), 393−402. (15) Van Durme, J.; Delgado, J.; Stricher, F.; Serrano, L.; Schymkowitz, J.; Rousseau, F. A graphical interface for the FoldX forcefield. Bioinformatics 2011, 27 (12), 1711−1712. (16) Guerois, R.; Nielsen, J. E.; Serrano, L. Predicting changes in the stability of proteins and protein complexes: a study of more than 1000 mutations. J. Mol. Biol. 2002, 320 (2), 369−387. (17) Christie, J. M.; Hitomi, K.; Arvai, A. S.; Hartfield, K. A.; Mettlen, M.; Pratt, A. J.; Tainer, J. A.; Getzoff, E. D. Structural tuning of the fluorescent protein iLOV for improved photostability. J. Biol. Chem. 2012, 287 (26), 22295−22304. (18) Olsson, M. H.; Søndergaard, C. R.; Rostkowski, M.; Jensen, J. H. PROPKA3: consistent treatment of internal and surface residues in empirical p K a predictions. J. Chem. Theory Comput. 2011, 7 (2), 525−537. (19) Hornak, V.; Abel, R.; Okur, A.; Strockbine, B.; Roitberg, A.; Simmerling, C. Comparison of multiple Amber force fields and development of improved protein backbone parameters. Proteins: Struct., Funct., Genet. 2006, 65 (3), 712−725. (20) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. A second generation force field for the simulation of proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc. 1995, 117 (19), 5179−5197. (21) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. Development and testing of a general amber force field. J. Comput. Chem. 2004, 25 (9), 1157−1174. (22) Becke, A. D. Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98 (7), 5648−5652. (23) Lee, C.; Yang, W.; Parr, R. G. Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B: Condens. Matter Mater. Phys. 1988, 37 (2), 785. (24) Vosko, S.; Wilk, L.; Nusair, M. Accurate spin-dependent electron liquid correlation energies for local spin density calculations: a critical analysis. Can. J. Phys. 1980, 58 (8), 1200−1211. (25) Hariharan, P. C.; Pople, J. A. The influence of polarization functions on molecular orbital hydrogenation energies. Theoretica chimica acta 1973, 28 (3), 213−222.

(26) Hehre, W. J.; Ditchfield, R.; Pople, J. A. Selfconsistent molecular orbital methods. XII. Further extensions of Gaussiantype basis sets for use in molecular orbital studies of organic molecules. J. Chem. Phys. 1972, 56 (5), 2257−2261. (27) Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A. Antechamber: an accessory software package for molecular mechanical calculations. Abstr. Pap. Am. Chem. Soc. 2001, 222, U403. (28) Case, D.; Berryman, J.; Betz, R.; Cerutti, D.; Cheatham, I.; Darden, T.; Duke, R.; Giese, T.; Gohlke, H.; Goetz, A. AMBER 2015; University of California: San Francisco, CA, 2015. (29) Frisch, M.; Trucks, G.; Schlegel, H. B.; Scuseria, G.; Robb, M.; Cheeseman, J.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. Gaussian 09, Revision A. 02; Gaussian Inc.: Wallingford, CT, 2009, 200. (30) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 1983, 79 (2), 926−935. (31) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. A smooth particle mesh Ewald method. J. Chem. Phys. 1995, 103 (19), 8577−8593. (32) Schrodinger, L. The PyMOL molecular graphics system, version 1.3 r1. 2010. (33) Humphrey, W.; Dalke, A.; Schulten, K. VMD: visual molecular dynamics. J. Mol. Graphics 1996, 14 (1), 33−38. (34) Hess, B.; Kutzner, C.; Van Der Spoel, D.; Lindahl, E. GROMACS 4: algorithms for highly efficient, load-balanced, and scalable molecular simulation. J. Chem. Theory Comput. 2008, 4 (3), 435−447. (35) ChemShell, a Computational Chemistry Shell; see www.chemshell. org. (36) Ahlrichs, R.; Armbruster, M.; Bar, M.; Baron, H.; Bauernschmitt, R.; Bocker, S.; Crawford, N.; Deglmann, P.; Ehrig, M.; Eichkorn, K. TURBOMOLE V6. 2 2010, a development of the University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989−2007, TURBOMOLE GmbH: since 2007. (37) Smith, W.; Todorov, I. T. A short description of DL_POLY. Mol. Simul. 2006, 32 (12−13), 935−943. (38) de Vries, A. H.; Sherwood, P.; Collins, S. J.; Rigby, A. M.; Rigutto, M.; Kramer, G. J. Zeolite structure and reactivity by combined quantum-chemical-classical calculations. J. Phys. Chem. B 1999, 103 (29), 6133−6141. (39) Kästner, J.; Carr, J. M.; Keal, T. W.; Thiel, W.; Wander, A.; Sherwood, P. DL-FIND: An Open-Source Geometry Optimizer for Atomistic Simulations. J. Phys. Chem. A 2009, 113 (43), 11856−11865. (40) Bauernschmitt, R.; Häser, M.; Treutler, O.; Ahlrichs, R. Calculation of excitation energies within time-dependent density functional theory using auxiliary basis set expansions. Chem. Phys. Lett. 1997, 264 (6), 573−578. (41) Furche, F.; Ahlrichs, R. Adiabatic time-dependent density functional methods for excited state properties. J. Chem. Phys. 2002, 117 (16), 7433−7447. (42) Schäfer, A.; Horn, H.; Ahlrichs, R. Fully optimized contracted Gaussian basis sets for atoms Li to Kr. J. Chem. Phys. 1992, 97 (4), 2571−2577. (43) Klaumünzer, B.; Krö ner, D.; Saalfrank, P. (TD-) DFT Calculation of Vibrational and Vibronic Spectra of Riboflavin in Solution. J. Phys. Chem. B 2010, 114 (33), 10826−10834. (44) Salzmann, S.; Silva-Junior, M. R.; Thiel, W.; Marian, C. M. Influence of the LOV domain on low-lying excited states of flavin: a combined quantum-mechanics/molecular-mechanics investigation. J. Phys. Chem. B 2009, 113 (47), 15610−15618. (45) Götze, J. P.; Karasulu, B.; Thiel, W. Computing UV/vis spectra from the adiabatic and vertical Franck-Condon schemes with the use of Cartesian and internal coordinates. J. Chem. Phys. 2013, 139 (23), 234108. (46) Karasulu, B.; Götze, J. P.; Thiel, W. Assessment of Franck− Condon Methods for Computing Vibrationally Broadened UV−vis Absorption Spectra of Flavin Derivatives: Riboflavin, Roseoflavin, and 5-Thioflavin. J. Chem. Theory Comput. 2014, 10 (12), 5549−5566. 3351


Article

The Journal of Physical Chemistry B (47) Marian, C. M.; Nakagawa, S.; Rai-Constapel, V.; Karasulu, B.; Thiel, W. Photophysics of flavin derivatives absorbing in the bluegreen region: thioflavins as potential cofactors of photoswitches. J. Phys. Chem. B 2014, 118 (7), 1743−1753. (48) Salzmann, S.; Tatchen, J.; Marian, C. M. The photophysics of flavins: What makes the difference between gas phase and aqueous solution? J. Photochem. Photobiol., A 2008, 198 (2), 221−231. (49) Silva-Junior, M. R.; Mansurova, M.; Gärtner, W.; Thiel, W. Photophysics of Structurally Modified Flavin Derivatives in the BlueLight Photoreceptor YtvA: A Combined Experimental and Theoretical Study. ChemBioChem 2013, 14 (13), 1648−1661. (50) Neiss, C.; Saalfrank, P.; Parac, M.; Grimme, S. Quantum chemical calculation of excited states of flavin-related molecules. J. Phys. Chem. A 2003, 107 (1), 140−147. (51) Sikorska, E.; Khmelinskii, I. V.; Prukala, W.; Williams, S. L.; Patel, M.; Worrall, D. R.; Bourdelande, J. L.; Koput, J.; Sikorski, M. Spectroscopy and photophysics of lumiflavins and lumichromes. J. Phys. Chem. A 2004, 108 (9), 1501−1508. (52) Wu, M.; Eriksson, L. A. Absorption Spectra of Riboflavin: A Difficult Case for Computational Chemistry. J. Phys. Chem. A 2010, 114 (37), 10234−10242. (53) Kammler, L.; van Gastel, M. Electronic Structure of the Lowest Triplet State of Flavin Mononucleotide. J. Phys. Chem. A 2012, 116 (41), 10090−10098. (54) Avila Ferrer, F. J. A.; Santoro, F. Comparison of vertical and adiabatic harmonic approaches for the calculation of the vibrational structure of electronic spectra. Phys. Chem. Chem. Phys. 2012, 14 (39), 13549−13563. (55) Santoro, F. FCclasses, a Fortran 77 code. See: http://village.pi. iccom.cnr.it 2011, [last accessed May 10, 2014]. (56) Zirak, P.; Penzkofer, A.; Mathes, T.; Hegemann, P. Photodynamics of roseoflavin and riboflavin in aqueous and organic solvents. Chem. Phys. 2009, 358 (1−2), 111−122. (57) Avila Ferrer, F. J.; Cerezo, J.; Stendardo, E.; Improta, R.; Santoro, F. Insights for an accurate comparison of computational data to experimental absorption and emission spectra: Beyond the vertical transition approximation. J. Chem. Theory Comput. 2013, 9 (4), 2072− 2082.

3352


Photophysics of the LOV-Based Fluorescent Protein Variant iLOV

Recommend Documents