Improving the Design of the Triple-Decker Motif in Red FluorescentProteinsMaria G. Khrenova,† Igor V. Polyakov,† Bella L. Grigorenko,†,‡ Anna I. Krylov,§
and Alexander V. Nemukhin*,†,‡
†Department of Chemistry, Lomonosov Moscow State University, Moscow, 119991, Russia‡Emanuel Institute of Biochemical Physics, Russian Academy of Sciences, Moscow, 119991, Russia§Department of Chemistry, University of Southern California, Los Angeles, California 90089-0482, United States
*S Supporting Information
ABSTRACT: We characterize computationally a red fluo-rescent protein (RFP) with the chromophore (Chro)sandwiched between two aromatic tyrosine rings in a triple-decker motif. According to the original proposal [J. Phys.Chem. Lett. 2013, 4, 1743], such a tyrosine-chromophore-tyrosine π-stacked construct can be accommodated in thegreen fluorescent protein (GFP). A recent study [ACS Chem.Biol. 2016, 11, 508] attempted to realize the triple-deckermotif and obtained an RFP variant called mRojoA-VYGV withtwo tyrosine residues surrounding the chromophore. Thecrystal structure showed that only a tyrosine-chromophore pairwas involved in π-stacking, whereas the second tyrosine wasoriented perpendicularly, edge-to-face with respect to thechromophore. We propose a more promising variant of this RFP with a perfect triple-decker unit achieved by introducingadditional mutations in mRojoA-VYGV. The structures and optical properties of model proteins based on the structures ofmCherry and mRojoA are characterized computationally by QM(DFT)/MM. The electronic transitions in the protein-boundchromophores are computed by high-level quantum chemical methods. According to our calculations, the triple-deckerchromophore unit in the new RFP variant is stable within the protein and its optical bands are red-shifted with respect to theparent mCherry and mRojoA species.
■ INTRODUCTION
Ongoing efforts to engineer efficient markers for multicolorimaging in living cells drive explorations of variousmodifications of the parent green fluorescent protein (GFP)chromophore and/or its immediate environment within theprotein barrel.1 One direction is based on an idea to exploit π-stacking of the conjugated chromophore with the neighboringaromatic groups. Such constructs are expected to have red-shifted electronic spectra and a decreased flexibility of thechromophore.2,3 Both features are highly desirable forapplications, especially for in vivo imaging,4 which depends onthe availability of bright red fluorescent proteins. Using π-stacking to achieve red-shifted absorption complements otherdesign strategies, such as modifications of the chromophore’sstructure, its protonation state, or the electrostatic field aroundthe chromophore.5,6
In the GFP-derived systems, this idea was first exploited inthe work that introduced yellow fluorescent protein (YFP).7,8
In YFP, 20 nm red shift in absorption and emission is achievedby the replacement of the threonine residue close to thechromophore in the wild-type GFP by tyrosine. The same idea,replacement of Thr by Tyr near the chromophore, was
executed in the red fluorescent proteins (RFPs), resulting inthe variants called mRojoA2 and mRojoA-VYGV.9 Thesemonomeric RFPs were engineered from the mCherryprecursor10 (crystal structure PDB ID: 2H5Q). We useabbreviation Chro for its chromophore, the π-conjugatedsystem of which is extended by N-acylimine group(C=N-C=O) as compared to the 4-(p-hydroxybenzylidene)-imidazolid-5-one (HBDI) moiety from GFP. By introducingsix-point mutations in mCherry, namely, Val16Thr/Ar-g125His/Gln163Leu/Val195Ala/Ile197Tyr/Ala217Cys, Chicaand co-workers2 designed an RFP variant called mRojoA(PDB ID: 3NEZ), in which the replacement Ile197Tyr led tothe stacked Chro-Tyr motif. The mRojoA-VYGV variant wascreated9 by introducing 4 mutations in mRojoA (Thr16Val/Pro63Tyr/Trp143Gly/Leu163Val). In this RFP, the valineresidue at position 16 was mutated back to threonine (as inmCherry) and proline at position 63 was replaced by tyrosine,with an aim to produce the triple-decker motif Tyr63-Chro-
Received: July 29, 2017Revised: October 31, 2017Published: November 1, 2017
Tyr197. Unfortunately, the desired parallel arrangement of thethree aromatic rings was not achieved, as clearly seen in thecrystal structure (PDFB ID: 5H89). Rather, only Chro-Tyr pairis π-stacked in this construct. Figure 1 illustrates the history ofmRojoA and mRojoA-VYGV starting from mCherry.The triple-decker tyrosine-chromophore-tyrosine motif was
originally proposed in ref 11 on the basis of a computationalstudy of the strategically designed GFP mutants. This constructappeared promising, but, as mentioned above, an attempt toimplement this π-stacked triple-decker motif in mRojoA9 hasnot achieved the goal. We agree with a conclusion9 that “furtherengineering [is] required to incorporate a triple-decker motifmade of exclusively parallel π-stacking interactions into RFPs”.In this work, we apply quantum-chemistry methods tocomputationally engineer and characterize a novel variant ofthe red protein mRojoA with the Tyr-Chro-Tyr (YChroY) π-stacking motif representing an improved design of the triple-decker chromophore.
■ MODELS AND METHODS
We begin our design of a new variant of the protein with thetriple-decker chromophore from the crystal structure of themRojoA-VYGV protein (PDB ID: 5H89) described in ref 9. Tobuild model systems, we added hydrogen atoms to the heavyatoms of crystal structures assuming positive charges of the sidechains of Lys, Arg, and negative charges of Glu and Asp. Uponexamining the local environment of all histidine residues, weprotonated them at either Nδ or Nε. The model systems werefully solvated by water. To equilibrate the latter, moleculardynamics simulations were carried out using NAMD12 with theCHARMM force fields13 by keeping protein molecules frozenas in the crystal.
In the QM/MM calculations using the NWChem program,14
the QM subsystems were treated at the DFT-D3/cc-pVDZlevel with the PBE0 functional15 augmented by the dispersioncorrection D3.16 The MM was described with the AMBERforce field parameters.17 The structures were obtained in seriesof unconstrained QM/MM minimizations. The QM partscomprised the chromophore, the side chains of Arg95, Glu215,Ser146, Tyr197 (for mRojoA and mRojoA-VYGV), Tyr63 (formRojoA-VYGV), and two water molecules.The computed structures of the chromophore-containing
pockets matched the corresponding crystal structures; there-fore, we applied the same computational protocol to designnovel variants of the protein with the triple-decker motif.Although we considered several variants, here we describe onlythe most promising ones. The inspection of the mRojoA-VYGVstructure prompted us to introduce the following replacements:Gly143Ser/Ile161Asn/Val163Ser/Val177Tyr/Leu199Ala.Every mutation had specific purpose. Serine residues atpositions 143 and 163 were introduced to form hydrogenbonds with the hydroxyl group of Tyr63, the prospectivecandidate to interact with Chro; tyrosine at position 177 andalanine at position 199 helped to keep the benzene ring ofTyr63 in a parallel arrangement with Chro; the replacement ofisoleucine by asparagine at position 161 was necessary to fixSer143. These mutations enabled to fix Tyr63 in a planarorientation with respect to Chro. The VMD program18 wasused to create the mutated structures. Figure 2 shows positionsof these replacements in the protein barrel.For this new structure (called here mRojoA-YChroY) we
carried out the same MD and QM/MM calculations as formCherry, mRojoA, and mRojoA-VYGV. As described in theResults, its structure accommodates a nearly perfect triple-decker Tyr-Chro-Tyr unit.
Figure 1. Previous attempts to create proteins with shifted absorption/emission bands via π-stacking of the chromophore with the tyrosine residueson the basis of mCherry (left panels). In mRojoA (central panels), the key replacement was Ile197Tyr; in mRojoA-VYGV (right panels), Pro63 wasmutated by Tyr. Other mutations were necessary to keep Tyr residues near Chro. The bottom row emphasizes the positions of the Tyr residues(shown in colored sticks) near Chro. In the balls and sticks representation here and in other figures, carbon atoms are colored green; oxygen, red;nitrogen, blue; sulfur, yellow; and hydrogen, white.
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
To verify whether such construct remains stable in a proteinwe run QM/MM MD trajectories for the mRojoA, mRojoA-VYGV and mRojoA-YChroY model systems using the CP2Kprogram.19 In these simulations, the same molecular groups asin QM/MM minimization were assigned to the QM parts. Itwas necessary to adjust computational protocols because CP2Kemploys a mixed Gaussian and plane waves basis. Specifically,we used PBE-D3/TZV2P-MOLOPT-GTH in QM, and theCHARMM force field in MM.To compute vertical transition energies between the ground
and excited states we considered molecular clusters composedof the QM parts of the model systems. According to ourexperience,11,20 this strategy delivers accurate results providedthat the clusters are large enough and include the chromophoreand several closest amino acid residues. Contributions from therest of the protein described by point charges from theconventional force fields do not significantly affect basicconclusions drawn from cluster calculations, especially, whenusing the extended multiconfigurational quasidegenerateperturbation theory (XMCQDPT2)21 based on the completeactive space self-consistent field (CASSCF) wave functions. Weused the Firefly program22 to carry out XMCQDPT2calculations. Our choice of molecular orbitals for CASSCFwas guided by preliminary estimates using configurationinteraction with single excitations method and visual inspectionof the orbitals. We included four lowest singlet states in state-averaging (SA-CASSCF) with the active space of 12 electronsdistributed over 12 orbitals.We used the XMCQDPT2/(SA(4)-CASSCF(12/12)) pro-
tocol to estimate the vertical excitation energies S0,min→S1 atthe QM/MM optimized structures of the ground state. Tofurther strengthen the conclusions, we also computed theexcitation energies using the SOS-CIS(D) method23 with theQChem package24 and TD-DFT with ωB97X-D25 with theORCA program.26
For the new variant mRojoA-YChroY we also estimated theenergy corresponding to emission from the excited stateS1,min→S0. To find the minimum-energy structure S1,min on theexcited state potential energy surface for a fairly large molecularcluster we used CASSCF and the Firefly program.22 Tocorrectly describe the immediate environment of the mostimportant part of the system composed of the chromophoreand the nearest residues (the QM-subsystem of the QM/MMmodel used in the ground-state optimization) we surroundedthe latter groups by the main chains of residues at positions 59,
60, 144, 145 and 198, side chains of residues at positions 64, 70,148, 161, 172, 199, and 217, and four water molecules.Coordinates of the terminal carbon atoms of the added groupswere kept frozen upon geometry optimization. To reduce thecomputational cost, these additional groups were describedwith a smaller basis set 3-21G*, whereas the primary part wastreated with the same basis set (cc-pVDZ) as in the ground-state optimization. The SA(2)-CASSCF(12/12) protocol wasused to scan the excited-state potential energy surface and tolocate the S1,min point. The vertical energy of the S1,min→S0transition was estimated by XMCQDPT2 as in excitationenergy calculations.
■ RESULTS AND DISCUSSIONPrototype Model Systems. Although our aim was to
design a π-stacked construct inside the protein, we begin byanalyzing a series of smaller model systems (Figure 3) with
such a motif in the gas phase. The chromophore from GFP(HBDI) in the anionic form served as a reference. Two watermolecules were added to saturate hydrogen bonds of thephenyl oxygen (Figure 3a).For the π-stacking constructs, we considered three model
systems with the following molecular fragments added to thereference structure: phenol in the head-to-tail orientation,which mimics that in mRojoA or YFP (Figure 3b); phenol in ahead-to-head orientation and additional water molecule (Figure3c); two phenol molecules on the both sides of thechromophore in opposite orientations and additional watermolecule (Figure 3d), which mimics the triple-decker construct.We optimized ground-state structures using the MP2/cc-pVDZlevel of theory to account for proper interaction between thebenzene rings. We computed the S0,min→S1 excitation energiesusing the same approach as for the model systems mimickingproteins, namely, SA(4)-CASSCF(12/12) followed byXMCQDPT2, as well as TD-DFT(ωB97X-D), and SOS-CIS(D). The cc-pVDZ basis was used in all calculations.Equilibrium structures of the model systems (panels b−d) in
Figure 3) clearly demonstrate π-stacking between the phenolmolecules and the HBDI chromophore. Table 1 lists thestructural parameters of the π-stacking including the distancesbetween the centroids of the benzene rings and the angles
Figure 2. Alignment of mRojoA-VYGV (pink) and mRojoA-YChroY(blue). Inset shows point mutations in mRojoA-VYGV improvingposition of Tyr63 necessary to achieve the triple-decker motif.
Figure 3. Prototype model systems: (a) HBDI, (b) HBDI and phenolin the head-to-tail (H-to-T) orientation, (c) HBDI and phenol in thehead-to-head (H-to-H) orientation, and (d) HBDI and 2 sandwichingphenol molecules. Water molecules are necessary to stabilizeconformations in the gas phase.
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
between their planes. In all considered systems, the benzenemoieties are nearly parallel with the smaller deviationcorresponding to the head-to-tail orientation of the phenolrelative to the HBDI molecule (Figure 3b). The distancesbetween the centroids of the benzene rings are slightly smallerin the case of the head-to-tail orientation. Thus, from thestructural point of view, we conclude that head-to-tailorientation is slightly more preferable for π-stacking, at leastin the small gas phase models. Also, the key geometryparameters, i.e., distances between centroids and anglesbetween planes, are almost the same in both models comprisingone or two phenol molecules.Table 2 summarizes the S0,min→S1 transition energies
calculated using different approaches. In all model systems, π-stacking results in the red shift of the S0,min→S1 transitions. Forstacking with a single phenol molecule, the shift is larger inhead-to-head orientation. The model system with HBDIchromophore and two phenol molecules (Figure 3d) exhibitsmore pronounced red shift, exceeding the sum of the shiftsfrom the phenols considered independently in model systemsshown in panels (b,c) in Figure 3.Figure 4 illustrates the changes of the electron densities upon
the S0→S1 excitation. All transitions involve the decrease of theelectron density on the phenolic moiety and the increase on the
imidazolinone moiety. These density difference maps suggestthat the computed red shifts can be attributed to the increasedcharge transfer character of the transition: as one can see, themore pronounced charge transfer is (panel d in Figure 4), thelarger is the red shift, here, for the system HBDI and 2sandwiching phenol molecules.The calculations of the model systems lead to the following
conclusions. The red shift for the S0,min→S1 transition isconsistently reproduced by all methods when the chromophoreis involved in π-stacking. The magnitude of red shift increasesin the triple-layered π-stacking relative to the double-layeredsystems. For all systems, the magnitude of the computed shiftsagree well among all three quantum chemistry methods,although TD-DFT considerably overestimates the transitionenergy relative to two other methods.
Protein Structures. Comparison of the QM/MMcomputed structure parameters of mCherry, mRojoA, andRojoA-VYGV model systems with the corresponding crystalstructures showed a good agreement between theory andexperiment: the RMSD values were 3.305 Å (heavy atoms) and1.192 Å (backbone atoms) for mCherry, 3.204 Å (heavy atoms)and 0.725 Å (backbone atoms) for mRojoA, and 3.196 Å(heavy atoms) and 0.684 Å (backbone atoms) for mRojoA-VYGV. In particular, our calculations show that the minimum-energy structure of RojoA-VYGV does not have a planar triple-decker motif Tyr197-Chro-Tyr63 (see right panels in Figure1); the side chain of Tyr63 assumed a perpendicular edge-to-face arrangement, as in the crystal structure.9
In contrast to this, the proposed structure mRojoA-YChroYshows a nearly perfect triple-decker motif for the Tyr-Chro-Tyrunit (Figure 5). The distances between the planes of thechromophore phenolate group and Tyr63 and Tyr167 rings(measured as centroid-to-centroid distances) constitute 3.7−3.8 Å. Importantly, this arrangement remains stable in thecourse of equilibrium dynamics; Table 3 shows fluctuation ofthe key structural parameters in mRojoA, mRojoA-VYGV, andmRojoA-YChroY along the QM/MM MD trajectories. First,we note a good agreement between the QM/MM values andthe corresponding parameters in the crystal structures formRojoA, mRojoA-VYGV. Second, the dynamical simulations ofthe chromophore-containing pocket in mRojoA-YChroYdemonstrate that the triple-decker motif remains stable insidethe protein barrel.Although we could afford relatively short dynamical
simulations with the QM/MM potentials (20 ps for mRojoA
Table 1. Distances Between the Centroids of the Benzene Rings and Angles Between the Planes (See Figure 3)
modelHBDI···H-to-T phenol
distance, ÅHBDI···H-to-T phenol angle,
degHBDI···H-to-H phenol
distance, ÅHBDI···H-to-H phenol angle,
deg.
HBDI and H-to-T phenol 3.43 4 n/a n/aHBDI and H-to-H phenol n/a n/a 3.59 8HBDI and 2 phenol molecules 3.44 2 3.55 6
Table 2. Excitation Energies in eV, the Corresponding Wavelengths in nm (in Parentheses), and the Shifts in These ValuesRelative to the HBDI Model
Figure 4. Electron density difference maps for the S0,min→S1 transitionof the HBDI chromophore in model systems: (a) HBDI, (b) HBDIand phenol in the head-to-tail orientation, (c) HBDI and phenol in thehead-to-head orientation, and (d) HBDI and 2 sandwiching phenolmolecules shown in Figure 3. Pink indicates the increase of theelectron density in the S1 state; violet indicates the decrease. Thecontour value is 0.003757.
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
and mRojoA-VYGV and 40 ps for mRojoA-YChroY), webelieve that the conclusions on the stability of thecorresponding structures are valid. To further test the stabilityof the optimized structures, we computed additional QM/MMtrajectories for the mRojoA-YChroY and mRojoA-VYGVstarting from strongly distorted structures. In these dynamicalsimulations, the distorted structures evolved back to theoriginal optimized structures, as illustrated in Figure 6. InmRojoA-YChroY, we twisted the Tyr63 side chain to a nearlyperpendicular orientation with respect to Chro (colored thicksticks in the left panel in Figure 6) and the system returnedback to the parallel orientation Tyr63-Chro (a bunch of orangethin sticks). In mRojoA-VYGV we twisted the Tyr63 side chainto a nearly parallel orientation with respect to Chro (coloredthick sticks in the right panel in Figure 6) and the systemreturned back to the perpendicular orientation Tyr63-Chro (abunch of orange thin sticks). Video files of QM/MM MDtrajectories illustrating the data shown in Figure 6 are presentedin Supporting Information.We also carried out the QM/MM geometry optimization for
mRojoA-YChroY starting from a strongly perpendicularorientation of Tyr63 relative to Chro and with different
arrangements of the surrounding side chains (the side chains ofSer143 and Ser163 (see Figure 5) were twisted). The results(shown in Supporting Information) illustrate that nearlyparallel arrangement of the Tyr63-Chro pair is restored,
Figure 5. Triple-decker motif in mRojoA-YChroY. The inset shows the key molecular groups from another perspective.
Table 3. Structural Parameters in mRojoA, mRojoA-VYGV, and mRojoA-YChroYa
structure method Tyr197−Chro distance, Å Tyr197−Chro angle, deg Tyr197−Chro distance, Å Tyr63−Chro angle, deg
aThe tyrosine−chromophore distance is defined as the centroid-to-centroid distance between the corresponding benzene rings, and the Tyr−Chroangle is defined as the angle between the two normal vectors to the benzene rings.
Figure 6. Results of QM/MM MD calculations for mRojoA-YChroY(left panel) and mRojoA-VYGV (right panel). Initial distortedstructures are shown in colored thick sticks. Bunches of orange thinsticks in both panels illustrate the final configurations.
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
although the optimization converged to a slightly higher (by 3kcal/mol) local minimum.Transition Energies. To compute the main features of
optical spectra of the new variant with the triple-decker motifmRojoA-YChroY we applied the same protocol as for mCherry,mRojoA and mRojoA-VYGV. Namely, we first computedvertical excitation energies S0,min→S1 for the experimentallystudied systems mCherry, mRojoA and mRojoA-VYGV, andthen used the same computational protocols (TD-DFT, SOS-CIS(D) and XMCQDPT2 in QM, as explained in Models andMethods) for mRojoA-YChroY. The convergence properties ofXMCQDPT2 with respect to the expansion of the effectiveHamiltonian is described in section S2 in the SupportingInformation. We note that the calculations of the oscillatorstrengths identify S1 as the bright state in all systems. Table 4collects the computed and experimental values for theexcitation energies S0,min→S1, the corresponding wavelengths,and the shifts relative to mCherry.As one can see, XMCQDTP2 excitation energies agree well
with observed absorption band maxima in mCherry, mRojoA,and mRojoA-VYGV. We note that errors in calculations of theabsolute electronic excitation energies for the GFP-typechromophores typically exceed 0.1 eV.27−30 Here, thediscrepancies between the experimental and XMCQDT2 valuesare fairly small (as well as in other organic chromophores, e.g.,ref 30). As expected, TD-DFT produces blue-shifted excitationenergies. The results of SOS-CIS(D) are closer to theexperimental values than those of TD-DFT, but are slightlyinferior to XMCQDPT2. Importantly, despite the discrepanciesin the absolute values of the excitation energies, all methods arein agreement regarding the sign and the magnitude of the shiftrelative to the parent mCherry.
Table 4 shows the results for the mRojoA-YChroY variant.All methods predict that this mutant should have red-shiftedabsorption, as compared to mCherry and mRojoA. However,the magnitude of the shift varies slightly, from 0.1 to 0.3 eV; thelargest red shift being predicted by XMCQDPT2.Electron density difference maps are much more complex
(Figure 7) in case of the molecular clusters mimicking proteinsthan in the prototype model systems (Figure 4). The mCherryvariant has the largest S0,min→S1 excitation energy andtransition energies in mRojoA and mRojoA-VYGV arecomparable. The electron density decrease is similar on Cα2and O2 in mRojoA and mRojoA-VYGV and much morepronounced in the mCherry model. Also, the electron densityincrease is larger on Cα1 and N1 in mCherry than in the twoother models. If we consider mRojoA-YChroY, there is almostno electron density decrease on O2, a small decrease on Cα2,and no electron density increase in the Cα1-N1 region.As explained in the Models and Methods section, we located
the minimum-energy structure on the excited-state potentialenergy surface (S1) for the new variant mRojoA-YChroY byconsidering a large molecular cluster cut out from the entireprotein model system. We computed the S1,min structure usingthe SA(2)-CASSCF(12/12) with the energy gradients of thesecond state (which has correct orbital character). To validatethe computational protocol we compare the changes instructural parameters of the electronically excited chromophoreto those of the ground state (Figure 8). We can also comparethese changes with the published results for the DsRedchromophore in the gas phase (the values in parentheses inFigure 8).31 We conclude that the changes in the GFP-typechromophore structures upon excitation5,31 are well repro-duced.We computed the S1,min→S0 vertical transition energy
following the same XMCQDPT2 protocol as for the S0,min→S1 excitation. The computed energy difference is 1.87 eV, andthe corresponding wavelength of the mRojoA-YChroYemission is 662 nm, showing the Stocks shift of about +20 nm.We conclude this section by noting that our results
demonstrate that π-stacking of the nearly planar chromophorewith the sandwiching aromatic groups leads to the desired redshifts of optical bands. The use of π-stacking of thechromophore phenolate group with tyrosine or other aromaticresidues is an effective strategy toward the design of fluorescentproteins with the longer absorption/emission wave-lengths.2,3,7−9,32−34
Recently, an efficient RNA mimics of green fluorescentprotein have been developed by introducing the fluorinatedGFP chromophore variant in double-stranded RNA environ-ment in spinach aptamer.35 The stacking interaction of thechromophore with the adjacent guanines in the G-quadruplexregion is one of the stabilizing factors in chromophore−RNAcomplex, which is responsible for optical properties. Simu-lations of electronic transitions in model complexes revealed an
Table 4. Excitation Energies in eV, the Corresponding Wavelengths in nm (in Parentheses), and the Shifts Relative to mCherry
ωB97X-D SOS-CIS(D) XMCQDPT2 experimental
method model S0→S1 shift S0→S1 shift S0→S1 shift S0→S1 shift
Figure 7. Electron density difference maps for the S0,min→S1 transitionin (a) mCherry, (b) mRojoA, (c) HBDI mRojoA-VYGV, and (d)mRojoA-YChroY. Pink indicates the increase of the electron density inthe S1 state, and violet indicates the decrease. The contour value is0.002609.
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
important role of π-stacking in shifting optical bands whilepointing out to other contributions as well.36,37
Another example illustrating that the optical shifts inmodified fluorophores are due to subtle interplay of variouscontributions38 can be found in the work by Chica and co-workers.2 The authors were able to design mRojoA, a proteinwith notably red-shifted optical bands from mCherry byintroducing 6 mutations, including the Ile197Tyr replacement;however, single Ile197Tyr mutation was considerably lesseffective, showing that other factors beyond π-stacking also playa role.
■ CONCLUSIONSHere we show that the triple-decker motif stabilized by π-stacking interaction between the aromatic rings of thechromophore and two nearby tyrosine residues sandwichingthe chromophore can be achieved in the red fluorescentproteins. We arrived to such a construct by properly modifyingthe structure of mRojoA-VYGV. Molecular dynamics simu-lations with the QM/MM potentials demonstrate that thearrangement of Tyr residues around the chromophore insidethe protein is stable. Importantly, we applied a uniform strategyto characterize computationally the structures of a series ofrelated protein models, mCherry, mRojoA, and mRojoA-VYGV, with the known crystal structures, and used the sameprotocol for the newly designed species, mRojoA-YChroY. Wealso used a uniform computational protocol to compute thetransitions between the electronic states. Our calculations usingquantum chemistry methods demonstrate an excellent agree-ment between the computed S0,min→S1 excitation energies forthe protein-bound chromophores and the band maxima inabsorption spectra in mCherry, mRojoA, and mRojoA-VYGV.This agreement lends support to our results for the designedvariant mRojoA-YChroY, which we characterized computation-ally using precisely the same protocol as for its precursors. Thecalculations predict that the mRojoA-YChroY protein with thetriple-decker chromophore unit should exhibit red-shiftedoptical bands relative to its precursors without π-stackedaromatic fluorophores.
■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.jpcb.7b07517.
Structures of the mRojoA mutants containing Tyr atposition 63 and convergence of the XMCQDPT2calculations (PDF)Video file 1 of the QM/MM MD trajectories (AVI)Video file 2 of the QM/MM MD trajectories (AVI)
■ AUTHOR INFORMATIONCorresponding Author*(A.V.N.) Telephone: +7-495-939-10-96. E-mail: [email protected]; [email protected] G. Khrenova: 0000-0001-7117-3089Anna I. Krylov: 0000-0001-6788-5016Alexander V. Nemukhin: 0000-0002-4992-6029NotesThe authors declare no competing financial interest.
■ ACKNOWLEDGMENTSWe thank Alexander Granovsky for helpful discussions oncomputational protocols. This work was supported by theRussian Science Foundation (Project No. 17-13-01051, M.G.K.,I.V.P., B.L.G.) and the U.S. National Science Foundation(CHE-1566428, A.I.K.). The research is carried out using theequipment of the shared research facilities of HPC computingresources at Lomonosov Moscow State University39 and of theJoint Supercomputer Center of the Russian Academy ofSciences.
■ REFERENCES(1) Nienhaus, K.; Ulrich Nienhaus, G. Fluorescent Proteins for Live-Cell Imaging with Super-Resolution. Chem. Soc. Rev. 2014, 43 (4),1088−1106.(2) Chica, R. A.; Moore, M. M.; Allen, B. D.; Mayo, S. L. Generationof Longer Emission Wavelength Red Fluorescent Proteins UsingComputationally Designed Libraries. Proc. Natl. Acad. Sci. U. S. A.2010, 107 (47), 20257−20262.(3) Eason, M. G.; Damry, A. M.; Chica, R. A. Structure-GuidedRational Design of Red Fluorescent Proteins: Towards DesignerGenetically-Encoded Fluorophores. Curr. Opin. Struct. Biol. 2017, 45,91−99.(4) Subach, F. V.; Verkhusha, V. V. Chromophore Transformationsin Red Fluorescent Proteins. Chem. Rev. 2012, 112 (7), 4308−4327.(5) Bravaya, K. B.; Grigorenko, B. L.; Nemukhin, A. V.; Krylov, A. I.Quantum Chemistry Behind Bioimaging: Insights from Ab InitioStudies of Fluorescent Proteins and Their Chromophores. Acc. Chem.Res. 2012, 45 (2), 265−275.(6) Acharya, A.; Bogdanov, A. M.; Grigorenko, B. L.; Bravaya, K. B.;Nemukhin, A. V.; Lukyanov, K. A.; Krylov, A. I. PhotoinducedChemistry in Fluorescent Proteins: Curse or Blessing? Chem. Rev.2017, 117 (2), 758−795.(7) Wachter, R. M.; Elsliger, M. A.; Kallio, K.; Hanson, G. T.;Remington, S. J. Structural Basis of Spectral Shifts in the Yellow-Emission Variants of Green Fluorescent Protein. Structure 1998, 6(10), 1267−1277.(8) Griesbeck, O.; Baird, G. S.; Campbell, R. E.; Zacharias, D. A.;Tsien, R. Y. Reducing the Environmental Sensitivity of YellowFluorescent Protein. J. Biol. Chem. 2001, 276 (31), 29188−29194.(9) Pandelieva, A. T.; Baran, M. J.; Calderini, G. F.; McCann, J. L.;Tremblay, V.; Sarvan, S.; Davey, J. A.; Couture, J.-F.; Chica, R. A.Brighter Red Fluorescent Proteins by Rational Design of Triple-Decker Motif. ACS Chem. Biol. 2016, 11 (2), 508−517.(10) Shaner, N. C.; Campbell, R. E.; Steinbach, P. A.; Giepmans, B.N. G.; Palmer, A. E.; Tsien, R. Y. Improved Monomeric Red, Orangeand Yellow Fluorescent Proteins Derived from Discosoma Sp. RedFluorescent Protein. Nat. Biotechnol. 2004, 22 (12), 1567−1572.
Figure 8. Selected structural parameters (Å) of Chro optimized in thechromophore-containing pocket in the ground state (S0,min), shown inblue, and in the excited state (S1,min), shown in red. The values inparentheses refer to the calculation results for the isolated DsRedchromophore.31
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
(11) Grigorenko, B. L.; Nemukhin, A. V.; Polyakov, I. V.; Krylov, A. I.Triple-Decker Motif for Red-Shifted Fluorescent Protein Mutants. J.Phys. Chem. Lett. 2013, 4 (10), 1743−1747.(12) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid,E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kale, L.; Schulten, K. ScalableMolecular Dynamics with NAMD. J. Comput. Chem. 2005, 26 (16),1781−1802.(13) Vanommeslaeghe, K.; Hatcher, E.; Acharya, C.; Kundu, S.;Zhong, S.; Shim, J.; Darian, E.; Guvench, O.; Lopes, P.; Vorobyov, I.;et al. CHARMM General Force Field: A Force Field for Drug-likeMolecules Compatible with the CHARMM All-Atom AdditiveBiological Force Fields. J. Comput. Chem. 2009, 31 (4), 671−690.(14) Valiev, M.; Bylaska, E. J.; Govind, N.; Kowalski, K.; Straatsma,T. P.; Van Dam, H. J. J.; Wang, D.; Nieplocha, J.; Apra, E.; Windus, T.L.; et al. NWChem: A Comprehensive and Scalable Open-SourceSolution for Large Scale Molecular Simulations. Comput. Phys.Commun. 2010, 181 (9), 1477−1489.(15) Adamo, C.; Barone, V. Toward Reliable Density FunctionalMethods without Adjustable Parameters: The PBE0Model. J. Chem.Phys. 1999, 110 (13), 6158.(16) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent andAccurate Ab Initio Parametrization of Density Functional DispersionCorrection (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010,132 (15), 154104.(17) 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.(18) Humphrey, W.; Dalke, A.; Schulten, K. VMD: Visual MolecularDynamics. J. Mol. Graphics 1996, 14 (1), 33−38.(19) VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.;Chassaing, T.; Hutter, J. Quickstep: Fast and Accurate DensityFunctional Calculations Using a Mixed Gaussian and Plane WavesApproach. Comput. Phys. Commun. 2005, 167 (2), 103−128.(20) Bravaya, K. B.; Khrenova, M. G.; Grigorenko, B. L.; Nemukhin,A. V.; Krylov, A. I. Effect of Protein Environment on ElectronicallyExcited and Ionized States of the Green Fluorescent ProteinChromophore. J. Phys. Chem. B 2011, 115 (25), 8296−8303.(21) Granovsky, A. A. Extended Multi-Configuration Quasi-Degenerate Perturbation Theory: The New Approach to Multi-StateMulti-Reference Perturbation Theory. J. Chem. Phys. 2011, 134 (21),214113.(22) Granovsky, A. A. Firefly version 8, http://classic.chem.msu.su/gran/firefly/index.html; last accessed July 2017.(23) Rhee, Y. M.; Head-Gordon, M. Scaled Second-OrderPerturbation Corrections to Configuration Interaction Singles:Efficient and Reliable Excitation Energy Methods. J. Phys. Chem. A2007, 111 (24), 5314−5326.(24) Shao, Y.; Gan, Z.; Epifanovsky, E.; Gilbert, A. T. B.; Wormit, M.;Kussmann, J.; Lange, A. W.; Behn, A.; Deng, J.; Feng, X.; et al.Advances in Molecular Quantum Chemistry Contained in the Q-Chem 4 Program Package. Mol. Phys. 2015, 113 (2), 184−215.(25) Chai, J.-D.; Head-Gordon, M. Long-Range Corrected HybridDensity Functionals with Damped Atom−atom Dispersion Correc-tions. Phys. Chem. Chem. Phys. 2008, 10 (44), 6615.(26) Neese, F. The ORCA Program System. Wiley Interdiscip. Rev.Comput. Mol. Sci. 2012, 2 (1), 73−78.(27) Georgieva, I.; Aquino, A. J. A.; Trendafilova, N.; Lischka, H.High-Level Ab Initio Absorption Spectra Simulations of Neutral,Anionic and Neutral+ Chromophore of Green Fluorescence ProteinChromophore Models in Gas Phase and Solution. Photochem.Photobiol. 2017, DOI: 10.1111/php.12778.(28) Pirojsirikul, T.; Gotz, A. W.; Weare, J.; Walker, R. C.; Kowalski,K.; Valiev, M. Combined Quantum-Mechanical Molecular MechanicsCalculations with NWChem and AMBER: Excited State Properties ofGreen Fluorescent Protein Chromophore Analogue in AqueousSolution. J. Comput. Chem. 2017, 38 (18), 1631−1639.(29) Das, A. K.; Hasegawa, J.-Y.; Miyahara, T.; Ehara, M.; Nakatsuji,H. Electronic Excitations of the Green Fluorescent Protein
Chromophore in Its Protonation States: SAC/SAC-CI Study. J.Comput. Chem. 2003, 24 (12), 1421−1431.(30) Martin, M. E.; Negri, F.; Olivucci, M. Origin, Nature, and Fateof the Fluorescent State of the Green Fluorescent ProteinChromophore at the CASPT2//CASSCF Resolution. J. Am. Chem.Soc. 2004, 126 (17), 5452−5464.(31) Olsen, S.; Smith, S. C. Radiationless Decay of Red FluorescentProtein Chromophore Models via Twisted Intramolecular Charge-Transfer States. J. Am. Chem. Soc. 2007, 129 (7), 2054−2065.(32) Strack, R. L.; Hein, B.; Bhattacharyya, D.; Hell, S. W.; Keenan,R. J.; Glick, B. S. A Rapidly Maturing Far-Red Derivative of DsRed-Express2 for Whole-Cell Labeling. Biochemistry 2009, 48 (35), 8279−8281.(33) Lin, M. Z.; McKeown, M. R.; Ng, H.-L.; Aguilera, T. A.; Shaner,N. C.; Campbell, R. E.; Adams, S. R.; Gross, L. A.; Ma, W.; Alber, T.;Tsien, R. Y. Autofluorescent Proteins with Excitation in the OpticalWindow for Intravital Imaging in Mammals. Chem. Biol. 2009, 16 (11),1169−1179.(34) Shen, Y.; Chen, Y.; Wu, J.; Shaner, N. C.; Campbell, R. E.Engineering of mCherry Variants with Long Stokes Shift, Red-ShiftedFluorescence, and Low Cytotoxicity. PLoS One 2017, 12 (2),e0171257.(35) Paige, J. S.; Wu, K. Y.; Jaffrey, S. R. RNA Mimics of GreenFluorescent Protein. Science (Washington, DC, U. S.) 2011, 333 (6042),642−646.(36) Li, X.; Chung, L. W.; Li, G. Multiscale Simulations on SpectralTuning and the Photoisomerization Mechanism in Fluorescent RNASpinach. J. Chem. Theory Comput. 2016, 12 (11), 5453−5464.(37) Bose, S.; Chakrabarty, S.; Ghosh, D. Electrostatic Origin of theRed Solvatochromic Shift of DFHBDI in RNA Spinach. J. Phys. Chem.B 2017, 121 (18), 4790−4798.(38) Park, J. W.; Rhee, Y. M. Emission Shaping in FluorescentProteins: Role of Electrostatics and π-Stacking. Phys. Chem. Chem.Phys. 2016, 18 (5), 3944−3955.(39) Voevodin, V. V.; Zhumatiy, S. A.; Sobolev, S. I.; Antonov, A. S.;Bryzgalov, P. A.; Nikitenko, D. A.; Stefanov, K. S.; Voevodin, V. V.Practice of “Lomonosov” Supercomputer. Open Syst. J. 2012, 7, 36−39.
The Journal of Physical Chemistry B Article
DOI: 10.1021/acs.jpcb.7b07517J. Phys. Chem. B 2017, 121, 10602−10609
