Bathochromic shift in green fluorescent protein: A puzzle for QM/MM approaches¶
Authors: Claudia Filippi, Francesco Buda, Leonardo Guidoni, Adalgisa Sinicropi
DOI: 10.1021/ct200704k
Takeaways:
Abstract¶
We present an extensive investigation of the vertical excitations of the anionic and neutral forms of wild-type green fluorescent protein using time-dependent density functional theory (TDDFT), multiconfigurational perturbation theory (CASPT2), and quantum Monte Carlo (QMC) methods within a quantum mechanics/molecular mechanics (QM/MM) scheme. The protein models are constructed via room-temperature QM/MM molecular dynamics simulations based on DFT and are representative of an average configuration of the chromophore–protein complex. We thoroughly verify the reliability of our structures through simulations with an extended QM region, different nonpolarizable force fields, as well as partial reoptimization with the CASPT2 approach. When computing the excitations, we find that wave function as well as density functional theory methods with long-range corrected functionals agree in the gas phase with the extrapolation of solution experiments but fail in reproducing the bathochromic shift in the protein, which should be particularly significant in the neutral case. In particular, while all methods correctly predict a shift in the absorption between the anionic and neutral forms of the protein, the location of the theoretical absorption maxima is significantly blue-shifted and too close to the gas-phase values. These results point to either an intrinsic limitation of nonpolarizable force-field embedding in the computation of the excitations or to the need to explore alternative protonation states of amino acids in the close vicinity of the chomophore.
Introduction¶
In this paper, we employ a variety of electronic structure approaches to compute the vertical excitations of the neutral A and anionic B forms of wild-type GFP. These are the protonation states of the chromophore responsible for the two room-temperature absorption peaks at 398 nm (3.12 eV) and 478 nm (2.59 eV), respectively.
Therefore, our results point at two possible sources of error still largely unexplored, namely, that some amino acids in the protein have a different protonation than commonly accepted or that the protein environment must be described beyond nonpolarizable force fields in the computation of the excitations.
Methods¶
We also investigate the stability of alternative protonation states of the chromophore and surrounding environment via additional QM/MM simulations. We test the possible stability of a solvated hydronium in proximity of the chromophore following the experimental suggestions of ref 42 as well as the existence of the zwitterionic and cationic forms in the presence of a protonated and deprotonated Glu222 residue. In these simulations, the QM region is extended to include the anionic chromophore; the side chains of Thr203, Ser205, and Glu222; and the five water molecules closest to the chromophore. One water molecule is protonated when investigating the stability of a hydronium.
Vertical Excitation Energies¶
Discussion¶
Structural Analysis of GFP Models¶
The structure of the chromophore is expected to play a very important role in tuning its excited state properties. In particular, the degree of bond-length alternation in the conjugate chain running through the chromophore will be correlated to the size of the excitation, which is here of π to π* character. In addition, the local features of the binding site of the chromophore can affect the spectral response of the chromophore by either tuning the internal geometrical structure of the chromophore or as a polarizing environment.