Towards molecular-level characterization of photo-induced decarboxylation of the green fluorescent protein: Accessibility of the charge-transfer states
B. Grigorenko, A.V. Nemukhin, D.I. Morozov, I. Polyakov, K.B. Bravaya, and A.I. Krylov
J. Chem. Theor. Chem. 8, 1912 – 1920
Irradiation of the green fluorescent protein (GFP) by intense violet or UV light leads to decarboxylation of the Glu222 side chain in the vicinity of the chromophore (Chro). This phenomenon is utilized in optical highlighters such as photoactivatable GFP (PA-GFP). Using state- of-the-art quantum-chemical calculations we investigate the feasibility of the mechanism proposed in the experimental studies [van Thor et al., Nature Struct. Biol. 2002, 9, 37–41; Bell et al., J. Am. Chem. Soc. 2003, 125, 37-41]. It was hypothesized that a primary event of this photoconversion involves population of a charge-transfer (CT) state, either via the first excited state S1 when using longer wavelength (404 nm and 476 nm) or via a higher excited state when using higher-energy radiation (254 nm and 280 nm). Following the results of electronic structure calculations we identify these critical CT states (produced by electron transfer from Glu to electronically excited Chro) and show that they are accessible via different routes, i.e., either directly, by one-photon absorption, or through a two-step excitation via S1. The calculations are performed for model systems representing the chromophore and the key nearby residues using two complimentary approaches: (i) the multiconfigurational quasidegenerate perturbation theory of second order (MCQDPT2) with the ORMAS scheme for configuration selection in the multiconfigurational self-consistent field (MCSCF) reference; (ii) the single-reference configuration interaction singles (CIS) method with perturbative doubles (SOS-CIS(D)) that does not involve active space selection. We examined electronic transitions with non-zero oscillator strengths in the UV and visible range between the electronic states involving the Chro and Glu residues. Both methods predict the existence of CT states with non-zero oscillator strength in the UV range, and a local excited state of the chromophore accessible via S1 that may lead to the target CT state. The results suggest several possible scenarios for the primary photoconversion event. We also demonstrate that the point mutation Thr203His exploited in PA-GFP results in shifting the light wavelength to access the CT up to 20 nm, which suggests a possibility of a rational design of photoactivatable proteins in silico.
Download this paper (PDF, NaN kB)
Understanding photoactive proteins in gas phase and in realistic environments