Proton Travel in Green Fluorescent Protein - Fluorescent Proteins: From Understanding to Design - page 172

conformations along which a proton may be translocated [ 26 ]. They applied this
algorithm to the proton transfer from the surface of cytochrome P450cam to its
buried active site and to carbonic anhydrase. We have implemented their algorithm
in a software termed QVADIS, where we also considered the kinetic barriers for
proton transfer in the Q-HOP framework [ 27 ]. Alternatively, Ullmann and cow-
orkers have considered the degree of evolutionary conservation combined with
analysis of hydrogen-bonding clusters in photosynthetic reaction centers [ 28 ].
Besides these approaches based on static conformations, also dynamic simula-
tion techniques have been developed to model PT pathways based on the molecular
dynamics technique [ 2 ]. Generally, molecular dynamics simulations have become a
standard tool for characterizing the native protein dynamics on time scales from
femtoseconds to microseconds [ 29 ]. For GFP, several authors have characterized
the dynamics of the stiff protein matrix and the flexible internal water hydrogen
bond network by molecular dynamics simulations [ 30 - 32 ]. On a nanosecond time
scale, several water molecules near the chromophore were tightly locked into
favorable coordinations, whereas others could almost freely rotate and alternate
between alternative roles as hydrogen bond donors and acceptors [ 30 ]. These
standard simulation techniques need to be modified to allow for bond breaking
and formation that are required to describe PT reactions.
According to the “proton-antenna” model discussed by Gutman [ 8 ], patches of
negatively charged residues on the surface of proteins may mark entry points for
protons from bulk solution. We therefore compared the proton capturing tendencies
of imidazole and acetic acid as side-chain analogs of histidine and of aspartic acid by
our Q-HOP molecular dynamics method [ 33 ]. We found a qualitatively different
protonation behavior of 4-methylimidazole compared to that of acetic acid. On one
hand, deprotonated, neutral 4-methylimidazole cannot as easily attract a freely diffus-
ing extra proton from solution. Once the proton is bound, however, it remains tightly
bound on a time scale of tens of nanoseconds. In a linear chain composed of acetic acid,
a separating water molecule, and 4-methylimidazole, an excess proton is equally
shared between 4-methylimidazole and water. When a water molecule is linearly
placed between two acetic acid molecules, the excess proton is always found on the
central water. On the other hand, an excess proton in a 4-methylimidazole-water-4-
methylimidazole chain is always localized on one of the two 4-methylimidazoles. This
suggests that aspartic and glutamic acid function as attractors for protons fromsolution,
whereas histidine residues may function as temporal reservoirs of proton storage.
We then mimicked the antenna effect of Gutman by a model system consisting
of two transmembrane helices connected by a short loop fragment. 1-3 Asp
residues were engineered into this loop and the helix-loop-helix fragment was
embedded in a DOPC lipid bilayer. During molecular dynamics simulations with
our hoppable proton model Q-HOP MD, we initially started all aspartate residues in
the deprotonated form and placed an excess proton in bulk solution. Via normal
Grotthuss-type diffusion, it is quite likely that the proton enters into the electrostatic
capture radius of the negatively charged aspartates. Then, it may hop to one of the
aspartates within a few picoseconds via a hydrogen-bonded water wire. We termed
these wires “tightly connected water wires” (TCW) [ 34 ].