Primary Photophysical Processes in Chromoproteins - Fluorescent Proteins: From Understanding to Design - page 55

suggests that such long-range proton transport (along a preformed wire) can be very
fast, as intermediate states in the chain were not detected. It has also proved possible to
modify the proton wire through mutagenesis. Single mutations can disrupt the proton
wire, trapping the chromophore in the A* state, which has a shorter lifetime than the I*
state [ 104 ]. It has also been shown that a second mutation, H148D, which places a
proton acceptor adjacent to the donor gives rise to a new proton-transfer reaction, presu-
mably to the asparagine carboxylate group [ 105 - 108 ]. This redirected proton transfer
has been shown to be extremely fast (

100 fs) [ 109 ]. It was suggested that this repre-
sents a case of proton transfer over a low barrier or barrierless hydrogen bond. Such low
barrier H-bonds are implicated in a number of key enzyme reactions [ 110 , 111 ].
Progress towards understanding the observed protein proton-transfer dynamics
requires theoretical modelling of the observed transient behaviour. Although accu-
rate quantum chemical calculations of excited states in proteins remain challenging,
there has been significant progress in modelling ESPT in avGFP. In an early work,
Lill and Helms used classical MD to simulate the proton transfer along the three-
step proton wire [ 112 ]. They concluded that after transfer was triggered by ejection
of the proton from the chromophore, the steps leading to protonation of E222 occur
on the tens of femtosecond timescale. Subsequently, quantum chemical calcula-
tions have been reported by two groups using the geometry of the proton-transfer
chain suggested by the protein structure, but in the absence of surrounding residues
[ 113 - 116 ]. Both groups calculated that the proton transfer occurred in a single
concerted step along a low barrier potential surface, with no stable intermediate
states. These calculations also suggested that the first proton to move in the
concerted process was the last in the chain (i.e., the proton protonating E222).
Zhang et al. found that the H148 residue, which is not part of the proton-transfer
chain but is H-bonded to the donor O atom, had a significant impact on the potential
surface, suggesting an important role for the surrounding residues [ 116 ].
Lluch and co-workers used molecular dynamics to study the structure and
stability of the proton relay chain in avGFP, and performed quantum chemical
calculations on a reduced set of residues, using the geometries obtained from the
MD simulations [ 117 - 119 ]. They investigated the proton-transfer surface for both
ground and excited states of the chromophore, and found that the photoactive state
was

<

pp ps

*. The potential energy surfaces were calculated to have
minima for the proton localised on the chromophore in the ground state and on
E222 in the excited state, in agreement with experiment. The ESPT has a small
barrier (ca. 2 kcal mol 1 ) and is strongly downhill for the S203 to E222 step, as
found in earlier calculations. This is consistent with a concerted but asynchronous
proton transfer, with the last proton 'leading'.
The majority of simulations suggest an effectively barrierless ultrafast ESPT in
the avGFP geometry, whereas experimental results show transfer on a picosecond
or tens of picoseconds timescale. Fang et al. recently reported a femtosecond-
stimulated Raman study of the first few picoseconds of ESPT in avGFP [ 120 ].
Their results suggest that fluctuations in the structure of the chromophore may
modulate the proton-transfer rate, and thus the frequency at which such fluctuations
result in the formation of a barrierless proton-transfer potential may control the

* rather than the