a triple mutant F64L I167T K238N of GFP was determined at 0.90 ˚ resolution
(Protein Data Bank entry 2WUR) that allowed for detection of 320 water mole-
cules, compared to 105 waters at 2.30 ˚
resolution (1 EML) and 74 at 2.50 ˚
resolution (1 EME) [ 20 ].
At first, there seems no necessity to involve further residues in proton transfer
apart from those involved in the multistep proton shuttle between the buried
chromophore and Glutamate Glu222 that is part of the fluorescence cycle of wild-
type GFP. The only time when proton transfer is functionally required is during
chromophore biosynthesis when protons need to be transferred to the protein
surface. NMR dispersion relaxation experiments of Halle and coworkers showed
that three internal buried water molecules in the bovine trypsin inhibitor exchange
with bulk solvent on time scales of milliseconds [ 21 ]. Therefore, the natural thermal
dynamics of GFP should also allow for the transfer of these protons to the protein
surface even more since the duration of this one-time process is not crucial.
However, Agmon and coworkers [ 22 , 23 ] identified an extensive proton wire
that connects the terminal chromophore residue Tyr66 via Glu222 to the surface of
the protein. The new data now allowed for a detailed analysis of this wire in GFP on
the basis of the ultra-high-resolution crystal structure [ 20 ]. The authors considered
all backbone, side chain, and water oxygen atoms as well as the protein nitrogen
atoms for the analysis of putative PT pathways. All O and N atoms within 3.0 ˚
distance were considered connected. It turned out that the GFP barrel contains two
large internal clusters on both sides of the chromophore throughout the entire
protein. Figure 2 shows one of these hydrogen-bonded clusters of putative pro-
ton-donating and -accepting groups. This “active-site wire” that connects to the
hydroxyl oxygen of Tyr66 of the chromophore contains 147 atoms in the 2WUR
crystal structure. Of these, 75 are protein atoms and 72 are water oxygens. This wire
contains two exit segments on partly hydrophobic parts of the protein surface.
According to the proton-antenna model, such surface patches are likely not
involved in capturing protons from solution. A similar reasoning was put forward
for the short Thr203-His148 exit [ 22 ]. A third exit connects the active-site wire to a
cluster of neighboring carboxylate residues (Glu5, Asp36, Glu34, Asp117, Glu6) on
the GFP surface. As discussed before, such an arrangement leads to a negative
electrostatic potential on the protein surface that should thus be able to capture
protons from solution.
The second large internal cluster (not shown in Fig. 2 ) located on the other face
of the chromophore is disconnected from the first cluster and from the protein
exterior. This wire does not seem to be involved in ESPT. It contains some key
residues near the chromophore such as His181, Thr62, Thr108, Tyr145, and Arg96.
Some of these residues are thought to be involved in chromophore biosynthesis.
Consequently, this cluster was termed the “biosynthesis cluster”. If Arg96 gets
temporarily deprotonated during the dehydration step, this wire may serve as a
proton storage. This is reminiscent of the situation at the exit pathway of the proton
pump bacteriorhodopsin for which Gerwert and coworkers have argued that several
buried water molecules that are held in place by two close-by carboxylate residues
function as a proton storage [ 24 ].