An alternative excited-state proton transfer pathway in green fluorescent protein variant S205V

Abstract

Wild-type green fluorescent protein (wt-GFP) has a prominent absorbance band centered at ∼395 nm, attributed to the neutral chromophore form. The green emission arising upon excitation of this band results from excited-state proton transfer (ESPT) from the chromophore hydroxyl, through a hydrogen-bond network proposed to consist of a water molecule and Ser205, to Glu222. Although evidence for Glu222 as a terminal proton acceptor has already been obtained, no evidence for the participation of Ser205 in the proton transfer process exists. To examine the role of Ser205 in the proton transfer, we mutated Ser205 to valine. However, the derived GFP variant S205V, upon excitation at 400 nm, still produces green fluorescence. Time-resolved emission spectroscopy suggests that ESPT contributes to the green fluorescence, and that the proton transfer takes place ∼30 times more slowly than in wt-GFP. The crystal structure of S205V reveals rearrangement of Glu222 and Thr203, forming a new hydrogen-bonding network. We propose this network to be an alternative ESPT pathway with distinctive features that explain the significantly slowed rate of proton transfer. In support of this proposal, the double mutant S205V/T203V is shown to be a novel blue fluorescent protein containing a tyrosine-based chromophore, yet is incapable of ESPT. The results have implications for the detailed mechanism of ESPT and the photocycle of wt-GFP, in particular for the structures of spectroscopically identified intermediates in the cycle.

Excitation of either of the two absorption bands of wild-type green fluorescent protein (wt-GFP) leads to green fluorescence. The major band at ∼395 nm, referred to as band A, is due to absorption of the neutral chromophore; while the minor band at ∼475 nm, referred to as band B, is due to absorption of the deprotonated anionic chromophore (Chattoraj et al. 1996). The green fluorescence upon excitation of band A is due to deprotonation of the neutral chromophore, induced by excited-state proton transfer (ESPT) (Chattoraj et al. 1996; Lossau et al. 1996). In wt-GFP, a hallmark of this process is that the decay of blue emission matches the rise time of green emission, and both are slowed upon deuteration of the sample. Upon illumination of band A, very slow photoconversion of species A into B is observed, on the timescale of several hours, while the ESPT process is on the timescale of several tens of picoseconds.

Chattoraj et al. (1996) proposed a photocycle for wt-GFP, postulating the existence of an intermediate excited state (I*). I* was proposed to form from the excited state of the neutral chromophore (A*) after excited-state proton transfer, and to be responsible for the green emission peaking at 508 nm, different from the green emission of the excited state of the anionic chromophore (B*) with maximum at 503 nm (Chattoraj et al. 1996; Lossau et al. 1996). The conversion between I* and B* is slow and infrequent. After emission of a green photon, I* decays to ground state I, which converts to the ground state A, forming a photocycle.

Based on crystal structures of wt-GFP and GFP S65T, a hydrogen-bonding network was proposed to promote rapid proton transfer from neutral chromophore, through an internal water molecule and Ser205, to Glu222 (Fig. 1A; Brejc et al. 1997^; Palm et al. 1997). The proposed ESPT pathway is supported by ultrafast vibrational spectroscopy of wt-GFP, which demonstrates that Glu222 is the terminal proton acceptor (Stoner-Ma et al. 2005). Upon illumination, the pK _a of the protonated chromophore drops substantially as a result of n → π* charge transfer from the phenol oxygen to the phenol and imidazolinone rings (Tolbert and Solntsev 2002), which triggers the proton transfer. The resulting excited-state anion is green fluorescent.

Figure 1.

Schematic diagram (A) of the proposed excited-state proton transfer pathway in wt-GFP (PDB ID 1EMB) and energy level diagram (B) for the photocycle of wt-GFP (see Chattoraj et al. 1996^; Kennis et al. 2004).

However, this simple model is insufficient to explain all of the spectroscopic data. Excitation of band B gives green emission with maximum at 503 nm, while excitation of band A gives green emission with maximum at 508 nm (Heim et al. 1994), implying some rearrangement distinguishing I* and B*. Recently, Kennis et al. (2004) expanded this photocycle and provided pump/dump/probe spectroscopic evidence for a second intermediate ground state (I₂; see Fig. 1B). Upon excitation of the ground state (A), the excited state (A*) either emits a blue photon or converts to an intermediate excited state (I*) through excited-state proton transfer. I* then emits a green photon, producing the first intermediate ground state I₁. Additional evidence suggests that I₁ rapidly (lifetime ∼3 ps in H₂O) converts to a second intermediate ground state (I₂), from which the ground state A is regenerated (lifetime ∼400 ps in H₂O). Kinetic isotope effects (KIE) upon substitution of exchangeable protons with deuterium are reported to be ∼2 for I₁ → I₂ and ∼12 for I₂ → A, suggesting multiple proton transfer in the latter process. The proton back-shuttle process of I₂ → A was proposed to be the reverse of the ESPT pathway shown in Figure 1A (Kennis et al. 2004). The structures corresponding to states I₁ and I₂ are not known.

Here, we investigate the role of Ser205 in ESPT within wt-GFP by using mutagenesis, time-resolved emission spectroscopy and X-ray crystallography. Our results strongly support the existence of an alternative excited-state proton transfer pathway, involving Thr203 in the GFP S205V mutant and possibly also in wt-GFP. This result suggests a possible structure for the second intermediate ground state I₂ proposed by Kennis et al. (2004), as well as an alternative proton back-shuttle pathway for the I₂ → A process in the expanded photocycle of wt-GFP. Our results also suggest that excited-state proton transfer within wt-GFP is more likely to be concerted than stepwise.

Results

Steady-state spectra of GFP variants

The absorption spectrum of GFP S205V exhibits a major band (A) at 395 nm, and a minor band (B) at 490–500 nm, which increases at pH above 10 (Fig. 2A). The chromophore pK _a could not be measured due to protein instability at high pH, but it is estimated to be above 11, consistent with the more hydrophobic environment resulting from the S205V substitution. While excitation of the B band leads to green emission with maximum at 508 nm (data not shown), excitation of the A band gives rise to weak blue emission and strong green emission peaking at 512 nm (green solid line, Fig. 2B). Furthermore, upon deuteration the blue emission increases significantly at the expense of the green emission (green dashed line, Fig. 2B). The excitation spectrum of green (512 nm) emission at pH 7.9 exhibits a major band also peaking at 395 nm (blue line, Fig. 2B). S205V has high quantum yield under excitation at 400 nm, 0.78 ± 0.01, equivalent to that of wt-GFP (Tsien 1998).

Figure 2.

Steady-state spectra of GFP S205V and S205V/T203V variants. (A) Absorption and (B) excitation and emission spectra of S205V. (C) Absorption and (D) emission spectra of S205V/T203V.

The absorbance and excitation spectra of the S205V/T203V variant (Fig. 2C) are very similar and show a single band (A), peaking at about 390 nm over a broad range of pH (5–10), which is characteristic of the neutral chromophore. At pH 5.2, this band is slightly red-shifted. S205V/T203V is blue fluorescent with the peak at 459 nm, but the emission band is very broad (Fig. 2D). As shown in Figure 2D, the fluorescence efficiency of S205V/T203V is pH dependent, decreasing at low pH (red curve, Fig. 2D). As it is extremely unlikely that the chromophore is anionic under any of the conditions tested (pK _a > 11; see above comment regarding S205V), we attribute the red shift in absorbance and the loss of fluorescence efficiency at low pH to the titration of an acidic residue, which alters the chromophore environment. The quantum yield of GFP S205V/T203V at pH 7.2 is reasonably high (0.29 ± 0.03), and is comparable to those of blue fluorescent proteins (BFPs) derived from GFP by the substitution Tyr66 → His (Tsien 1998). Thus, S205V/T203V is a novel BFP, containing a tyrosine-based chromophore.

Time-resolved emission spectra

Upon excitation at 400 nm in H₂O, S205V emits blue fluorescence (peaking at 450 nm), which has a rise time shorter than the instrument response function (IRF, ∼30 ps). The emission lifetime is ∼400 ps (open circles, upper panel, Fig. 3). In D₂O the decay of blue fluorescence is slowed significantly with emission lifetime ∼800 ps (open circles, lower panel, Fig. 3), suggesting the involvement of proton transfer.

Figure 3.

Time-resolved emission spectra of S205V. (A) Time-resolved blue (450 nm) and green (510 nm) fluorescence in H₂O and (B) D₂O upon excitation at 388 nm. The raw data points are represented by circles, and fitted curves by solid lines.

However, the kinetics of green fluorescence are more complicated. In H₂O, time-resolved emission studies (λ_det = 510 nm) reveal two amplitude components: one arising on a timescale similar to the decay of the blue fluorescence, with the other appearing instantaneously within the IRF (open circles, upper panel, Fig. 3). Upon deuteration, the amplitude of the fast component increases significantly (open circles, lower panel, Fig. 3). Therefore, in contrast to wt-GFP, in S205V green fluorescence appears to arise more quickly upon deuteration. Such behavior has been observed previously during studies on dual emission GFPs (deGFPs) (Hanson et al. 2002; McAnaney et al. 2002). In those cases, the rapidly rising amplitude component of 510 nm fluorescence could be attributed to the broad red tail of the blue emission band. That analysis may apply to S205V, in which the red tail of the blue emission band appears to extend beyond 510 nm (see Fig. 2B). In support of this argument, the blue emission band of S205V can be accurately fit to the broad blue emission of S205V/T203V (See supplemental material for supporting information [SI], Supplemental Fig. S1), which does extend significantly beyond 510 nm (Fig. 2D). Based on this assumption, the time-resolved emission data can be well fitted by an ESPT model (solid lines in Fig. 3). For details of the fitting model, the reader is referred to Supplemental scheme S1. The kinetic parameters from the ESPT model (Supplemental Table S1) indicate that the excited-state proton transfer rate in H₂O is ∼2.4 ns⁻¹, but in D₂O the rate decreases to ∼0.55 ns⁻¹, for a KIE ∼4, which is close to that (∼5) reported for wt-GFP (Chattoraj et al. 1996; Kennis et al. 2004). From these data we conclude that the proton transfer rate for ESPT within S205V is ∼30 times slower than in wt-GFP.

Crystal structure of GFP S205V

To investigate the structural basis of green fluorescence of GFP S205V upon excitation at 400 nm, the crystal structure of this variant was solved at 1.59 Å. The variant crystallized in space group P2₁, with four monomers in the asymmetric unit. Residues 4–231 are visible in the electron density map; however, density for residues 155–158 is weak for all chains, indicating partial disorder. Data collection and refinement statistics for the atomic model, which are excellent, are presented in Supplemental Table S2. PROCHECK (Laskowski et al. 1993) reveals that there are no residues in the disallowed regions of the Ramachandran diagram. Upon superposition of the α-carbons with wt-GFP, RMS deviations are <0.30 Å for each of the four monomers in the asymmetric unit. The atomic model is thus very similar to wt-GFP, indicating that the protein matrix is insensitive toward variation at position 205.

The chromophore environment of GFP S205V (Fig. 4A,B) is similar to that of wt-GFP; however, two residues are rearranged compared to those of wt-GFP (shown in cyan, PDB ID 1EMB). First, Glu222 rotates away from Val205 and forms hydrogen bonds with Thr203 and a water molecule, possibly due to steric interference with Val205. Second, Thr203 adopts the least favored rotamer conformation with χ₁ ∼ −180° (based on the rotamer library of Coot) (Emsley and Cowtan 2004), presumably stabilized by the hydrogen bond with Glu222. In wt-GFP, the side chain of Thr203 is partially disordered (Brejc et al. 1997) and occupies the two more favored rotamer configurations (χ₁ ∼ +60 and −60°). In one conformation, the Thr203 hydroxyl forms a hydrogen bond with the anionic chromophore of wt-GFP (both colored cyan in Fig. 4A). Despite the rearrangement of Glu222 and Thr203, a bridging water molecule (hydrogen bonded with the chromophore hydroxyl) occupies almost the same position as one found in wt-GFP. This presumed water is also hydrogen bonded with carbonyl oxygens of Asn146 and Thr203. However, the relatively long distance (3.3 Å, averaged over four monomers) between the bridging water molecule and Thr203 hydroxyl oxygen leads us to conclude that, on average, either there is no hydrogen bond between the water molecule and the threonine hydroxyl, or it is weak.

Figure 4.

Chromophore environment of GFP S205V variant. (A) Stereo pair depicts rearrangement of residues Glu222 and Thr203 of GFP S205V (green), compared to those of wt-GFP (magenta). Hydrogen bonds are shown by dashed lines. (B) (2F_o − F_c) Electron density map contoured at 1 σ level. Shown are superimposed ball-and-stick models of the chromophore, two water molecules, and the side chains of Thr203, Glu222, and Val205.

Discussion

Despite the recent discovery of a great variety of fluorescent proteins from marine sources and the generation of novel red fluorescent proteins by directed evolution (Shaner et al. 2004; Shu et al. 2006), the GFP from Aequorea victoria remains unique. At present, it is the only biological system for which ESPT has been well characterized and the molecular structure, together with the proposed excited-state proton transfer pathways, has been determined at atomic resolution. Thus, GFP constitutes an ideal system for study of intramolecular proton transfer.

We tested the role of Ser205 in the proposed ESPT pathway by mutating it to valine, which cannot support proton transfer. However, time-resolved emission spectroscopy and deuterium kinetic isotope effects provided strong evidence that upon excitation at 400 nm, the green (510 nm) fluorescence of S205V, is nevertheless due to excited-state proton transfer, albeit at a much slower rate. The crystal structure analysis of S205V suggested an alternative ESPT pathway involving the chromophore hydroxyl, a water molecule, and Thr203, which is effectively eliminated by substitution of Thr203 with valine. The 30-fold slower rate observed for proton transfer within S205V is attributed to a very long hydrogen bond between Thr203 and the water molecule. These results allow us to speculate that proton transfer within GFP occurs via a concerted, rather than stepwise mechanism. Finally, we propose a more detailed model for the photocycle of wt-GFP.

Excited-state proton transfer

Unlike wt-GFP, the steady-state spectra of the S205V variant indicate a significant component of blue fluorescence, and upon deuteration of the sample, the intensity of blue fluorescence increases significantly at the expense of green emission. In both cases deuteration increases the decay lifetime of blue fluorescence by a factor of 4–5, suggesting that the green emitting species arises from the blue emitting species via proton transfer, as proposed for wt-GFP (Chattoraj et al. 1996; Lossau et al. 1996). This can be explained by the existence of an alternative ESPT pathway with a significantly reduced rate of proton transfer. Indeed, kinetic measurements show a blue fluorescence lifetime of ∼400 ps, in contrast to wild type, where the blue fluorescence lifetime is ∼12 ps (Chattoraj et al. 1996; Lossau et al. 1996; Kennis et al. 2004), suggesting that proton transfer within S205V is ∼30 times slower than in wt-GFP. However, it is important to note that given these long lifetimes, radiative decay of the blue emitting state competes effectively with proton transfer kinetics, so that a more detailed kinetic model for the emission is required (see Supplemental material).

In S205V, time-resolved emission at 510 nm is less straightforward to interpret, but is consistent with our analysis of the decay of blue fluorescence. As described in Results, emission at 510 nm can be resolved into two amplitude components: fast (within IRF) and slow. The fast component shows increasing amplitude upon deuteration, which is interpreted as originating from the red tail of the blue emission band of the protonated chromophore. Due to significantly slowed proton transfer rates, the protonated chromophore population, and hence the amplitude of blue emission, increases significantly upon deuteration (Fig. 2D) at the expense of the green component. Given this assumption, the ESPT model successfully fits the time-resolved 510 nm emission data, and the kinetic parameters again suggest that within S205V, ESPT is ∼30 times slower than that of wt-GFP. We now turn to the atomic model for an explanation of these effects.

Alternative ESPT pathway

The high-resolution crystal structure of GFP S205V reveals a new hydrogen-bonding network that involves the chromophore hydroxyl, a bridging water molecule, Thr203, and Glu222, formed by rearrangement of Glu222 and Thr203. This hydrogen-bonding network is similar to that in wt-GFP, involving the chromophore, a bridging water molecule at almost the same position, Ser205, and Glu222. The substitution T203V in the S205V background leads to elimination of green fluorescence and production of bright blue fluorescence, which strongly suggests a role for Thr203 in ESPT. We propose that this hydrogen-bonding network is an alternative ESPT pathway (Fig. 5).

Figure 5.

Schematic diagram of proposed alternative pathway for excited-state proton transfer in GFP S205V. The dashed line between the bridging water and Thr203 hydroxyl oxygen does not necessarily imply a hydrogen bond, because of their observed large average separation of ∼3.3 Å.

However, an important feature of the new ESPT pathway identified in S205V distinguishes it from that found in wt-GFP. The average distance from the bridging water molecule to Thr203 hydroxyl oxygen is 3.3 Å (see Fig. 5), suggest that either there is no hydrogen bond between the water molecule and the hydroxyl or the hydrogen bond is weak. In contrast, the distance between the bridging water and the hydroxyl of Ser205 in wt-GFP—2.6 Å in PDB ID 1EMB (Brejc et al. 1997), 2.7 Å in PDB ID 1GFL (Yang et al. 1996)—is normal. It is likely that the large gap between the bridging water molecule and Thr203 results in infrequent proton transfer. Furthermore, if the rate of proton transfer from the chromophore to the bridging water molecule is limited by proton transfer from the bridging water to Thr203, the overall process of proton transfer within S205V may be concerted.

Concerted versus stepwise excited-state proton transfer

Previously, in wt-GFP, the results of ultrafast vibrational spectroscopy (Stoner-Ma et al. 2005) revealed that the protonation rate of Glu222 is statistically indistinguishable from the deprotonation rate of the chromophore, as determined by ultrafast emission spectroscopy (Chattoraj et al. 1996; Lossau et al. 1996). This seems to imply that the ESPT within wt-GFP is concerted. However, theoretical studies (Lill and Helms 2002; Stoner-Ma et al. 2005) suggested that if the proton is first transferred from the chromophore to the bridging water molecule, then the subsequent proton transfer from the hydronium ion to Ser205 and from Ser205 to Glu222 should occur within 100 fs, which is much faster than the time resolution of the instrumentation. Therefore, ultrafast vibrational spectroscopy does not rule out a stepwise mechanism for ESPT. Here, we define excited-state proton transfer to be stepwise if the bridging water molecule becomes temporarily protonated and thus positively charged during the course of ESPT.

The simplest explanation for the dramatically slowed decay rate of blue fluorescence in S205V is that in both wt-GFP and S205V, excited-state proton transfer is concerted, and in the latter case, proton transfer from the bridging water to Thr203 limits the deprotonation rate of the chromophore. Conversely, if the proton transfer in both wt-GFP and S205V is stepwise, this would suggest a similar decay rate of blue fluorescence in both proteins. In either case, stepwise proton transfer seems energetically very unfavorable, because the hydronium ion formed would have a pK _a much lower than that of protonated glutamate. We conclude that the evidence favors concerted proton transfer in both wt-GFP and S205V.

Implications for the photocycle of wt-GFP

Replacement of Ser205 with valine led to obvious structural rearrangements within the hydrogen-bond network adjacent to the chromophore, presumably in order to satisfy the hydrogen bonding requirements of Glu222. This demonstrates structural flexibility on the part of conserved Glu222, which has been observed and discussed in the context of other fluorescent proteins as well as GFP (van Thor et al. 2005; Shu et al. 2006). Therefore, it is reasonable to assume that in wild-type GFP, an equivalent hydrogen-bond network can form, if energetically favorable.

On this basis, we propose detailed structures for intermediates in the expanded photocycle of wt-GFP (Fig. 6). Upon photo-excitation, the proton at the chromophore hydroxyl transfers through the bridging water molecule and Ser205 to Glu222, resulting in the deprotonation of the chromophore and protonation of Glu222. We identify this as the excited intermediate state (I*). After the emission of a green photon or through a so-called “green dump” (a stimulated emission process) which transfers I* → I₁ on a timescale of femtoseconds (Kennis et al. 2004), a rearrangement becomes energetically favorable. One possibility is for protonated Glu222 and Thr203 to rearrange slightly as seen in the S205V variant and to form a new hydrogen bond. This conformation (lower right in Fig. 6) is proposed to correspond to the second intermediate ground state (I₂). One would not expect a large KIE for this step, which is consistent with the observed KIE ∼2 for the I₁ → I₂ transition (Kennis et al. 2004). On the other hand, the reported lifetime of I₁ is only 3 ps, which may rule out motion of atoms heavier than hydrogen, that is, the rotation of Thr203 suggested in Figure 6. Thus, the I₁ → I₂ transition may simply involve rearrangement of the protons, with Thr203 rotation taking place in the second step. An example of such proton motion might be the required anti to syn flip of the newly arrived proton on Glu222 (a rearrangement not shown in Fig. 6).

Figure 6.

Schematic diagram of proposed extended photocycle of wt-GFP. The excited state (A*) of the neutral chromophore, photo-excited from the ground state (A), is transformed into an intermediate excited state (I*) (Brejc et al. 1997) after concerted ESPT, which upon emission forms the first intermediate ground state (I₁) (Kennis et al. 2004). The rearrangement of Thr203 and Glu222 (upper right) forms the second intermediate ground state (I₂) with a different hydrogen-bonding network. In the figure bond rotations are indicated by the 270° semicircles terminating with arrowheads. After the proton transfer from Glu222 to the chromophore (lower right) and rearrangement of Thr203, Glu222, the bridging water molecule and Ser205 (lower left), the ground state (A) of the neutral chromophore is regenerated.

In the final step, the protonated chromophore is regenerated (I₂ → A) by multiple proton transfer from Glu222 to the deprotonated chromophore via Thr203 and the bridging water molecule, followed by further rearrangement of Thr203, Glu222, the bridging water molecule, and Ser205. The multiple proton transfers and the extensive hydrogen-bonding rearrangements during the process of I₂ → A are consistent with the large KIE (∼12) observed for this step (Kennis et al. 2004). The second intermediate state I₂ has a much longer lifetime than I₁, ∼400 ps versus ∼3 ps in H₂O. We suggest that this long lifetime could be due to a similarly long distance between the bridging water molecule and Thr203 hydroxyl oxygen as observed in the S205V mutant, so that the proton transfer from Thr203 to the water limits the protonation rate of the chromophore. Interestingly, the decay rate reported by Kennis et al. (2004) of I₂ (∼400 ps) is very similar to the deprotonation rate of the chromophore for S205V.

As a final note, absorbance data suggest that during the photocycle of wt-GFP, the Thr203 hydroxyl is unlikely to form a hydrogen bond with the deprotonated chromophore (as observed for S65T GFP) (Ormo et al. 1996). This follows from the observation that intermediate states I₁ and I₂ have absorption maxima at 497 and 500 nm, respectively, which is similar to GFP variants T203V and T203I (∼500 nm) (Kummer et al. 2000). In those cases, the observed red shift of the absorption maximum of the anionic chromophore compared to wt-GFP was attributed to destabilization of the ground state due to absence of a hydrogen bond between the anionic chromophore and Thr203 (Kummer et al. 2000).

In conclusion, substitution of Ser205 and Thr203 by valine demonstrates that in green fluorescent protein these residues are key players in excited-state proton transfer. Time-resolved vibrational spectroscopy of the S205V variant should be helpful to confirm Glu222 as the proton acceptor, and may provide clues as to whether excited-state proton transfer in S205V is concerted or stepwise.

Materials and Methods

Mutagenesis, protein expression, and crystallization.

Mutagenesis was performed using the QuikChange method (Stratagene). Protein was expressed in Escherichia coli strain JM109 (DE3), by use of the PRSET_B His-tagged expression system. Protein was purified by Ni²⁺-affinity chromatography over Ni-NTA agarose (Qiagen), and then buffer exchanged with PD-10 Sephadex columns (Amersham Pharmacia) into 50 mM HEPES (pH 7.9). Crystals of GFP S205V were obtained by hanging-drop vapor diffusion: 2 μL protein solution (50 mM HEPES pH7.9, A₂₈₀ = 28) with 2 μL well solution (100 mM Imidazol pH 8.0, 1.1 M Na Citrate) after 6–7 mo.

Steady-state spectroscopy

Absorbance spectra (400 μg/mL protein in 100 mM buffers) were recorded with a HP 8453 UV-visible spectrophotometer. Fluorescence spectra were taken with a Perkin-Elmer LS-55 fluorometer. Protein samples were diluted to 10–50 μg/mL in 2 mL 100 mM buffers. Buffers include citric acid, HEPES, or CHES as appropriate. The quantum yields (QY) of emission were measured by referencing to the standards 9-aminoacridine and fluorescein.

Time-resolved emission spectroscopy

For excitation we used a cavity-dumped mode-locked Ti:sapphire femtosecond laser (Mira Coherent), which provides short, 80 fs, pulses of variable repetition rate. We used the SHG frequency, over the spectral range of 380–440 nm. Time-resolved fluorescence was acquired using the time-correlated single-photon counting (TCSPC) techniques. The TCSPC detection system is based on a Hamamatsu 3809U, photomultiplier and Edinburgh instruments TCC 900 computer module for TCSPC. The overall instrumental response was about 30 ps (fwhm). Measurements were taken at 10-nm spectral width. The large dynamic range of the TCSPC system (more than four orders of magnitude) enabled us to accurately determine the nonexponential photoluminescence decay profiles of the wt-GFP fluorescence.

The excitation pulse energy was reduced by neutral density filters to about 10 pJ. We checked the sample absorbance prior to and after time-resolved measurements and could not find noticeable changes in the absorption spectra due to sample irradiation. The time-resolved emission decay curves of the S205V samples were the same after repeated experiments. We thus, conclude that, under our irradiation protocol, no sample deterioration could be detected.

Data collection and structure solution

Diffraction data were collected on a Quantum-315 CCD on beamline 14-BM-C at the Advanced Photon Source. Data were reduced using HKL2000 (HKL Research). Molecular replacement was performed with EPMR (Kissinger et al. 1999) using the GFP S65T (PDB entry 1EMA) as a search model, providing unambiguous identification of the space group and unique solutions to the rotation and translation problems. Crystallographic refinement was performed using TNT (Tronrud et al. 1987), and model building was conducted with Coot (Emsley and Cowtan 2004) in several stages of increasing resolution. Atomic coordinates for GFP/S205V are available from the Protein Data Bank under access code 2QLE.