The Chromophore of GFP

---

GFP is unique among fluorescent proteins in that its fluorophore is not a seperately synthesized prostethic group but composed of modified amino acid residues within the polypeptide chain.

Chemical Structure of the Chromophore

Upon limited papain digestion of GFP, a hexapeptide with the same spectral properties as denatured GFP is released. The amino acid sequences of the hexapeptides from Aequorea and Renilla are: [3]

Aequorea:	FSYGVQ

Renilla:	FSYGDR

The chromophore itself is a p-hydroxybenzylidene-imidazolidone (green background). It consists of residues 65-67 (Ser - dehydroTyr - Gly) of the protein. The cyclized backbone of these residues forms the imidazolidone ring. The peptide backbone trace is shown in red.

Although the amino acid sequence SYG can be found in a number of other proteins as well, it is neither cyclized in any of these, nor is the Tyrosine oxidized, nor are these proteins fluorescent. This implies that the tendency to form such a chromophore is no intrinsic property of this tripeptide.

Biosynthesis of the Chromophore

Heim et al. [2] have proposed the following biosynthetic scheme for the chromophore. (R1 signifies residues 1-64, R2 residues 68-238.)

The first step is a nucleophilic attack of the amino group of Gly67 onto the carbonyl group of Ser65 (upper left). Subsequent elimination of water results in formation of the imidazolidinone ring (upper right). In a second step the $Calpha-Cbeta$ bond of Tyr66 is oxidized to give a large delocalized pi-system (bottom). The proposed biosynthetic scheme is based upon the following observations:

• Expression of recombinant GFP (e.g. in E. coli) is possible, i.e. fluorophore formation does not require any specific enzymes from A. victoria .
• If soluble GFP is expressed in E. coli under anaerobic conditions, it is nonfluorescent but otherwise indistinguishable from native GFP on a denaturing SDS-gel.
• Fluorescence of anaerobically expressed GFP gradually develops after admission of air, even in very dilute cell lysates (i.e. oxidation is not enzymatically catalized).
• Imidazolin-5-ones are known to undergo autoxidative formation of double bonds at the 4-position, which in this case would complete the fluorophore.
• After readmission of atmospheric oxygen to anaerobically expressed, nonfluorescent GFP, development of fluorescence exhibits pseudo-first order kinetics with a rate constant of$0.24\; \pm \; 0.06\; hr-1$.

Excitation and Fluorescence Emission Spectra of GFP

Denatured GFP is not fluorescent and its absorbtion spectrum is significantly different from native GFP [1], which implies that noncovalent interactions of the chromophore with its local environment have a great influence on the spectral characteristics and that fluorescence is mediated by amino acids close to the chromophore in the tertiary structure of GFP.

The excitation spectrum of native GFP from A. victoria (blue) has two excitation maxima at 395 nm and at 470 nm. The fluorescence emission spectrum (green) has a peak at 509 nm and a shoulder at 540 nm [3]. (Spectra after [4])

Recombinant A. victoria-GFP generated in E. coli has the same spectral properties [4], the excitation spectrum of R. reniformis-GFP is different, whereas its emission spectrum is the same [3]. The absorption spectrum of A. victoria-GFP, is not significantly affected by titration of external pH up to pH 10, which is just below the threshold of denaturation [2, 11].

Förster cycle

GFP is the first known example of a Förster cycle within the core of a protein [5,6,7]:

Depending on whether Tyr66 is in its hydroxyl form (upper left) or in its phenolate form (lower left), the fluorophore absorbs light at 395 nm or at 470 nm, respectively. As phenols are known to be more acidic in their excited state than in their ground state, it is postulated that the protonated excited form of the fluorophore (upper right) converts to the excited phenolate (lower right), which is the only fluorescent species and emits light at 509 nm. As a result, a cycle is formed, in which the flourophore absorbs a photon, then loses a proton, emits a photon and finally takes up a proton, returning to its original state.
Interestingly, excitation at 395 nm leads to a decrease over time of the 395 nm peak and a reciprocal increase of the 475 nm excitation band [6].

After the structures of wild type GFP [6] and of the S65T variant of GFP [7] have been solved by X-ray crystallography, some GFP variants, which have one excitation band of wild type GFP deleted and retain the other without a significant wavelength shift, can be understood in view of the 3-dimensional structure and the Förster cycle:
The fluorophore inside the protein is not solvent accessible. It interacts with the sidechains of many surrounding residues via hydrogen-bonds:

• The basic residues of His148, Gln94 and Arg96 presumably stabilize and and possibly further delocalize the charge on the fluorophore [6].
• The hydroxyl of Thr203 donates a hydrogen-bond to Tyr66, stabilizing the phenolate form [7].
• Ser65 may donate a hydrogen bond to Glu222, stabilizing it as an anion, whose charge inhibits ionization of Tyr66, thus favouring the hydroxyl form [7] for electrostatic reasons.

Click on the small picture for a larger illustration (8k) of some of the fluorophore's interactions with the surrounding sidechains. Hydrogen-bonds are shown in blue, except for the one presumably stabilizing the respective form of the fluorophore, which is shown in red.

Mutations affecting the $pK$_a of Tyr66

Stabilization of the phenolate or hydroxyl form of Tyr66 by specific hydrogen-bonds can explain why disruption of these hydrogen-bonds leads to an altered absorption spectrum in GFP-variants:

• The T203I variant [8] has a single absorption maximum at 400 nm and the same fluorescence emission spectrum as the wild type. This indicates that Thr203 indeed stabilizes the phenolate form of Tyr66, which is responsible for the 470 nm absorption band. Since Ile cannot form the hydrogen-bond, Tyr66 is predominantly in its hydroxyl form, absorbing light at 400 nm but not at 470 nm.
• The S65G, S65A and S65V variants all have a single excitation maximum at 470-490 nm [7]. In each of these variants the hydrogen-bond between residue 65 and Glu222 is removed, which presumably leads to a protonation of Glu222, as its ionized form is no longer stabilized. Uncharged Glu222 possibly favours ionization of Tyr66, and the phenolate form absorbs light at 470 nm but not at 395 nm.
• The E222G variant [8] has a single excitation maximum at 481 nm. It lacks the negative charge of Glu222, so that presumably the phenolate form of Tyr66 is favoured.
• The single excitation maximum of the S65T variant at 489 nm [7] is more difficult to explain. The authors point out that Thr65 may adopt a different conformation than Ser65, so that the hydroxyl of Ser65 donates a hydrogen bond to Glu222, stabilizing its ionized form, whereas the hydroxyl of Thr65 acts as a hydrogen-bond acceptor, stabilizing the uncharged form of Glu222. Once again, uncharged Glu222 favours the phenolate form of Tyr66, which absorbs light at 470 nm.

Mutations of the Chromophore Forming Residues

There is a vast amount of different GFP variants described in the literature, many of which have quite interesting characteristics. But discussion of all of these is far beyond the scope of this project and besides, there is no rational explanation yet for quite a few of the observed features. So I will limit further discussion of GFP variants to those which bear a mutation in the fluorophore itself.

• The mutations of Ser 65 have been discussed above.
• Since the aromatic system of Tyr 66 is part of the fluorophore's delocalized pi-system, a substitution of Tyr 66 with a non-aromatic amino acid cannot be expected to be fluorescent. The only fluorescent variants that have been found are T66H and T66W [2]. T66H shows a blue shift in excitation and emission in comparison to the wild-type with a much weaker fluorescence. T66W has a red-shifted excitation maximum and a blue-shiftet emission maximum, but is only weakly fluorescent as well. T66F gives no detectable fluorescence [2].
• Gly 67 is the only residue that is absolutely conserved [5]. Quite possibly, any side chain of residue 67 would sterically hinder cyclization of the chromophore.

[Introduction] [Index] [GFP Structure]