Understanding GFP posttranslational chemistry: structures of designed variants that achieve backbone fragmentation, hydrolysis, and decarboxylation

The green fluorescent protein (GFP) creates a fluorophore out of three sequential amino acids by promoting spontaneous posttranslational modifications. Here, we use high-resolution crystallography to characterize GFP variants that not only undergo peptide backbone cyclization but additional denaturation-induced peptide backbone fragmentation, native peptide hydrolysis, and decarboxylation reactions. Our analyses indicate that architectural features that favor GFP peptide cyclization also drive peptide hydrolysis. These results are relevant for the maturation pathways of GFP homologues, such as the kindling fluorescent protein and the Kaede protein, which use backbone cleavage to red-shift the spectral properties of their chromophores. We further propose a photochemical mechanism for the decarboxylation reaction, supporting a role for the GFP protein environment in facilitating radical formation and one-electron chemistry, which may be important in activating oxygen for the oxidation step of chromophore biosynthesis. Together, our results characterize GFP posttranslational modification chemistry with implications for the energetic landscape of backbone cyclization and subsequent reactions, and for the rational design of predetermined spontaneous backbone cyclization and cleavage reactions.     

Structural evidence for an enolate intermediate in GFP fluorophore biosynthesis

The Aequorea victoria green fluorescent protein (GFP) creates a fluorophore from its component amino acids Ser65, Tyr66, and Gly67 through a remarkable post-translational modification, involving spontaneous peptide backbone cyclization, dehydration, and oxidation reactions. Here we test and extend the understanding of fluorophore biosynthesis by coupling chemical reduction and anaerobic methodologies with kinetic analyses and protein structure determination. Two high-resolution structures of dithionite-treated GFP variants reveal a previously uncharacterized enolate intermediate form of the chromophore that is viable in generating a fluorophore (t1/2 = 39 min-1) upon exposure to air. Isolation of this enolate intermediate will now allow specific probing of the rate-limiting oxidation step for fluorophore biosynthesis in GFP and its red fluorescent protein homologues. Such targeted characterizations may lead to the design of faster maturing proteins with enhanced applications in biotechnology and cell biology. Moreover, our results reveal how the GFP protein environment mimics enzyme systems, by stabilizing an otherwise high energy enolate intermediate to achieve its post-translational modification.     

Reaction progress of chromophore biogenesis in green fluorescent protein

The mature form of green fluorescent protein (GFP) is generated by a spontaneous self-modification process that is essentially irreversible. A key step in chromophore biosynthesis involves slow air oxidation of an intermediate species, in which the backbone atoms of residues 65-67 have condensed to form a five-membered heterocycle. We have investigated the kinetics of hydrogen peroxide evolution during in vitro GFP maturation and found that the H2O2 coproduct is generated prior to the acquisition of green fluorescence at a stoichiometry of 1:1 (peroxide/chromophore). The experimental progress curves were computer-fitted to a three-step mechanism, in which the first step proceeds with a time constant of 1.5 (+/-1.1) min and includes protein folding and peptide cyclization. Kinetic data obtained by HPLC analysis support a rapid cyclization reaction that can be reversed upon acid denaturation. The second step proceeds with a time constant of 34.0 (+/-1.5) min and entails rate-limiting protein oxidation, as supported by a mass loss of 2 Da observed for tryptic peptides derived from species that accumulate during the reaction. The final step in GFP maturation proceeds with a time constant of 10.6 (+/-1.2) min, suggesting that this step may contribute to overall rate retardation. We propose that under highly aerobic conditions, the dominant reaction path follows a cyclization-oxidation-dehydration mechanism, in which dehydration of the heterocycle is facilitated by slow proton abstraction from the Tyr66 beta-carbon. In combination, the results presented here suggest a role for molecular oxygen in trapping the cyclized form of GFP.     

Computational prediction of absorbance maxima for a structurally diverse series of engineered green fluorescent protein chromophores

By virtue of its self-sufficiency to form a visible wavelength chromophore within the confines of its tertiary structure, the Aequorea victoria green fluorescent protein (GFP) is single-handedly responsible for the ever-growing popularity of fluorescence imaging of recombinant fusion proteins in biological research. Engineered variants of GFP with altered excitation or emission wavelength maxima have helped to expand the range of applications of GFP. The engineering of the GFP variants is usually done empirically by genetic modifications of the chromophore structure and/or its environment in order to find variants with new photophysical properties. The process of identifying improved variants could be greatly facilitated if augmented or guided by computational studies of the chromophore ground and excited-state properties and dynamics. In pursuit of this goal, we now report a thorough investigation of computational methods for prediction of the absorbance maxima for an experimentally validated series of engineered GFP chromophore analogues. The experimental dataset is composed of absorption maxima for 10 chemically distinct GFP chromophore analogues, including a previously unreported Y66D variant, measured under identical denaturing conditions. For each chromophore analogue, excitation energies and oscillator strengths were calculated using configuration interaction with single excitations (CIS), CIS with perturbative correction for double substitutions [CIS(D)], and time-dependent density functional theory (TD DFT) using several density functionals with solvent effects included using a polarizable continuum model. Comparison of the experimental and computational results show generally poor quantitative agreement with all methods attempted. However, good linear correlations between the calculated and experimental excitation energies (R2>0.9) could be obtained. Oscillator strengths obtained with TD DFT using pure density functionals also correlate well with the experimental values. Interestingly, most of the computational methods used in this work fail in the case of nonaromatic Y66S and Y66L protein chromophores, which may be related to a significant contribution of double excitations to their excited-state wavefunctions. These results provide an important benchmark of the reliability of the computational methods as applied to GFP chromophore analogues and lays a foundation for the computational design of GFP variants with improved properties for use in biological imaging.     

Cis-trans photoisomerization of fluorescent-protein chromophores

Photochromic variants of fluorescent proteins are opening the way to a number of opportunities for high-sensitivity regioselective studies in the cellular environment and may even lead to applications in information and communication technology. Yet, the detailed photophysical processes at the basis of photoswitching have not been fully clarified. In this paper, we used synthetic FP chromophores to clarify the photophysical processes associated with the photochromic behavior. In particular, we investigated the spectral modification of synthetic chromophore analogues of wild-type green fluorescent protein (GFP), Y66F GFP (BFPF), and Y66W GFP (CFP) upon irradiation in solutions of different polarities. We found that the cis-trans photoisomerization mechanism can be induced in all the chromophores. The structural assignments were carried out both by NMR measurements and DFT calculations. Remarkably, we determined for the first time the spectra of neutral trans isomers in different solvents. Finally, we calculated the photoconversion quantum yields by absorption measurements under continuous illumination at different times and by a nanosecond laser-flash photolysis method. Our results indicate that cis-trans photoisomerization is a general mechanism of FP chromophores whose efficiency is modulated by the detailed mutant-specific protein environment.     

Cellular Synthesis and X-ray Crystal Structure of a Designed Protein Heterocatenane

Herein, we report the biosynthesis of protein heterocatenanes using a programmed sequence of multiple post-translational processing events including intramolecular chain entanglement, in situ backbone cleavage, and spontaneous cyclization. The approach is general, autonomous, and can obviate the need for any additional enzymes. The catenane topology was convincingly proven using a combination of SDS-PAGE, LC-MS, size exclusion chromatography, controlled proteolytic digestion, and protein crystallography. The X-ray crystal structure clearly shows two mechanically interlocked protein rings with intact folded domains. It opens new avenues in the nascent field of protein-topology engineering.     

One-electron photooxidation and site-selective strand cleavage at 5-methylcytosine in DNA by sensitization with 2-methyl-1,4-naphthoquinone-tethered oligonucleotides

Photosensitized one-electron oxidation was applied to discriminate a specific base site of 5-methylcytosine (mC) generated in DNA possessing a partial sequence of naturally occurring p53 gene, using a sensitizing 2-methyl-1,4-naphthoquinone (NQ) chromophore tethered to an interior of oligodeoxynucleotide (ODN) strands. Photoirradiation and subsequent hot piperidine treatment of the duplex consisting of mC-containing DNA and NQ-tethered complementary ODN led to oxidative strand cleavage selectively at the mC site, when the NQ chromophore was arranged so as to be in close contact with the target mC. The target mC is most likely to be one-electron oxidized into the radical cation intermediate by the sensitization of NQ. The resulting mC radical cation may undergo rapid deprotonation and subsequent addition of molecular oxygen, thereby leading to its degradation followed by strand cleavage at the target mC site. In contrast to mC-containing ODN, ODN analogs with replacement of normal cytosine, thymine, adenine, or guanine at the mC site underwent less amount of such an oxidative strand cleavage at the target base site, presumably due to occurrence of charge transfer and charge recombination processes between the base radical cation and the NQ radical anion. Furthermore, well designed incorporation of the NQ chromophore into an interior of ODN could suppress a competitive strand cleavage at consecutive guanines, which occurred as a result of positive charge transfer. Thus, photosensitization by an NQ-tethered ODN led to one-electron oxidative strand cleavage exclusively at the target mC site, providing a convenient method of discriminating mC in naturally occurring DNA such as human p53 gene as a positive band on a sequencing gel.     

Cellular Synthesis and X‐ray Crystal Structure of a Designed Protein Heterocatenane

Herein, we report the biosynthesis of protein heterocatenanes using a programmed sequence of multiple post‐translational processing events including intramolecular chain entanglement, in situ backbone cleavage, and spontaneous cyclization. The approach is general, autonomous, and can obviate the need for any additional enzymes. The catenane topology was convincingly proven using a combination of SDS‐PAGE, LC‐MS, size exclusion chromatography, controlled proteolytic digestion, and protein crystallography. The X‐ray crystal structure clearly shows two mechanically interlocked protein rings with intact folded domains. It opens new avenues in the nascent field of protein‐topology engineering.     

Does Electron Capture Dissociation Cleave Protein Disulfide Bonds?

Peptide and protein characterization by mass spectrometry (MS) relies on their dissociation in the gas phase into specific fragments whose mass values can be aligned as ‘mass ladders’ to provide sequence information and to localize possible posttranslational modifications. The most common dissociation method involves slow heating of even-electron (M+n H)ⁿ⁺ ions from electrospray ionization by energetic collisions with inert gas, and cleavage of amide backbone bonds. More recently, dissociation methods based on electron capture or transfer were found to provide far more extensive sequence coverage through unselective cleavage of backbone N–Cα bonds. As another important feature of electron capture dissociation (ECD) and electron transfer dissociation (ETD), their unique unimolecular radical ion chemistry generally preserves labile posttranslational modifications such as glycosylation and phosphorylation. Moreover, it was postulated that disulfide bond cleavage is preferred over backbone cleavage, and that capture of a single electron can break both a backbone and a disulfide bond, or even two disulfide bonds between two peptide chains. However, the proposal of preferential disulfide bond cleavage in ECD or ETD has recently been debated. The experimental data presented here reveal that the mechanism of protein disulfide bond cleavage is much more intricate than previously anticipated.     

19F NMR studies of the native and denatured states of green fluorescent protein

Biosynthetic preparation and (19)F NMR experiments on uniformly 3-fluorotyrosine-labeled green fluorescent protein (GFP) are described. The (19)F NMR signals of all 10 fluorotyrosines are resolved in the protein spectrum with signals spread over 10 ppm. Each tyrosine in GFP was mutated in turn to phenylalanine. The spectra of the Tyr --> Phe mutants, in conjunction with relaxation data and results from (19)F photo-CIDNP (chemically induced dynamic nuclear polarization) experiments, yielded a full (19)F NMR assignment. Two (19)F-Tyr residues (Y92 and Y143) were found to yield pairs of signals originating from ring-flip conformers; these two residues must therefore be immobilized in the native structure and have (19)F nuclei in two magnetically distinct positions depending on the orientation of the aromatic ring. Photo-CIDNP experiments were undertaken to probe further the structure of the native and denatured states. The observed NMR signal enhancements were found to be consistent with calculations of the HOMO (highest occupied molecular orbital) accessibilities of the tyrosine residues. The photo-CIDNP spectrum of native GFP shows four peaks corresponding to the four tyrosine residues that have solvent-exposed HOMOs. In contrast, the photo-CIDNP spectra of various denatured states of GFP show only two peaks corresponding to the (19)F-labeled tyrosine side chains and the (19)F-labeled Y66 of the chromophore. These data suggest that the pH-denatured and GdnDCl-denatured states are similar in terms of the chemical environments of the tyrosine residues. Further analysis of the sign and amplitude of the photo-CIDNP effect, however, provided strong evidence that the denatured state at pH 2.9 has significantly different properties and appears to be heterogeneous, containing subensembles with significantly different rotational correlation times.     

Kinetics of oxidation of bilirubin and its protein complex by hydrogen peroxide in aqueous solutions

A comparative study of oxidation reactions of bilirubin and its complex with albumin was carried out in aqueous solutions under the action of hydrogen peroxide and molecular oxygen at different pH values. Free radical oxidation of the pigment in both free and bound forms at pH 7.4 was shown not to lead to the formation of biliverdin, but to be associated with the decomposition of the tetrapyrrole chromophore into monopyrrolic products. The effective and true rate constants of the reactions under study were determined. It was assumed that one possible mechanism of the oxidation reaction is associated with the interaction of peroxyl radicals and protons of the NH groups of bilirubin molecules at the limiting stage with the formation of a highly reactive radical intermediate. The binding of bilirubin with albumin was found to result in a considerable reduction in the rate of the oxidation reaction associated with the kinetic manifestation of the protein protection effect. It was found that the autoxidation of bilirubin by molecular oxygen with the formation of biliverdin at the intermediate stage can be observed with an increase in the pH of solutions.     

Structural features responsible for GFPuv and S147P-GFP’s improved fluorescence

Green fluorescent protein (GFP) is used as a biological marker. It is a protein in the jellyfish, Aequorea victorea, which is found in the cold Pacific Northwest. Mature GFP, i.e. fully fluorescent GFP, is most efficiently formed at temperatures well below 37 °C. The GFPuv (F99S/M153T/V163A) and S147P-GFP mutants mature more efficiently at room temperature than wild-type GFP, and therefore result in increased fluorescence at room temperature. Computational methods have been used to examine whether the low-energy precyclized forms of these improved GFP-mutants are preorganized so that they can more efficiently form the chromophore than the wild-type and S65T-GFP. All mutations examined (S147P, F99S, M153T, V163A and F99S/M153T/V163A) more efficiently preorganize the immature precyclized forms of GFP for chromophore formation than immature wild-type GFP. It has been proposed that Arg96 is involved in chromophore formation. Our calculations suggest that the M153T and V163A mutations in GFPuv maybe partially responsible for the increased maturation efficiency observed in GFPuv because they improve the Arg96–Tyr66 interaction. The same is true for the S147P mutation in S147P-GFP.     

Are the Fluorescent Properties of the Cyan Fluorescent Protein Sensitive to Conditions of Oxidative Stress?

The modifications induced by reactive oxygen species (ROS) on fluorescent proteins (FPs) may have important implications for live cell fluorescence imaging. Using quantitative γ-radiolysis, we have studied the ROS-induced biochemical and photophysical perturbations on recombinant cyan fluorescent protein (CFP). After oxidation by the ˙OH radical, the protein displays a modified RP-HPLC elution profile, while the CFP fluorescence undergoes pronounced decreases in intensity and lifetime, without changes in its excitation and emission spectra. Meanwhile, the Förster resonant energy transfer (FRET) between the single W⁵⁷ and the chromophore remains unperturbed. These results rule out a direct oxidation of the CFP chromophore and of W⁵⁷ as well as major changes in the protein 3D structure, but show that new fluorescent forms associated to a higher level of dynamic quenching have been generated. Thus, strict in situ controls are required when CFP is to be used for FRET studies in situations of oxidative activity, or under strong illumination.     

Identification of leucine-enkephalin radical oxidation products by liquid chromatography tandem mass spectrometry

The oxidation of the peptide leucine-enkephalin (YGGFL) induced by the hydroxyl radical (HO*), formed under Fenton-like conditions [Cu (II)/H(2)O(2)], was studied and monitored by LC-MS. The oxidation products identified included products resultant from (a) the insertion of oxygen atoms (1-5), (b) peptide backbone cleavage (short-chain products formed by diamide pathway) and (c) radical-radical crosslinking reactions. In order to identify the modified residues, LC-MS/MS spectra were obtained. The insertion of oxygen atoms into the peptide originated hydroxide, di-hydroxide and/or hydroperoxide derivatives. In addition it was found that the aromatic amino acids are most susceptible to being hydroxylated, while the aliphatic amino acids are more prone to forming hydroperoxides. Oxidation products with double bonds were also identified. The short chain products resulted from the alpha-carbon radical of terminal amino acids (Tyr and Leu). Products resulting from cross-linking reactions between intact carbon-centered peptide radical (with and without one HO group) and a side chain radical (*C(7)H(7)O) were identified. It was found that, although all amino acids residues of the peptide undergo modifications, the N-terminal seems to be prone to oxidative modifications under these conditions.     

Theoretical Computer‐Aided Mutagenic Study on the Triple Green Fluorescent Protein Mutant S65T/H148D/Y145F

Green fluorescent protein (GFP) mutant S65T/H148D has been proposed to host a photocycle that involves an excited‐state proton transfer between the chromophore (Cro) and the Asp148 residue and takes place in less than 50 fs without a measurable kinetic isotope effect. It has been suggested that the interaction between the unsuspected Tyr145 residue and the chromophore is needed for the ultrafast sub‐50 fs rise in fluorescence. To verify this, we have performed a computer‐aided mutagenic study to introduce the additional mutation Y145F, which eliminates this interaction. By means of QM/MM molecular dynamics simulations and time‐dependent density functional theory studies, we have assessed the importance of the Cro–Tyr145 interaction and the solvation of Asp148 and shown that in the triple mutant S65T/H148D/Y145F a significant loss in the ultrafast rise of the Stokes‐shifted fluorescence should be expected.     

Facile identification of photocleavable reactive metabolites and oxidative stress biomarkers in proteins via mass spectrometry

Described herein is a method which combines bond selective fragmentation by photodissociation with online liquid chromatographic separation and mass spectrometric analysis. Photoexcitation of proteins or peptides with 266-nm light does not normally yield abundant fragmentation; however, incorporation of a suitable carbon-sulfur or carbon-halogen bond that is proximal to a chromophore allows access to direct dissociation pathways, resulting in homolytic cleavage of these bonds. Radicals generated through this process can cause further dissociation of the peptide backbone, which is useful for site specifically identifying the point of modification. Two specific applications of this technique for peptide analysis in model systems are presented: (1) identification of reactive metabolites which covalently modify cysteine residues, and (2) characterization of halogenated tyrosine residues which are biomarkers related to oxidative stress. In both cases, these naturally occurring post translational modifications create photocleavable bonds which can be fragmented by 266-nm light. The selectivity offered by photodissociation allows facile identification of the peptides of interest even in complex mixtures, and subsequent selective radical directed backbone fragmentation pinpoints the site of modification. This combination greatly simplifies data analysis and provides more confident assignments.     

Origin, nature, and fate of the fluorescent state of the green fluorescent protein chromophore at the CASPT2//CASSCF resolution

Ab initio CASPT2//CASSCF relaxation path computations are employed to determine the intrinsic (e.g., in vacuo) mechanism underlying the rise and decay of the luminescence of the anionic form of the green fluorescent protein (GFP) fluorophore. Production and decay of the fluorescent state occur via a two-mode reaction coordinate. Relaxation along the first (totally symmetric) mode leads to production of the fluorescent state that corresponds to a planar species. The second (out-of-plane) mode controls the fluorescent state decay and mainly corresponds to a barrierless twisting of the fluorophore phenyl moiety. While a "space-saving" hula-twist conical intersection decay channel is found to lie only 5 kcal mol(-1) above the fluorescent state, the direct involvement of a hula-twist deformation in the decay is not supported by our data. The above results indicate that the ultrafast fluorescence decay observed for the GFP chromophore in solution is likely to have an intrinsic origin. The possible effects of the GFP protein cavity on the fluorescence lifetime of the investigated chromophore model are discussed.     

GFP family: structural insights into spectral tuning

Proteins homologous to green fluorescent protein (GFP) span most of the visible spectrum, offering indispensable tools for live cell imaging. Structural transformations, such as posttranslational autocatalytic and photo-induced modifications, chromophore isomerization, and rearrangements in its environment underlie the unique capacity of these proteins to tune their own optical characteristics. A better understanding of optical self-tuning mechanisms would assist in the engineering of more precisely adapted variants and in expanding the palette of GFP-like proteins to the near-infrared region. The latest advances in this field shed light upon multiple features of protein posttranslational chemistry, and establish some important basic principles about the interplay of structure and spectral properties in the GFP family.     

Anthraquinone-sensitized photooxidation of 5-methylcytosine in DNA leading to piperidine-induced efficient strand cleavage

One-electron photooxidations of 5-methyl-2'-deoxycytidine (d(m)C) and 5-trideuteriomethyl-2'-deoxycytidine ([D(3)]d(m)C) by sensitization with anthraquinone (AQ) derivatives were investigated. Photoirradiation of an aerated aqueous solution containing d(m)C and anthraquinone 2-sulfonate (AQS) afforded 5-formyl-2'-deoxycytidine (d(f)C) and 5-hydroxymethyl-2'-deoxycytidine (d(hm)C) in good yield through an initial one-electron oxidation process. The deuterium isotope effect on the AQS-sensitized photooxidation of d(m)C suggests that the rate-determining step in the photosensitized oxidation of d(m)C involves internal transfer of the C5-hydrogen atom of a d(m)C-tetroxide intermediate to produce d(f)C and d(hm)C. In the case of a 5-methylcytosine ((m)C)-containing duplex DNA with an AQ chromophore that is incorporated into the backbone of the DNA strand so as to be immobilized at a specific position, (m)C underwent efficient direct one-electron oxidation by the photoexcited AQ, which resulted in an exclusive DNA strand cleavage at the target (m)C site upon hot piperidine treatment. In accordance with the suppression of the strand cleavage at 5-trideuterio-methylcytosine observed in a similar AQ photosensitization, it is suggested that deprotonation at the C5-methyl group of an intermediate (m)C radical cation may occur as a key elementary reaction in the photooxidative strand cleavage at the (m)C site. Incorporation of an AQ sensitizer into the interior of a strand of the duplex enhanced the one-electron photooxidation of (m)C, presumably because of an increased intersystem crossing efficiency that may lead to efficient piperidine-induced strand cleavage at an (m)C site in a DNA duplex.     

Excited state proton transfer in strongly enhanced GFP (sGFP2)

Proton transfer is an elementary process in biology. Green fluorescent protein (GFP) has served as an important model system to elucidate the mechanistic details of this reaction, because in GFP proton transfer can be induced by light absorption. We have used pump-dump-probe spectroscopy to study how proton transfer through the 'proton-wire' around the chromophore is affected by a combination of mutations in a modern GFP variety (sGFP2). The results indicate that in H(2)O, after absorption of a photon, a proton is transferred (A* → I*) in 5 ps, and back-transferred from a ground state intermediate (I → A) in 0.3 ns, similar to time constants found with GFPuv, although sGFP2 shows less heterogeneous proton transfer. This suggests that the mutations left the proton-transfer largely unchanged, indicating the robustness of the proton-wire. We used pump-dump-probe spectroscopy in combination with target analysis to probe suitability of the sGFP2 fluorophore for super-resolution microscopy.