Vibrational spectroscopy and mode assignments for an analog of the green fluorescent protein chromophore

https://doi.org/10.1016/S0022-2860(00)00825-5Get rights and content

Abstract

Infrared absorption (IR), Raman, and resonance Raman spectra have been obtained from 500 to 1700 cm−1 for 4-hydroxybenzylidene-2,3-dimethyl-imidazolinone (HBDI), an analog of the green-fluorescent protein (GFP) chromophore. Numerous transitions are evident in both the IR and Raman spectra, with the resonance Raman spectrum of HBDI dominated by a subset of transitions in the 1430–1700 cm−1 region. Assignment of the transitions in this frequency region to the corresponding normal coordinates is accomplished through computational studies employing density functional and Hartree–Fock theory. The computational results indicate that the vibrational transitions in this frequency range are dominated by in-plane stretching modes that are localized to the imidazolinone or tyrosine portions of the chromophore, rather than being delocalized over the entire chromophore. No evidence is obtained for significant excited-state structural evolution along the O–H stretching coordinate. The implications of these findings with respect to the excited-state proton transfer dynamics of GFP are discussed.

Introduction

Green fluorescent protein (GFP) from the jellyfish Aequorea victoria has become an extremely useful spectroscopic tool in biotechnology [1], [2]. The utility of this protein arises from the post-transcriptional formation of a p-hydroxybenzylidene–imidazolinone chromophore through cyclization and oxidation of the Ser65-Tyr66-Gly67 amino-acid sequence [1], [3], [4]. Irradiation of the chromophore with near-ultraviolet light results in green fluorescence max=508nm) with high efficiency [2], [5]. These photophysical characteristics enable GFP to serve as a genetically encodable spectroscopic label that has found wide use as a marker for gene and protein expression [1], [2]. Current interest in GFP involves identifying the photophysical parameters and chromophore–protein interactions that define its spectroscopic properties [6]. This information will help guide the development of GFP mutants with unique spectroscopic properties, thereby increasing the utility of GFP labeling [2].

Recent studies of GFP photophysics have employed time-resolved spectroscopic techniques in an attempt to elucidate the excited- and ground-state dynamics responsible for its unusual spectroscopic properties [6], [7], [8], [9]. The static absorption spectrum of GFP is characterized by two transitions with maxima at 398 and 478 nm (Fig. 1). The relative intensities of these transitions vary with pH and ionic strength. These variations are consistent with assignments of the high- and low-energy transitions to protonated and unprotonated states of the chromophore, respectively [7], [10], [11], [12], [13], [14], [15], [16]. Fluorescence upconversion and picosecond time-resolved fluorescence experiments have demonstrated that following photoexcitation into the high-energy absorption band, the optically-prepared excited-state undergoes decay with a time constant of ∼3 ps [7], [8]. The excited-state decay kinetics are sensitive to deuterium substitution consistent with excited-state proton transfer (ESPT) [7], [8]. High-resolution crystallographic studies of GFP have been interpreted in terms of a chromophore-to-protein proton transfer mechanism [13], [17], [18]; however, the details of this mechanism are still unclear [14], [19]. The fluorescence from GFP originates mainly from the excited-state of the deprotonated chromophore, and recent time-resolved fluorescence measurements have revealed two deprotonated excited-state species with similar lifetimes (3.3 versus 2.8 ns) [9]. The difference between these excited-state species has been ascribed to protein conformational relaxation [7], [8], [20], [21].

Although the photophysical properties of GFP are apparently consistent with ESPT, recent studies have provided evidence for competing photochemical processes. Confocal fluorescence microscopy studies of single GFP molecules have shown that under continuous irradiation, periods of nonemissivity occur, demonstrating the formation of non-fluorescent states [22], [23]. Studies of GFP mutants have demonstrated that the modification of protein residues near the chromophore can result in large changes in the fluorescence spectrum and quantum yield [2], [6], [18]. Studies of synthetic GFP chromophore analogs and chromophore-containing peptide fragments obtained by proteolysis have shown that in the absence of protein, the fluorescence quantum yield is significantly reduced at room temperature [24], [25], [26]. Computational studies of the isolated chromophore suggest that conformational evolution due to internal rotation may be responsible for the apparent increase in the non-radiative decay rate of the excited chromophore [27], [28]. Currently at issue are what processes occur following photoexcitation of GFP and the isolated chromophore, and in what way these processes are affected by environment.

We have recently employed resonance Raman spectroscopy to explore the chromophore structure and excited-state dynamics of GFP [29]. Resonance Raman intensities can be used to develop a detailed description of the excited-state structural evolution that occurs upon photoexcitation [30], [31]. The interpretation of resonance Raman intensities requires knowledge of the chromophore's normal modes; therefore, we present here a vibrational spectroscopic study and computational analysis designed to provide normal mode assignments for a synthetic GFP chromophore. Infrared, Raman, and resonance Raman spectra have been obtained from 500 to 1700 cm−1 for 4-hydroxybenzylidene-2,3-dimethyl-imidazolinone (HBDI), an analog of the GFP chromophore in which the aliphatic carbons that link the chromophore to the protein are replaced by methyl groups (see inset Fig. 1). Numerous transitions are observed in the resonance Raman spectrum, demonstrating that substantial excited-state structural evolution occurs along multiple coordinates following photoexcitation. The resonance Raman spectra are dominated by a subset of transitions in the 1430–1700 cm−1 region. Assignment of the modes is accomplished using computational techniques employing density functional and Hartree–Fock theory. The computational results demonstrate that the transitions in this frequency region correspond to in-plane stretching of the chromophore consistent with substantial evolution in bond order upon photoexcitation. However, these modes are not delocalized over the entire chromophore, but are relatively localized to the tyrosine or imidazolinone components of the chromophore. Finally, no evidence is observed for significant motion along the O–H stretching coordinate.

Section snippets

Synthesis

4-hydroxybenzylidene-2,3-dimethyl-imidazolinone was synthesized by published methods [25]. Purity was demonstrated by TLC and comparison of 1H NMR (500 MHz, methanol-d4: δ 2.42 (3 H, s), δ 3.22 (3 H, s), δ 6.87 (2 H, d, J=8.3 Hz), δ 7.05 (1 H, s), δ 8.03 (2 H, d, J=8.3 Hz)) and FTIR spectra of the compound with literature data [32]. In addition, the electrospray-MS spectrum of HBDI dissolved in methanol/water/1% acetic acid had a single peak at mass 217.15 consistent with protonation of the

Absorption spectra

The electronic absorption spectra of GFP and HBDI are presented in Fig. 1. The band observed at ∼280 nm in the GFP spectrum originates from the aromatic amino acids of the protein. As mentioned above, the bands at 398 and 476 nm are attributed to the protonated and deprotonated forms of the chromophore in GFP [7], [13], [14], [15], [16]. In comparison, the spectrum of HBDI in EtOH exhibits a single maximum at 372 nm consistent with a single protonation state. The absorption spectrum of HBDI

Comparison of HF and DFT results

Fig. 3, Fig. 4 demonstrate that the Raman spectra of HBDI are dominated by transitions in the 1430–1700 cm−1 frequency region. Transitions in this region also dominate the resonance Raman spectrum of GFP [29]. In addition, no Raman or infrared transitions were observed between this region and the C–H stretching region (∼3000 cm−1). Therefore, we have focused our analysis on the 1430–1700 cm−1 frequency region.

To facilitate presentation of the computational results, Fig. 5, Fig. 6 show schematic

Discussion

The experimental and computational results presented here demonstrate that for HDBI, the normal coordinates in the 1430–1700 cm−1 region involve in-plane deformation of the chromophore. One interesting result of this analysis is that these deformations are not delocalized over the entire chromophore. Because the optical transition of interest is π–π in character, electronic excitation results in significant bond-order changes for bonds contributing to the delocalized π-electron density. Given

Acknowledgements

We thank Craig Beeson, Michael Hart, and Bart Kahr for their assistance in the synthesis of HBDI, and the National Science Foundation (grant MCB-9904618) for support. PJR is a Cottrell Fellow of the Research Corporation and an Alfred P. Sloan Fellow. We also thank A. Voityuk, A. Kummer, and M.-E. Michel-Beyerle for useful discussions.

References (43)

  • A.B. Cubitt et al.

    Trends Biochem. Sci.

    (1995)
  • O. Shimomura

    FEBS Lett.

    (1979)
  • A.D. Kummer et al.

    Chem. Phys.

    (1998)
  • H. Lossau et al.

    Chem. Phys.

    (1996)
  • A.A. Voityuk et al.

    Chem. Phys. Lett.

    (1997)
  • C. Scharnagl et al.

    Biophys. J.

    (1999)
  • S. Kojima et al.

    Tetrahedron Lett.

    (1998)
  • A.A. Voityuk et al.

    Chem. Phys. Lett.

    (1998)
  • S. Kojima et al.

    Tetrahedron Lett.

    (1997)
  • B.M. Bode et al.

    J. Mol. Graphics Mod.

    (1998)
  • I. Bytheway et al.

    Chem. Phys. Lett.

    (1998)
  • M. Pfeiffer et al.

    Chem. Phys. Lett.

    (1997)
  • R.Y. Tsien

    Annu. Rev. Biochem.

    (1998)
  • C.W. Cody et al.

    Biochemistry

    (1993)
  • H. Morise et al.

    Biochemistry

    (1974)
  • M. Chattoraj et al.

    Proc. Natl Acad. Sci. USA

    (1996)
  • G. Striker et al.

    J. Phys. Chem. B

    (1999)
  • W.W. Ward et al.

    Photochem. Photobiol.

    (1980)
  • W.W. Ward et al.

    Photochem. Photobiol.

    (1982)
  • M.A. Esliger et al.

    Biochemistry

    (1999)
  • K. Brejc et al.

    Proc. Natl Acad. Sci. USA

    (1997)
  • Cited by (41)

    • Normal modes and the Duschinsky mixing of the ground- and excited-state vibrations of the green fluorescent protein chromophore

      2013, Chemical Physics Letters
      Citation Excerpt :

      In the PEDs, ‘Ph’ indicates the phenol-group moiety, ‘imid’ indicates the imidazolinone-group moiety, and ‘exo’ indicates the bridging (linking) atoms between the phenol and imidazolinone groups. As arises from the Table 2, most of the calculated frequencies correspond well with the experimental ones [13,15]. Deviations from experimental values are most probably caused by a combination of different factors, including the size the of active space used in the CASSCF calculation, the size of the basis set, the neglect of anharmonicity, the nature of the experimental conditions etc.

    • Gas-phase infrared spectrum of the anionic GFP-chromophore

      2012, International Journal of Mass Spectrometry
      Citation Excerpt :

      Mode assignments were made by isotopic substitution and have been “translated” here into the vibrational mode descriptions used in Table 1. The bands at 1583 and 1499 cm−1 indicated in Fig. 1e with ph were referred to as “phenol” modes [22], being mainly phenol CC stretches [24]. Calculated spectra of the anion in solution – generated using the polarisable continuum model (PCM), the 6-31G** and 6-31+G** basis sets and assuming frequency scale factors identical to those used for the gas-phase calculations [36] – are also presented in Fig. 1.

    • Ultrafast vibrational dynamics of parallel excited state proton transfer reactions in the Green Fluorescent Protein

      2012, Vibrational Spectroscopy
      Citation Excerpt :

      The polarised optical measurements resulted in one solution which was oriented 1.5° relative to the calculation at the TD-DFT B3LYP/6-311+G(d,p) level, indicating that there is only a small uncertainty for the direction of the electronic transition dipole tensor [21]. Vibrational mode assignments for data shown in Fig. 2 used harmonic frequency calculations of the chromophore at the DFT and CIS levels, based on assignments previously reported [31,32]. The comparison of the experimental dichroism and the corresponding mode assignment with theoretical calculation used dipole gradients calculated [21] with Gaussian 03.

    • Chromophore Packing Leads to Hysteresis in GFP

      2009, Journal of Molecular Biology
    • Balance between ultrafast parallel reactions in the green fluorescent protein has a structural origin

      2008, Biophysical Journal
      Citation Excerpt :

      Molecular interpretation is additionally affected, even for relatively localized vibrations such as the C-O stretching mode of heme-bound CO in myoglobin, where bond-stretching causes a dipole reorientation and deviations between the transition dipole moment vector and the corresponding bond vector (13). The photoselection results of GFP provide ultrafast structural information on responsive amino acids E222 and Q69; these results also test existing mode assignments for the ground state proposed from Density Functional Theory (DFT) normal mode calculations in addition to isotope substitution strategies with chromophore model compounds (14,15). Excited-state assignments were previously proposed on the basis of expected bond order changes (4) and, in this study, are evaluated from calculation of frequencies and infrared transition dipole vectors using single excitation configuration interaction (CIS).

    View all citing articles on Scopus
    1

    Current address: Institut für Physikalische and Theoretische Chemie, Technische Universität München, Lichtenbergstr. 4, D-85748 Garching, Germany.

    View full text