Regular article
Structure-fluorescence correlations in a single tryptophan mutant of carp parvalbumin: solution structure, backbone and side-chain dynamics1

https://doi.org/10.1006/jmbi.2000.3549Get rights and content

Abstract

Heterogeneous fluorescence intensity decays of tryptophan in proteins are often rationalized using a model which proposes that different rotameric states of the indole alanyl side-chain are responsible for the observed fluorescence lifetime heterogeneity. We present here the study of a mutant of carp parvalbumin bearing a single tryptophan residue at position 102 (F102W) whose fluorescence intensity decay is heterogeneous and assess the applicability of a rotamer model to describe the fluorescence decay data. We have determined the solution structure of F102W in the calcium ligated state using multi-dimensional nuclear magnetic resonance (NMR) and have used the minimum perturbation mapping technique to explore the possible existence of multiple conformations of the indole moiety of Trp102 of F102W and, for comparison, Trp48 of holo-azurin. The maps for parvalbumin suggest two potential conformations of the indole side-chain. The high energy barrier for rotational isomerization between these conformers implies that interwell rotation would occur on time-scales of milliseconds or greater and suggests a rotamer basis for the heterogeneous fluorescence. However, the absence of alternate Trp102 conformers in the NMR data (to within 3 % of the dominant species) suggests that the heterogeneous fluorescence of Trp102 may arise from mechanisms independent of rotameric states of the Trp side-chain. The map for holo-azurin has only one conformation, and suggests a rotamer model may not be required to explain its heterogeneous fluorescence intensity decay. The backbone and Trp102 side-chain dynamics at 30 °C of F102W has been characterized based on an analysis of 15N NMR relaxation data which we have interpreted using the Lipari-Szabo formalism. High order parameter (S2) values were obtained for both the helical and loop regions. Additionally, the S2 values imply that the calcium binding CD and EF loops are not strictly equivalent. The S2 value for the indole side-chain of Trp102 obtained from the fluorescence, NMR relaxation and minimum perturbation data are consistent with a Trp moiety whose motion is restricted.

Introduction

Fluorescence spectroscopy is a popular tool for investigating protein structure and dynamics. Unavoidably, however, the fluorescence of a protein can report only indirectly; it is always necessary to infer the physicochemical behavior of the environment from the measured photophysical parameters predicated on the assumption that the photophysical responses to the environment are understood a priori. Inferences regarding the dynamics of the protein matrix and solvent are generally made from measurements of either the fluorescence intensity decays (lifetimes) or, from presumably direct determination of the rotational or librational motions of the fluorophore itself; such determinations being based on inferences drawn from fluorescence anisotropy decays and construction of plausible physical models.

The intensity decay of tryptophan (Trp) fluorescence in proteins and peptides, even those containing a single Trp moiety, is almost invariably heterogeneous requiring multi-exponential decay kinetics to describe the data (Beechem & Brand, 1985). A similar situation holds even for free Trp in aqueous solution at neutral pH (Szabo & Rayner, 1980) although the fluorescence of N-acetyl-tryptophanamide is generally agreed to be mono-exponential. A variety of models have been proposed in an attempt to rationalize the physical meaning of such heterogeneity, most of which, in one way or another, are based on the dynamics of the indole side-chain, hence by inference, the dynamics of the protein matrix. The rotamer model of Szabo & Rayner (1980) is fundamentally an extension of earlier work (Gauduchon & Wahl, 1978). The model proposes that different rotameric states of the alanyl side-chain of Trp are responsible for the observed heterogeneity, the proximity of the indole ring to the amino- and carboxy-functional groups being the principal determinants of the fluorescence lifetime. In contrast, the models of James et al. (1985) and Alcala et al. (1987) posit the existence of a quasi-continuous distribution of states which arises because the dynamics of the fluorophore allows it to sample multiple physical states each with a unique set of physicochemical interactions with the excited fluorophore giving rise to non-exponential decay. The width of the lifetime distribution is thought to reflect the number of unique microenvironments (conformational substates) sampled by the fluorophore and the rate of interconversions among these substates during the fluorescence lifetime. A third model that attempts to rationalize the fluorescence intensity decay of Trp is that of Bajzer & Prendergast (1993) which builds on the earlier findings of Petrich et al. (1983). At its core is the notion that heterogeneous fluorescence intensity decays may be obtained without the need to invoke the existence of different unique conformational states, or side-chain motions of the fluorophore per se as the heterogeneity is thought to result from quenching of the excited state by a variety of mechanisms not dependent on dynamics, the most probable being electron transfer. van Gilst et al. (1994) have proposed a fourth model which, in essence, is similar to that of Bajzer and Prendergast but includes specifically back transfer of an electron to a transient radical indole cation.

All of the above models can provide credible explanations for the existence of heterogeneous Trp fluorescence intensity decays and, given the complexity of protein structure, there is at present no justification a priori for choosing a particular model to explain the data in the absence of a clear structural basis for doing so. Ultimately, the selection of a particular model should depend on the availability of corroborating experimental or computational data. Thus, before invoking the rotamer model, for example, evidence must be adduced supporting the existence of the aforementioned rotamers for the system in question. Such evidence has been provided in peptides from high resolution nuclear magnetic resonance (NMR) data (Ross et al., 1992). Identification of relevant rotameric states in proteins may be difficult given that such states will not be detected by either NMR or X-ray crystallographic techniques unless they are sufficiently long lived and significantly populated. However, if the three-dimensional structure of the protein is known, computer simulations can be used to probe the possible existence of rotamers and should, at least in principle, allow quantitative assessment of the applicability of a rotamer model to explain the fluorescence decay data.

For experimental studies, the model chosen is obviously key. The least ambiguous data are likely to arise from proteins bearing a single Trp residue and preferably devoid of tyrosine, whose fluorescence intensity decay is heterogeneous. The tertiary structure of the protein should be known and preferably, the physicochemical properties should allow for measurement of the dynamics by both fluorescence and NMR spectroscopy. We have chosen to study a mutant form of carp parvalbumin which meets these criteria. Parvalbumins (Mr≈12,000), of which there are two classes, α and β, are acidic proteins which are predominantly found in fast-twitch muscle, usually at milli-molar concentrations intra-cellularly. Although their function remains uncertain, it appears that α-parvalbumins such as that from carp, serve as cytosolic calcium buffers; the β-isoforms are known to have immunomodulating properties (Brewer et al., 1989). The X-ray crystal structure of carp parvalbumin Kretsinger et al 1971, Moews and Kretsinger 1975, Swain et al 1989 reveals a compact globular protein having six α-helices denoted A-F whose packing delineate a hydrophobic core. Parvalbumins have three helix-loop-helix or EF-hand motifs (Kretsinger & Nockolds, 1973) which are numbered 2, 3 and 4 because the latter two domains are almost congruent with those of the homologous proteins calmodulin and troponin C (Kawasaki & Kretsinger, 1994). Domains 3 and 4, which are capable of binding both calcium and magnesium ions (Kawasaki & Kretsinger, 1994) are located between the C–D and E–F helices, respectively. The second domain, which does not bind metal ions, forms a cap covering the hydrophobic surface of lobes 3 and 4. The parvalbumin mutant used in this work is designated F102W and has the phenylalanine residue at position 102 in the wild-type protein replaced by Trp. The introduced Trp residue is immediately after the last metal-ion coordinating residue in site 4 (−z + 1 of the calcium binding loop) and is in the same location as the naturally occurring Trp residue in whiting (Joassin & Gerday, 1977) and cod parvalbumin (Hutnik et al., 1990b). Indeed, the wavelength of maximum fluorescence emission of F102W is similar to that reported for these two proteins Hutnik et al 1990a, Hutnik et al 1990b, Eftink and Wasylewski 1989. Furthermore, the absence of tyrosine residues in carp parvalbumin allows the unambiguous assignment of the fluorescence signal to the Trp residue.

This study was motivated primarily by one issue, namely, whether we could identify rotameric states of Trp from which we could rationalize heterogeneous fluorescence lifetimes. The value of such rationalization lies in defining the factors within the protein matrix which influence Trp dynamics. The ultimate objective is to engender confidence in our ability to use fluorescence techniques for detecting and quantifying protein dynamics. We determined the solution structure of F102W using multi-dimensional NMR spectroscopy to enable inferences between the environment of the indole side-chain and its fluorescence properties to be made. We then assess the likelihood that a rotamer model can explain the fluorescence intensity decay and order parameters of the indole side-chain of the introduced Trp residue of the mutant parvalbumin. A comparison to Trp48 of azurin (of Pseudomonas aeruginosa) is also made. The probability of side-chain rotamers was obtained by using the minimum perturbation mapping technique Shih et al 1985, Haydock 1993 which allows calculation of the potential energy surface for the indole side-chain as a function of its χ1 and χ2 dihedral angles. Additionally, we have sought to determine if the reported difference in the value of the overall rotational correlation time determined by NMR and fluorescence spectroscopy Palmer et al 1993, Kemple et al 1994, Kemple et al 1997, Yuan et al 1996 is applicable to this parvalbumin mutant.

Section snippets

NMR spectra, assignment and description of the structure

Figure 1 shows a representative 2D 1H-15N HSQC spectrum of calcium-saturated F102W at 30 °C. In general, the spectra were of high quality. Consequently, almost complete side-chain assignments of the residues in the protein were possible (Table A of Supplementary Material).

The three-dimensional structure of calcium saturated F102W was determined from the NMR data using a simulated annealing protocol from a total of 1039 NMR derived restraints. The 30 structures comprising the final ensemble all

Discussion

Tryptophan fluorescence is widely used to make inferences about the structure and dynamics of proteins. This is due in part to the sensitivity of the fluorescence technique and the relative ease with which fluorescence data can be acquired. However, despite recent advances in calculating the wavelength of fluorescence emission of Trp residues in proteins (Callis, 1997), the interpretation of fluorescence data remains largely qualitative because the manner in which the protein matrix influences

Protein preparation

The entire cDNA of carp parvalbumin was synthesized by the overlap fill-in method (Sambrook et al., 1989) based on the published sequence (Coffee & Bradshaw, 1973). A Trp codon (TGG) was introduced by thermal cycling to replace Phe (TTC) at position 102. The mutated DNA was cloned into the NcoI-BamHI cloning site of the pET3d expression plasmid. Mutations were verified by the standard dideoxy DNA sequencing method using the Sequenase II reagents (United States Biochemical, USA). Protein

Acknowledgements

We thank Dr Christopher Haydock for many helpful discussions and for calculating the 1La and 1Lb fluorescence order parameters from the minimum perturbation data. We also thank Dr Art Palmer III of Columbia University for making the MODELFREE program available. This work was supported by NIH grants GM34847 to F.G.P. and AR37701 to J.D.P.

References (80)

  • Y. Harpaz et al.

    Volume changes on protein folding

    Structure

    (1994)
  • C.M. Hutnik et al.

    Comparison of metal ion-induced conformational changes in parvalbumin and oncomodulin as probed by the intrinsic fluorescence of tryptophan 102

    J. Biol. Chem.

    (1990)
  • M. Ikura et al.

    Three dimensional NOESY-HMQC spectroscopy of a 13C-labelled protein

    J. Magn. Reson.

    (1990)
  • D.R. James et al.

    Distributions of fluorescence lifetimesconsequences for the photophysics of molecules adsorbed on surfaces

    Chem. Phys. Letters

    (1985)
  • L. Joassin et al.

    The amino acid sequence of the major parvalbumin of the whiting Gadus merlangus

    Comp. Biochem. Physiol.

    (1977)
  • L.E. Kay et al.

    New methods for the measurement of NH-CαH coupling constants in 15N-labelled proteins

    J. Magn. Reson.

    (1990)
  • L.E. Kay et al.

    Practical aspects of 3D heteronuclear NMR of proteins

    J. Magn. Reson.

    (1989)
  • L.E. Kay et al.

    Three-dimensional triple-resonance NMR spectroscopy of isotopically enriched proteins

    J. Magn. Reson.

    (1990)
  • M.D. Kemple et al.

    13C NMR and fluorescence analysis of tryptophan dynamics in wild-type and two single-trp variants of Escherichia coli thioredoxin

    Biophys. J.

    (1994)
  • K. Kinosita et al.

    A theory of fluorescence polarization decay in membranes

    Biophys. J.

    (1977)
  • R.H. Kretsinger et al.

    Carp muscle calcium-binding protein. II. Structure determination and general description

    J. Biol. Chem.

    (1973)
  • J. Kuszewski et al.

    The impact of direct refinement against proton chemical shifts on protein structure determination by NMR

    J. Magn. Reson.

    (1995)
  • A.M. Mandel et al.

    Backbone dynamics of Escherichia coli ribonuclease HIcorrelations with structure and function in an active enzyme

    J. Mol. Biol.

    (1995)
  • P.C. Moews et al.

    Terbium replacement of calcium in carp muscle calcium-binding parvalbuminan X-ray crystallographic study

    J. Mol. Biol.

    (1975)
  • H. Nar et al.

    Crystal structure analysis of oxidized Pseudomonas aeruginosa azurin at pH 5.5 and pH 9.0. a pH-induced conformational transition involves a peptide bond flip

    J. Mol. Boil.

    (1991)
  • A. Pardi et al.

    Calibration of the angular dependence of the amide proton-Cα proton coupling constants, 3JHNα, in a globular protein. Use of 3JHNα for identification of helical secondary structure

    J. Mol. Biol.

    (1984)
  • T.L. Pauls et al.

    Efficient complementary DNA amplification and expression using polymerase chain reaction technology

    Methods Enzymol.

    (1993)
  • A.L. Swain et al.

    Restrained least squares refinement of native (calcium) and cadmium-substituted carp parvalbumin using X-ray crystallographic data at 1.6 Å resolution

    J. Biol. Chem.

    (1989)
  • K.K. Turoverov et al.

    The environment of the tryptophan residue in Pseudomonas aeruginosa azurin and its fluorescence properties

    Biophys. Chem.

    (1985)
  • P. Yuan et al.

    Structure and dynamics of melittin in lysomyristoyl phosphatidylcholine micelles determined by nuclear magnetic resonance

    Biophys. J.

    (1996)
  • G. Zhu et al.

    Improved linear prediction of damped NMR signals using modified “forward-backward” linear prediction

    J. Magn. Reson.

    (1992)
  • Z. Zolnai et al.

    Magnetization exchange network editingmathematical principles and experimental demonstration

    Chem. Phys.

    (1995)
  • Z. Zolnai et al.

    Zooming, a practical strategy for improving the quality of multi-dimensional NMR spectra

    J. Magn. Reson. ser. A

    (1996)
  • A. Abragam

    Principles of Nuclear Magnetism

    (1961)
  • H. Akima

    A method of bivariate interpolation and smooth surface fitting for irregularly distributed data points

    ACM Trans. Math. Software

    (1978)
  • M. Akke et al.

    Effects of ion binding on the backbone dynamics of calbindin D9k determined by 15N NMR relaxation

    Biochemistry

    (1993)
  • T. Alattia et al.

    Assignment of 13C resonances and analysis of relaxation properties and internal dynamics of pike parvalbumin by 13C-NMR at natural abundance

    Eur. J. Biochem.

    (1996)
  • J.R. Alcala et al.

    Interpretation of fluorescence decays in proteins using continuous distributions

    Biophys. J.

    (1987)
  • C. Baldellon et al.

    15N-NMR relaxation studies of calcium-loaded parvalbumin show tight dynamics compared to those of other EF-hand proteins

    Biochemistry

    (1998)
  • G. Barbato et al.

    Backbone dynamics of calmodulin studied by 15N relaxation using inverse detected two-dimensional nmr spectroscopythe central helix is flexible

    Biochemistry

    (1992)
  • Cited by (0)

    1

    Edited by P. E. Wright

    View full text