11286
J. Phys. Chem. B 2000, 104, 11286-11295
Nanosecond Fluorescence Dynamic Stokes Shift of Tryptophan in a Protein Matrix Michel Vincent,*,† Anne-Marie Gilles,‡ Ine` s M. Li de la Sierra,‡,§ Pierre Briozzo,⊥ Octavian Baˆ rzu,‡ and Jacques Gallay† LURE Baˆ timent 209D, UMR 130 CNRS, UniVersite´ Paris-Sud, 91898 Orsay Cedex, France, and Laboratoire de Chimie Structurale des Macromole´ cules, URA 1129 CNRS, Institut Pasteur, 75724 Paris Cedex 15, France, and Laboratoire de Biochimie Structurale, URA 1129 CNRS, Institut Pasteur, 75724 Paris Cedex 15, France, and Laboratoire de Chimie Biologique, U206 INRA Paris-Grignon, 78850 ThiVerVal-Grignon, France ReceiVed: February 18, 2000; In Final Form: July 20, 2000
The fluorescent dynamic Stokes shift (FDSS) method has emphasized a time-dependent dipolar relaxation process around the single tryptophan residue (Trp31) in cytidine monophosphate kinase from E. coli (CMPK). This Trp residue, located close to the protein surface in a hydrophobic pocket, is weakly accessible to acrylamide, a water-soluble quencher. It exhibits fluorescence characteristics suitable for a detailed study of dipolar relaxation: (i) a fluorescence decay almost monoexponential and (ii) a fluorescence emission maximum of 329 nm, in a wavelength range intermediate between those of a completely polar environment and a strongly apolar one. This emission maximum is shifted to 320 nm by decreasing the temperature to 230-240 K with glycerol as cryoprotectant. A time constant (∼100 ps) affected by a negative preexponential, evidenced in the red-edge fluorescence intensity decays, supports the existence of an excited-state reaction. A multiphasic FDSS (with time constants ranging from ∼100 ps to several nanoseconds) with a total amplitude between 130 and 340 cm-1 (0.4-1 kcal.mol-1) is observed in the temperature range 293-232 K and not below.
Introduction Dielectric screening of charges or dipoles of proteins by solvent molecules or other polar groups of the protein likely influences many features of the protein structure and reactivity.1-3 This electrostatic aspect of protein dynamics has been tackled by different experimental approaches. They include molecular dynamic simulations (see Simonson4 for a review), physical measurements such as pK shifts,5,6 energy transfer,7 and dielectric dispersion of dry protein powders.8 Time-resolved fluorescence of polarity-sensitive probes can provide valuable information on solvation processes in the picoto nanosecond time scale in polar liquids.9-13 In a protein matrix, the solvent dynamics properties are not appropriately described by such an oversimplified picture.14-18 A polarity-sensitive probe should display a dipole moment value significantly higher in the excited state than in the ground state. Fluorescent molecules such as indole are potential reporters of dipolar relaxation.19,20 They are particularly interesting since one of its derivatives, the amino acid tryptophan, is naturally present in proteins. Their solvatochromic properties have their origins in the enhancement of the permanent dipole moment of the excited 1La state (5.5 D for indole) with respect to the one of the ground state (2 D) (see Callis21). The 1Lb state has a small dipole change accompanying the excitation and the issue of this light emission affects only the limited wavelength range below 310 nm.22 The peak emission values, depending on the solvent, range within a wide energy gap between 28 200 and 32 800 cm-1 for water and hexane, respectively.23-28 The * Corresponding author: Fax/telephone: 33 1 64 46 80 82. E-mail:
[email protected]. † Universite ´ Paris-Sud. ‡ Laboratoire de Chimie Structurale de Macromole ´ cules, Institut Pasteur. § Laboratoire de Biochimie Structurale, Institut Pasteur. ⊥ Laboratoire de Chimie Biologique, U206 INRA Paris-Grignon.
expected maximal spectral shift is as much as 4600 cm-1, at least in the situation where both emission bands are involved. This last value for the expected shift has to be lowered if one considers only the environment-sensitive 1La band. For instance, in glass-forming polar solvents, a temperature-dependent spectral shiftsby 3000 cm-1 or moreshas been observed from fluid to glass media. Studies of dielectric relaxation have been reported with indole and derivatives from the collection of time-resolved emission spectra in viscous dipolar solvents such as mono- or polyalcohols26,29-32 and micellar systems.33 Whatever the fluorescent probe, this process is revealed by the existence of rise time components (a time component coupled with a negative preexponential term) in the fluorescence intensity decays measured in the red-edge of the emission spectrum and by a time-dependent spectral shift of the center of mass of the fluorescence emission spectrum.34 While such measurements are quite straightforward in alcoholic media since the emission decay is monoexponential, this is not the case in proteins, where monoexponential decays are the exception.35,36 Ground-state heterogeneity associated with the existence of rotamers is a generally accepted interpretation.37-39 Excited-state reactions can also occur, although this has been rarely reported.40 Emission decay heterogeneity complicates the observation of the characteristic features of dipolar relaxation. The apparent rate constants characteristic of the underlying dipolar relaxation processes can be undistinguishable from the emission decay constants due to their overlap. The multiexponential behavior of a protein tryptophan decay may partly be a misinterpretation of a dipolar relaxation process in terms of ground-state heterogeneity. It can be especially true when such fluorescence emission is analyzed by using a decay associated spectra approach (DAS): the invariance of the time-component parameters of the fluorescence intensity has to be checked as a function
10.1021/jp000638x CCC: $19.00 © 2000 American Chemical Society Published on Web 10/28/2000
Stokes Shift of Tryptophan
Figure 1. Ribbon representation of the three-dimensional structure of the CMPK from E. coli in the close vicinity of the Trp31. The side chains of amino acid residues Trp31, Leu29, Pro124, and Pro6 as well as two water molecules (W51 and W91) are represented as ball-andsticks. C and N are the C- and N-termini amino acid residues, respectively. (A figure generated by MOLSCRIPT86.)
of the emission wavelength. An additional complexity stems from the existence of several polar groups of different chemical natures and dynamics which are usually present in the close vicinity of the indole ring. Thus, whenever these groups are involved in the dipolar relaxation process, the associated relaxation rates will be distributed over a wide time scale.17 In this work, we have focused on a single tryptophancontaining protein, the cytidine monophosphate kinase from E. coli (CMPK).41 It displays a key role in the metabolism of nucleic acids and of phospholipids. It is a monomeric protein which contains 227 amino acid residues and belongs to the class of “long chain” nucleotide monophosphate kinase family (NMPK).42 Its crystal structure was recently solved at 1.75-Å resolution,43 allowing the knowledge at the atomic level of the Trp31 environment. This residue is located at the opposite side of the nucleotide-binding sites, in a partially hydrophobic region, but not very far from the protein surface (Figure 1). It is shown to display spectroscopic characteristics suitable for the detailed study of dipolar relaxation by time-resolved fluorescence spectroscopy. The dipolar dynamics around the Trp31 residuesin the subnanosecond and nanosecond time rangeswas monitored by time-resolved fluorescence emission spectra measurements as a function of temperature with glycerol as a cryoprotectant at different concentrations. Large range temperature effects (from ambient to cryogenic) on the steady-state and time-resolved fluorescence properties of the Trp31 residue are reported and in particular a FDSS. The presence of glycerolsfrom 52 to 75% (w/w)snot only allows exploration of the cryogenic range of temperature but permits also an easier description of the FDSS by slowing down the relaxation processes. Besides, the existence of a thermal-induced transition at 220-230 K, affecting the dynamics of the environment of Trp31 residue, is reported and its implication is discussed with regards to the FDSS. Materials and Methods Chemicals. All chemicals were of the highest grade commercially available.
J. Phys. Chem. B, Vol. 104, No. 47, 2000 11287 Protein Preparation. The enzyme from the overproducing strain was purified as previously described and its activity was determined by a coupled spectrophotometric assay.41,42,44 The protein was diluted in a phosphate buffer, 0.01 M, pH 7.4, containing or not containing glycerol. Protein concentration was comprised between 8 and 30 µM. Fluorescence Measurements. Steady-state fluorescence emission spectra, fluorescence intensity, and anisotropy decays were obtained by the time-correlated single photon counting technique from the polarized components Ivv(t) and Ivh(t) on the experimental setup installed on the SB1 window of the synchrotron radiation machine Super-ACO (Anneau de Collision d’Orsay).45 A Hamamatsu microchannel plate R3809U-52 was utilized to detect the fluorescence photons. Data for Ivv(t) and Ivh(t) were stored in separate 2K memories of a plug-in multichannel analyzer card (Canberra) in a DESKPRO microcomputer (Compaq). Time resolution was routinely chosen between ∼12 and 25 ps per channel, depending on the mean excited-state lifetime value. The instrumental response function was automatically collected in alternation with the parallel and perpendicular components of the polarized fluorescence decay by measuring the scattering at the sample emission wavelength of a glycogen solution. The automatic sampling of the data was driven by the microcomputer. The steady-state fluorescence spectrum was obtained on the same apparatus. The storage ring provides a light pulse with a full width at half-maximum (fwhm) of typically 700 ps when working at a frequency of 8.33 MHz in the double bunch mode. It has been already demonstrated by the construction of simulated data32 that such a broad instrument response function did not preclude the recovery of time components as short as 50 ps. The only restrictive condition for such a recovery is a high stability of the instrument response function, a condition that is totally fulfilled with the synchrotron radiation pulsed light source. Numerical methods used for data analysis are also important to determine the effective time resolution of a fluorescence decay experiment. The importance of the convolution integral algorithms required for the analysis of time-resolved fluorescence data has been previously underlined.46 These results show that a reduction of the width of the instrument response function does not necessarily lead to better time resolution, unless a suitably accurate interpolation is used in the iterative convolution process. Cryogenic experimental conditions were obtained with a variable-temperature pourfill Janis cryostat VPF-100 (Janis Research Co., Wilmington, MA). The system consists of a coldfinger, a radiation shield, an aluminum vacuum shroud, and an electrical feedthrough port. It uses liquid nitrogen in conjunction with a thermal impedance and a built-in heater to operate at any stabilized temperature (within 0.5 K) between 83 and 320 K.13 Either in the steady-state or time-resolved modes, blanks were subtracted in the same experimental conditions when needed and never exceeded 5% as expressed in total number of counts for the noise-to-signal ratios. Data Analysis of Fluorescence Lifetime and Rotational Correlation Time Distributions. Analysis of the fluorescence intensity decay data as a sum of 150 exponential terms was performed by using the maximum entropy method (MEM).47 The program calls MEMSYS 5 (MEDC Ltd., U.K.) as a library of subroutines. Optionally, the program can handle a 150dimensioned vector, without any a priori assumption on each amplitude sign and on the number of significant vector components remaining at the end of the analysis.32 Then, these significant vector components are collected in families which
11288 J. Phys. Chem. B, Vol. 104, No. 47, 2000
Vincent et al.
number, at least in this work, will never exceed three terms. Neither baseline nor shift of the instrumental response was applied during the data analysis procedure. Only an extra time component, of zero value and accounting for elastic light scattering, was used in addition to the 150 other ones. Analysis of the polarized fluorescence decay data was performed with MEM using the classical model for the anisotropy, in which each rotational motion is assigned to each fluorescence lifetime, as previously described.45 Time-Resolved Fluorescence Emission Spectra (TRES) of Trp31: Collection and Analysis. TRES were reconstructed in each experimental condition from 14 individual decays as a function of the emission wavelength from 305 to 370 nm (bandwidth 8 nm) with a 5-nm step (27 recorded spectra with a 2.5-nm step in some experiments). Each individual curve was deconvoluted with the MEM program using the negative amplitude option. The integral of each impulse decay curve was normalized to the corresponding steady-state fluorescence emission wavelength recorded on the same instrument with identical optical conditions. By collecting the vertical (Ivv) and horizontal (Ivh) fluorescence intensity components and by taking into account a β correction factor, the calculated impulse as well as the steady-state fluorescence intensities (Ivv + 2βIvh) are de facto corrected for the difference in the transmission of the polarized light by the optics. Steps of ∼15 ps were used for the spectral shift reconstruction. For the quantitative description of the spectral shift, the centers of mass in frequency were computed from the raw spectra. The shift function C(t) was defined classically11 as
C(t) )
νj(t) - νj∞ νj0 - νj∞
where νj(t), νj0, and νj∞ are the center of mass values in frequency at time t, 0, and ∞, respectively. Relaxation times were determined from a MEM analysis of C(t), as a sum of exponential terms:12,17,32
C(t) )
∑i aie-t/τ
Ri
Analytical Description of the Transient Spectra from TRES. As early established,48 the shape of the absorption spectra of complex organic compounds can be satisfactorily analyzed by the use of a four-parameter log-normal function (a skewed Gauss equation). Later, it was demonstrated that fluorescence spectra of indole, tryptophan and some derivatives could also be accurately approximated versus the frequency scale (wavenumber ν) by the same function, but in its mirrorsymmetric form:28
I(ν) ) Im exp{ln(2)/ln2F) ln2[(a - ν)/(a - νpeak)]} (at ν < a) I(ν) ) 0 (at ν g a) where Im ) I (νpeak) is the maximal fluorescence intensity, νm is the wavenumber of the spectrum (in cm-1) at its maximum, F is the asymmetry parameter, and a is the function limiting point such as
a ) νm + fwhm‚F/(F2 - 1) where fwhm (in cm-1) is the spectral full width at half of its maximum value.
Figure 2. Time-resolved fluorescence intensity decay of Trp31 in CMPK at pH 7.4 and at 273 K. Upper part: the experimental decay (b) and the instrumental response function (O). Inset: plot of the deviation function (weighted residuals). Lower part: the corresponding excited-state lifetime distribution recovered by MEM. Excitation wavelength: 292 nm (bandwidth 2 nm). Emission wavelength: 325 nm (bandwidth 5 nm). Protein concentration: 30 µM.
The time-resolved fluorescence emission spectra were analyzed in shape with a model involving one log-normal function, using a classical nonlinear least-squares regression algorithm. The starting parameters (Im, a, and νm) where introduced by the user for the first spectrum (at time t ) 0) and the resulting fitted values (Im, a, and νm) at time t where used, in a recurrent way, as starting parameters to fit the spectrum at time t + δt. δt was chosen as the time resolution used for the record of the fluorescence decays, i.e., ∼15 ps per channel. The F value (the asymmetry parameter) was chosen as 1.2428 and kept constant during the fit procedure. Calculation of the Solvent Accessibility of Trp31. Solvent accessibility was calculated with AREAIMOL 3.149 with 1.4 Å for the probe radius assuming a single radius for each element. Results Steady-State and Time-Resolved Fluorescence Emission of Trp31 in CMPK from E. coli in Neutral Buffer. In phosphate buffer pH 7.4, the fluorescence emission spectrum of Trp31 in CMPK from E. coli exhibits a maximum value of 329 nm, indicating an environment of the indole ring of low polarity, largely screened from the dielectric properties of the bulk solvent as compared to 350 nm, the value for indole in water.24 This is fully compatible with the three-dimensional structure of the protein43 showing that the indole ring is located within a crevice surrounded by apolar residues mainly. Two water molecules (W51 and W91) are nevertheless observed in the crystal at respective distances of 5.8 and 5.0 Å from the N atom of the indole ring (Figure 1). The fluorescence intensity decay of Trp31 emission, at the maximum emission wavelength, is not monoexponential, as it can be seen on the experimental data curve (Figure 2, upper part). Nevertheless, the analysis by MEM shows that it can be described by one major broad population of lifetimes centered on a mean value of 1.4 ns, which represents 90% of the excited state populations, and by two minor broad peaks centered around
Stokes Shift of Tryptophan 0.3 and 4 ns (Figure 2, lower part). The short value of the major lifetime relative to that of indole or N-acetyltryptophanamide in aqueous solutions (respectively 4.7 and 3.0 ns)32,50 is likely due to the quenching effect of three peptide bonds of Pro6, Leu29, and Pro124. Stern-Volmer quenching experiments with a series of model compounds containing one or two primary or secondary amides showed that the peptide bond quenches indole fluorescence by an electron-transfer mechanism.51 For Leu29 and Pro124, the oxygen atom of each carbonyl group is located at 3.7 or 4.04 Å of the N atom of the indole ring, respectively. For Pro6, the oxygen atom is located at 3.81 Å of the Cζ3 atom of the indole ring.43 Accessibility of the Trp31 Residue to the Solvent: Fluorescence Quenching by Acrylamide. The localization of the Trp31 residue within an hydrophobic cleft in the CMPK matrix, close to the protein/water interface in the crystal structure, raises the question of its accessibility to the solvent. This accessibility was estimated by fluorescence quenching with acrylamide as a water-soluble fluorescent quencher.52 Stern-Volmer representation of the lifetime value versus acrylamide concentration gives a good linear representation (regression coefficient: 0.994). The slope corresponds to a Stern-Volmer quenching constant (Ksv) value of 1.29 M-1, from which a bimolecular quenching constant value of ∼109 M-1 s-1 is calculated. The estimated area accessible to the quencher is about 10-15%.52,53 Calculations of the solvent accessibility, performed using the crystal structure of the protein,43 provide an area accessible to the solvent of 46 Å2 compared to 378 Å2 for tryptophan in solution. A relative accessibility of 12% is obtained, in good agreement with the value estimated above by acrylamide quenching experiments. In conclusion, this low accessibility is compatible with the relatively blue fluorescence emission of Trp31. Steady-State and Time-Resolved Fluorescence Emission of Trp31 in Glycerol/Water Mixtures. Our concern was to study the local dynamicssin the subnanosecond and nanosecond time rangesaround the Trp residue in this protein, for which we needed to explore a wide range of temperatures. For this purpose, we had to employ cryogenic cosolvents such as glycerol. We therefore evaluated the effect of this cosolvent at different concentrations on the fluorescence properties of the Trp31 residue and on the local conformation and dynamics of the protein. At room temperature, the modification of the solvent composition by glycerol addition influences the fluorescence properties of Trp31. The quantum yield gradually increases from ∼0.06 in buffer to 0.14 upon addition of cosolvent up to 75% (w/w). In the same glycerol concentration range, the maximum emission wavelength is shifted to the blue by 4 nm (from 329 to 325 nm). At the maximum emission wavelength, the major lifetime value increases in the same proportion as the quantum yield, i.e., from 1.4 ns in buffer to 2.1 ns, and further to 2.9 ns in glycerol/water of 52, 60, and 75% (w/w), respectively. The lifetime profile is not strongly changed in the presence of glycerol: it is always characterized by a major broad lifetime distribution but with one additional small short component in the glycerol concentration range of 52-75% (data not shown). With pure indole dissolved in these glycerol/water mixtures, opposite changes on the lifetime values are observed. The lifetime value of indole in solvents decreases from 4.5 ns in pure water, to 3.9 ns and further to 3.8 ns in 60 and 75% glycerol, respectively (data not shown). Such changes of the Trp31 excited-state properties are probably not the result of a
J. Phys. Chem. B, Vol. 104, No. 47, 2000 11289 TABLE 1: Fluorescence Anisotropy Decay Parameters of Trp31 in E. coli CMPK under Different Conditionsa samplea
θ1, ns (β1)
θ2, ns (β2)
CMPK 293 K no glycerol CMPK 293 K glycerol 75% CMPK 273 K no glycerol CMPK 270 K glycerol 75% CMPK 232 K glycerol 75%
1.3 0.025 5.5 0.025 1.7 0.016 3.2 0.011 4.3 0.005
15 0.146 ∞ 0.148 21 0.158 ∞ 0.157 ∞ 0.198
At)0
ωmax deg 18
0.171 18 0.173 14 0.174 12 0.168 7 0.203
a A multi-exponential model was assumed, each of them being associated with a time-constant θi and weighted by βi. Glycerol concentrations in w/w. Protein concentration was ∼8 µM. Excitation wavelength: 292 nm. The values of the wobbling-in-cone angle describing the Trp31 internal motion were calculated from Kinosita et al.54 as β2/At)0 ) [1/2 cos ωmax (1 + cos ωmax)]2. At)0, the anisotropy at zero time, is the sum over the βi terms.
direct effect of the cosolvent either on the bulk solvent polarity or on the accessibility of the Trp31 to the glycerol/water solvent. Time-Resolved Fluorescence Anisotropy Decay Measurements of the Trp31 Residue. The almost unimodal fluorescence intensity decay of Trp31 (Figure 2) is an indication of a weak motion inside the protein matrix. This is confirmed by the fluorescence anisotropy decay measured at the fluorescence emission maximum, which is dominated by the Brownian motion of the protein molecule in a monomeric form. The measured rotational correlation time of 15 ns corresponds exactly to the expected value for a spherical protein hydrated at 50% (w/w) and containing 227 amino acids (with a mean weight of 110 D for each amino acid). An internal motion of the Trp31 of small amplitude is, however, also detected in the nanosecond time range (Table 1). This mobility is slowed in the presence of glycerol at 293 K, in agreement with the increase of the major excited-state lifetime value provoked by glycerol addition. The amplitude of the motion, estimated by the wobbling-in-cone angle ωmax,54 is further strongly reduced by decreasing the temperature (Table 1). Effect of Temperature on the Fluorescence Emission Parameters of the Trp31 Residue: Existence of a Dynamical Transition from a Glassy to a Flexible State. The steadystate fluorescence emission spectrum of the protein was recorded in a glycerol/water mixture (75% w/w) as a function of temperature (from 293 to 133 K). As a control, the plot of the maximum emission value of pure indole in the same medium is also reported as a function of the temperature (Figure 3). A gradual blue shift of ∼5 nm of the maximum emission, without significant change in the spectral shape (not shown), is observed upon decreasing the temperature. The variation with the temperature of the spectral shift in the protein is not linear and strongly differs from that of pure indole. The broad change which initiates at 280 K and ends up at 230 K (Figure 3) suggests the existence of a dynamical transition in this protein, in this temperature range. Temperature also strongly affects the fluorescence intensity decay parameters of the Trp31 residue. The value of the major excited-state lifetime is enhanced by about 2.5 times from 2.2 ns to reach 5.3 ns (Figure 4), like the fluorescence intensity (not shown), at a glycerol/water ratio of 60% (w/w) when the temperature decreased to 150 K. This variation is, however, not linear with temperature. The lifetime value increases sharply from 290 to 230 K, after which the slope decreases strongly from 230 to 133 K (Figure 5). The transition occurs in the range
11290 J. Phys. Chem. B, Vol. 104, No. 47, 2000
Figure 3. Steady-state fluorescence emission maximum of Trp31 in CMPK (b) in glycerol 75% (w/w) as a function of temperature. Control experiment with indole (O) under the same experimental conditions. Excitation wavelength: 292 nm (bandwidth 2 nm).
Figure 4. Temperature effect on the lifetime distribution of the Trp31 of the CMPK (30 µM) dissolved in glycerol 60% (w/w) as a function of temperature. The excitation wavelength was 295 nm (bandwidth 5 nm), and the emission was set at 325 nm (bandwidth 10 nm).
220-230 K, whatever the glycerol content, as for the fluorescence intensity (not shown) and for the spectral position (Figure 3). It appears therefore that temperature lowering affects the internal dynamics of the Trp31 environment in CMPK in such a way that the dynamic quenching by close carbonyl groups of probably Leu29, Pro6, and Leu124 peptide bonds, is considerably reduced. In the glassy state, the protein is frozen into a distribution of conformational states. Mobility is restrained to group rotations and small-amplitude vibrational and torsional motions. In the flexible state, segmental motions, and conformational rearrangement become possible. These effects are characteristic features of a “dynamic” transition.55 TRES of the Trp31 Residue as a Function of Temperature and Glycerol. The blue shift of the steady-state fluorescence emission that we observed upon decreasing the temperature in
Vincent et al.
Figure 5. Variation of the major excited-state lifetime of Trp31 in CMPK as a function of temperature in CMPK (30 µM) dissolved in the presence of 52% (O), 60% (4), and 75% (0) glycerol (w/w). Excitation wavelength: 295 nm (bandwidth 5 nm). Emission wavelength: 330 nm (bandwidth 10 nm).
the absence or in the presence of glycerol led us to assume the existence of a relaxation process of the environment around the indole excited-state dipole. A characteristic feature of the presence of an excited state reaction such as dipolar relaxation (assumed as a two-states or a continuous model) is the existence of a buildup of the fluorescence signal in the red-edge of the fluorescence emission spectrum. We systematically tried to detect such a feature. An excitation wavelength of 292 nm was used in order (i) to optimize the signal of indole with the lowest possible tyrosine fluorescence contamination, (ii) to minimize the red-edge effect, which is known to cancel the negative components owing to the inhomogeneous characteristics of the absorption spectrum,32 and (iii) to record the decay as close as possible to the blueedge emission. The fluorescence decays were therefore systematically recorded as a function of the emission wavelength, in different conditions of glycerol concentration and/or temperature. The analyses of the data at all emission wavelengths were systematically performed by MEM using the model allowing the recovery of negative preexponential terms, without any a priori assumption on their existence. This allowed us to reconstruct TRES and to evaluate the time constants characterizing the fluorescence shift (see Materials and Methods). Control experiments were also performed with indole in the same bulk solvent, at comparable temperatures (Table 2). For CMPK in neutral buffer and at room temperature, no negative preexponential terms were detected (not shown). At a lower temperature of 273 K, however, a short time component of ∼100 ps associated with a negative preexponential term is clearly evidenced in the red-edge of the emission spectrum, whatever the glycerol content (Figure 6). A fit using a model allowing only the presence of positive preexponential terms does not describe satisfactorily the experimental data. The goodness of the fit criteria (comparison of the χ-squared values and visual inspection of the weighted residues, the deviation function) clearly indicates the necessity of the presence of negative terms (Figure 7). The occurrence of negative preexponential terms is much more pronounced in the case of Trp31 than for indole in the same solvent mixture and temperature conditions (Figure 6). It can also be remarked that at the emission maximum, an additional short time constant of ∼150-300 ps, associated with a positive amplitude, can be detected on the lifetime profile
Stokes Shift of Tryptophan
J. Phys. Chem. B, Vol. 104, No. 47, 2000 11291
TABLE 2: C(t) Parameters (Defined in Materials and Methods) of Trp31 in E. coli CMPK and Indole under Different Conditions of Temperature at a Glycerol/Water Ratio of 75% (w/w)a T, K
τR1, ns (a1)
τR2, ns (a2)
τR3, ns (a3)
∆ν, cm-1
ν0, cm-1
ν∞, cm-1
νss, cm-1
293# 281 270 261 254# 232# 203 270#,b 233b
0.07 (0.14) 0.07 (0.13) 0.16 (0.22) 0.13 (0.14) 0.20 (0.22)
0.26 (0.52) 0.43 (0.39) 0.70 (0.28) 0.61 (0.27) 0.97 (0.18) 0.56 (0.16)
0.7 (0.34) 6.0 (0.48) 11.4 (0.50) 15.3 (0.59) 17.6 (0.60) 25.7 (0.84)
123 213 283 326 337 321
30 404 30 462 30 494 30 505 30 523 30 526
30 281 30 249 30 211 30 179 30 186 30 205
0.025 (0.03) 0.13 (0.13)
0.30 (0.53) 0.48 (0.22)
440 926
29 989 30 945
29 549 30 019
30 315 30 336 30 313 30 346 30 384 30 448 30 512 29 607 30 205
0.81 (0.44) 6.3 (0.65)
a For the sake of comparison, νss is the center of mass of the steady-state spectrum. Excitation wavelength: 292 nm. Protein concentration: ∼8 µM. Time-resolved fluorescence emission spectra are reconstructed from the deconvolution of the individual decays collected at 5 or 2.5 (labeled #) nm. b Indole in glycerol/water 75% (w/w).
Figure 6. Comparative effect of glycerol content on the excited-state lifetime distribution of Trp31 in CMPK and indole measured in the red part of the fluorescence spectrum (340 and 350 nm, respectively). The CMPK is dissolved in 0.01 M phosphate buffer, pH 7.4, at 273 K (a), in the presence of 52% (w/w) glycerol at 271 K (b) and in the presence of 75% (w/w) glycerol at 271 K (c). Pure indole is dissolved in water at 273 K (d), in the presence of 52% (w/w) glycerol at 271 K (e), and in the presence of 75% glycerol at 271 K (f). Excitation wavelength 292 nm for Trp31 (bandwidth 5 nm).
measured in the temperature range of 273-244 K which almost disappears at lower temperature (Figure 4). We have previously shown by simulation data that our instrumentation was able to recover time constants as short as 50-80 ps.32 An example of variation of the lifetime amplitude profiles as a function of the emission wavelength in the presence of glycerol 75% (w/w) at 270 K is shown in Figure 8. The major peak of the CMPK lifetime distribution profile remains quite stable from 305 to 370 nm, with a population mainly centered around a value of 3.7 ( 0.1 ns. In the extreme blue-edge of the fluorescence emission spectrum, a short lifetime peak associated with a positive amplitude is, however, present. It disappears at longer emission wavelength. A negative peak is increasingly visible when going more to the red-edge part of the fluorescence emission spectrum, and the associated apparent lifetime value is centered on ∼100 ps (Figure 8). The existence of the fluorescence build-up kinetics in the long wavelength edge of the fluorescence emission spectrum is an indication of a FDSS. Such a Stokes shift is observed after
Figure 7. Example of recovery of the time constant distribution maximum of Trp31 in CMPK (30 µM) in 0.01 M phosphate buffer, pH 7.4, with 75% glycerol (w/w) at 232 K by MEM using a model of 150 exponential terms. Upper part: only positive preexponential terms are allowed (χ2 ) 1.19). Lower part: both negative and positive preexponential terms are allowed (χ2 ) 1.03). In the inset are presented the corresponding plot of the weighted residuals. Excitation wavelength: 292 nm (bandwidth 2 nm). Emission wavelength 350 nm (bandwidth 10 nm).
reconstruction of the fluorescence emission spectra as a function of time as shown for instance at the temperature of 270 K (Figure 9, lower part). It is interesting to note that the negative preexponentials observed above 340 nm are associated with faster time constants than any of the positive amplitudes observed below 340 nm (Figure 8). This implies that the fluorescence spectrum is gaining amplitude on the red side, without losing any on the blue side, and should thus become broader. Such a broadening is effectively observed in the fwhm time evolution computed from a Gaussian log-normal fit of each of the transient spectra (Figure 9, upper part). On the average, the spectral shift is small but measurable. Its value is comprised between ∼120 and ∼340 cm-1. This corresponds to a stabilization energy of the excited-state ranging from ∼0.4 to ∼1 kcal‚mol-1. The value of the center of mass at infinite time (νj∞) remains constant whatever the temperature (30218 ( 39 cm-1) whereas the value at zero time (νj0) increases upon decreasing the temperature (Table 2). The value of the center of mass of the steady-state emission spectra stays between these two (νj∞ and νj0) values. It is closer to νj∞ at room
11292 J. Phys. Chem. B, Vol. 104, No. 47, 2000
Figure 8. Excited-state lifetime distribution of Trp31 in CMPK (30 µM) in 0.01 M phosphate buffer, pH 7.4 with glycerol 75% (w/w) as a function of the emission wavelength. Excitation wavelength: 292 nm (bandwidth 5 nm). Temperature: 270 K.
temperature and reaches the νj0 value at cryogenic temperature (Table 2). This is explained by the progressively slower kinetics of the spectral shift position with decreasing temperature. This indicates that the relaxation process is fully described in the nano- and in the subnanosecond time range. The time shift of the center of mass defining the position of the emission spectrum is not monoexponential. Analysis by MEM of the C(t) function (defined in Materials and Methods) with a sum of exponential terms as a model gives three time constants ranging from hundreds of picoseconds to several nanoseconds (Table 2). The proportion of the slow nanosecond process increases when the temperature decreases from 293 to 232 K. The fastest component which represents 14-20% of the decrease of the excited-state energy remains invariant with temperature. The intermediate component increases from ∼0.3 to 1 ns upon decreasing the temperature. Its amplitude decreases from 52% at room temperature to 16% at 203 K to the benefit of the slowest component, which becomes dominant at low temperature where it represents 84% of the excited-state energy loss. The values of the slowest component also increase with decreasing temperature from 0.7 to 26 ns. At temperatures lower than 203 K, no spectral shift can be observed any more and the fluorescence decays of the red-edge part of the emission spectra do not exhibit any negative peaks (data not shown). TRES experiments were also performed in other cryogenic water/glycerol mixtures containing lower amounts of cosolvent (20 and 52%). The observed FDSS were of the same order of magnitude. The kinetics of the spectral shift were multiphasic. Three temperature sensitive time constants were observed by MEM analysis of the data as in the presence of 75% glycerol. The slope of an Arrhenius representation for the longest relaxation process measured in the presence of 75% glycerol, in the temperature range where the variations are significant,
Vincent et al.
Figure 9. Lower part: the CMPK fluorescence spectral evolution (after the deconvolution of each individual decay for each emission wavelength) as a function of the time following the illumination. Experimental wavelengths used (converted in wavenumber) (O) and fitted spectra (s) by using a Gaussian log-normal model (see Materials and Methods). The impulse transient spectra shown (sorted by the decrease in their intensity with time) correspond to times of 0, 0.075, 0.149, 0.224, 0.298, 0.447, 0.596, 0.746, 0.895, 1.044, 1.342, 1.566, 2.162, 2.684, 3.206, 3.727, 5.069, and 8.126 ns after excitation. Upper part: the time evolution of the fwhm parameter computed from the fit of each transient spectrum with a Gaussian log-normal model.
Figure 10. Arrhenius representation of the longest time component (O) describing the C(t) function as a function of 1/T.
provides an estimation of the activation energy value of 3.5 kcal‚mol-1 (Figure 10). Discussion The mobile nature of protein structure has been recognized some 25 years ago by Lakowicz and Weber,56 who showed that the fluorescence of buried tryptophan residues in proteins was quenched by molecular oxygen. This observation implied that conformational fluctuations, with rates of the order of magnitude of that of the fluorescence emission (ns-1), occurred in the protein matrix, allowing diffusion of molecular oxygen inside the protein molecule. The concept of protein dynamics is now totally included in all the models of enzyme action.57-61 Different modes of motions can be studied by various physical methods including NMR,62,63 neutron scattering,64,65 analysis of temperature factors in the X-ray diffraction data,66,67 and computer simulations.68,69 An important aspect of protein
Stokes Shift of Tryptophan dynamics concerns the polar fluctuations3,70 that can be tackled by fluorescence techniques. This work reports a detailed fluorescence study of the environment and dynamics of the region of the single tryptophan residue (Trp31) of the CMPK from E. coli. Several important points can be underlined, which concern the existence of a FDSS of the Trp31 emission spectrum and the existence of a “glassy transition” near 230-240 K that abolishes it. The existence of dipolar relaxation around the Trp31 residue in CMPK is suggested by several experimental observations, in agreement with the indole solvatochromic properties (for a review, see, Callis21). They are (i) the detection in the red-edge of the fluorescence emission spectrum of a time constant affected by a negative amplitude in different glycerol concentrations and temperature, (ii) the observation of a FDSS (120-340 cm-1), which is suppressed below a critical temperature range of 230240 K, and (iii) the progressive blue shift of the steady-state fluorescence emission spectrum upon decreasing the temperature (380 cm-1). The amplitude of the FDSS experimentally observed for indole in polar viscous solvent is on the order of 1500-1800 cm-1.25,30-32,71 Estimation by molecular dynamics simulations of the energy values of the dipolar interaction between the indole excited-state dipole and water molecules are approximately 2000 cm-1.21 These simulations show that the shift arises from a maximum number of 20 water molecules lying within 6 Å of the indole ring center, each water molecule contributing for 100-200 cm-1. Proteins contain many dipolar groups associated with different structural entities. Apart from highly charged moieties (i.e., protonated amino or acidic side chains) responsible for dipole moment values not larger than 1-3 D, some structural groups are able to bear larger dipoles. The order of magnitude of the dipole moment of a peptide bond has been evaluated as 3.5-4 D.72-74 However, these values can be strongly amplified when these individual groups are implicated in oriented structures such as R-helices or to a lesser extent like β-strands. A large macrodipole of 100 D or more can especially result for R-helical structures.72-74 The examination of the three-dimensional structure of the protein shows that the Trp31 residue is largely screened from the bulk solvent except for the N atom of the indole ring, which is directed toward the outside of the indole pocket (Figure 1). The accessibility of the aromatic ring to the aqueous solvent corresponds to a small fraction of its surface area. Few polar groups (essentially peptide bonds and two water molecules) are located in the close vicinity of the indole ring inside this pocket (Figure 1). The hydrophobic barriers shaping the Trp31 site in CMPK are principally constituted by two antiparallel R-helices respectively of 14 and 20 amino acids in length: helix-1 (Gly17Gln30) and helix-9 (Ser208-Leu227) (Figure 1). These two amphipathic helices interact by hydrophobic contacts implying stretches of six to seven apolar amino acid side-chains. These TRES measurements of the Trp31 residue of CMPK allow observing relaxation processes within the same time range as, for instance, fluctuations of protein parts in cytochrome c, as indicated by molecular dynamic simulations.4 These fluctuations of protein parts, other than the charged side-chains, are characterized by a large number of modes, occurring in the picosecond-nanosecond time scale.4 Higher time resolution of our instrumentation would have allowed faster detection of relaxation processes. The low mean value of the activation energy of the local dipolar perturbation following the photonic excitation of the fluorophore suggests the existence of low-
J. Phys. Chem. B, Vol. 104, No. 47, 2000 11293 energy barriers between different conformations of the local dipoles in equilibrium.75 Such low-energy barriers are consistent with an efficient adaptation of the protein matrix to an electrostatic perturbation like those which can be produced by the transient appearance of a partial charge on a transitory state in the course of a catalytic reaction.76 This value can be, however, an average of several kinetic processes with different activation energies. Our data suggest that fast relaxations with high activation energy dominate the TRES at room temperature while slow relaxations with low activation energy are more preponderant upon decreasing the temperature. Strongly limited rotational motions of the indole ring are expected from the relatively low-temperature factor values provided for Trp31 (B factor: 29.3 Å2) and its closest residues (average side chain B factor: 36.6 Å2) by the three-dimensional structure of the protein.43 In agreement with these structural data, the fluorescence anisotropy decay measurements show that the indole ring in CMPK does not exhibit motion with large amplitude but does not remain strictly immobilized. A very fast motion could explain the difference observed in our data between the anisotropy values at zero time (Table 1) and the expected theoretical ones (0.2) for an excitation wavelength of 292 nm.77 This phenomenon can account for the dipole reorganization and, thus, as a way for energy dissipation inside the protein matrix after the photon absorption. However, a multidipolar interaction can also be assumed and is suggested by the heterogeneity of the FDSS: time constants ranging from pico- to nanoseconds are detected as also reported for the 2′(N,N-dimethylamino)-6-naphthoyl-4-trans-cyclohexanoic acid extrinsic probe in apomyoglobin.17 In this study, glycerol addition was necessary in order to detect safely the existence of the FDSS. This raises the question of the effect of cosolvent on the protein structure and dynamics. We have observed three time components describing the FDSS of indole in glycerol/water mixtures and in pure glycerol.71 In the temperature range where the FDSS can be observed in CMPK, its amplitude is, however, weaker for the protein than for the solvent and the average activation energy is smaller. From room temperature to 240 K, the viscosity of glycerol/ water mixtures of 50% and 75% increases by a factor of 30 and of 100, respectively. This is not reflected in the variation of the relaxation time constants detected by FDSS of Trp31 in CMPK. Moreover, the dynamic processes in the protein giving rise to the FDSS and to the dynamic quenching of Trp31 is suppressed at 230-240 K where a dynamic transition is likely to occur. The temperature range of the observed transition does not correspond, however, to the melting transition of the cryogenic mixtures employed. This transition occurs at 250 and at 238 K for 52% and 60% glycerol, respectively. Moreover, no melting transition has been reported for 75% glycerol where only glass formation occurs at 169 K.78 We observe in fact the same transition at 230-240 K in 75% glycerol as in the presence of lower amounts of cosolvent. Therefore, this transition is independent of the solvent composition. This supports the hypothesis that the effect of glycerol on the internal dynamics of the protein is only indirect. The glycerol cosolvent is wellknown to potentially govern indirectly the protein internal dynamics. By increasing the solvent viscosity, it can reduce the motion of the amino acid side chains at the protein surface close to the indole ring and in turn the motions of inner residues by inducing a contraction of the protein molecule.79-81 Glycerol can also affect protein dehydration by means of displacing some water molecules of the protein solvation shell. Preferential hydration has also been suggested.82
11294 J. Phys. Chem. B, Vol. 104, No. 47, 2000 The dynamic transition observed at 230-240 K occurs in a narrower range of temperature than in other proteins that have been studied so far by other techniques such as X-ray diffraction, neutron scattering, or Mo¨ssbauer spectroscopy (for a review see Gregory55). This may be due to the fact that we observe only a particular region of the protein and not its structure as a whole. It is in fact possible that the dynamic state of a protein may differ from one region to another, therefore smearing out the transition detected by a global technique. A study of the nucleotide-binding site of the protein using a fluorescent nucleotide inhibitor of high affinity41 is currently being developed to examine this hypothesis. Conclusions From its fluorescence properties, the Trp31 residue in CMPK, located close to the protein-water interface, but clearly buried from the bulk aqueous phase, appears to report a dielectric relaxation processsin the pico/nanosecond time ranges characterized by low activation energy. Such a process has been scarcely reported in the literature using Trp fluorescence owing to several reasons that are summarized in the following. The frequent existence of multiexponential decays in proteins greatly complicates the detection of the FDSS, or at least its interpretation. The existence of several ground-state conformers, giving eventually rise to heterogeneity of excited state lifetimes with different spectral distributions, can create an apparent FDSS. This can be erroneously interpreted as an authentic spectral shift due to an excited state reaction of the dipolar type. There is globally no clear-cut correlations between a blue emission and a short lifetime value for protein tryptophan emission process.27,52 Nevertheless, indirect approaches allow the spectral resolution of individual lifetime components. For instance, decay associated spectra83 or selective quenching84,85 experiments can be performed. It comes out that short lifetimes are concentrated more in the blue region of the fluorescence emission spectrum than in the red one. Indeed, buried tryptophans are often in interaction with quencher groups such as peptide bonds and other electron or proton scavengers inside proteins.84 Besides this trivial case, there is a strong incidence of the internal dynamics of the Trp residue on the timedependent Stokes shift range. Solvent-exposed Trp residues can, in principle, be good candidates to report dipolar relaxation of the bound solvent molecules or charged amino acid side chains. However, not only are they often very mobile but the solvent molecules in the hydration shell may also be very mobile in the fluorescence lifetime scale. Accordingly, the dielectric relaxation process may be too fast to be fully monitored with a conventional pulsed light source. This is likely the situation for CMPK at room temperature in the absence of glycerol where a large part of the process is lost in the short time scale. Nevertheless, the intrinsic properties of Trp31 in CMPK are such that these restrictive conditions are partly fulfilled. Acknowledgment. The technical staff of LURE is acknowledged for running the synchrotron machine during the beam sessions. I.L.S. acknowledges financial support from her laboratory during the course of this work. M.V. acknowledges the Institut National de la Sante´ et de la Recherche Me´dicale for its continual financial support. References and Notes (1) Nakamura, H. Q. ReV. Biophys. 1996, 29, 1-90. (2) Warshel, A. Proc. Natl. Acad. Sci. U.S.A. 1978, 75, 5250-5254. (3) Warshel, A.; Åqvist, J. Annu. ReV. Biophys. Biophys. Chem 1991, 20, 267-298.
Vincent et al. (4) Simonson, T.; Perahia, D. Faraday Discuss. 1996, 103, 71-90. (5) Russell, A. J.; Fersht, A. R. Nature 1987, 328, 496-500. (6) Tishmack, P. A.; Bashford, D.; Harms, E.; Etten, R. L. V. Biochemistry 1997, 36, 11984-11994. (7) Northrup, S. H.; Wensel, T. G.; Meares, C. F.; Wendoloski, J. J.; Matthew, J. B. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 9503-9507. (8) Bone, S.; Pethig, R. J. Mol. Biol. 1985, 181, 323-326. (9) Maroncelli, M.; Fleming, G. R. J. Chem. Phys. 1987, 86, 62216239. (10) Castner, J. E. W.; Maroncelli, M.; Fleming, G. J. Chem. Phys. 1987, 86, 1090-1097. (11) Bagchi, B. Annu. ReV. Phys. Chem. 1989, 40, 115-141. (12) Horng, M. L.; Gardecki, A.; Papazyan, A.; Maroncelli, M. J. Phys. Chem. 1995, 99, 17311-17337. (13) Viard, M.; Gallay, J.; Vincent, M.; Meyer, O.; Robert, B.; Paternostre, M. Biophys. J. 1997, 73, 2221-2234. (14) Gafni, A.; DeToma, R. P.; Manrow, R. E.; Brand, L. Biophys. J. 1977, 17, 155-168. (15) MacGregor, R. B.; Weber, G. Nature 1986, 319, 70-73. (16) Bashkin, J. S.; McLendon, G.; Mukamel, S.; Marohn, J. J. Phys. Chem. 1990, 94, 4757-4760. (17) Pierce, D. W.; Boxer, S. G. J. Phys. Chem. 1992, 96, 5560-5566. (18) Weber, G. Methods Enzymol. 1997, 278, 1-15. (19) Demchenko, A. P. In Topics in Fluorescence Spectroscopy; Lakowicz, J. R., Ed.; Plenum Press: New York, 1992; Vol. 3, pp 65-111. (20) Demchenko, A. P. Biochim. Biophys. Acta 1994, 1209, 149-164. (21) Callis, P. R. Methods Enzymol. 1997, 278, 113-151. (22) Callis, P. R.; Burgess, B. K. J. Phys. Chem. 1997, 101, 94299432. (23) Kawski, A.; Sepiol, J. Bull. Acad. Pol. Sci. Ser. Math. Asr. Phys 1972, 20, 707-715. (24) Tatisheff, I.; Klein, R. Photochem. Photobiol. 1975, 22, 221-229. (25) Gonzalo, I.; Escudero, J. L. J. Phys. Chem. 1982, 86, 2896-2899. (26) Lami, H.; Glasser, N. J. Chem. Phys. 1986, 15. (27) Eftink, M. R. Methods Biochem. Anal. 1991, 35, 127-205. (28) Burstein, E. A.; Emelyanenko, V. I. Photochem. Photobiol. 1996, 64, 316-320. (29) Lakowicz, J. R.; Szmacinski, H.; Gryczynski, I. Photochem. Photobiol. 1988, 47, 31-41. (30) Chapman, C. F.; Fee, R. S.; Maroncelli, M. J. Phys. Chem. 1995, 99, 4811-4819. (31) Demchenko, A. P.; Gallay, J.; Vincent, M. In 5th International Conference on Laser Applications in Life Science, Proceedings; Apanasevich, A., Koroteev, N. I., Kruglik, S. G., Zadkov, V. N., Eds.; SPIE proceedings; 1995; Vol. 2370, pp 693-700. (32) Vincent, M.; Gallay, J.; Demchenko, A. P. J. Phys. Chem. 1995, 99, 14931-14941. (33) De Foresta, B.; Gallay, J.; Sopkova, J.; Champeil, P.; Vincent, M. Biophys. J. 1999, 77, 3071-3084. (34) Easter, J. H.; Toma, R. P. D.; Brand, L. Biophys. J. 1976, 16, 571583. (35) James, D. R.; Ware, W. R. Biochemistry 1985, 20, 5517-5526. (36) Szabo, A. G.; Stepanik, T. M.; Wayner, D. M.; Young, N. M. Biophys. J. 1983, 41, 233-244. (37) Chen, R. F.; Knutson, J. R.; Ziffer, H.; Porter, D. Biochemistry 1991, 30, 5184-5195. (38) Ross, J. B.; Wyssbrod, H. R.; Porter, R. A.; Schwartz, G. P.; Michaels, C. A.; Laws, W. R. Biochemistry 1992, 31, 1585-1594. (39) Willis, K. J.; Szabo, A. G.; Kracjarski, D. T. Chem. Phys. Lett. 1991, 182, 614-616. (40) Sopkova, J.; Gallay, J.; Vincent, M.; Pancoska, P.; Lewit-Bentley, A. Biochemistry 1994, 33, 4490-4499. (41) Bucurenci, N.; Sakamoto, H.; Briozzo, P.; Palibroda, P.; Serina, L.; Sarfati, R. S.; Labesse, G.; Briand; Danchin, A.; Baˆrzu, O.; Gilles, A. M. J. Biol. Chem 1996, 271, 2856-2862. (42) Noda, L. In The Enzymes 3rd Edition; Boyer, P. D., Ed.; Academic Press: New York, 1973; Vol. 8, pp 279-305. (43) Briozzo, P.; Golinelli-Pimpaneau, B.; Gilles, A. M.; Gaucher, J. F.; Burlacu-Miron, S.; Sakamoto, H.; Janin, J.; Baˆrzu, O. Structure 1998, 6, 1517-1527. (44) Blondin, C.; Serina, L.; Wiesmuller, L.; Gilles, A. M.; Baˆrzu, O. Anal. Biochem. 1994, 220, 219-221. (45) Rouvie`re, N.; Vincent, M.; Craescu, C. T.; Gallay, J. Biochemistry 1997, 36, 7339-7352. (46) Bajzer, Z.; Zelic, A.; Prendergast, F. G. Biophys. J. 1995, 69, 11481161. (47) Brochon, J. C. Methods Enzymol. 1994, 240, 262-311. (48) Siano, D. B.; Metzler, D. E. J. Chem. Phys. 1969, 51, 1856-1861. (49) Brick, P. Acta Crystallogr. D 1994, 50, 760-763. (50) Gallay, J.; Vekshin, N.; Sopkova, J.; Vincent, M. In Time-ResolVed Laser Spectroscopy in Biochemistry IV; Lakowicz, J. R., Ed.; SPIE proceedings: 1994; Vol. 2137, pp 390-400.
Stokes Shift of Tryptophan (51) Chen, Y.; Liu, B.; Yu, H.-T.; Barkley, M. D. J. Am. Chem. Soc. 1996, 118, 9271-9278. (52) Eftink, M. R. In Topics in Fluorescence Spectroscopy; Lakowicz, J. R., Ed.; Plenum Press: New York, 1991; Vol. 2, pp 53-126. (53) Johnson, D. A.; Yguerabide, J. Biophys. J. 1985, 48, 949-955. (54) Kinosita, K. J.; Kawato, S.; Ikegami, A. Biophys. J. 1977, 20, 289305. (55) Gregory, R. B. In Protein-SolVent Interactions; Gregory, R. B., Ed.; Marcel Dekker: New York, 1995; pp 191-264. (56) Lakowicz, J. R.; Weber, G. Biochemistry 1973, 12, 4161-4170. (57) Carreri, G.; Fasella, P.; Gratton, E. Annu. ReV. Biophys. Bioeng. 1979, 8, 69-97. (58) Frauenfelder, H.; Petsko, G. A.; Tsernoglou, D. Nature 1979, 280, 558-563. (59) Elber, R.; Karplus, M. Science 1987, 235, 318-321. (60) Lumry, R. In Protein-SolVent Interactions; Gregory, R. B., Ed.; Marcel Dekker: New York, 1995; pp 1-142. (61) Frauenfelder, H.; McMahon, B. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 4970-4975. (62) Wuthrich, K. Biochem. Soc. Symp. 1981, 46, 17-37. (63) Kay, L. E. Nature Struct. Biol. NMR Suppl. 1998, 5, 513-516. (64) Doster, W.; Cusack, S.; Petry, W. Nature 1989, 337, 754-756. (65) Kneller, G. R.; Smith, J. C. J. Mol. Biol 1994, 242, 181-185. (66) Tilton, R. F. J.; Dewan, J. C.; Petsko, G. A. Biochemistry 1992, 31, 2469-2481. (67) Rasmussen, B. F.; Stock, A. M.; Ringe, D.; Petsko, G. A. Nature 1992, 357, 423-424. (68) Van Gunsteren, W. F.; Hu¨nenberger, P. H.; Kovacs, H.; Mark, A. E.; Schiffer, C. A. Philos. Trans. R. Soc., London B 1995, 348, 49-59.
J. Phys. Chem. B, Vol. 104, No. 47, 2000 11295 (69) Duan, Y.; Wang, L.; Kollman, P. A. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 9897-9902. (70) Pethig, R. Dielectric and electronic properties of biological materials; John Wiley: New York, 1979. (71) Vincent, M.; Gallay, J.; Unpublished results. (72) Wada, A. AdV. Biophys. 1976, 9, 1-63. (73) Hol, W. G. AdV. Biophys. 1985, 19, 133-165. (74) Hol, W. G. Prog. Biophys. Mol. Biol. 1985, 45, 149-195. (75) Frauenfelder, H.; Parak, F.; Young, R. D. Annu. ReV. Biophys. Chem 1988, 17, 451-479. (76) Simonson, T.; Perahia, D.; Bricogne, G. J. Mol. Biol. 1991, 218, 859-886. (77) Valeur, B.; Weber, G. Photochem. Photobiol. 1979, 25, 441-444. (78) Luyet, B.; Rasmussen, D. Biodynamica 1968, 10, 167-191. (79) Priev, A.; Almagor, A.; Yedgar, S.; Gavish, B. Biochemistry 1996, 35, 2061-2066. (80) Gekko, K.; Timasheff, S. N. Biochemistry 1981, 20, 4677-4686. (81) Gekko, K.; Timasheff, S. N. Biochemistry 1981, 20, 4667-4676. (82) Oliveira, A. C.; Gaspar, L. P.; Poian, A. T. D.; Silva, J. L. J. Mol. Biol. 1994, 240, 184-187. (83) Knutson, J. R.; Walbridge, D. G.; Brand, L. Biochemistry 1982, 21, 4671-4679. (84) Eftink, M. R.; Wasylewski, Z.; Ghiron, C. A. Biochemistry 1987, 26, 8338-8346. (85) Ross, J. B.; Schmidt, C. J.; Brand, L. Biochemistry 1981, 20, 43694377. (86) Kraulis, P. J. J. Appl. Crystallogr. 1991, 24, 946-950.