5494
J. Phys. Chem. B 2007, 111, 5494-5502
Tyrosyl Rotamer Interconversion Rates and the Fluorescence Decays of N-Acetyltyrosinamide and Short Tyrosyl Peptides Jay R. Unruh, Mangala Roshan Liyanage, and Carey K. Johnson* Department of Chemistry, UniVersity of Kansas, 1251 Wescoe DriVe, Lawrence, Kansas 66044 ReceiVed: July 16, 2006; In Final Form: February 27, 2007
It has long been recognized that the fluorescence lifetimes of amino acid residues such as tyrosine and tryptophan depend on the rotameric configuration of the aromatic side chain, but estimates of the rate of interchange of rotameric states have varied widely. We report measurements of the rotameric populations and interchange rates for tyrosine in N-acetyltyrosinamide (NATyrA), the tripeptide Tyr-Gly-Gly (YGG), and the pentapeptide Leu-enkephalin (YGGFL). The fluorescence lifetimes were analyzed to determine the rotameric interchange rates in the context of a model incorporating exchange among three rotameric states. Maximum entropy method analysis verified the presence of three fluorescence decay components for YGGFL and two for YGG and NATyrA. Rotameric exchange between the gauche(-) and trans states occurred on the nanosecond time scale, whereas exchange with the gauche(+) state occurred on a longer time scale. Good agreement was obtained with rotameric populations and exchange rates from molecular dynamics simulations. Quenching by iodide was used to vary the intrinsic fluorescence lifetimes, providing additional constraints on the determined interchange rates. The temperature dependence was measured to determine barriers to exchange of the two most populated rotamers of 3, 5, and 7 kcal/mol for NATyrA, YGG, and YGGFL, respectively.
Introduction The importance of intrinsic fluorescence for probing protein and peptide structure and dynamics is difficult to overstate. The ability to probe native tryptophans or tyrosines in polypeptides or incorporate them through biosynthesis without causing significant changes in structure opens a window into properties of polypeptides that is often not accessible by other biophysical techniques. Often, small biomolecules contain one or two fluorescent amino acids, making dynamic studies on the native protein feasible. Despite these advantages, the photophysical properties of fluorescent amino acids are complex. One of the key nonradiative decay pathways in these systems is charge-transfer quenching of fluorescent amino acids.1-7 In recent years, progress has been made in the understanding of these decay processes.8-11 Since the rate of charge transfer depends on the proximity to a quenching group, for example, an R-carbonyl group, the fluorescence lifetime of tyrosine depends on the rotameric conformation of the aromatic side chain3,6 (see Figure 1). Similar phenomena arise for the fluorescence decays of tryptophan in peptides.12-15 Estimates of the rates of rotamer interchange for tyrosine have varied widely. In 1978, Gauduchon and Wahl pointed out that if the rates of rotamer interchange are similar to the fluorescence lifetimes, then the observed decay will be multiexponential with lifetimes and amplitudes that are a function of the rotamer interchange rates, the excited-state decay rates, and the rotamer populations.3 They found amplitudes of the fluorescence decay terms that were significantly different from the apparent rotamer * To whom correspondence should be addressed. E-mail: ckjohnson@ ku.edu.
populations and therefore concluded that the rotamer interconversion rate must be on the same order of magnitude as the fluorescence lifetimes.3 In the 1980s, Ross and co-workers used improved fluorescence fitting methods to suggest that the amplitudes of the fluorescence lifetimes for tyrosyl peptides matched the rotamer populations and that the interconversion rate was therefore significantly slower than the fluorescence lifetimes.4,6,16,17 In contrast, in molecular dynamics (MD) simulations for N-acetyl tyrosinamide in SPC water, multiple rotamer transitions were observed within several 100-ps simulations.18 This work suggested that tyrosyl rotamer interconversion was much faster than the fluorescence lifetime. As the discussion above indicates, experimental measurements of interconversion rates are needed. Indeed, several recent experiments have addressed the rate of rotamer interconversion in NATyrA. Guzow et al. found that rotamer populations from simulated potentials of mean force and from NMR measurements do not correspond to the amplitudes in multiexponential fluorescence decays.11 The simulated barriers for rotamer interconversion in N-acetyltyrosinamide (NATyrA) suggested that two of the rotamers interconvert rapidly on the fluorescence time scale, but conversion to and from the third rotamer is too slow to be observed in a fluorescence experiment.11,19 In another recent study, Noronha et al. derived expressions for the fluorescence decay rates that incorporated interchange between two rotameric states.10 The measured fluorescence lifetimes suggested rotameric exchange on the time scale of the fluorescence lifetime. For NATyrA in water, two fluorescence lifetimes were observed, although the NMR results appeared to support the presence of all three rotamers. In this study, we report measurements of rotamer interconversion rates for NATyrA and for the tripeptide Tyr-Gly-Gly
10.1021/jp0645059 CCC: $37.00 © 2007 American Chemical Society Published on Web 04/25/2007
Tyrosyl Rotamers, NATyrA, Short Tyrosyl Peptides
J. Phys. Chem. B, Vol. 111, No. 19, 2007 5495
Figure 1. Rotameric conformations of tyrosine about the CR-Cβ bond. Quenching by the carbonyl group is least in the gauche(-) rotamer.
(YGG) and the opioid pentapeptide leucine enkephalin, TyrGly-Gly-Phe-Leu (YGGFL). We demonstrate that only two rotamers are populated for NATyrA, as supported by a new analysis of NMR data. This significantly simplifies the solution to the kinetic differential equations. From an Arrhenius analysis of the interconversion rates, we determined the energy barriers for rotation, which are in good agreement with previous studies.10,11 We have also extended this approach to YGG, and YGGFL, in which all three rotamers are populated. For YGGFL, three lifetimes were observed, allowing for determination of all rotamer interconversion rates. Materials and Methods All peptides and amino acids used in this study were purchased from Sigma (St. Louis, MO). Buffers were of the highest quality available. For fluorescence studies, the peptide or amino acid concentrations were ∼300 µM. For NMR measurements, the concentrations were 1 mM. Phosphate concentrations of 1 mM were used to stabilize pH at ∼7. Higher phosphate concentrations were avoided to prevent tyrosinephosphate interactions. (Quenching by phosphate becomes significant at concentrations on the on the order of 100 mM.20,21) For the sucrose and iodide titration experiments, the pH was adjusted before the addition of sucrose or iodide. A check verified that the pH did not change by more than 0.5 after addition of sucrose. Time-Resolved Fluorescence. The time-resolved fluorescence studies (other than the iodide quenching experiments) were carried out as previously described22 with a mode-locked, Nd:YAG laser (Coherent Antares) synchronously pumping a cavity-dumped rhodamine 6G dye laser, which was frequencydoubled to 285 nm for excitation. The sample was excited 1 mm from the edge of the fluorescence cuvette to avoid significant reabsorption of emission. Emission was collected through a monochromator set at 303 nm for all samples with a band pass of 10 nm. All tyrosine decays were collected to at least 10 000 peak counts. All other decays were collected to at least 20 000 peak counts. The cavity-dumped repetition rate was 7.6 MHz, and the count rates were kept below 0.1% of the laser repetition rate to avoid pulse pile-up. The instrument response function (IRF) collected with a dilute colloidal solution of nondairy creamer had a full-width at half-maximum of 100 ps. Possible contributions to the recorded decays from scattered excitation light were checked in a blank sample consisting of the buffer used for the samples and were found to be negligible in the collection wavelength region. For iodide quenching experiments, the sample was excited with a cavity-dumped Ti: sapphire laser system (Mira 900F with Pulse Switch accessory, Coherent Inc., Santa Clara, CA). Excitation at 285 nm was generated by third-harmonic generation (5-050 Ultrafast Har-
monic Generator, Inrad Northvale, NJ). The monochromator (model 9030, Sciencetech Inc, Concord, ON, Canada) was operated at 8-nm bandpass. Signals for iodide quenching experiments were processed by a PC board (Becker and Hickl, SPC-830, Berlin, Germany). The IRF had a fwhm of 50 ps. Analysis of Fluorescence Decays. Analysis of multiexponential decays by nonlinear least-squares fitting of the data can be problematic. First, analysis by nonlinear least-squares requires an assumption for the number of exponential decay components, which may not be known. Second, nonlinear least-squares fits are susceptible to the problem of multiple minima in the χ2 surface; i.e., there may be multiple sets of parameters that fit the data set reasonably well so that it is difficult to determine the best set of decay times. Therefore, as at least a preliminary step in data analysis, we chose to analyze these data by the maximum entropy method (MEM), which fits the data to a distribution of lifetimes by minimizing both χ2 and the roughness of the amplitude surface for the distribution.23-26 The MEM minimization provides a useful means for characterizing the minimum number of fluorescence lifetime components without a priori assumptions about the number of components. Peak count levels were collected to greater than 20 000 counts, and the decay was recorded down to the noise level, as recommended for MEM fits.25 The channel shift was allowed to vary to account for a color-dependent shift, but the shape of the distribution was not highly dependent on this parameter. Although the value of the lifetime may co-vary with color shift, this effect is small relative to the extent of rotamer quenching. The amplitude did not significantly co-vary with the color shift. Decay amplitudes were determined by fitting the MEM peaks to multiple Gaussians either on the linear time scale for linearly spaced data or on the log time scale for logarithmically spaced data. The amplitudes recovered in this way were checked against amplitudes from discrete nonlinear least-squares fits, which were not significantly different. Comparison of MEM fits with linearly or logarithmic spaced lifetimes showed that the differences in the shape and position of the lifetime peaks were not significant. Peak positions of MEM distributions were obtained by fitting the peak of the curve to a second-order polynomial function and calculating the lifetime at which the derivative of this function is zero. Fluorescence decays for NATyrA and YGG were also analyzed by fitting to discrete decay components by nonlinear least-squares fits. Nonlinear least-squares fitting of the fluorescence decays was accomplished with a nonlinear least-squares iterative reconvolution software program written in-house. The fitting model did not include a contribution from scattered excitation light, which was found to be negligible. Uncertainties were obtained by support plane analysis.27 This analysis fixes the parameter of interest at a series of values near the χ2
5496 J. Phys. Chem. B, Vol. 111, No. 19, 2007
Unruh et al.
minimum and varies the other parameters to create a χ2 surface for the parameter in question. The error limits at 1 standard deviation were then determined by the increase in χ2 specified by the F-statistic. The fluorescence decays for YGGFL were further analyzed by target analysis.28 Target analysis fits the decay data directly to a proposed kinetic model. We applied target analysis to YGGFL for a rotamer interchange kinetic model with three rotameric states. The rotamer interchange rates in the model fitting function were adjusted to miminize the error parameter χ2 by the simplex method. 1H NMR Measurements. NMR measurements were carried out with a 600 MHz Varian Unity Innova spectrometer. YGG solutions were 1 mM in D2O and adjusted to pD ≈ 6.3. HRHβ coupling constants were determined by assuming that the HβR protons were upfield from the HβS protons following Laws et al.4 The equilibrium rotamer populations, pr (where r denotes the rotamer), were calculated by two different methods. First, following previous workers,4 the rotamer populations were calculated by the method of Pachler:29
p(g-) ) [J(HR-HβR) - Jg]/[Jt-Jg]
(1a)
p(t) ) [J(HR-HβS) - Jg]/[Jt-Jg]
(1b)
where p(g-) and p(t) denote the populations of the gauche(-) and trans rotamers, J(HR-β) are the HR-Hβ coupling constants for protons on the R and β C atoms, R and S denote the stereoisomer, and Jt and Jg are the expected average coupling constants for the trans and gauche states, respectively. Pachler measured the average coupling constant by several methods and used the Jt/Jg ratio determined by Abraham and McLauchlan30 to calculate average amino acid coupling constants. The trans coupling constant was estimated to be 13.56 Hz, whereas the gauche coupling constant was 2.60 Hz.29 The method of Pachler assumes that most amino acids have identical constituents on the R and β carbons. Additionally, it assumes the existence of a peptide bond on either side of the amino acid. Finally, it assumes that the coupling constants for the g+ and g- states are identical. Since the work of Pachler, a great amount of effort has been focused on the prediction of coupling constants (reviewed recently31). We chose to utilize the approach of Altona and coworkers, as it accounts for the protonation state of the amino and carboxy groups of the amino acid as well as the nature of the solvent.32 This approach uses the empirical equation suggested by Haasnoot and co-workers33 to fit the coupling constants of 299 molecules with known dihedral angles and substituents as follows:
J(HR-Hβ) ) 14.63 cos2(φ) + 0.78 cos(φ) + 0.60 + λi{0.34 - 2.31 cos2[siφ + 18.4|λi|]} (2)
∑i
where φ is the HR-CR-Cβ-Hβ dihedral angle; λi is the difference in the electronegativity of substituent group, i, from the electronegativity of hydrogen measured in the appropriate solvent; and si is either -1 or +1, depending on the position of substituent, i, relative to the coupled proton.33 All electronegativities were taken from Altona et al., except that of the aromatic ring of tyrosine, which was taken from Inamoto and Masuda,34 since Altona et al. did not report the electronegativity of this group. Using eq 2 and the φ values (+60°, 180°, -60°) for the
TABLE 1: Measured 1H NMR Coupling Constants and Rotamer Populations by Two Different Methods method 1a
method 2b
J
J
sample
(HR-HβR)
(HR-HβS)
g-
t
g+
g-
t
g+
NATyrA4 NATyrA11 NATyrA10 YG4 YGG (this study) YGGFM39
9.15 8.74 8.72 7.63 7.35
6.10 6.11 6.08 6.71 7.35
0.60 0.56 0.56 0.43 0.43
0.32 0.32 0.32 0.38 0.43
0.08 0.12 0.12 0.19 0.13
0.64 0.59 0.59 0.43 0.42
0.34 0.34 0.33 0.37 0.44
0.03 0.07 0.08 0.20 0.14
7.80
0.36 0.48 0.16 0.34 0.49 0.17
a
6.58
Pachler’s method. and Methods).32
29 b
Haasnoot and Altona method (see Materials
three rotamers, it was possible to predict the coupling constants expected for each proton in each rotamer configuration. The experimentally observed coupling constants for the R and S protons were assumed to be population-weighted averages of the coupling constants for each proton in each rotamer configuration. These were then compared with the measured average coupling constants for each species to calculate relative rotamer populations from
Jexp(R) ) p(g+)Jcalc(HR-Hβ, R, g+) + p(g-)Jcalc(HR-Hβ, R, g-) + (1 - p(g+) - p(g-))Jcalc(HR-Hβ, R, t) (3) (and a similar equation for the S stereoisomer). The rotamer populations obtained in this way are tabulated in Table 1 and compared with the populations computed by the method of Pachler. Molecular Dynamics Simulations. MD simulations were carried out using the program CHARMM35 as previously described.36 The zwitterionic peptide, YGG, was constructed using the Sybyl 6.9 (The Tripos Associates, St. Louis, MO, 2004) peptide builder. The peptide, generated in an extended initial conformation, was placed in an orthorhombic cell of dimensions 40 × 40 × 40 Å. The cell was then overlaid with 1057 water molecules for 3209 total simulation atoms. The simulation took place at constant temperature (298 K) and pressure (1 atm). After a preparatory period of 100 ps of equilibration of solvent in the presence of fixed peptide and 100 ps equilibration of the full system, a molecular dynamics trajectory of 20 ns length was generated. Theory The reactions involved in excited-state rotamer exchange are shown in Figure 2. Here, we have assumed that the rotamer interchange rates are not affected by the electronic excitation of tyrosine. We further assume that the shape and position of the fluorescence spectra and the intrinsic radiative rates are the same for each rotamer.10 The rate equations are given by
( )
[g-] d [t] ) dt [g+]
(
)
-kg- - kg-,t - kg-,g+ kt,gkg+,gkg-,t -kt - kt,g- - kt,g+ kg+,t × kg-,g+ kt,g+ -kg+ - kg+,g- - kg+,t
( ) [g-] [t] [g+]
(4)
Tyrosyl Rotamers, NATyrA, Short Tyrosyl Peptides
J. Phys. Chem. B, Vol. 111, No. 19, 2007 5497
Figure 2. Reaction scheme for rotamer interconversion among the gauche(-) rotamer, the trans rotamer, and the gauche(+) rotamer. kJ is the fluorescence decay rate of species J in the absence of interconversion. kIJ is the rate for the reaction from species I to species J (I, J ) g-, t, or g+).
where kI (I ) g-, t, or g+) are the fluorescence rate constants in rotamers g-, t, and g+, and kIJ are the rotamer interchange rate constants. The solution for this system is
( )
[g-](t) [t](t) ) [g+](t) B2 exp(-λ2t) + cV B3 exp(-λ3t) (5) aV B1 exp(-λ1t) + bV
where λi are the eigenvalues of the 3 × 3 matrix in eq 4, and B Vi are the eigenvectors corresponding to eigenvalues, λi. The values of a, b, and c were set by the conditions that at time zero, the fractional concentrations of rotamers g-, t, and g+ must equal their equilibrium concentrations and that [g-] + [t] + [g+] ) 1. This system of equations can be solved analytically to generate the concentration of each species as a function of time, but it is easier to generate numerical expressions for the eigenvalues and eigenvectors, given initial guesses of the rotamer exchange rates and the intrinsic fluorescence decay rates in the absence of exchange and to optimize these guesses until the experimental fluorescence lifetimes and amplitudes are reproduced. The expected experimental amplitudes are given by (for the first amplitude)
Amplitude1 ) a
∑i V1i
(6)
where i represents the component of eigenvector, B V1. The forward and reverse rates between each species (kIJ and kJI) are related by the appropriate equilibrium constant as follows:
[J] kIJ ) KJI ) [I] kJI
(7)
Thus, we only have to determine three rotamer interchange rates and three fluorescence lifetimes in eq 4 if we already know the values of the equilibrium constants. The equilibrium constants were determined by NMR spectroscopy as discussed in Materials and Methods. As shown in the Results Section, only two rotamers are significantly populated for NATyrA. This simplifies the above treatment to a two-state system. For YGG, there are apparently three rotamers present, but the two with short lifetimes (t and g+) were not resolved. Therefore, we also treated this system with the two-state model. The resulting rates are, therefore,
Figure 3. Maximum entropy lifetime distributions at room temperature in aqueous solution, pH ∼ 7 for tyrosine, NATyrA, YGG, and YGGFL (leucine enkephalin). The discrete lifetimes and relative amplitudes determined by three-exponential nonlinear regression for YGGFL are also shown for comparison (vertical lines).
weighted averages of all the interchange rates of the system. Fortunately, as our MD analysis shows, the rates of interchange to the g+ state are low in YGG, indicating that the observed rates were dominated by the g- to t and t to g- transitions. For NATyrA and YGG, we used iodide as a quencher and measured lifetimes as a function of iodide concentration in order to obtain the rotamer interconversion rates more accurately by varying the fluorescence lifetimes. Classic Stern-Volmer theory describes quenching by
kn ) k0n + kq[Q]
(8)
where kn is the observed decay rate of species n, k0n is the decay rate of species n in the absence of quenching, and kq is the Stern-Volmer quenching rate. With this equation, along with the two-state form of eq 5, it was possible to fit the observed lifetimes as a function of iodide concentration to determine the rotamer interconversion rates. Results Fluorescence Lifetime Distributions. In order to analyze the number of decay components for the species studied in this work, we used the MEM method as described in Materials and Methods. The lifetime distributions for tyrosine, NATyrA, YGG, and YGGFL in aqueous solution at room temperature are shown in Figure 3. It is apparent that narrowly distributed lifetimes are present for NATyrA and tyrosine, whereas broader distributions are present for the peptides. For YGGFL, three peaks were observed in the lifetime distribution, but for YGG, only two broad lifetime distributions were resolved. All observed peptide lifetimes were shorter than the lifetime of free tyrosine. MEM analysis inherently reports a distribution of lifetimes, even for discrete lifetime decays. The widths of the MEM lifetime distributions reflect, at least in part, noise in the data; therefore, the analysis cannot distinguish between discrete and
5498 J. Phys. Chem. B, Vol. 111, No. 19, 2007
Unruh et al.
TABLE 2: Amplitudes and Lifetimes of Decay Components NATyrAa YGGa YGGFLb a
a1
τ1 (ns)
a2
τ2 (ns)
a3
τ3 (ns)
0.11 ( 0.02 0.46 ( 0.01 0.26
0.39 ( 0.13 0.25 ( 0.01 0.13
0.89 ( 0.02 0.54( 0.01 0.26
1.58 ( 0.01 1.30 ( 0.01 0.42
0.48
1.46
From fits to discrete decay components at 25 °C. b From MEM fits at 25 °C.
distributed lifetimes. It is not possible at this level of analysis to determine whether the increased width in the MEM distributions for YGG and YGGFL compared to Tyr and NATyrA is a result of intrinsic inhomogeneity in the peptide systems or increased uncertainty in multiexponential decay components. Therefore, the kinetic model used for analysis of rotamer interchange assumes discrete interchange rates, consistent with previous treatments. For analysis based on MEM distributions, we used the peaks of the MEM distributions for time constants in the context of the rotamer interconversion model. For YGG, only two components were resolved in the MEM analysis shown in Figure 3, even though NMR results showed that three rotamer states were populated (see below). It may be that the two short lifetimes are close enough for YGG that they are not resolved, given the low amplitude (0.14) of the g+ state (see Table 1). This may also be the reason that only two lifetimes were resolved for YGGFL at temperatures of 35 °C and higher. It seems likely, therefore, that the broad short-lifetime distribution for YGG encompasses unresolved decays of both the t and g+ states. On the basis of the identification of two decay components in the MEM analysis, the fluorescence decays for NATyrA and YGG were fit by nonlinear analysis to two decay components. The fitting parameters are given in Table 2. The fluorescence decay amplitudes do not match the NMR rotamer populations in Table 1 and, therefore, are not consistent with a rotamer model without exchange. For example, for NATyrA, the NMR analysis yielded a population of 0.59-0.64 for the g- rotamer (expected to experience the least quenching), as compared to an amplitude of 0.89 for the long fluorescence decay component. For YGG, the g- NMR population was 0.42, as compared to an amplitude of 0.54 for the long decay component. Previous studies of YGGFL also recovered three lifetimes by analysis with discrete lifetime components,37,38 but the lifetimes recovered are not in agreement with the MEM distributions shown in Figure 3. Analysis of the present results for YGGFL with discrete lifetimes produced fits that assimilate the two shortest MEM lifetime distributions into a single discrete lifetime (shown for comparison in Figure 3). A second lifetime characterized the longer lifetime distribution with a third lifetime between the short lifetime distributions and the long lifetime distribution. Analysis with only two exponentials eliminated the intermediate lifetime but increased the χ2 value quite significantly. In contrast, MEM analysis predicted three lifetimes at temperatures of 25 °C and lower. This finding suggests limitations for fits to multiple discrete exponential components by nonlinear regression. As this analysis illustrates, discrete lifetime components recovered by nonlinear regression may not accurately represent the distribution of decay components present. The peak of the lifetime distributions and the amplitudes predicted by MEM analysis for YGGFL are given in Table 2. The fluorescence decay amplitudes for YGGFL differ from NMR populations determined for the related peptide YGGFM,39 again suggesting rotamer interchange. NMR Determination of Rotamer Populations. Previous measurements of tyrosyl rotamer populations via NMR coupling constants have relied on the method of Pachler.29 In the years
since Pachler’s publication, several methods have been developed to predict rotamer populations that take into account differences in substituent electronegativity as well as differences in the g+ and g- coupling constants. Table 1 compares the resulting rotamer populations calculated using Pachler’s approach with those calculated by the approach described in Materials and Methods for the coupling constants reported by several groups for NATyrA. In all cases, Pachler’s approach appeared to overestimate the population of the g+ state of NATyrA by ∼5%. For example, when the new approach was applied to the coupling constants measured by Laws and Ross,4 the population predicted for the g+ component was negligible within error limits ((5%). Other measurements, when reanalyzed by this method, still predicted a population around 7%.10,11 For YGG, we measured NMR coupling constants as described in Materials and Methods. The coupling constants were not dependent on temperature, indicating that the rotamer populations are not very sensitive to temperature. Unlike NATyrA, YGG displayed a significant population of the g+ state (Table 1). Although an NMR measurement of rotamer populations for YGGFL was not available, rotamer populations for Met-enkephalin (YGGFM),39 a peptide similar to YGGFL in which the C-terminal leucine is replaced by methionine, are listed in Table 1. Iodide Quenching of Fluorescence Lifetimes. In order to accurately assess the rotamer interconversion rate and its effect on the fluorescence lifetimes of tyrosine, we carried out iodide quenching experiments on NATyrA and YGG. The fluorescence decays for NATyrA and YGG were fit to discrete lifetimes by nonlinear least-squares fitting with the number of decay times fixed to two on the basis of the MEM analysis, which identified two components for both NATyrA and YGG (see Figure 3). Iodide quenching allows the fluorescence lifetime of tyrosine to be altered systematically, as shown in Figure 4. It can be noted that discrete analysis assigned a somewhat longer value (0.2 ns) to the short decay component in comparison to the MEM analysis (55 ps). As the iodide concentration was increased, the amplitudes of the NATyrA lifetimes did not change significantly, despite the fact that both fluorescence lifetimes were shortened relative to the interconversion rate. For YGG, the amplitudes changed only slightly. To ensure that the changes observed in fluorescence lifetime were not the result of the increase in ionic strength upon addition of iodide, we measured the fluorescence decays in the presence of 0.25 M KCl. The observed fluorescence parameters at high ionic strength did not differ significantly from those in simple aqueous solution (data not shown). The observed fluorescence lifetimes in the presence of iodide were predicted as a function of quencher concentration on the basis of the theory for the dependence of the fluorescence decay on the rotamer interconversion described above by adjusting the rotamer interchange rates. For this treatment, the expected fluorescence lifetimes in the absence of rotamer interchange were assumed to follow the standard Stern-Volmer quenching expression. The best-fit rotamer interchange parameters are given in Table 3, and the fitting parameters from least-squares fits are shown in Figure 4. The rotamer populations were fixed
Tyrosyl Rotamers, NATyrA, Short Tyrosyl Peptides
J. Phys. Chem. B, Vol. 111, No. 19, 2007 5499
Figure 4. Fluorescence decay parameters of NATyrA (left) and YGG (right) as a function of KI concentration. The amplitude of the short decay component is plotted in the top panel. Black lines show fits determined for a model with rotamer exchange and the rotamer populations fixed to the values determined by NMR (see Table 1). The quenching constants determined from the fits were 4.2 ns-1 for NATyrA, 5.1 ns-1 for the grotamer of YGG, and 12.4 ns-1 for the t rotamer of YGG. The error limits were determined by support plane analysis (see Materials and Methods). The dashed lines show fits to a model without rotamer exchange.
TABLE 3: Rotamer Interchange Kinetics from Iodide Quenching Experiments and MD Simulations kg-t (ns-1) NATyrAa YGGa simulation
+0.04 0.52-0.07
0.39 ( 0.05 0.25-0.5c
ktg- (ns-1) 1.0b 0.28b 0.34-0.74c
τg- (ns)d +0 3.3-0.4
2.4 ( 0.3
τt (ns)d
kQg- (ns-1)
kQt (ns-1)
0.34 ( 0.03 0.28 ( 0.02
4.2 ( 0.3 5.1 ( 0.2
12 ( 3
a Determined from fits with rotamer populations constrained to the NMR values. b Determined from kg-t and Ktg- with eq 7. c 90% confidence limits for the interchange rate for an exponential probability distribution of persistence times. d Intrinsic fluorescence lifetimes of rotameric states from fits to rotamer exchange model. These parameters were constrained to be less than the lifetime of tyrosine alone and greater than the observed short lifetime, respectively.
TABLE 4: Rotamer Interchange Rates for YGGFL
kg-t ktg-a kg-g+ kg+g-a ktg+ kg+ta
MEM analysis (ns-1)
target analysis (ns-1)
simulationsb (ns-1)
0.44 0.31
