J. Phys. Chem. B 2008, 112, 12801–12815
12801
Fluorescence Quenching Induced by Conformational Fluctuations in Unsolvated Polypeptides Xiangguo Shi, Denis Duft, and Joel H Parks* The Rowland Institute at HarVard, 100 Edwin H. Land BouleVard, Cambridge, Massachusetts 02142 ReceiVed: April 17, 2008; ReVised Manuscript ReceiVed: June 30, 2008
Time-resolved measurements were conducted to relate the fluorescence lifetimes of dye-derivatized polypeptides to local conformational dynamics in trapped, unsolvated peptide ions. This research was performed to better understand the intramolecular interactions leading to the observed increase of fluorescence quenching with temperature and, in particular, how this quenching is related to conformational fluctuations. Dye-derivatized polyproline ions, Dye-[Pro]n-Arg+-Trp, are formed by electrospray ionization and trapped in a variabletemperature quadrupole ion trap where they are exposed to a pulsed laser which excites fluorescence. Lifetime data exhibit fluorescence quenching as a result of an interaction between the dye and tryptophan (Trp) side chain. This result is consistent with solution measurements performed for comparison. The lifetime temperature dependence is closely fit over the range 150-463 K by an Arrhenius model of the ensemble averaged quenching rate, kq. Model fits of the measured lifetimes yield a frequency prefactor of ∼1011 s-1 for kq characteristic of collective motions of the side chains identified in molecular dynamics (MD) simulations. The data fits also yield activation barriers of ∼0.3 eV, which are comparable to intramolecular electrostatic interactions calculated between the unshielded charge on the Arg residue and the dye. As a result, the quenching rate appears to be determined by the rate of conformational fluctuations and not by the rate of a specific quenching mechanism. The peptide sequence of Dye-Trp-[Pro]n-Arg+ was also studied and identified a dependence of the quenching rate on the electrostatic field in the vicinity of the dye, Trp pair. Molecular dynamics simulations were performed over the range of experimental measurements to study trajectories relevant to the quenching interaction. The MD simulations indicate that as the temperature is increased, conformational fluctuations in the presence of strong electrostatic fields of the charged Arg+ residue can result in both (a) an increased number of dye and Trp separations 105 collisions which equilibrate them with the bath gas which is maintained at the temperature of the trap. After exposure to laser excitation, ions are ejected, and the mass spectrum is obtained using an electron multiplier. 2.1. Time-Resolved Fluorescence. Fluorescence lifetime measurements are performed by exciting the trapped ions of interest at 532 nm with a mode-locked, diode-pumped, solidstate Nd:YVO4 laser (Vanguard 2000-HM532, Spectra-Physics) for 50–300 ms at an average power of 15 mW. At this average power, absorption was linear with intensity and multiphoton processes did not contribute to the measured lifetimes. The pulse width was 12 ps (fwhm) and the repetition rate was reduced from 80 to 26.7 MHz through pulse picking by a transverse field modulator (Conoptics, Danbury, CT). Approximately 50 replicate measurements are performed to determine each lifetime. As indicated in Figure 1, a 50 cm focal length lens (LT) used to focus the Gaussian laser beam on the ion cloud through a ring electrode aperture of 1.5 mm diameter. This reduces the beam diameter to ∼170 µm which eliminates scattering on trap apertures and electrodes. The laser-ion cloud overlap is further
Conformational Fluctuations in Unsolvated Polypeptides
Figure 2. Chemical structure of the peptide Pron (n ) 4 or 10) derivatized with the dye BoTMR through a dye linker. A peptide without the 5 carbon linker for n ) 4, Pro4 sans X, has also been studied.
optimized by alignment of the laser and adjustment of a DC bias voltage applied to the trap endcaps which centers the cloud position on the laser cross section. The laser interacts with ∼200 ions (∼10% of trapped ions) depending on trap operating parameters and temperature. The emitted fluorescence was collected (LC) through a second 1.5 mm diameter aperture and is focused by a triplet lens (LD) onto a pinhole entry (A) to a GaAs photomultiplier (H7421-40, Hamamatsu, Japan). A sharp cutoff long wave filter (F) (Chroma, Rockingham, VT) is required to minimize detection of scattered laser radiation. A histogram-accumulating real-time processor (TimeHarp 200, PicoQuant, Berlin, Germany) is used for time-correlated single photon counting. The average detected count rate is maintained below 1% of the excitation rate to maintain single photon counting statistics (i.e., to ensure a low probability of detecting more than one photon per cycle). The resulting fluorescence decay curves are deconvoluted and fit by a stretched exponential model implemented in the FluoFit data analysis package (version 4.0, PicoQuant). 2.2. Materials. Derivatized peptides are commercially synthesized (BioMer Technology, Concord, CA) and purified by reversed-phase HPLC to a stated purity of >70% prior to shipment. The BODIPY analog of tetramethylrhodamine (BoTMR) is obtained from Invitrogen (Carlsbad, CA) with or without an aminohexanoate linker (cat. no. D6117 and B10002, respectively; the former is used unless otherwise noted) and the C-termini of the peptides were amidated. The BoTMR excitation and emission spectra have been published elsewhere.19,20 Attachment of BoTMR to peptides does not significantly alter the emission or excitation spectra (data not shown). Measurements were performed on the (M+H)+ charge state of the derivatized peptides BoTMR-(Pro)n-Arg-Trp (n ) 4 or 10; “Pro4” and “Pro10”, respectively). The chemical structure for this peptide is shown in Figure 2. BoTMR is neutral in solution and consequently Arg is the most probable protonation site. In addition, the sequences BoTMR-(Pro)4-Arg without Trp and BoTMR-(Pro)4-Arg-Trp without an aminohexanoate linker (“Pro4 sans Trp” and “Pro4 sans X”, respectively) were studied. Measurements were also performed on the (M+H)+ charge state of the derivatized peptides BoTMR-Trp-(Pro)n-Arg (n ) 0, 4 and 10; “DW”, “DW-Pro4”, and “DW-Pro10”, respectively). These peptides were synthesized without the linker and their chemical structure is shown in Figure 3. Electrospray solutions contain the derivatized peptides at 10 µM in 50% acetonitrile/ 50% water. Acetonitrile (>99.9%) and distilled, deionized water were obtained from Fisher Scientific (Fair Lawn, NJ) and VWR (West Chester, PA), respectively. 2.3. Computational Methods. All-atom MD simulations are performed with version 9.0 of the MacroModel software package (Schro¨dinger, Portland, OR) using the OPLS/AA force field.21 We present results only for the Pro4 peptide (M+H)+ in this
J. Phys. Chem. B, Vol. 112, No. 40, 2008 12803
Figure 3. Chemical structure of the peptides DW Pron (n ) 0, 4 or 10) derivatized with the dye BoTMR without a dye linker.
paper. The peptide structure was constructed in an extended conformation within the Maestro graphical interface (Schro¨dinger). The Lennard-Jones parameters for carbon are used for the boron atom in BoTMR because parameters for boron are not readily available in the force field. (The exact parameters for this particular atom are not expected to be crucial for these MD simulations because the atom is at the center of a tetrahedral arrangement of atoms.) For the (M+H)+ ion, the proton is located on the guanido group of the Arg side chain. No cutoff is used for charge-charge or charge-dipole interactions. A constant dielectric electrostatic treatment is used with a molecular dielectric constant, ε ) 1.0, unless otherwise noted. No intrinsic or explicit solvent is used for these simulations which are in gas phase. Bonds to hydrogen are constrained using the SHAKE algorithm22 and a time step of 1.5 fs is used. Energy minimization is performed using the Polak-Ribiere Conjugate Gradient23 method with a derivative convergence criterion of 0.05 kJ/Å · mol. Multiple cycles of simulated annealing (SA)24 are used to generate multiple starting structures for simulations. For SA, the molecule is energy minimized and cooled from 900 to 300 K over a period of 10.5 ns. The resulting structure is then saved and used as the initial structure for the next cycle of SA. For MD simulations, a starting structure resulting from SA is minimized and followed by 6 ns of simulation during which structures are sampled at 0.1 ps intervals. The distances between (a) the BoTMR dye and the Trp indole side chain (DyeTrp), (b) the dye and the charged guanido group of Arg (DyeArg+), (c) Trp and the charged guanido group of Arg (TrpArg+), and (d) the CR positions of Pro1 and Pro4 (Pro1-Pro4) residues were monitored throughout the simulations. Computations are performed on a 2.2 GHz Opteron (AMD, Sunnyvale, CA) processor. 3. Results and Discussion The quenching measurements performed on peptide species in gas phase are accompanied here by the results of molecular dynamic simulations performed to provide a basis for discussing the relationship of quenching to conformational fluctuations. The fluctuating structures derived from MD trajectories include specific conformers which can enhance the quenching process. These conformers occur for particular configurations of the dye and residues and will be shown to depend on the intramolecular electrostatic fields. The trajectories show evidence of these fields and will be used to estimate the field strengths at relevant positions within the conformer structure. Finally, the electrostatic energy can increase the exothermicity of an electron transfer reaction of the dye and Trp. The possibility that quenching is occurring via photoinduced electron transfer will be considered. Consequently, the quenching temperature dependence will be related to the rate of fluctuations which increase the probability of such conformations.
12804 J. Phys. Chem. B, Vol. 112, No. 40, 2008
Shi et al.
Figure 4. Fluorescence lifetime (red squares) and quenching rate (blue squares) vs temperature for the (M+H)+ ions of (a) Pro4 sans W showing the decay without Trp; (b) Pro4 with Trp and the flexible linker; (c) Pro4 with Trp but without the flexible linker; (d) Pro10 with Trp. Best-fits to the quenching model are shown by lines through each set of data points.
3.1. Fluorescence Lifetime Measurements. (a) Lifetime Temperature Dependence. Fluorescence lifetime measurements of the dye derivatized peptides have been used to probe the conformational dynamics as a function of temperature. These peptides were designed to study the dependence of the quenching rate on (i) the Dye-Trp separation (Figure 2) and (ii) on the electrostatic interactions introduced by the Arg protonation site (Figure 3). Previous measurements15 performed over a smaller temperature range (300-440 K) identified that interactions between the dye and Trp were responsible for variation of the dye lifetime with temperature. Furthermore it was observed that in the absence of Trp the charge played no role in the lifetime variation. In the present study, the temperature range has been extended to 150-463 K to obtain a more complete understanding of the physical processes determining the temperature dependence of the lifetime. The lifetime versus temperature data for Pro4 sans Trp, Pro4, Pro4 sans X, and Pro10 are shown in Figure 4a-d, respectively. Assuming the fluorescence is emitted by a population decaying from a single excited electronic state, the lifetime, τ, can be shown from a rate equation analysis for the excited-state population to be given by 1/τ ) 1/τ0 + kq where τ0 is the unperturbed lifetime and the nonradiative rate kq defines the quenching rate constant. The temperature dependence of the quenching rates, kq versus T, is also shown in Figure 4 for each species. It is important to point out that fragmentation was not observed in any of the fluorescence lifetime measurements as a result of the low average laser power (15 mW) and short exposure time used to measure decay times.
(b) Quenching Model. The quenching interactions in these peptide structures are occurring in the presence of conformational fluctuations which not only lead to large changes in the intramolecular separations of Dye-Trp, but also in the electrostatic interactions of Arg+ with the dye and side chain polarizations. In the presence of spatial fluctuations, an unshielded charge can produce electrostatic field strengths the order of 107-108 V/cm at the position of nearby side chains. As a result, the dye or side chains can become trapped in fluctuating potential well depths of ∼0.1-0.6 eV. Consequently, the temperature dependence of the fluctuations is expected to strongly influence the quenching rates since these rates will depend on the Dye-Trp separation and possibly the strength of electrostatic interactions. A model of the lifetime temperature dependence will be based on the assumption that the quenching rate, kq, is limited by the rate of conformational fluctuations, kf, and not by a specific quenching interaction, for example a rate of charge transfer. As a result, the model does not include details of the Dye-Trp interaction responsible for the quenching. The constraints imposed by molecular conformations on relaxation processes involving electron transfer have been studied experimentally for gas phase organics,25-27 protein complexes in solution,28,29 and also considered theoretically.30,31 Almost exclusively, these previous studies have shown that interconversion among a relatively small number of dominant conformers was the rate limiting-step in the electron transfer reaction. In this regard, the polypeptides studied here represent a more complex situation involving structures fluctuating among a much larger number
Conformational Fluctuations in Unsolvated Polypeptides
J. Phys. Chem. B, Vol. 112, No. 40, 2008 12805
of conformers comparable to those observed in recent gas phase mobility experiments.32 Consequently, the assumption that fluctuations are the rate-limiting kinetic process is reasonable and consistent with prior studies. The following phenomenological model relates the quenching process to the peptide intramolecular dynamics by representing the temperature dependence of the fluorescence quenching rate, kq, by a first-order Arrhenius process
kq ) kf ) Af exp(-Ebf /kT)
(1)
where kf is the rate of fluctuations which result in quenching conformers. Af represents an ensemble averaged rate characterizing those fluctuations responsible for achieving these quenching conformations. It is likely that Af represents collective side chain fluctuations having rates ∼1011-1012 s-1 or slower backbone motions ∼1010 s-1 but not single atom fluctuations. The prefactor Af is assumed to depend only weakly on temperature relative to the exponential term and is taken to be constant in the fitting procedure. The spatial fluctuations leading to quenching conformers are constrained by energy barriers on the peptide potential energy surface having an ensemble average Ebf. The temperature T is the thermal equilibrium temperature established by collisions of trapped peptide ions with a background He gas. The model for kq given in eq 1 is used in τ ) τ0/(1 + τ0 kq), to determine τ0 by fitting τ to the lifetime versus temperature data. The model parameters Af and Ebf are then determined by directly fitting eq 1 to the kq versus temperature data. The quenching rate model fits the lifetime and data very closely for each peptide species studied as shown in Figures 4 and 5. These fit parameters Af and Ebf are listed in Table 1 for all peptides. Pro4 sans Trp and Pro4. In the absence of Trp (Pro4 sans Trp), the lifetime is essentially unchanged as shown in Figure 4a, exhibiting a decrease of e10% even at the highest temperatures. This slight decrease in lifetime is probably related to weaker interactions of the dye with the remaining residues resulting in quenching rates 0.35 eV and only partial electron transfer resulted for ∆ < 0.2 eV. The value of ∆ ) 0.52 eV is estimated for the interaction of BoTMR and Trp indicating that the quenching interaction is dominated by photoinduced electron transfer in water solution. Electron Transfer in Gas Phase. Photoinduced electron transfer in isolated, unsolvated supermolecules composed of donor and acceptor molecules has been considered in a series of publications by Jortner and co-workers reviewed in ref.81 A principal conclusion of these theoretical analyses82 is that the relevant electronic-vibrational coupling supplied by the solvent for electron transfer in condensed phase83,84 can be provided by intramolecular modes85,86 for reactions in unsolvated species. Consequently, such a molecular limit can be expected to lead to electron transfer in the absence of a solvent. Measurements in supersonic expansions by Zewail and co-workers,87,88 Verhoeven and co-workers25-27 and Levy and co-workers89 have provided evidence for the occurrence of PET in gas phase between intramolecular bridged donor and acceptor molecules. In these studies, PET was identified by detecting fluorescence emitted by the electron transfer state. The observation of a reaction energy threshold indicated that the overall electron transfer reaction was determined by the rate of conformational change and not by the rate of electron transfer. Previous gas phase studies are encouraging in that they clearly identify PET driven by conformational changes in organic molecules; however those studies cannot be relied upon to confirm the presence of electron transfer reactions in our measurements. The structure and dynamics of a biomolecule are not only considerably more complex, the unshielded charge introduces intramolecular electrostatic interactions which have to be considered. The quenching interactions involving energy transfer and intersystem crossing that usually compete with electron transfer are not expected to be significant in the BoTMR-Trp interaction occurring in gas phase. Since this dye is neutral in aqueous solution, the electronic states not involving charge exchange such as ionization and attachment will be relatively unperturbed by the surrounding water molecules. In this case, solution measurements are expected to be useful guides to gas phase quenching of the neutral dye. Energy transfer cannot contribute significantly to the observed quenching process since there are no singlet or triplet energy levels of Trp90 near the first singlet excited-state of the dye at ∼2.3 eV. The intersystem crossing rate for S1fT0 of the dye core molecule BODIPY has been measured in solution91 to be kISC ∼106 s-1 and as a result does not contribute to quenching on the time scale of the dye lifetime
in aqueous solution. Consequently, the conditions necessary for electron transfer to be an important quenching process will be considered. The exothermic free energy measured in solution does not imply that the interaction will yield an exothermic PET reaction under solvent-free conditions. This will be shown to be strongly dependent on the electrostatic environment in which the DyeTrp interaction occurs. Previous calculations92,93 and experimental studies57,94 of the effect of externally applied electric fields on electron transfer in proteins have investigated the consequences of these fields interacting with the large dipole moment of the separated ion pair. These interactions were shown to modify ∆G, the change in free energy driving the electron transfer reaction. However, protonated biomolecules in gas phase introduce intramolecular electrostatic fields many orders of magnitude stronger for which interactions with residue dipoles and polarizability need to be considered. Jortner82,95,96 has developed the theoretical basis for longrange electron transfer in large solvent-free molecules containing electron donor, D, and electron acceptor, A, subcomponents. In the absence of solvent, the occurrence of PET relies on intramolecular electronic-vibrational coupling between the D*A and D+A- pairs to provide curve crossings and internal conversion. Fluorescence measurements in supersonic jets have identified PET in solvent-free molecules by detecting exciplex emission from the D+A- state.25,27,87-89 Several conditions for occurrence of solvent-free electron transfer were specified82 including (1) the level structure must satisfy S0(DA) < S1(D+A-) < S2(D*A) for an exothermic reaction, (2) sufficiently large electronic coupling and vibrational overlap between S1 and S2, and (3) a sufficiently large density of effectively coupled vibronic levels for rapid internal conversion of S1 to the groundstate vibrational manifold. The conditions under which a PET reaction in unsolvated Pro4 is exothermic97 will be considered here. To be specific, Figure 13 describes the Pro4 electronic states formed during the interaction of the ground and excited states of the dye with the Trp side chain. Figure 13 shows potential energy curves of the ground (DT), locally excited (D*T), and charge-transfer (D-T+) diabatic states and their asymptotes. The DT and D*T potential curves are assumed to represent weakly bound states consistent with the solution results. The photoexcitation occurs at 532 nm and the multiple dashed curves representing (D-T+) indicate various exothermicities depending on the intramolecular electrostatic interactions. The energy, ∆E, is the difference between the minima of the D-T+ and D*T states and an exothermic reaction is defined by the condition ∆E ) E(D-T+)E(D*T) 0. If we include the interaction energies of the unshielded charge on Arg+ with the averaged dipole moment, 〈µj〉, electronic and
Conformational Fluctuations in Unsolvated Polypeptides
J. Phys. Chem. B, Vol. 112, No. 40, 2008 12811
Figure 12. Trajectories of (a) the Dye-Trp separation, (b) the DyeArg+ separation from a simulation at 500 K with starting structure str7. Calculation of (c) the energy difference, ∆E, defined in eq 6. Horizontal lines are indicators for discussion in the text.
Figure 13. Potential curves of the ground (DT), locally excited (D*T), and charge-transfer (D-T+) diabatic states associated with photoinduced electron transfer reactions. The photoexcitation occurs at 532 nm and the multiple curves representing (D-T+) (blue, dashed) correspond to variations induced by fluctuating electrostatic interactions.
nuclear polarizabilities, Rej , Rnj, of each residue, the resulting energy separation is expressed by
∆E ) ∆E0 - e2/εRdR + e2/εRTrpR -
∑ PjEj
(6)
j
where Pj ) (Rej + Rnj) EjR + 〈µj〉 , Ej is the net field at each residue j from Arg+, Trp+, and Dye-, and the sum is taken over all residues. Dipole moments and polarizabilities are estimated using values calculated in refs 54 and 55 and the DyeTrp, Dye-Arg+, Trp-Arg+ separations, RdTrp , RdR and RTrpR, are obtained from the MD trajectories. The Dye-Trp and Dye-Arg+ trajectories from MD simulations for starting structure str7 at 500 K are shown in Figure 12a,b, respectively. The exothermicity, ∆E, calculated for these trajectories show quite clearly in Figure 12c that fluctuations of the dye and Trp can result in quenching conformations for which PET becomes an exothermic reaction, ∆E < 0, and simultaneously have Dye-Trp separations
