PROTEINS: Structure, Function, and Genetics 37:253–263 (1999) Fluorescence Quenching in the DsbA Protein from Escherichia coli: Complete Picture of the Excited-State Energy Pathway and Evidence for the Reshuffling Dynamics of the Microstates of Tryptophan Alain Sillen,1 Jens Hennecke,2 Daniela Roethlisberger,2 Rudi Glockshuber,2 and Yves Engelborghs1* 1Laboratory of Biomolecular Dynamics, University of Leuven, Leuven, Belgium 2Institut für Molekularbiologie und Biophysik, Eidgenössische Technische Hochschule Hönggerberg, Zürich, Switzerland ABSTRACT The disulfide oxidoreductase DsbA is a strong oxidant of protein thiols and is required for efficient disulfide bond formation in the bacterial periplasm. DsbA contains two tryptophans: W76 and W126. The fluorescence of W76 changes upon reduction of the disulfide bridge, as analyzed previously (Hennecke et al., Biochemistry 1997;36:6391– 6400). The fluorescence of W126 is highly quenched. The only two potential side chain quenchers are Q74 and N127, and these were replaced by alanine, resulting in a threefold increase in fluorescence intensity. The fluorescence intensity increase is not due to the removal of dynamic quenchers but to an increase in the population with the longest lifetime. In this report, the possibility of a change in the conformation of W126 is investigated theoretically by using molecular mechanics and dynamic simulations and experimentally by using a reaction with N-bromosuccinimide. This reacts preferably with the most exposed microstate of tryptophan, which is responsible for the longest lifetime. The simulations and the experimental results reveal that the amino acid replacements allow W126 to increase the population of its antiperpendicular conformation. The selectivity of the N-bromosuccinimide reaction allows the visualization of the reshuffling kinetics at exhausting reagent concentration. To the authors’ knowledge, this is the first time that the kinetics of Trp population reshuffling have been measured. Proteins 1999;37:253–263. r 1999 Wiley-Liss, Inc. Key words: protein dynamics; molecular mechanics; time-resolved fluorescence; nuclear magnetic resonance structure; X-ray structure; N-bromosuccinimide INTRODUCTION Internal quenching of tryptophan fluorescence is a phenomenon appreciated among biochemists who are interested in the study of the conformational changes in proteins.1 The nature of internal quenching of tryptophan fluorescence in proteins is mainly dynamic. Possible quenchers in proteins are the side chains of the amino acids histidine,2,3 cysteine and its disulfide bridge,4 arginine and lysine,5 and the peptide backbone.6 Static quenchr 1999 WILEY-LISS, INC. ing also is observed in proteins, but it is possible that this is due to dynamic quenching of very high frequency that escapes detection by nanosecond (ns) lifetime analysis.7 Some proteins possess tryptophans that are very strongly quenched, e.g., W126 in DsbA.8 It is of interest to determine the cause of this strong quenching, because the interaction with the quencher can be used to monitor conformational changes. Removal of the quencher by sitedirected mutagenesis can reveal the interaction with other residues and/or substrates and eventually can make the tryptophan a better reporter group. DsbA from Escherichia coli is a member of the thioldisulfide oxidoreductase (TDOR) enzyme family and is a monomeric, periplasmic protein of 21,130 Da (189 amino acids) required for efficient disulfide bond formation.9,10 TDOR enzymes are involved in numerous processes in prokaryotic and eukaryotic cells.11–13 All TDOR enzymes catalyze the formation or reduction of structural, regulatory, or catalytic disulfide bridges in target proteins by disulfide exchange reactions with their substrates. The C-X-X-C motif of the active-site is characteristic for all TDORs. Reduction of the catalytic disulfide bridge in DsbA has been shown to cause a strong increase in tryptophan fluorescence.8 The three-dimensional X-ray structure of oxidized DsbA14 has revealed that the enzyme possesses a thioredoxin-like domain (residues 1–62 and 139–189), a motif that is found in all known structures of disulfide oxidoreductases.15 The sequence of the thioredoxin-like domain of DsbA, however, is only 10% identical with E. coli thioredoxin. DsbA possesses a second domain (residues 63–138) of unknown function that is inserted into the thioredoxin motif and contains the only two tryptophans of DsbA: W76 and Abbreviations: TDOR, thiol-disulfide oxidoreductase; DTT, dithiothreitol; NBS, N-bromosuccinimide; DTNB, dithionitrobenzoic acid; IPTG, isopropyl-␤-D-thiogalactoside; DAS, decay associated spectra; MD, molecular dynamics. Grant sponsor: Fund for Scientific Research (Belgium/Flanders); Grant number: G.0242.96; Grant sponsor: ETH Zürich (Switzerland). Jens Hennecke’s present address is the Structural Molecular Biology Laboratory, Department of Molecular and Cellular Biology, Howard Hughes Medical Institute, Harvard University, 7 Divinity Avenue, Cambridge, MA 02138. *Correspondence to: Yves Engelborghs, Celestijnenlaan 200D, B-3001 Leuven, Belgium. E-mail: email@example.com Received 19 January 1999; Accepted 10 May 1999 254 A. SILLEN ET AL. Fig. 1. Molscript representation56 of the X-ray structure of oxidized DsbA from Escherichia coli. W126. W76 is responsible for the large fluorescence change upon reduction of the disulfide bridge8 and is located next to the thioredoxin-like domain of DsbA.14 W126 is located deeper in the second domain, and its fluorescence is strongly quenched. In this study, we investigated the cause of this strong fluorescence quenching, and we increased the fluorescence intensity of W126 by site-directed mutagenesis. This made W126 a good probe with which to monitor possible changes in the second domain upon reduction of DsbA. A closer look at the X-ray structure of DsbA revealed that Q76 and N127 are both in close contact with W126 (Fig. 1), within 4.3 Å and 5.0 Å, respectively. The quenching ability of these amide groups was interesting to investigate, because some authors have claimed that amide groups are weak quenchers,4 others have claimed that amide groups are not quenchers at all,16 and others have claimed that amide groups can quench in special circumstances with the aid of other amide groups or with the aid of the peptide backbone.6 MATERIALS AND METHODS Materials Dithiothreitol (DTT), N-acetyltryptophanamide, and tryptophan were obtained from Sigma Chemical Co. (Buchs, Switzerland) The buffer solution used was a 0.05 M KH2PO4 buffer adjusted with NaOH to pH 7.0. The ionic strength of the solutions was always 0.1 M adjusted with KCl. All buffer reagents were of analytical grade and were obtained from either Sigma, Janssen Chimica (Geel, Belgium), or Fluka (Buchs, Switzerland). Ampicillin, dithionitrobenzoic acid (DTNB), and polymyxin B sulfate were purchased from Sigma. Isopropyl-␤-D-thio galactoside (IPTG) was from AGS GmbH (Heidelberg, Germany), DE52- and CM52-cellulose were purchased from Whatman (Maidstone, United Kingdom), and Phenyl-Superose HR 10/10 column was obtained from Pharmacia (Uppsala, Sweden). All other chemicals were from Merck (Darmstadt, Germany) or Fluka and were of the highest purity available. Solutions were filtered (0.22 µm), and assayed for spectral purity. Mutagenesis For construction and expression of the DsbA variants Q74A and N127A, the phasmid pDsbA2 was used, harboring the dsbA gene under control of the trc-promotor.17 Site-directed mutagenesis was performed according to Kunkel et al.18,19 by using the helper phage M13KO720 and the kit supplied by Bio-Rad (Hercules, CA). The following oligonucleotides were used: Q74A: 58-GCC ATC GCC ACT GCC CAT GCC GCT GTC AGG TCT TTG CCC AGG TC-38 N127A: 58-GAG ATT TCA CCA CAA AGC TAG CCC ACG CTG CAT CAT ACT CTT CAC CTT TA-38. The DsbA variants Q74A/W76F, W76F/N127A, and Q74A/W76F/N127A were constructed by using polymerase chain reaction (PCR). By using the plasmid pDSBA2 as a template, the 58-oligonucleotide (Q74A/W76F) 58-CGC ATG GAA TTC AGA TCT GAC TGC GGC ATT TGC TGT GGC 255 FLUORESCENCE QUENCHING IN DSBA GAT GGC G-38 (the BglII restriction site is indicated in italics) and the 38-oligonucleotide 58-CTG CGC ACC GGT ACC GCC ACC TTT TTT CTC GGA CAG ATA TTT C-38 parts of the dsbA gene were amplified by PCR, and a 316 base pair (bp) fragment was cloned through the BglII and PstI restriction sites into the plasmid pDsbA2, generating pDSBA2-Q74A/W76F. The plasmid pDSBA2 Q74A/W76F/ N127A was constructed essentially in the same manner, but pDSBA2-N127A was used as the template for PCR. For construction of pDSBA2-W76F/N127A, the plasmid pDSBA2-N127A was used as the template, and the 58oligonucleotide (W76F) 58-CGC ATG GAA TTC AGA TCT GAC TCA GGC ATT TGC TGT GGC GAT GGC G-38 (the BglII restriction site is indicated in italics) and the same 38-oligonucleotide described above were used for the PCR. Again, the 316-bp fragment was cloned through the BglII and PstI restriction sites. Mutants were identified by restriction analysis and verified by sequencing of the whole mutated genes using the T7 sequencing kit (Pharmacia). intermediate frequency of 455 kHz by external crosscorrelation and is then filtered and amplified further. The phase shift is measured in the low-frequency domain (700 Hz) by using a second cross-correlation step. In this way, fluorescence lifetime measurements are performed by measuring the phase shift of the modulated emission at 50 frequencies ranging from 1.6 MHz to ⬇1 GHz. Either N-acetyltryptophanamide (in water filtered by using the MilliQ system; Millipore Corp., Bedford, MA) with a fluorescence lifetime of 3.059 ns or p-terphenyl in cyclohexane with a lifetime of 1.04 ns26 was used as a reference fluorophore. The measured phase shifts () at a modulation frequency () of the exciting light are related to the fluorescence decay in the time domain, as described previously.27 Data analysis was performed as described by Vos et al.25 Quantum yields were determined relative to tryptophan in water according to the method of Parker and Rees.28 Qprot ⫽ Expression and Purification of DsbA Variants For expression of DsbA variants in the bacterial periplasm, the E. coli strain THZ221 was used. The methods used have been described previously.8,17 All DsbA variants were obtained in the oxidized form after purification, as shown by the lack of free thiol groups,22 and their correct molecular mass was confirmed by MALDI mass spectrometry. Ultraviolet Absorption Measurements Ultraviolet absorption was measured on a Kontron 940 spectrophotometer (Uvikon). The molar absorption coefficients at 295 nm are calculated by taking the ratio of the absorbance at 295 nm and at 280 nm multiplied by the molar absorption coefficients at 280 nm obtained from Hennecke et al.8 Steady-State Fluorescence Measurements Steady-state fluorescence was measured with an SPEX spectrofluorometer (Fluorolog 1691), as described by Sillen et al.,23 with excitation and emission slits providing band passes of 7.2 nm and 3.6 nm, respectively. The cuvette holder was thermostated at 22°C. The excitation wavelength was 295 nm to ensure that the measured fluorescence was due only to tryptophan fluorescence. Fluorescence Lifetime Data Measurements Fluorescence lifetime data were determined by using an automated multifrequency phase fluorometer. The instrument is similar to the one described by Lakowicz et al.,24 except for the use of a high-gain photomultiplier (Hamamatsu H5023) instead of a microchannel plate. The detection part was described previously by Vos et al.25 The excitation source consists of a mode-locked, titaniumdoped sapphire laser (Tsunami; Spectra Physics) pumped by a Beamlok 2080 Ar⫹-ion laser (2080; Spectra Physics). After frequency tripling (GWU; Spectra Physics), the excitation wavelength is 295 nm. A single harmonic component of the exciting light pulse train is converted first to an 兰I 兰I Prot ATrp QTrp, (1) Trp AProt where 兰I is the integrated intensity over the wavelength region 300–450 nm, A is the absorbance at 295 nm, and the quantum yield (QTrp) for tryptophan in water is 0.14.29 Decay-Associated Spectra Decay-associated spectra were constructed by multiplying the intensity fraction with the intensity of the emission spectra at the respective wavelengths.30 A log-normal function31 is fitted to the associated intensities to obtain the decay-associated spectra: 3 I() ⫽ Im exp ⫺ ln 2 ln 2 ln2 1 (a ⫺ 1/) 24 (a ⫺ 1/m) . (2) Here, Im ⫽ I(m) is the maximal fluorescence intensity; m is the wavelength of the band maximum; ⫽ (1/m ⫺ 1/⫺)/(1/⫹ ⫺ 1/m) is the band asymmetry parameter; a ⫽ 1/m ⫹ (1/⫹ ⫺ 1/⫺) /(2 ⫺ 1) is the function-limiting point; and ⫹ ⫽ 107/(0.830 · 107/m ⫹ 7,071) and ⫺ ⫽ 107/(1.1768 · 107/m ⫺ 7,681) are the wavelength positions of half-maximal amplitudes. The average radiative rate constant is calculated by dividing the quantum yield by a wavelength-independent amplitude average lifetime:32 7kr8 ⫽ Q ⌺i␣i . (3) With ⌺i␣i the average lifetime, ␣i is a wavelengthindependent amplitude fraction and is defined as32,33 ␣i ⫽ 兰I 0i() d ⫽ l⫽n 兺兰I l⫽1 0l() d 兰 I () d . I () d 兰 0i 0 (4) 256 A. SILLEN ET AL. The fluorescence intensity at time zero (I0) of the corresponding lifetime is integrated over all wavelengths and is then normalized. Calculation of the Change in Fluorescence Intensity Due to Static and Dynamic Quenching and by a Change in the Population of Microconformations We suggest splitting the ratio of the quantum yield of different variants relative to a reference protein [e.g., wild type (WT)] into a factor (fkr) representing the change in kr, or homogenous static quenching, a factor (fPR) reflecting heterogenous static quenching or population reshuffling, and a factor (fDQ) representing pure dynamic quenching:32 Q Q0 ⫽ 7kr8 ⌺␣ii 7kr08⌺␣0i0i ⫽ 7kr8 ⌺␣i0i ⌺␣ii 7kr08 ⌺␣0i0i ⌺␣i0i ⫽ fkr ⫻ fPR ⫻ fDQ. (5) The factor fPR is affected by static quenching only if the static quenching is heterogenous. If there is static quenching and an increase in the fluorescence due to population reshuffling, then fPR is the minimum increasing factor of fluorescence intensity due to a change in microconformations. Energy Map To characterize the possible conformations of tryptophan, a minimum perturbation map was calculated.34 Minimum perturbation mapping of W126 was performed by using the CHARMM22 package.35 Map calculation was done by using the all-hydrogen CHARMM22 parameter and topology file. The system contained no water. Minimum perturbation maps36 are calculated by fixing the Trp side chain at a particular 1, 2 point and allowing the residues nearby Trp to conformationally relax to achieve an energy minimization in which all of the other residues are fixed to their crystallographic or nuclear magnetic resonance (NMR) coordinates. The nearby residues included K70, D71, L72, T73, Q74, A75, W76, A77, V78, A79, A81, L82, A105, R109, F112, I117, E120, E121, Y122, D123, A124, A125, N127, S128, F129, V130, K131, K132, S133, L134, V135, and A136. The minimization procedure, used at each grid point on the 1, 2 map, consists of 100 steps by the steepest descent method and 1,000 steps by the Powell method with a tolerance of 0.1. This is repeated in 10° steps over the whole angular space of 1 and 2. Throughout this article, 1 is defined as the torsion angle around C␣-C␤ by the bond connectivity N-C␣-C␤-C␥, and 2 is defined as the torsion angle around C␤-C␥ by the bond connectivity C␣-C␤-C␥-C␦1. These maps display areas where Trp is in an energetically favored position. Molecular Dynamics The three-dimensional X-ray structure of oxidized DsbA14 was obtained from the Brookhaven Protein Data Bank.37 The GROMOS 8738 package was obtained from Biostructure S.A. (France). The input files for the GROMOS 87 package were generated by using the program WHATIF,39 which also was used to make the mutations in the threedimensional NMR structure. The structures were placed in a truncated octahedral box of SPC water40 with the counter ions (Cl⫺ and Na⫹), where a minimum distance of 7 Å was kept between the protein and the border of the box. The energies of the protein and the water were then minimized for 500 steps by using a steepest-descent algorithm.41 The velocities of the atoms were assigned following a Maxwellian velocity distribution at 100 K. The system was warmed up to 300 K in two consecutive steps of 1.3 ps in total, and a free molecular dynamics (MD) simulation was performed for 200 ps using a constant pressure of 1 MPa and a constant temperature of 300 K. The temperature of the protein and the solvent were coupled separately to a water bath42 using a coupling constant of 0.1 ps. The pressure was kept constant by coupling to an external pressure bath42 with a coupling constant of 0.5 ps. The conditions of the MD simulation were the following: the time step employed was 2 fs, the integration of the equations of motion and energy were done by using a leap-frog algorithm included in the GROMOS package, the bond lengths were constrained by using the SHAKE routine,43 and cut-offs of 8 Å for nonbonded interactions (van der Waals interactions) and 11 Å for the electrostatic interactions were used. For analysis, the coordinates were saved every ps. The calculations were performed by using an Indigo 2 workstation (Silicon Graphics) equipped with an MIPS R10000 processor. The data were analyzed by using the programs WHATIF42 and SIMLYS.44 Reaction With N-Bromosuccinimide The accessibility of the tryptophan conformation was probed with N-bromosuccinimide (NBS) according to the method of Spande and Witkop,45 as described by Sillen et al.23 Measurements were done at pH 7.0 (0.1 M sodium phosphate buffer). Titration of DsbA W76F/N127A/Q74A with increasing amounts of NBS was performed by adding aliquots of a 4.9-mM stock solution of NBS to the cuvette containing the protein (95 µM), after which the changes in the fluorescence lifetimes and corresponding amplitude fractions were monitored. In a second experiment, the steady-state fluorescence intensity was monitored after the addition of a tenfold excess of NBS over DsbA W76F/ N127A/Q74A mutant concentration. A fluorescence decrease followed by a fluorescence increase was observed. The curve was fitted with the following equation: Fss(t) ⫽ A(1 ⫺ exp (⫺krect)) ⫹ B exp (⫺kNBSt), (6) where A is the amplitude of fluorescence recovery, krec is the rate constant of recovery that is attributed to population reshuffling, kNBS is a rate constant of reaction with NBS, and B is the amplitude of the fluorescence decrease, omitting the first 4 seconds to avoid mixing artifacts. FLUORESCENCE QUENCHING IN DSBA 257 Fig. 2. Decay-associated spectra of DsbA variants in the oxidized state (ox) at 22°C and pH 7.0 (excitation, 295 nm). The deviation at low wavelength between the calculated and the measured spectra is due to the contribution of scattering. Circles, short lifetime 1; squares, middle lifetime 2; triangles, long lifetime 3; solid lines, measured emission spectra; dotted lines, sum of the decay-associated spectra. RESULTS Fluorescence Spectra and Quantum Yield of the DsbA Variants suggesting the absence of conformational changes in the second domain upon reduction of the disulfide bond. Analysis of the fluorescence spectra (Fig. 2) and the quantum yields (Table I) of the DsbA variants W76F, W76F/N127A, W76F/Q74A, and W76F/N127A/Q74A in both oxidized and reduced states reveals that the variant W76F has a strongly quenched tryptophan fluorescence and that removing N127 or Q74 leads to a fluorescence increase in the oxidized state by 15% and 123%, respectively. This suggests that the amide groups in Q74 and N127 indeed are potential quenchers of tryptophan fluorescence.46 Reducing the disulfide bridge, on the other hand, does not significantly change the fluorescence properties of the DsbA variants in which W74 is removed, strongly Time-Resolved Fluorescence Parameters The fluorescence decay of W126 in the variants W76F, W76F/N127A, W76F/Q74A, and W76F/N127A/Q74A was measured at emission wavelengths ranging from 330 nm to 380 nm in 10-nm intervals. A single or double exponential fit of the data yielded unacceptably high values of R2 and a significant nonrandomness in the autocorrelation function of the weighted residuals as a function of the frequency. The best fit with lowest R2 and no systematic deviations in the autocorrelation function of the weighted residuals was obtained with a triple exponential fit. Different starting values for the fitting procedure resulted in the 258 A. SILLEN ET AL. TABLE I. Molar Absorption Coefficients, Quantum Yields, Average Lifetimes, and Radiative Rate Constants of the DsbA Variants† ⑀295 (M⫺1 cm⫺1) Variant W76F ox W76F/N127A ox W76F/Q74A ox W76F/N127A/ Q74A ox N127A ox Q74A ox W76F red W76F/N127A red W76F/Q74A red W76F/N127A/ Q74A red Q 78␣ 7kr8 (ns) (ns⫺1) 2,656 ⫾ 496 2,572 ⫾ 247 2,442 ⫾ 166 0.0132 ⫾ 0.003 0.26 0.051 0.0152 ⫾ 0.002 0.29 0.052 0.0295 ⫾ 0.002 0.64 0.046 2,877 ⫾ 143 6,172 ⫾ 320 6,057 ⫾ 30 2,999 ⫾ 420 2,734 ⫾ 141 2,627 ⫾ 540 0.036 ⫾ 0.005 0.036 ⫾ 0.002 0.046 ⫾ 0.008 0.0118 ⫾ 0.003 0.0147 ⫾ 0.001 0.0291 ⫾ 0.004 2,718 ⫾ 295 0.0409 ⫾ 0.012 0.89 0.046 0.86 0.77 0.89 0.27 0.28 0.61 0.042 0.047 0.047 0.044 0.053 0.048 molar absorption coefficient; Q, quantum yield; 78␣ (ns), average lifetime in nanoseconds; 7kr 8 (ns⫺1 ), average radiative rate constant. †⑀ 295 (M⫺1 cm⫺1 ), same end values. Single decay analysis did not show a systematic increase in the value of the lifetimes at higher wavelengths, indicating that relaxation processes of the environment influencing the lifetimes are unlikely. To improve the recovery of the decay parameters, a global analysis of all of the phase measurements at the different wavelengths was performed. The result of this global fit is summarized in Table II. All of the DsbA variants have approximately the same fluorescence lifetimes: Only the amplitude fractions change in the different variants. Energy Mapping Minimum perturbation map calculations of W126 were performed on the energy minimized average NMR structure (PDB ID:1A23).47 The energy map of W126 in WT (Fig. 3) shows only two stable conformations for the tryptophan side chain, i.e., 1 ⫽ 169° and 2 ⫽ 37° (the perpendicular conformation) is the minimum energy conformation that corresponds exactly to the minimized average NMR conformation, and the second minimum is at 1 ⫽ 129° and 2 ⫽ ⫺123° (the antiperpendicular conformation) and has an energy that is 60.3 Kcal/mol higher; this second minimum is relatively close to the X-ray conformation48 (PDB ID: 1A2J), with 1 ⫽ 178.3° and 2 ⫽ ⫺99.6°. There are 17 out of the 20 NMR structures that position W126 in the perpendicular conformation and three structures that position W126 in the antiperpendicular conformation. Apparently, the two conformers are possible in the solution; however, one minimum is preferred, and the other is preferred in the crystal form (these experimental facts prove that the absolute value of the energy difference of 60 Kcal/mol is largely overestimated, because calculations are done at 0°K for a single molecule; see Discussion). The energy map of the W76F variant has the same two regions, with the minimum at 1 ⫽ 150° and 2 ⫽ 60° and the second minimum at 1 ⫽ 130° and 2 ⫽ ⫺110°, with an energy of 70.4 kcal/mol higher. The energy map of the W76F/Q74A variant also has two stable conformations; however, here, the perpendicular conformation (1 ⫽ 169° and 2 ⫽ 77°) is only 1.5 kcal/mol higher in energy than the other minimum (1 ⫽ 139° and 2 ⫽ ⫺103°). Molecular Dynamics Simulations MD simulations in water were performed on the W76F/ Q74A variants of DsbA: one with the starting structure of W126 in the perpendicular conformation (1 ⫽ 169° and 2 ⫽ 77°) and the other in the antiperpendicular position (1 ⫽ 139° and 2 ⫽ ⫺103°). The 1 and 2 angles were monitored during a 200-ps simulation. The only intention of such a short simulation is to relax the immediate environment of the tryptophan residues. In both simulations, the angles remained practically in their starting position. The distance between the CE3 atom and the C atom (PDB atom notation) of W126 is monitored during the MD simulation and averaged. Charge transfer from the excited indole moiety to the carbonyl group of the peptide backbone is probably the main nonradiative pathway in proteins.6 In the perpendicular position, the CE3 atom of W126 has a high frequency of approaching (3.9 Å) the carbonyl group of the backbone, whereas, in the antiperpendicular position, the distance is greater (4.3 Å), and the frequency is lower. Selective Quenching of Fluorescence by Reaction With NBS To analyze the orientation of W126, its reaction with NBS was studied. NBS reacts irreversibly with tryptophan, generating a totally nonfluorescent oxindole product.49 For NBS to react with Trp, it is essential that the CG and CD1 atoms are solvent-accessible.50 This is the case in the antiperpendicular conformation (Table III). In the perpendicular conformations, these atoms are shielded far more from the solvent. Therefore, the reaction with NBS should be able to report on the conformation of W126. The result of the reaction with low NBS concentration is shown in Figure 4. The amplitude fraction of the largest lifetime decreases strongly, whereas the amplitude fraction of the middle lifetime increases. The amplitude fraction of the smallest lifetime decreases to only a limited extent. Thus, the conformation responsible for the largest lifetime is the most reactive toward NBS, whereas the conformation of the middle lifetime is not reactive or reacts very little to NBS. The smallest lifetime shows only a limited reactivity. Also, the steady-state fluorescence decreases strongly upon reaction with low NBS concentration (Fig. 5); however, upon exhaustion of NBS, the fluorescence intensity increases again, indicating the reappearance of a long lifetime with a rate constant of 0.0142 ⫾ 0.0002 s⫺1 (Fig. 5, inset b). This process of fluorescence recovery is attributed to the reshuffling of the microstates. The calculated rate constant is only an estimate of the recovery rate, because the mechanism of fluorescence recovery potentially is more complicated due to the different accessibilities and reshuffling rates of the three microstates of tryptophan. 259 FLUORESCENCE QUENCHING IN DSBA TABLE II. Lifetimes, Wavelength Independent Amplitude Fractions, and R2 as Obtained by Global Analysis and DAS of the Fluorescence Decay of DsbA and its Variants at pH 7.0 in the Oxidized and Reduced States† ␣1 1 (ns) ␣2 2 (ns) ␣3 3 (ns) R2 a 0.94 ⫾ 0.04 0.92 ⫾ 0.02 0.77 ⫾ 0.02 0.75 ⫾ 0.02 0.46 ⫾ 0.02 0.41 ⫾ 0.02 0.93 ⫾ 0.02 0.93 ⫾ 0.03 0.80 ⫾ 0.01 0.75 ⫾ 0.02 0.14 ⫾ 0.01 0.12 ⫾ 0.01 0.14 ⫾ 0.01 0.13 ⫾ 0.01 0.17 ⫾ 0.02 0.18 ⫾ 0.02 0.14 ⫾ 0.03 0.12 ⫾ 0.01 0.14 ⫾ 0.01 0.15 ⫾ 0.01 0.050 ⫾ 0.001 0.037 ⫾ 0.005 0.07 ⫾ 0.008 0.05 ⫾ 0.004 0.48 ⫾ 0.02 0.49 ⫾ 0.02 0.06 ⫾ 0.01 0.03 ⫾ 0.008 0.06 ⫾ 0.01 0.04 ⫾ 0.02 1.81 ⫾ 0.03 1.33 ⫾ 0.2 0.83 ⫾ 0.1 1.03 ⫾ 0.1 1.03 ⫾ 0.03 1.00 ⫾ 0.03 1.73 ⫾ 0.1 1.49 ⫾ 0.2 0.78 ⫾ 0.1 0.77 ⫾ 0.3 0.01 ⫾ 0.04 0.04 ⫾ 0.02 0.15 ⫾ 0.02 0.20 ⫾ 0.02 0.06 ⫾ 0.01 0.10 ⫾ 0.005 0.01 ⫾ 0.02 0.04 ⫾ 0.03 0.15 ⫾ 0.01 0.21 ⫾ 0.002 3.94 ⫾ 0.01 3.16 ⫾ 0.1 3.07 ⫾ 0.06 3.57 ⫾ 0.06 3.21 ⫾ 0.1 3.28 ⫾ 0.1 3.96 ⫾ 0.1 3.10 ⫾ 0.1 3.09 ⫾ 0.1 3.53 ⫾ 0.4 3.9 3.8 3.1 2.4 2.3 3.6 2.8 3.1 3.5 2.6 Variant W76F ox W76F/N127A ox W76F/Q74A ox W76F/N127A/Q74A ox N127A ox Q74A ox W76F red W76F/N127A red W76F/Q74A red W76F/N127A/Q74A red †␣, wavelength independent amplitude fraction; , lifetime; ox, oxidized; red, reduced. high R2 is due to the very low intensity. (See the quantum yields.) aThe Fig. 3. Calculated energy map of W126 (1 and 2) in DsbA wild type (WT) and W76F/Q74A variant. Energy is expressed in kcal/mol and is relative to the lowest energy. TABLE III. Accessible Surface Å2 Calculated on the Basis of the Nuclear Magnetic Resonance Structure of Reduced DsbA of Reactive Atoms of W126 in DSBA W76F/Q74A With N-Bromosuccinimide Atom Perpendicular Antiperpendicular CG1 CD1 Total 2.49 3.02 5.51 1.32 8.63 9.95 DISCUSSION Quenching Analysis Compared with Trp in solution, which has a quantum yield of 0.14, the fluorescence of W126 is highly quenched in both the oxidized state (Q ⫽ 0.013) and the reduced state (Q ⫽ 0.012) of DsbA. This seems to be due largely to an increase of dynamic quenching, because the apparent radiative rate constant is the same as the radiative rate constant of tryptophan, 0.053 ns⫺1,51 whereas the nonradiative rate constant is 3.8 ns⫺1 compared with 0.33 ns⫺1 for Trp in solution. We therefore looked for dynamic quenchers in the neighborhood of W126. The only two candidates within collisional distances were amide groups of Q74 and N127. The quenching constants related to Q74 and N127 are calculated in Table IV. Replacing Q74 and N127 by alanine indeed reduced the nonradiative rate constant to lower values, and the quenching constants of Q74 and N127 clearly are additive in the variant W76F, indicating parallel pathways of decay. The remaining questions are 1) how does N127 and Q74 quench the fluorescence of W126, and 2) why is the remaining nonradiative rate constant still quite high (1.12 ns⫺1)? Inspection of the lifetime data (Table II) and detailed quenching analyses (Table V) reveal that the lifetimes themselves hardly change upon replacement of Q74 or N127. Upon removal of the amide of Q74, there is a decrease in intensity due to static quenching of 14% (fkr), and replacement of N127 increases the intensity by only 2%. Replacement of Q74 or N127 has approximately the same effect on dynamic quenching of the fluorescence of W126; in both cases, the intensity decreases by ⫾20%. This indicates that amide groups are not 260 A. SILLEN ET AL. TABLE IV. Rate Constant of Fluorescence Decay, Intrinsic Fluorescence Decay Rate, and Apparent Dynamic Quenching Caused by N127 and Q74 Variant 78␣ (ns) W76F/N127A/ Q74A W76F/Q74A W76F/N127A W76F 0.86 1.16 0.64 1.56 0.29 3.44 0.26 3.84a and 3.84b k (ns⫺1) kint kq(N127) kq(Q74) (ns⫺1) (ns⫺1) (ns⫺1) 1.16 1.16 1.16 1.16 — 0.4 — 0.4 — — 2.28 2.28 k, fluorescence decay rate constant; kint, intrinsic fluorescence decay rate constant; kq, apparent dynamic quenching constant. aexperimentally observed. bcalculated on the basis of the single variants. TABLE V. Relative Quenching Analysis: Static Quenching, Dynamic Quenching, and Decrease in the Fluorescence Intensity of W126 by the Change of Microconformations With W76F as Reference State† Fig. 4. N-bromosuccinimide (NBS) titration of DsbA at pH 7.0 and 22°C W76F/Q74A/N127A ox at a protein concentration of 95 µM. The change in the amplitude fraction (␣i) at 340 nm is given as a function of initial NBS concentration. Circles, short lifetime 1; squares, middle lifetime 2; triangles, long lifetime 3. Variant Q/Q0 fkr fPR fDQ W76F/N127A W76F/Q74A W76F/N127A/Q74A 1.15 2.23 2.73 1.05 0.93 0.82 1.35 3.16 3.76 0.81 0.76 0.88 †f kr , static quenching; fPR , change of microconformations; fDQ , dynamic quenching. lated. The increase in fluorescence intensity due to a change in microconformations can be calculated by fPR. A similar phenomenon in which a thermally induced increase in fluorescence intensity of Trp-X peptides is due to the higher population of microconformation with the longer lifetime has been reported previously.52 Another (and apparently the most important) dynamic quencher in the vicinity of tryptophan is the carbonyl group of the peptide bond,6 which is very close (3.9 Å) to the indole group in the perpendicular conformation and, thus, is responsible for the short lifetime. Molecular Mechanics and Dynamics Fig. 5. The change in the steady-state fluorescence intensity on reaction of NBS with the variant W76F/Q74A/N127A ox as function of time. Inset b: Fitting of Equation 11 to the fluorescence intensity (see text). direct quenchers of tryptophan fluorescence. The cause of the strong quenching of W126 is due to the high population of the microconformation with the shortest lifetime. Replacement of Q74 or N127 makes it possible for microconformations with higher lifetimes to become more popu- The microconformation with the shortest lifetime is 95% populated. Replacement of N127 increases the population of the longest lifetime with 4%, replacement of Q74 increases the population of the longest lifetime with 16%, and replacements of both increase with the sum (20%). To determine whether it is possible for Trp to change its conformation, an energy map was calculated. The energy map calculations reveal that there are two energy minima of W126 in DsbA in both the WT and the W76F/Q74A variant. The energy difference between the two minima is very large in the WT and W76F variant, but the energy fluctuations of the MD simulations of both conformations are even larger than the energy difference of the two minima. This indicates that energy maps provide only a qualitative idea of the possible energy minima. They also are made at 0°K rather than at room temperature, and the backbone of the protein is fixed in these energy maps. All of this also could explain why there are only two energy minima on the map although three lifetimes have been 261 FLUORESCENCE QUENCHING IN DSBA Fig. 6. Scheme 1: The scheme of the total excited-state energy pathway in DsbA. k, Rate constant of fluorescence decay; kint, intrinsic fluorescence decay rate; kQ, apparent dynamic quenching; Q74 and N127, side-chain quenchers; W126 and W76, tryptophans. measured. The middle lifetime must result from another conformation that was not highly populated in the experimentally determined structures or in the calculations (compare with ⫾5% in the fluorescence measurements). However, it is reassuring to note that, in the X-ray structure, one energy minimum is populated (antiperpendicular) and the other is populated in the NMR structure (perpendicular). To investigate the possibility of fluorescence quenching by the carbonyl group of the peptide bond in both microconformations, an MD simulation was performed, and the distance between the CE3 of the indole group and the carbonyl carbon of the peptide bond was monitored. In the MD simulation, W126 remains in the original conformations. In the perpendicular conformation, the indole group is closer to the carbonyl of the peptide group compared with the antiperpendicular conformation. Thus, the perpendicular conformation is quenched more than the antiperpendicular and, thus, is associated with the lowest lifetime. Linking of the Conformations With the Lifetimes NBS reacts preferably with the tryptophan that has the most exposed pyrrole ring. Analysis of the NMR structure in the reduced DsbA reveals that, in the antiperpendicular conformation, the pyrrole ring of W126 is the most exposed. Lifetime determinations of DsbA W76F/N127A/ Q74A reacting with increasing amounts of NBS reveal that the amplitude fraction of the longest lifetime decreases the greatest. Thus, reaction with NBS identifies the longest lifetime with the antiperpendicular conformation (1 ⫽ 139° and 2 ⫽ ⫺103°). Also, in the steady-state experiment, at exhausting concentration of NBS (when the modified Trp redistributes over the different microconformations), a fraction of the long lifetime reappears and explains the recovery of the fluorescence intensity (Fig. 5). After some time, the limited excess of NBS has disappeared due to reaction and to hydrolyses of NBS. The MD simulation of the variant W76F/Q74A reveals that the carbonyl carbon of the backbone of W126 is closer in the perpendicular conformation (1 ⫽ 169° and 2 ⫽ 77°) Consequently, the lifetime of this conformation may be lower. Thus, this conformation is linked to the smallest lifetime and/or the middle lifetime. Influence of Reduction on the Second Domain in DsbA Upon reduction of DsbA, there are no changes in the fluorescence properties of W126. Changing W126 or the environment of W126 has an influence on the fluorescence of W76 that is different in the oxidized state and the reduced state. Upon oxidation, a static quenching is removed for W76 if N127 or Q74 are changed; whereas, if W126 is replaced by a Phe, then the static quenching is removed upon reduction. This is an additional influence of the oxidation state of the first domain on the W76 fluorescence that depends on the mobility of the W76 in the second domain. Overall Scheme of the Quenching in DsbA The overall scheme of rate constants (Fig. 6) provides a picture of the fluorescence decay pathway of the two tryptophans in DsbA. The scheme of the rate constants that effect the fluorescence of W76 is taken from Hennecke et al.8 The other rate constants are taken from Table IV. CONCLUSIONS The increase in fluorescence intensity of W126 upon removal of the amides is not due to the removal of collisional quenchers but to a reshuffling of microstates toward a population that is less quenched. The high knr in the triple mutant is due to the fact that the lowest lifetime is still highly populated (75%). The origin of this lowest lifetime is that the conformation associated with this lifetime is very close to the carbonyl group of the peptide bond. The amide groups do not quench tryptophan fluorescence directly.6,16 Reduction of the disulfide bridge does not seem to lead to important conformational changes in the second domain of DsbA, because neither W76F nor the triple mutant showed any fluorescence change upon reduction. Our results indicate that the multiple exponential fluorescence decay is 262 A. SILLEN ET AL. caused by multiple microconformations of tryptophan in the protein matrix53 that slowly interchange from one conformation to the other, and not by the existence of multiple acceptors of fluorescence emission.54 Energy transfer to different acceptors is a decay process with parallel paths from the same source; therefore, the decay rates are additives that give rise to only one exponential decay. Other possible explanations for the nonexponential decay of tryptophan fluorescence in proteins are not applicable to the fluorescence of W126. If the nonexponential decay originates from, e.g., solvent relaxation,55 then reaction with NBS would not affect one lifetime more than the others, and, as a result, the steady-state fluorescence would not increase after the time when the NBS is exhausted. The same reasoning is applicable to any explanation for the nonexponential decay of tryptophan fluorescence that does not take into account the conformational heterogeneity of the tryptophan. Our NBS experiments provide an idea of the time scale during which large amino acids like tryptophan that are buried partially in the protein matrix change their conformation. In this protein, the time scale of the process is in the seconds range. To our knowledge, this is the first report that has measured the kinetics of Trp population reshuffling. ACKNOWLEDGMENTS The authors thank Dr. Dı́az (Leuven) for his help with the programs SIMLYS and WHATIF/GROMOS and P. James (Zürich) for performing mass spectrometry. REFERENCES 1. Dı́az JF, Sillen A, Engelborghs Y. Equilibrium and kinetic study of the conformational transition towards the active state of Ha-rasp21, induced by the binding of BeF3- to the GDP-bound state, in the absence of GAP’s. J Biol Chem 1997;272:23138–23143. 2. Vos R, Engelborghs Y. A fluorescence study of tryptophanhistidine interaction in the peptide anantin and in solution. Photochem Photobiol 1994;60(1):24–32. 3. Shinitzky M, Goldman R. Fluorometric determination of histidinetryptophan complexes in peptides and proteins. Eur J Biochem 1967;3:139–144. 4. Cowgill RW. Fluorescence and protein structure XI. Fluorescence quenching by disulfide and sulfhydryl groups. Biochim Biophys Acta 1967;140:37–44. 5. Bushueva TL, Busel EP, Burstein EA. The interaction of protein functional groups with indole chromophore. III. amine, amide and thiol groups. Stud Biophys 1975;52:41–52. 6. Chen Y, Liu B, Yu H-T, Barkley MD. The peptide bond quenches indole fluorescence. J Am Chem Soc 1996;118:9271–9278. 7. Webber SE. The role of time-dependent measurements in elucidating static versus dynamic quenching processes. Photochem Photobiol 1997;65(1):33–38. 8. Hennecke J, Sillen A, Huber-Wunderlich M, Engelborghs Y, Glockshuber R. Quenching of tryptophan fluorescence by the active-site disulfide bridge in the DsbA protein from Escherichia coli. Biochemistry 1997;36:6391–6400. 9. Bardwell JCA, McGovern K, Beckwith J. Identification of a protein required for disulfide bond formation in vivo. Cell 1991;67: 581–589. 10. Kamitani S, Akiyama Y, Ito K. Identification of an Escherichia coli gene required for the formation of correctly folded alkaline phosphatase, a periplasmic enzyme EMBO J 1992;11:57–62. 11. Gilbert HF. Molecular and cellular aspects of thiol-disulfide exchange. Adv Enzymol Relat Areas Mol Biol 1990;63:69–172. 12. Bardwell JCA, Beckwith J. The bonds that tie: catalysed disuldide bond formation. Cell 1993;74:769–771. 13. Loferer H, Hennecke H. Protein disulphide oxidoreductase in bacteria. Trends Biochem Sci 1994;19:169–171. 14. Martin JL, Bardwell JCA, Kuriyan J. Crystal structure of the DsbA protein required for disulphide bond formation in vivo. Nature 1993;365:464–468. 15. Martin JL. Thioredoxin—a fold for all reasons. Structure 1995;3: 245–250. 16. Ricci WR, Nesta M. Inter- and intra-molecular quenching of indole fluorescence by carbonyl compounds. J Phys Chem 1976;80:974– 980. 17. Hennecke J, Spleiss C, Glockshuber R. Influence of acidic residues and the kink in the active-site helix on the properties of the disulfide oxidoreductase DsbA. J Biol Chem 1997;272:189–195. 18. Kunkel TA. Rapid and efficient site-specific mutagenesis without phenotypic selection. Proc Natl Acad Sci USA 1985;82:488–492. 19. Kunkel TA, Roberts JD, Zakour RA. Rapid and efficient sitespecific mutagenesis without phenotypic selection. Methods Enzymol 1987;154:367–382. 20. Vieira J, Messing J. Production of single-stranded plasmid DNA. Methods Enzymol 1987;153:3–11. 21. Grauschopf U, Winther JR, Korber P, Zander T, Dallinger P, Bardwell JCA. Why is DsbA such an oxidizing disulfide catalyst? Cell 1995;83:947–955. 22. Ellman GL. Tissue sulfhydryl groups. Arch Biochem Biophys 1959;82:70–77. 23. Sillen A, Vos R, Engelborghs Y. Fluorescence study of the conformational properties of recombinant tick anticoagulant peptide (Ornithodorus moubata) using multifrequency phase fluorometry. Photochem Photobiol 1996;64:785–791. 24. Lakowicz JR, Laczko G, Gryczinski I. 2-GHz frequency-domain fluorometer. Rev Sci Instrum 1985;57:2499–2506. 25. Vos R, Strobbe R, Engelborghs Y. Gigahertz phase fluorometry using a fast high-gain photomultiplier. J Fluorescence 1997;7(1): 33S–35S. 26. Desie G, Boens N, De Schryver FC. Study of the time-resolved tryptophan fluorescence of crystalline ␣-chymotrypsin. Biochemistry 1986;25:8301–8308. 27. Weber G. Resolution of the fluorescence lifetimes in a heterogeneous system by phase and modulation measurements. J Phys Chem 1981;85:949–953. 28. Parker CA, Rees WT. Corrections of fluorescence spectra and the measurement of fluorescence quantum efficiency. Analyst 1960;85: 587–600. 29. Kirby EP, Steiner RFS. The influence of solvent and temperature upon the fluorescence of indole derivatives. J Phys Chem 1970;74: 4480–4490. 30. Ross JA, Schmid CJ, Brand L. Time-resolved fluorescence of the two tryptophans in horse liver alcohol dehydrogenase. Biochemistry 1981;20:4369–4377. 31. Burstein EA, Emelyanenko VI. Log-normal description of fluorescence spectra of organic fluorophores. Photochem Photobiol 1996; 64(2):316–320. 32. Sillen A, Engelborghs Y. The correct use of ‘‘average’’ fluorescence parameters. Photochem Photobiol 1998;67(5):475–486. 33. Willis KJ, Szabo AG. Conformation of parathyroid hormone: time-resolved fluorescence studies. Biochemistry 1992;31:8924– 8931. 34. Silva ND, Prendergast FG. Tryptophan dynamics of the FK506 binding protein: time-resolved fluorescence and simulations. Biophys J 1996;70:1122–1137. 35. Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Swaminathan S, Karplus M. CHARMM: a program for macromolecular energy, minimization and dynamics calculations. J Comput Chem 1983;4: 187–217. 36. Haydock C. Protein side chain rotational isomerisation: a minimum perturbation mapping study. J Chem Phys 1993;98:8199– 8214. 37. Bernstein FC, Koetzle TF, Williams GJB, Meyer EF Jr., Brice MD, Rodgers JR, Kennard O, Shimanouchi T, Tasumi M. The Protein Data Bank: a computer-based archival file for macromolecular structures. J Mol Biol 1977;112:535–542. FLUORESCENCE QUENCHING IN DSBA 38. Van Gunsteren WF, Berendsen HJC. Program system GROMOS 87. Groningen: Laboratory of Physical Chemistry, University of Groningen [distributed by Biomos Biomolecular Software b.v.]; 1987. 39. Vriend G. WHAT IF: a molecular modeling and drug design program. J Mol Graphics 1990;8:52–56. 40. Berendsen HJC, Postma JPM, Van Gunsteren WF, Hermans J. Interaction models for water in relation to protein hydratation. In: Pullman B, editor. Intramolecular forces. Dordrecht: Reidel; 1981. p 331–342. 41. Levitt M, Lifson S. Refinement of protein conformations using a macromolecular energy minimization procedure. J Mol Biol 1969; 46:269–279. 42. Berendsen HJC, Postma JPM, Van Gunsteren WF, Dinola A, Haak JR. Molecular dynamics with coupling to an external bath. J Chem Phys 1984;81:3684–3690. 43. Ryckaert JP, Ciccotti G, Berendsen HJC. Numerical integration of Cartesian equations of motion of a system with constrains: molecular dynamics of n-alkanes. J Comput Phys 1977;23:327– 341. 44. Krüger P, Lüke M, Szameit A. SIMLYS—a software package for trajectory analysis of molecular dynamics simulations. Comput Phys Commun 1991;62:371–380. 45. Spande TF, Witkop B. Reactivity toward N-bromosuccinimide as a criterion for buried and exposed tryptophan residues in proteins. In: CHW Hirs, editor. Methods in Enzymology, Vol. XI. New York: Academic Press; 1967. p 528–532. 46. Van Gilst M, Hudson S. Histidine-tryptophan interactions in T4 lysozyme: ‘‘anomalous’’ pH dependence of fluorescence. Biophys Chem 1996;63:17–25. 263 47. Schirra HJ, Renner C, Czisch M, Huber-Wunderlich M, Holak TA, Glockshuber R. Structure of reduced DsbA from Escherichia coli in solution. Biochemistry 1998;37:6263–6276. 48. Guddat LW, Bardwell J, Martin JL. Crystal structures of reduced and oxidized DsbA: investigation of domain motion and thiolate stabilization. Structure 1998;6:757–767. 49. Imoto T, Foster LS, Ruplay JA, Tanaka F. Fluorescence of lysozyme: emission from tryptophan residues 62 and 108 and energy migration. Proc Natl Acad Sci USA 1972;69:1151–1155. 50. Green NM, Witkop B. Oxidation studies of indoles and the tertiary structure of proteins. Trans NY Acad Sci 1964;26:659–669. 51. Eftink MR, Jia Y, Hu D, Ghiron CA. Fluorescence studies with tryptophan analogues: excited state interactions involving the side chain amino group. J Phys Chem 1995;99:5713–5723. 52. Brancaleon L, Gasparini G, Manfredi M, Mazzini A. A model for the explanation of the thermally induced increase of the overall fluorescence in tryptophan-X peptides. Arch Biochem Biophys 1997;348:125–133. 53. Dahms TES, Willis KJ, Szabo AG. Conformational heterogeneity of tryptophan in a protein crystal. J Am Chem Soc 1995;117:2321– 2326. 54. Bajzer Ž, Prendergast FG. A model for multiexponential tryptophan fluorescence intensity decay in proteins. Biophys J 1993;65: 2313–2323. 55. Lakowicz JR, Cherek H. Dipolar relaxation on the nanosecond timescale observed by wavelength-resolved phase fluorometry of tryptophan fluorescence. J Biol Chem 1980;255:831–834. 56. Kraulis PJ. MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J Appl Crystalogr 1991;24: 946–950.