Quantum Yield Measurements of Short-Lived Photoactivation

J. Phys. Chem. A 2010, 114, 3207–3214

3207

Quantum Yield Measurements of Short-Lived Photoactivation Intermediates in DNA Photolyase: Toward a Detailed Understanding of the Triple Tryptophan Electron Transfer Chain† Martin Byrdin,‡,§,3 Andras Lukacs,|,⊥,O Viruthachalam Thiagarajan,‡,§ Andre´ P. M. Eker,# Klaus Brettel,*,‡,§ and Marten H. Vos*,|,⊥ CEA, IBITECS, Laboratoire de Photocatalyse et Biohydroge`ne, Gif sur YVette, F-91191, France, CNRS, URA2096, Gif sur YVette, F-91191, France, Laboratoire d’Optique et Biosciences, CNRS, Ecole Polytechnique, F-91128 Palaiseau, France, INSERM U696, F-91128 Palaiseau, France, and Department of Cell Biology and Genetics, Medical Genetics Centre, Erasmus UniVersity Medical Centre, P.O. Box 2040, 3000 CA Rotterdam, The Netherlands ReceiVed: September 29, 2009; ReVised Manuscript ReceiVed: NoVember 17, 2009

The light-dependent DNA repair enzyme photolyase contains a unique evolutionary conserved triple tryptophan electron transfer chain (W382-W359-W306 in photolyase from E. coli) that bridges the ∼15 Å distance between the buried flavin adenine dinucleotide (FAD) cofactor and the surface of the protein. Upon excitation of the semireduced flavin (FADH°), electron transfer through the chain leads to formation of fully reduced flavin (FADH-; required for DNA repair) and oxidation of the most remote tryptophan residue W306, followed by its deprotonation. The thus-formed tryptophanyl radical W306°+ is reduced either by an extrinsic reductant or by reverse electron transfer from FADH-. Altogether the kinetics of these charge transfer reactions span 10 orders of magnitude, from a few picoseconds to tens of milliseconds. We investigated electron transfer processes in the picosecond-nanosecond time window bridging the time domains covered by ultrafast pump-probe and “classical” continuous probe techniques. Using a recent dedicated setup, we directly show that virtually no absorption change between 300 ps and 10 ns occurs in wild-type photolyase, implying that no charge recombination takes place in this time window. In contrast, W306F mutant photolyase showed a partial absorption recovery with a time constant of 0.85 ns. In wild-type photolyase, the quantum yield of FADH- W306°+ was found at 19 ( 4%, in reference to the established quantum yield of the long-lived excited state of [Ru(bpy)3]2+. With this yield, the optical spectrum of the excited state of FADH° can be constructed from ultrafast spectroscopic data; this spectrum is dominated by excited state absorption extending from below 450 to 850 nm. The new experimental results, taken together with previous data, allow us to propose a detailed kinetic and energetic scheme of the electron transfer chain. Introduction DNA photolyase (PL) is a light driven DNA repair enzyme that can revert some of the UV-induced damages to DNA. The most frequently encountered type are cyclobutane pyrimidine dimers (CPD) that, if not repaired, can block processing of the DNA and induce mutations and ultimately cell death.1-3 According to the current view,1 photolyase binds in the dark to the damaged DNA site and repair is initiated by formation of the excited state of the catalytic cofactor: a flavin adenine dinucleotide (FAD) in the doubly reduced state FADH-. This happens by absorption of a 300-500 nm photon, either by FADH- directly or by a UV-absorbing antenna molecule, where it is followed by energy transfer to the flavin. The excited state †

Part of the “Benoît Soep Festschrift”. * Corresponding authors. K.B.: tel, +33169089869; fax, +33169088717; e-mail, [email protected]. M.H.V.: tel, +33169335066; fax, +33169335084; e-mail, [email protected]. ‡ Laboratoire de Photocatalyse et Biohydroge`ne. § CNRS. 3 Present address: Institut de Biologie Structurale Jean-Pierre Ebel, CEA, CNRS, UJF, 41 rue Jules Horowitz, 38027 Grenoble, France. | Ecole Polytechnique. ⊥ INSERM. O Present address: School of Chemical Sciences and Pharmacy, University of East Anglia, Norwich NR4 7TJ, U.K. # Erasmus University Medical Centre.

FADH-* decays intrinsically in the nanosecond range;4 in PL binding CPD, catalytic electron transfer from the flavin (yielding FADH°) to the CPD efficiently competes with this decay.5,6 After splitting of the CPD, reverse electron transfer reforms FADH-, and the substrate is released within 50 µs,7 which closes the catalytic cycle. Apart from this photorepair activity, photolyase also displays a second, independent photoreaction, referred to as “photoactivation”.1 Here, visible light (up to 650 nm) can be used to trigger reduction of the flavin from its radical (singly reduced) state FADH° (in which it is typically found in purified PL) to the catalytically competent FADH-. The electron ultimately comes from a reductant in the solvent, a process involving transient oxidation of a surface-exposed tryptophan (W306 in E. coli PL)8 located 15 Å from the flavin.9 A triple tryptophan electron “nanowire” is used to transfer an electron within tens of picoseconds10 from W306 to photoexcited FADH°. This triad of tryptophans (also involving W382 adjacent to the flavin and the intermediate W359 in E. coli PL), conserved in photolyases and the related cryptochromes,11 has been noted in the structure9 and was proposed to act as the electron transfer chain for flavin reduction a decade ago on the basis of theoretical12 and experimental11 studies. Unraveling the processes occurring in this chain has been the subject of a number of time-resolved

10.1021/jp9093589  2010 American Chemical Society Published on Web 12/02/2009

3208

J. Phys. Chem. A, Vol. 114, No. 9, 2010

Figure 1. Disposition of the FADH°-W382-W359-W306 chain in DNA photolyase from E. coli (PDB entry 1DNP9) and scheme of forward electron transfer and deprotonation reactions.

absorption spectroscopy studies.10,11,13-20 Challenges in such studies arise from difficulties in the detection of electron transfer between chemically identical species and the fact that the processes involved (forward electron transfer/deprotonation/ recombination) cover a time window of ten decades (a few picoseconds or less to tens of milliseconds) and require multiple transient absorption techniques covering different time windows. The main forward reaction steps are schematized in Figure 1. Very briefly, there is evidence that, upon formation of the excited state FADH°*, electron transfer from the surfaceexposed W306 to FADH° is rate-limited by the first step, oxidation of the close-by W382, experimentally determined at 30 ps,11,13,18 and that electron transfer between the tryptophans occurs with a faster rate for both the second18 and the third step.10 The W306 cation radical (306TrpH°+; where appropriate, we designate tryptophan by TrpH, where H stands for the N1 proton, and the deprotonated tryptophan radical by Trp°) releases a proton to the solvent in ∼200 ns11 and, if no extrinsic reductant is available, recombination with FADH- occurs with a pH dependent overall time constant of ∼1 ms (pH 5.5) to ∼20 ms (pH 9).11,14 Among the open issues are reliable determinations of quantum yields of the radical intermediate states at different time scales. Even after it became clear10 that forward electron transfer from the surface-exposed W306 to FADH° is complete within the 100 ps time range, i.e., before the end of the time window accessible to pump-probe transient absorption, the question remained open whether recombination processes intervened on this time scale, which could equally be masked by the ratelimiting first step in forward electron transfer. Indeed, a subunity quantum yield for the initial step can be deduced from the 80 ps lifetime of FADH°* in the absence of the primary electron donor W382,13 and recombination from the primary charge pair (FADH- 382TrpH°+) was estimated to occur in less than 4 ps.18 Furthermore, an (FADH- TrpH°+) recombination phase with a time constant of 3 ns has been proposed on the basis of ultrafast measurements,20 i.e., at the limit of the window that is open to observation with classical flash absorption techniques. Finally, the quantum yield of the overall process in E.coli PL is reported to be low (10% at 1 µs after excitation21). Reliable determinations of quantum yields of different intermediates are required to obtain a full kinetic and energetic picture of the reactions in this important electron transfer chain. Here we address the issue by (a) re-examining the absolute quantum yield on the nanosecond scale, (b) connecting transient absorption experiments on picosecond and nanosecond set-ups by the use of a reference dye solution of known behavior in both time ranges (“kinetic overlap actinometry”), and (c) applying a dedicated “classical” setup with improved time

Byrdin et al. resolution at selected wavelengths22 to ensure actual overlap of resolvable time scales. Our results demonstrate that, in wildtype E. coli PL, substantial losses occur prior to formation of the ultimate (FADH- 306TrpH°+) charge pair but that there is no significant kinetic development in the 100 ps-to-10 ns time window. In mutant photolyase lacking the ultimate intrinsic electron donor (terminal tryptophan replaced by redox inert phenylalanine; W306F), an important deprotonation/recombination process with a time constant of 0.85 ns is identified. Combination of the kinetic information thus obtained with all the kinetic data obtained earlier on wildtype photolyase and various electron transfer chain mutants, allows for drafting a tentative scheme of the reaction rates and free energies required for efficient functioning of the triple tryptophan chain. Finally, an improved absorption spectrum of the excited state of the flavin radical FADH°* can be deduced from our measurements. Materials and Methods Expression and purification of WT and W306F mutant E. coli PL and sample preparation were as described.15,18 Ultrafast transient absorption spectroscopy was performed with pump pulses centered at 620 nm and continuum probe pulses as described.18 “Classical” (quasi-continuous probe) flash transient absorption experiments with a time resolution of 2 ns were carried out with a wavelength selectable probe beam essentially as described,15 with modifications detailed below. A 300 ps time resolution for transient absorption measurements at 488 nm upon excitation with 100 ps flashes at 532 nm was achieved using a newly developed dedicated setup, employing a chopped continuous wave laser probe beam, as described,22 omitting the electronic amplifier. The same setup, but with 5 ns excitation flashes of ∼10 mJ at 620 nm from an OPO (Rainbow, Quantel, France), was used for kinetic overlap actinometry (see below). In all experiments the probe beam polarization was at magic angle (54.7°) with respect to that of the pump pulses. We emphasize that the excitation wavelengths in all experiments, either ∼620 or 532 nm, were such that only the FADH° form of the flavin was excited. Samples were kept on ice between and at 10 °C during measurements. Kinetic simulations were performed by solving the master equation for population dynamics using home written Mathematica (Wolfram Research, Champaign, IL) code. Absolute Quantum Yield Actinometry. Solutions of ruthenium complexes have proven to be convenient chemical actinometers, both for their stability and for their photophysical properties: ultrafast formation of a long-lived metal to ligand charge transfer (MLCT) state.23 We used [Ru(bpy)3]2+ that can conveniently be excited at 532 nm (where FADH° also absorbs) and has an IUPAC recommended24 φRu∆ε450 ) 10 ( 0.9 mM-1 cm-1,25 where φRu is the quantum yield of MLCT state formation of [Ru(bpy)3]2+ and ∆ε450 is the accompanying differential molar extinction coefficient at 450 nm. Absorption changes were measured with a time resolution of 2 ns in the same geometry and under identical flash excitation for PL and [Ru(bpy)3]2+ samples in an optical cell of 2 × 10 mm rectangular section. Deviating from the setup described in ref 15, excitation was with 300 ps flashes of ∼10 mJ at 532 nm (Nd/YAG laser YG501-10 from Quantel, France) along the 2 mm path and absorption changes were detected along the 10 mm optical path, using an oscilloscope bandwidth of dc to 600 MHz. Absorbance at 532 nm was low (OD2mm < 0.04) and comparable for the two samples. Flash excitation was in the linear regime (as verified by a saturation curve).

Photoactivation Intermediates in DNA Photolyase

J. Phys. Chem. A, Vol. 114, No. 9, 2010 3209

TABLE 1: Results of Kinetic Overlap Actinometry photolyaseb transient absorption method

l/mm

picosecond pump probe (final amplitude) nanosecond quasi-continuous probe (initial amplitude)

1 10

a

RuN3c in DMF

∆A488/mOD

I /au

∆A488/mOD

Id/au

excitation corrected amplitude ratio (∆A488/I)PL/(∆A488/I)RuN3

-1.1 ( 0.1 -13.7 ( 0.6

1 1

-5 ( 1 -58.7 ( 1.8

0.98 1.19

0.22 ( 0.06 0.28 ( 0.02

d

a

Path length of probe light. b 460 and 33 µM in pico- and nanosecond methods, respectively. c 780 and 86 µM in pico- and nanosecond methods, respectively. d Convolution of pump pulse spectrum and sample absorption spectrum.

The quantum yield φPL for formation of the (FADH+ 306TrpH° ) state was calculated from the observed absorbance change of photolyase at 490 nm, ∆A490, that of [Ru(bpy)3]2+ at 450 nm, ∆A450, as well as the optical densities, AX, of the samples at the excitation wavelength according to

φPL ) φRu[∆A490/(AX∆ε490)]PL /[∆A450/(AX∆ε450)]Ru

(1) The differential absorption coefficient for formation of (FADH+ -1 -1 was taken from 306TrpH° ) in photolyase ∆ε490) 3.2 mM cm ref 16. Kinetic Overlap Actinometry. [Ru(bpy)3]2+ does not absorb at 620 nm where we excited the flavin radical in the pump probe experiments. We therefore used a modified complex, RuN3 (cisbis(isothiocyanato)bis(2,2′-bipyridyl-4,4′-dicarboxylato)ruthenium)26 from Solaronix (Aubonne, Switzerland), where two of the bipyridyls are doubly carboxylated and the third one is replaced by two isothiocyanate ligands. These substitutions shift the absorption tail of the ruthenium complex up to 800 nm. Unfortunately, the quantum yield for excited state formation has not been reported for this complex in solution. However, the excited state lifetime (20 ns in dimethylformamide (DMF)27) is long enough to be detectable also by “classical” flash absorption, so that RuN3 can be used as a common reference for experiments on photolyase on both setups. Measurement of kinetic traces on photolyase samples were “sandwiched” between two series of measurements on the dye. To compensate for not exactly equal sample absorptions between photolyase and dye at the wavelength of excitation (∼620 nm), for both setups the absorption spectra Ai(λ) of the two samples (the index i stands for PL or Ru) were convoluted with the spectral profile of the excitation pulse E(λ) and the respective ratios I of the integrals

I)

∫ Ai(λ) E(λ) dλ ∫ APL(λ) E(λ) dλ

(2)

were applied as correction factors for each setup (Table 1). Signal amplitudes were compared at the probe wavelength 488 nm. Results and Discussion Quantum Yield Determination on Different Time Scales. Figure 2 compares nanosecond transient absorption measurements of PL at 490 nm and [Ru(bpy)3]2+ at 450 nm under conditions of equal flash excitation at 532 nm. The initial PL signal is attributed to the (FADH- 306TrpH°+) state; the small decay on a 200 ns scale is assigned to deprotonation of 11 + 306TrpH° , yielding the long-lived state (FADH 306Trp°). The 2+ [Ru(bpy)3] signal decays on the 1 µs scale as expected.28 On the basis of the initial amplitudes of these two traces and using eq 1, we obtain a quantum yield of 18.6% for the formation of (FADH- 306TrpH°+). Experiments at 2 and 4 times lower excitation energies gave virtually identical quantum yields

Figure 2. Actinometric quantum yield determination of photolyase photoactivation. Absorbance changes upon identical flash excitation at 532 nm were recorded at 450 nm for 150 µM [Ru(bpy)3]2+ (upper trace) and at 490 nm for 45 µM photolyase (lower trace). The optical densities at 532 nm over the 2 mm excitation path were 0.0348 and 0.0212, respectively. A total of 64 signals were averaged for each trace. Small arrows indicate the signal amplitudes extrapolated to time zero (signal halfrise) by fits to an exponential decay plus a constant.

(18.1% and 18.7%, respectively). Taking into account the (9% error of the published φRu∆ε450 value of the actinometer (see Materials and Methods) and an estimated error of (10% for ∆ε490 of photolyase, the quantum yield for the formation of (FADH- 306TrpH°+) should be included between 15 and 23%. This yield is significantly higher than an earlier estimate of 10 ( 2%.21 The latter value was obtained with reference to 360 nm absorption changes of [Ru(bpy)3]2+, employing ε360 ) 104 M-1 cm-1 for its excited state. The latter value appears to be underestimated by a factor of ∼2,29 explaining the discrepancy with our quantum yield determination. Our results imply that at the onset of our nanosecond experiments, a quantum yield loss of ∼80% with respect to total excited FADH° has already occurred. We therefore went further to investigate the yield at earlier times by use of relatiVe actinometry, i.e., comparison of the photolyase signal with a reference signal on both the ultrafast and slower time scales (kinetic overlap actinometry), using as a relative reference the red-aborbing dye RuN3 that has an excited state lifetime of 20 ns in DMF. The amplitude ratio between dye and photolyase signal (Table 1) after 100 ps (ultrafast pump probe) agreed within error ((30%) with that at the earliest time accessible with the “classical” transient absorption setup (time resolution limited by the 5 ns duration of the excitation pulse; the observed signal was extrapolated to time zero). Acquisition of a new laser as excitation light source (100 ps pulse length fwhm at 532 nm), as well as use of continuous laser light for monitoring and an optimized detection system, recently allowed us to extend the time window accessible to the classical technique into the subnanosecond time range.22

3210

J. Phys. Chem. A, Vol. 114, No. 9, 2010

Byrdin et al.

Figure 3. Detection of subnanosecond kinetics in photolyase on a newly developed transient absorption setup.22 Absorption changes at 488 nm upon excitation by 100 ps flashes of ∼5 mJ at 532 nm were recorded for 50 µM of wildtype (lower trace) and of W306F mutant photolyase (upper trace). A total of 64 signals were averaged for each trace. Fitting the upper trace by an exponential decay plus a constant yielded a decay time constant of 0.85 ns (smooth line).

Flash induced absorption changes detected at 488 nm on wildtype and W306F mutant PL using this apparatus are shown in Figure 3. No significant decay of the initial bleaching is visible in the 10 ns time window in wildtype PL. We emphasize that in W306F mutant PL under exactly identical experimental conditions a clear phase with an 0.85 ns lifetime is observed (Figure 3, discussed below), highlighting that the lack of decay in the WT signal is not due to limited time resolution of the instrument. From both the “kinetic overlap actinometry” and the direct subnanosecond measurements, we conclude that for wildtype PL, there is no significant evolution of signal amplitude in the previously poorly characterized time window from 100 ps to 10 ns. Therefore, no FADH- TrpH°+ charge recombination processes take place in this time scale. This assessment contrasts with an earlier suggestion that a major charge recombination takes place in ∼3 ns, based on extrapolation of a small signal observed to linearly decay using a 1 ns optical delay line.20 As the residue harboring the radical state at ∼100 ps was identified as W306 in polarized absorption experiments,10 the ensemble of our results imply that FADH- 306TrpH°+ recombination does not play a role on the time scale 100 ps to 10 ns. Spectrum of the Excited State FADH°*. The unambiguous determination of the quantum yield of 19 ( 4% for (FADHTrpH°+) at ∼100 ps, after the 30 ps phase11 reflecting electron transfer and competing processes (see below), allows for a renewed analysis of the initial transient spectrum (prior to the 30 ps phase) reflecting (FADH°* minus FADH°). The asymptotic spectrum (t ) 100 ps) can well be modeled by (FADHTrpH°+ minus FADH° TrpH).10 Using the above quantum yield, we can establish the amplitude of the ground state bleaching of FADH° with respect to the initial difference spectrum, and determine the spectrum associated with FADH°* (Figure 4). This analysis indicates that the FADH° bleaching accounts for most of the structure in the visible part of the initial spectrum. This result deviates from an analysis of this spectrum we made five years ago,30 which only took into account quantum yield loss during the first charge separation step; as discussed below, the ensemble of now-available data allows us to assess the presence of another sub-30 ps loss process. Our new analysis shows that the excited flavin radical FADH°* has a broad and

Figure 4. Construction of the absolute optical spectrum associated with the FADH°* state from ultrafast spectroscopic data. (A) Transient absorption spectra associated with the state prior to the 30 ps decay phase of FADH°* (initial, red) and after completion of the 30 ps phase (final, blue), obtained from a global analysis (as in re 30) of ultrafast spectroscopic data taken from refs 18 (visible) and 30 (near-infrared). The final spectrum corresponds to the state (FADH- 306TrpH°+). The contribution of the bleaching of the FADH° ground state (FADH°, green) to the final spectrum is superimposed. The difference between this spectrum and the final spectrum at the blue side is due to contributions of TrpH°+ and FADH- absorption. (B) Absolute spectra of FADH° bleaching (FADH°, green) and of initial state formation (initial, red) were obtained by scaling the corresponding spectra in panel A using ε580 ) 5 mM-1 cm-1 for FADH° 39 and the final state quantum yield of 0.19 for the concentration ratio of final and initial states. The FADH°* spectrum (brown solid) is calculated by subtracting the green spectrum from the red one. The dotted and dashed brown lines correspond to calculations with quantum yields of 15% and 23%.

rather featureless absorption band (molar extinction around 4 mM-1 cm-1 in the 450-650 nm range). The rather steep lowering of the FADH°* spectrum above ∼630 nm may be due to the superimposition of stimulated emission (which gives rise to a negative ∆A signal) to the broad excited state absorption spectrum; fluorescence of the flavin radical in a flavodoxin was reported to rise at ∼630 nm and peak at ∼700 nm.31 W306F Mutant. As mentioned above, in contrast to wildtype photolyase, in W306F mutant photolyase there is a clear decay phase with a time constant of 0.85 ns (Figure 3). As in this mutant, prior to the 0.85 ns phase, 359TrpH°+ is present,10 this phase reflects decay of the FADH- 359TrpH°+ state. Because our “classical” subnanosecond setup is limited to wavelengths where sufficiently stable continuous lasers are available, a

Photoactivation Intermediates in DNA Photolyase

J. Phys. Chem. A, Vol. 114, No. 9, 2010 3211

Figure 5. Kinetic and energetic scheme of the triple tryptophan electron transfer chain in E. coli photolyase. Time constants (A) and standard reaction free energies (B) taken from previous studies are printed in black, those obtained from the present study in blue. Energies are not drawn to scale. See text and Table 2 for details.

complete spectrum of the kinetic phase could not be obtained. However, we previously already predicted the presence of a phase in the 200 ps to 10 ns range, reflecting two parallel processes: (FADH- 359TrpH°+) charge recombination and + 359TrpH° deprotonation, the latter with a ∼3.5 times lower rate and yield than the overall process.10,15 This prediction was based on transient spectra recorded at 150 ps10 and at 10 ns.15 The proton released from 359TrpH°+ may not escape to the bulk solution but rather remain in the protein interior during the lifetime of the state (FADH- 359Trp°) (see ref 15 for a discussion of possible intraprotein proton acceptors). The present determination of the time constant of the overall process of concomitant charge recombination and deprotonation at 0.85 ns allows us to determine the time constants of 359TrpH°+ deprotonation at 3 ns and of (FADH- 359TrpH°+) charge recombination at 1.2 ns. This recombination may proceed directly or via the proximal tryptophan W382, and hence the 1.2 ns time constant is a lower limit for each of these two channels (see Figure 5A). Once the deprotonation of 359TrpH°+ to 359Trp° completes, further recombination in W306F mutant photolyase can occur either via reprotonation to 359TrpH°+ followed by electron transfer from FADH- (direct or via W382) or in the inverse order, by electron transfer from FADH- followed by reprotonation of 359Trp- (downward arrow from state 4 to state 0 in Figure 5A). We cannot distinguish as yet between these two channels, which might also occur concomitantly. The overall decay of the (FADH- 359Trp°) state takes place on the time scale of ∼0.5 µs and is not strictly single exponential.15 As the decay channel of (FADH- 359Trp°) cannot be faster than the observed ∼0.5 µs, recombination of this state via reprotonation (reaction 4 f 3 in Figure 5A) followed by recombination of FADH- and 359TrpH°+ (which takes 1.2 ns, see above) must be slower than 0.5 µs. From this constraint we can estimate limiting values for the equilibrium constant for the deprotonation of 359TrpH°+, K34 ) k34/k43, and, hence, for

the standard reaction free energy ∆G°34 ) -kBT ln K34, where kB is Boltzmann’s constant and T ) 283 K in our experiments. As k43 must be much smaller than both k34 (referring to the states in Figure 5, kij denotes the rate constant of reaction i f j) and the overall rate of recombination from state 3, we obtain k43 < (0.85 ns)-1 (0.5 µs)-1/(1.2 ns)-1 ) (0.35 µs)-1. Therefore, with k34 ) (3 ns)-1 and k43 < (0.35 µs)-1, it follows that K34 >120. We conclude that the energetic stabilization due to deprotonation of 359TrpH°+ is at least 0.12 eV. Kinetic and Energetic Scheme. Along with the body of previously published kinetic data, the newly determined quantum yield of (FADH- 306TrpH°+) formation in wildtype PL, and the determination of the rates of back electron transfer from FADHto 359TrpH°+ and deprotonation of the latter in W306F mutant PL (that cannot be directly determined in wildtype PL due to the very rapid rereduction of 359TrpH°+ by 306TrpH), allow us to set constraints to the time constants (Figure 5A) and standard free energies (Figure 5B) of most of the reactions involved in the photoactivation of wildtype photolyase. In developing this scheme, we made the following simplifying assumptions: (1) replacement of a member of the tryptophan triad by phenylalanine does only affect reactions involving the replaced residue. (2) All reactions are kinetically and energetically homogeneous. (3) The states considered in the scheme do not undergo energetic relaxations during their lifetimes. Time constants and standard reaction free energies that were taken from previous studies are printed in black in Figure 5; the origins of these values are indicated in Table 2. Quantities that were obtained from the present study are printed in blue in Figure 5, and their deduction is discussed below. Our present results on the W306F mutant photolyase provided an estimate of 3 ns for the time constant of deprotonation of + 359TrpH° (reaction 3 f 4). The 1.2 ns time constant for the overall charge recombination of (FADH- 359TrpH°+) is a lower limit for the direct recombination channel (reaction 3 f 0). Furthermore, it could be concluded that deprotonation of

3212

J. Phys. Chem. A, Vol. 114, No. 9, 2010

Byrdin et al.

TABLE 2: Origin of Time Constants and Reaction Free Energies Shown in Figure 5 reaction

τ

1f0

80 ps

k/s-1 1.25 × 10

-∆G°/eV 10

1f2

45 ps

2.2 × 1010

1f3

25 ns

4 × 107

2f0

2.5 × 1011

2f3

1.1 × 1011

1.9 0.11

3f0

>1.2 ns

3f4

3 ns

1.39 >0.12

4f0

>0.5 µs

0.24

5f6

200 ns

5 × 106 0.18 (at pH 7)

>25 µs

5f0

(9 ps)-1, K23 ) k23/k32 > 90 and ∆G°23 < -0.11 eV. We

ref 13 13, 18 13 18 this work this work this work this work, 15 this work 15 10 this work, 14 11 11, 32 14 32, 38 14

conclude that the driving force of electron transfer from 359TrpH to 382TrpH°+ is at least 0.11 eV. Finally, recombination of (FADH- 306TrpH°+) via repopulation of 359TrpH°+ cannot be faster than the overall recombination time constant of 25 µs deduced from the pH dependence of the lifetime of the state (FADH- 306Trp°).14 As the equilibrium between (FADH- 306TrpH°+) and (FADH- 359TrpH°+) is installed in less than 30 ps, i.e., much faster than recombination and deprotonation of the (FADH- 359TrpH°+) state (1.2 and 3 ns, respectively; see above), we can approximate the situation by a fast pre-equilibrium that is depopulated via recombination of (FADH- 359TrpH°+). We obtain (1.2 ns)-1/(K35 + 1) < (25 µs)-1; hence K35 > 2.1 × 104 and ∆G°35 < -0.24 eV. We conclude that the driving force of the third electron transfer step (3 f 5) is at least as high as 0.24 eV. The above analysis shows that the free energy of charge separated states that are composed of chemically identical species (FADH- TrpH°+) can differ by more than 0.35 eV, and hence that the TrpH°+/TrpH redox potential can vary by more than 0.35 V for different tryptophans within the protein. This remarkably high variation must be due to subtle tuning of the protein environment. A difference between the redox potentials of the last two tryptophans of ∼0.2 V has indeed been predicted in a theoretical study32 (our analysis >0.24 V) and can be ascribed to the fact that W306 is surface-exposed. In contrast, the substantial (>0.11 V) difference between the first two tryptophans, as deduced from our analysis, was not predicted32

Photoactivation Intermediates in DNA Photolyase

Figure 6. Simulation of the population dynamics according to the scheme shown in Figure 5. Colors and numbering correspond to the states in that figure. Rate constants and reaction free energies were set according to the (limiting) values indicated in Figure 5 and Table 2, except that we set k30 ) k40 ) k50 ) 0, which is a consequence of choosing the limiting values for ∆G°23, ∆G°34 and ∆G°35. The reverse reaction rates k21, k32, k43, k53, and k65 were set according to ∆G°ij ) -kBT ln(kij/kji). An alternative simulation with k30, k40, and k50 fixed according to the limiting values in Figure 5 and Table 2 and k32 ) k43 ) k53 ) 0 yielded virtually identical population dynamics (not shown).

but allows the reaction chain to be effectively downhill for all steps. This difference may be related to the proximity of the flavin and W382, although we note that recent Stark spectroscopy experiments give no indication of charge transfer interactions between their aromatic rings.33 The driving force of the first electron transfer step (from 382TrpH to excited FADH°, reaction 1 f 2) can now be estimated as the difference between the total free energy available from excited FADH° (1.9 eV corresponding to the 0-0 transition at 650 nm) and the sum of the free energy gaps of the subsequent steps 2 f 3 (>0.11 eV), 3 f 5 (>0.24 eV), and 5 f 0 (1.15 eV; see Figure 5B and Table 2). We conclude that actually there can be up to 0.4 eV driving force available for the first step in the flavin reducing triple tryptophan intraprotein electron transfer chain of E. coli. Our analysis has led to the net downhill reaction scheme of Figure 5, with gradually stabilizing charge pairs. The considerable (∼80%) quantum yield loss due to competing decay processes from the two highest energy states might be the drawback of using intrinsic protein residues as redox factors, rather than more specialized cofactors and their often complex integration systems, as for instance in photosynthetic reaction center electron transfer chains. One might also speculate that for PL and the homologous family of cryptochrome blue light receptor proteins, the “nanowire” properties (overall very fast electron transfer) are more important for signaling-type functions than quantum yield considerations. On the other hand, it is also possible that in members of the cryptochrome family employing oxidized FAD as an electron acceptor (see below), quantum yield losses, both due to competing excited state decay processes and due to charge recombination from the primary charge pair, are substantially lower. Concluding Remarks. Finally, Figure 6 shows the time evolution of the population of the different states in wildtype PL based on the scheme in Figure 5 (see legend for the choice

J. Phys. Chem. A, Vol. 114, No. 9, 2010 3213 of the parameters for which only limits are available; as mentioned above, we assumed the mutant enzymes differ only from WT for reactions involving the replaced residue). For all but two states they correspond to the populations observed in time-resolved experiments. For (FADH- 382TrpH°+) and (FADH- 359TrpH°+) they correspond to the detection limits (and may actually be lower than indicated), because the direct observation of the fast electron hopping between the three tryptophans in wildtype PL is hindered by the relatively slow initial charge separation step. This situation appears to be different in cryptochromes where the resting state of the flavin is the oxidized form (FADox) and in E. coli PL prepared with FADox. In these systems, FADox* decay, presumably via reduction by the tryptophan chain, has been reported to occur on the order of 1 ps, substantially faster than FADH°* decay in PL,34-36 which may be related to the higher driving force of the initial electron transfer step. Such proteins, although difficult to handle, may therefore provide good systems to detect and characterize sizable populations of the intermediate tryptophan radical states. Whereas electron transfer rates within the tryptophan triad in these systems have been suggested on the basis of one-color experiments probing at 710 nm,35 i.e., out of the spectral range of absorption of tryptophan radicals,37 potential identification of the intermediate TrpH°+ states must await full spectral studies in the TrpH°+ absorption band combined with site-directed mutagenesis. In conclusion, our new data and present analysis allow setting constraints on the kinetics and energetics of all the relevant electron transfer reactions within the tryptophan triad in PL. In particular, our results demonstrate that all three forward electron transfer steps are associated with a substantial loss in free energy. For the second and third step, this is remarkable in view of the fact that the redox states are chemically identical and may help in determining the efficiency of the signaling pathway in the homologeous family of cryptochrome blue light receptors where the three tryptophans are conserved.11 A main perspective will be the possibility to fully kinetically unravel the chain in cryptochromes bearing the flavin cofactor in the oxidized form. Acknowledgment. We thank Drs. Sandrine Villette, Agathe Espagne, and Jie Pan for their contributions to obtaining the data in Figures 2-4, respectively. This work was supported by grant ANR-05-BLAN-0304. References and Notes (1) Sancar, A. Chem. ReV. 2003, 103, 2203. (2) Weber, S. Biochim. Biophys. Acta 2005, 1707, 1. (3) Yasui, A.; Eker, A. P. M. DNA photolyases. In DNA damage and repair; Nickoloff, J. A., Hoekstra, M. F., Eds.; Humana Press Inc.: Totowa, NJ, 1998; Vol. 2, pp 9. (4) Enescu, M.; Lindqvist, L.; Soep, B. Photochem. Photobiol. 1998, 68, 150. (5) Kao, Y.-T.; Saxena, C.; Wang, L.; Sancar, A.; Zhong, D. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 16128. (6) MacFarlane, A. W.; Stanley, R. J. Biochemistry 2003, 42, 8558. (7) Espagne, A.; Byrdin, M.; Eker, A. P. M.; Brettel, K. ChemBioChem 2009, 10, 1777. (8) Li, Y. F.; Heelis, P. F.; Sancar, A. Biochemistry 1991, 30, 6322. (9) Park, H.-W.; Kim, S.-T.; Sancar, A.; Deisenhofer, J. Science 1995, 268, 1866. (10) Lukacs, A.; Eker, A. P. M.; Byrdin, M.; Brettel, K.; Vos, M. H. J. Am. Chem. Soc. 2008, 130, 14394. (11) Aubert, C.; Vos, M. H.; Mathis, P.; Eker, A. P. M.; Brettel, K. Nature 2000, 405, 586. (12) Cheung, M. S.; Daizadeh, I.; Stuchebrukhov, A. A.; Heelis, P. F. Biophys. J. 1999, 76, 1241. (13) Byrdin, M.; Eker, A. P. M.; Vos, M. H.; Brettel, K. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 8676. (14) Byrdin, M.; Sartor, V.; Eker, A. P. M.; Vos, M. H.; Aubert, C.; Mathis, P.; Brettel, K. Biochim. Biophys. Acta 2004, 1655, 64.

3214

J. Phys. Chem. A, Vol. 114, No. 9, 2010

(15) Byrdin, M.; Villette, S.; Eker, A. P. M.; Brettel, K. Biochemistry 2007, 46, 10072. (16) Byrdin, M.; Villette, S.; Espagne, A.; Eker, A. P. M.; Brettel, K. J. Phys. Chem. B 2008, 112, 6866. (17) Kapetanaki, S. M.; Ramsey, M.; Gindt, Y. M.; Schelvis, J. P. J. Am. Chem. Soc. 2004, 126, 6214. (18) Lukacs, A.; Eker, A. P. M.; Byrdin, M.; Villette, S.; Pan, J.; Brettel, K.; Vos, M. H. J. Phys. Chem. B 2006, 110, 15654. (19) Saxena, C.; Sancar, A.; Zhong, D. J. Phys. Chem. B 2004, 108, 18026. (20) Wang, H.; Saxena, C.; Quan, D.; Sancar, A.; Zhong, D. J. Phys. Chem. B 2005, 109, 1329. (21) Heelis, P. F.; Sancar, A. Biochemistry 1986, 25, 8163. (22) Byrdin, M.; Thiagarajan, V.; Villette, S.; Espagne, A.; Brettel, K. ReV. Sci. Instrum. 2009, 80, 043102. (23) Damrauer, N. H.; Cerullo, G.; Yeh, A.; Boussie, T. R.; Shank, C. V.; McCusker, J. K. Science 1997, 275, 54. (24) Kuhn, H. J.; Braslavsky, S. E.; Schmidt, R. Pure Appl. Chem. 2004, 76, 2105. (25) Yoshimura, A.; Hoffman, M. Z.; Sun, H. J. Photochem. Photobiol. A 1993, 70, 29. (26) Nazeeruddin, M. K.; Kay, A.; Rodicio, I.; Humphrybaker, R.; Muller, E.; Liska, P.; Vlachopoulos, N.; Gra¨tzel, M. J. Am. Chem. Soc. 1993, 115, 6382. (27) Argazzi, R.; Bignozzi, C. A.; Heimer, T. A.; Castellano, F. N.; Meyer, G. J. Inorg. Chem. 2002, 33, 5741.

Byrdin et al. (28) Juris, A.; Balzani, V.; Barigelletti, F.; Campagna, S.; Belser, P.; von Zelewsky, A. Coord. Chem. ReV. 1988, 84, 85. (29) Lachish, U.; Infelta, P. P.; Gra¨tzel, M. Chem. Phys. Lett. 1979, 62, 317. (30) Pan, J.; Byrdin, M.; Aubert, C.; Eker, A. P. M.; Brettel, K.; Vos, M. H. J. Phys. Chem. B 2004, 108, 10160. (31) Kao, Y.-T.; Saxena, C.; He, T.-F.; Guo, L.; Wang, L.; Sancar, A.; Zhong, D. J. Am. Chem. Soc. 2008, 130, 13132. (32) Popovic, D. M.; Zmiric, A.; Zaric, S. D.; Knapp, E.-W. J. Am. Chem. Soc. 2002, 124, 3775. (33) Kodali, G.; Siddiqui, S. U.; Stanley, R. J. J. Am. Chem. Soc. 2009, 131, 4795. (34) Brazard, J. Ph.D. Thesis, Universite´ Pierre et Marie Curie, Paris, 2009. ¨ ztu¨rk, N.; Li, J.; Wang, L.; (35) Kao, Y.-T.; Tan, C.; Song, S.-H.; O Sancar, A.; Zhong, D. J. Am. Chem. Soc. 2008, 130, 7695. (36) Shirdel, J.; Zirak, P.; Penzkofer, A.; Breitkreuz, H.; Wolf, E. Chem. Phys. 2008, 352, 35. (37) Solar, S.; Getoff, N.; Surdhar, P. S.; Armstrong, D. A.; Singh, A. J. Phys. Chem. 2002, 95, 3639. (38) Balland, V.; Byrdin, M.; Eker, A. P. M.; Ahmad, M.; Brettel, K. J. Am. Chem. Soc. 2009, 131, 426. (39) Wang, B.; Jorns, M. S. Biochemistry 1989, 28, 1148.

JP9093589