Distinguishing Chromophore Structures of Photocycle Intermediates of

9118

J. Phys. Chem. B 2008, 112, 9118–9125

Distinguishing Chromophore Structures of Photocycle Intermediates of the Photoreceptor PYP by Transient Fluorescence and Energy Transfer§ Daniel Hoersch,† Harald Otto,† Michael A. Cusanovich,‡ and Maarten P. Heyn*,† Biophysics Group, Department of Physics, Freie UniVersita¨t Berlin, Arnimallee 14, 14195 Berlin, Germany, and Department of Biochemistry and Molecular Biophysics, UniVersity of Arizona, Tucson, Arizona, 85721 ReceiVed: February 8, 2008; ReVised Manuscript ReceiVed: April 18, 2008

The cinnamoyl chromophore is the light-activated switch of the photoreceptor photoactive yellow protein (PYP) and isomerizes during the functional cycle. The fluorescence of W119, the only tryptophan of PYP, is quenched by energy transfer to the chromophore. This depends on the chromophore’s transition dipole moment orientation and spectrum, both of which change during the photocycle. The transient fluorescence of W119 thus serves as a sensitive kinetic monitor of the chromophore’s structure and orientation and was used for the first time to investigate the photocycle kinetics. From these data and measurements of the ps-fluorescence decay with background illumination (470 nm) we determined the fluorescence lifetimes of W119 in the I1 and I1′ intermediates. Two coexisting distinct chromophore structures were proposed for the I1 photointermediate from time-resolved X-ray diffraction (Ihee, H., et al. Proc. Natl. Acad. Sci. U.S.A., 2005, 102, 7145): one with two hydrogen bonds to E46 and Y42, and a second with only one H-bond to Y42 and a different orientation. Only for the first of these is the calculated fluorescence lifetime of 0.22 ns in good agreement with the observed one of 0.26 ns. The second structure has a predicted lifetime of 0.71 ns. Thus, we conclude that in solution only the first I1 structure occurs. The high resolution structure of the I1′ intermediate, the decay product of I1 at alkaline pH, is still unknown. We predict from the observed lifetime of 1.3 ns that the chromophore structure of I1′ is quite similar to that of the I2 intermediate, and I1′ should thus be considered as the alkaline (deprotonated) form of I2. 1. Introduction Photoactive yellow protein (PYP) is the structural prototype for the PAS domain family of signaling proteins and a model system for signal transduction in photoreceptors.1 The PYP from the halophilic purple phototrophic bacterium Halorhodospira halophila is a small 14-kDa soluble cytoplasmic protein. The blue light absorption of PYP (λmax ≈ 446 nm) is due to its p-hydroxycinnamoyl chromophore that is covalently bound via a thioester linkage to cysteine-69. In the dark, the phenol group of the chromophore is deprotonated, and the C7dC8 bond is trans. The ionization of the chromophore and the hydrogen bonding of its O- group to E46 and Y42 are largely responsible for the observed spectral tuning in the dark state. Photoisomerization around the C7dC8 bond is rapid (2 ps)2,3 and is followed by a sequence of slower thermal relaxations that leads to recovery of the dark state in less than 1 s. The first long-lived intermediate relevant for this study is I1 (λmax ≈ 460 nm) which is formed in several ns and in which the C7dC8 bond of the chromophore is cis. In the photoisomerization the thioester group of the chromophore is flipped over, breaking the hydrogen bond between the carbonyl of the chromophore and the amino group of cysteine 69.2,4 I1 decays in several hundred microseconds to the I2 intermediate, which has a protonated chromophore and a § Abbreviations: FRET, fluorescence resonance energy transfer; SVD, singular value decomposition; PAS domain, acronym formed from the names of the first three proteins recognized as sharing this sensory domain; PYP, photoactive yellow protein; LED: light emitting diode; fwhm, full width at half-maximum. * Corresponding author. Phone: 40-30-83856160; fax: 49-30-83856299; e-mail: [email protected]. † Freie Universita ¨ t Berlin. ‡ University of Arizona.

blue-shifted absorption maximum at ∼370 nm.5–7 In several milliseconds I2 decays to I2′, which absorbs at ∼350 nm 5–7 and is believed to be the signaling state. The I2 to I2′ transition is associated with a major conformational change, which has been characterized by NMR,8,9 CD,10 small-angle X-ray scattering,11 and FTIR.12,13 Formation of I2′ is associated with exposure of a hydrophobic patch,14 presumably the recognition and binding site for a response regulator. Experiments with hydrophobic dyes showed that these bind transiently to I2′ but not to I2.15 The existence of an equilibrium between the I2 and I2′ intermediates was shown by photoreversal experiments.16 At alkaline pH I1 decays to I1′ (λmax ≈ 425 nm) rather than to I2.17 The photocycle kinetics of PYP have been investigated in great detail by transient absorption spectroscopy.18,19,6,7,17 The current model of a sequential cycle with back reactions and equilibria is shown in Figure 1A for the alkaline pH range. The rates and equilibria are pH,6,7,17 and salt 20,21 dependent. Proton release and uptake are tightly coupled to the cycle kinetics.15 The spectra of the photocycle intermediates were determined by the extrapolated difference method 22,6,17 and are shown in Figure 1B. Additional insight on the protonation state and on the chromophore structure has been obtained using time-resolved FTIR difference spectroscopy 12,13 and time-resolved resonance Raman spectroscopy.23 A great advantage of PYP as a model system for photoreceptors is that high resolution structures are available for most intermediates from time-resolved X-ray diffraction and from diffraction studies with trapped intermediates.24–29 In this way the time courses of the structural changes of PYP at room temperature have been determined by diffraction methods from nanoseconds to seconds.26

10.1021/jp801174z CCC: $40.75  2008 American Chemical Society Published on Web 07/08/2008

Transient Tryptophan Fluorescence in PYP

J. Phys. Chem. B, Vol. 112, No. 30, 2008 9119

Figure 2. Set-up for the measurement of the transient tryptophan fluorescence changes during the PYP photocycle. The sample is illuminated by an LED emitting at 280 nm from above. The fluorescence light is collected under an angle of 90° and detected by a photomultiplier tube (PM). The 10 ns excitation pulse at 460 nm that triggers the PYP photocycle enters the cuvette (path length 3 mm in both directions) from the side also with 90° geometry. To protect the PM from stray light at 280 nm and from the scattered excitation pulse, the filters WG320 or WG305 and UG11 were placed in front of the PM.

Figure 1. (A) Photocycle of PYP at alkaline pH together with the apparent time constants. (B) Absorption spectra of the dark state P and the photointermediates I1, I1′, I2′ (from 17) and tryptophan emission spectrum.5

PYP contains a unique tryptophan W119, positioned at the interface between the central β-sheet and the N-terminal domain. Its emission spectrum is shown in Figure 1B, and its steadystate30 and time-resolved5 fluorescence have been characterized. The fluorescence of W119 is strongly quenched by energy transfer to the hydroxycinnamoyl chromophore.5 One requirement for FRET is spectral overlap between the emission spectrum of the donor and the absorption spectrum of the acceptor. Figure 1B shows that this requirement is satisfied for the dark state P and intermediates, and that the spectral overlap differs in the various intermediates. Measurements of the timeresolved fluorescence depolarisation of donor and acceptor showed that they are immobilized on the fluorescence time scale.5 Because of the isomerization and rotation of the chromophore during the photocycle, the transition dipole moment geometry of the donor/acceptor pair changes in each intermediate (κ2 factor, see eq 3 and Figure 3). Since the rate of energy transfer is proportional to the product of the spectral overlap integral J and the angular factor κ2 (see eq 2 below), it differs in the various intermediates leading to different fluorescence lifetimes.5 This implies that the steady-state fluorescence also differs significantly in each intermediate. Here we apply transient fluorescence for the first time to monitor the photocycle kinetics and to obtain information on the fluorescence lifetimes and chromophore structures of the I1 and I1′ intermediates. The availability of high resolution structures of a number of key intermediates provides a unique opportunity to compare the measured fluorescence lifetime with their values calculated on the basis of the chromophore structure and the spectral overlap. For each intermediate the κ2 factor can be calculated from the transition dipole moment orientations derived from the available

X-ray structures. In previous work we showed that there was excellent agreement between the measured fluorescence lifetimes of W119 in the dark state P, the I2 intermediate, and the bleached form of the mutant E46Q and the lifetimes calculated from their high resolution structures and spectral overlap integrals.5 In this way our approach was validated. The intermediate lifetimes were measured in a photostationary intermediate mixture produced by background illumination at 470 nm. These studies were at acid and neutral pH where only the P, I2, and I2′ intermediates accumulate.5 In the present work we extend these fluorescence measurements to the I1 and I1′ intermediates which only accumulate with background illumination at alkaline pH.17 We focus on the following two questions concerning these intermediates: (1) According to time-resolved X-ray diffraction experiments at room temperature two I1 structures with quite distinct chromophore structures coexist in parallel photocycle branches.26 Can we detect these two I1 chromophore structures also in solution on the basis of their different fluorescence lifetimes, and do the measured and structure-based fluorescence lifetimes agree? (2) What is the chromophore structure of the I1′ intermediate, whose high-resolution X-ray diffraction structure is unknown ? Can we make a plausible structure prediction consistent with the observed fluorescence lifetime ? 2. Experimental Methods Protein Production and Purification. Recombinant H. halophila holo-PYP was produced by coexpression with the biosynthetic enzymes TAL and pCL and subsequently purified from E. coli BL21(DE3) as described.31 Transient Fluorescence Spectroscopy/Kinetics. These measurements were performed with a modification of our setup for transient absorption spectroscopy22 shown in Figure 2. The three orthogonal directions available in the cuvette are used to excite the tryptophan, to excite the photocycle, and to observe the tryptophan emission. The tryptophan fluorescence is excited by an LED emitting at 280 nm with a half-width of 10 nm, and the emission is detected by a photomultiplier (PM). The time

9120 J. Phys. Chem. B, Vol. 112, No. 30, 2008

Hoersch et al.

TABLE 1: Fluorescence Lifetimes of W119 Calculated from X-ray Diffraction Dataa PDB file

intermediate

RDA (Å)

R (deg)

δ (deg)

β (deg)

κ2

J (10-15 cm3 M-1)

τmax (ns)

τmin (ns)

1TS7b

P (A) I1 (B) I1 (C) I1′ (D) I1′

16.3 16.1 17.1 18.7 18.4

139 145 114 91 93

31 30 28 29 29

108 115 91 65 68

2.65 2.91 1.18 0.22 0.26

6.35 4.08 4.08 6.54 6.54

0.17 0.22 0.71 2.41 2.10

1.45 1.30

1TS7b 1TS7b 1TS6b 1S4Sc

a Angles and distances between the transition dipole moments of W119 and the chromophore for the X-ray structures of various photointermediates. The angles R, β, and δ are defined in Figure 3. κ2 was calculated from eq 3. J is the numerically calculated overlap integral between the W119 emission and the chromophore absorption spectra of Figure 1B. τmin and τmax are the minimum and maximum tryptophan fluorescence lifetimes, respectively. A, B, C, and D refer to the panels of Figure 6 where these structures are shown. b Reference 26. c Reference 29.

Figure 3. Transition dipole moment geometry of the donor (µbD) and acceptor (µbA) in P (blue) and I2 (purple) from.24 RbDA is the vector from donor to acceptor. The orientations and origins of µbA and µbD in the molecular framework of the acceptor and donor are discussed in the text. For µbA(I2) two positions are shown. The dashed arrow is the vector connecting the ends of the conjugated system. The dotted arrow is constructed by parallel displacement of the dashed arrow to the origin of µbA(P). The values of R, β, δ, and RbDA derived from the X-ray diffraction structures are listed in Table 1.

course of this signal is recorded from 0.1 µs to seconds after exciting the sample with a 20 ns blue flash at 460 nm from an excimer-dye laser.22 In order to protect the PM from stray light from the LED and the excitation flash, the filters WG305 or WG320 and UG11 were placed in front of it. In this way the evolution of the steady-state fluorescence of W119 during the photocycle is recorded. The cuvette was placed in a thermostatted sample holder. All measurements were performed at 23 °C. The PM current I is proportional to the number of detected photons which is itself proportional to the product of the number of the excited fluorophores ni and the fluorescence quantum yield Φi in the intermediate i in the photocycle:

I∝

∑ Φini ∝ ∑ Φi(1 - 10-A ) ) ∑ Φi(1 - 10-A) i

i

i

i

In the last step we assumed that the absorption Ai of W119 does not change during the photocycle. This was verified by measuring the transient absorption changes at 280 nm, which were found to be small (data not shown). Before the flash, PYP is in the ground-state P with a fluorescence lifetime τP of around 0.18 ns.5 After the flash, a fraction of the initial ground-state population has entered the photocycle with several intermediates i with fluorescence lifetimes τi. In our plots we show the normalized fluorescence change ∆F/F0 ) (F - F0)/F0 ) (I I0)/I0 which represents the time-resolved fluorescence difference. Using the expression for I above and the fact that Φi is proportional to τi we find:

∆F(t) ) F0

before the flash. The sum extends over all intermediates. From measurements of the fluorescence decay we know that our preparations contain a small amount of apo-PYP which lacks the chromophore and which consequently has a very long lifetime τA of 4.8 ns (see next section). Its concentration cA is estimated to be 4-7% of c0 from the long fluorescence decay component in the dark of Figure 5A and from ref 5. Because of its long lifetime this small population makes a large contribution to F0. It does not contribute to the difference ∆F, however. We therefore have to correct the above expression for the apo-PYP contribution as follows:

∆F(t) ) F0

∑ ci(t)(τi - τP) i

c0τP + cAτA

1 ) fC

∑ ci(t)(τi - τP) i

c0τP

(1)

with the correction factor fC ) 1 + cAτA/c0τP. The sum Σi ci/c0 is called the fraction cycling. With our configuration its value is typically between 0.3 and 0.4. Fluorescence Lifetime Experiments. These were performed as described32 by time-correlated single-photon counting using a ps Ti-sapphire laser system. The instrumental response function had a width (fwhm) of about 70 ps at the excitation wavelength. For excitation of tryptophan at 298 nm, a frequency tripler was used. An LED emitting at 470 nm with a fwhm of 30 nm (Conrad, 10 mW) was used to generate a photostationary mixture of P and the I1, I1′, and I2′ intermediates in the fluorescence cuvette (3 × 3 mm, from Helma (105.251-QS)). A UG11 band filter from Schott was used to prevent the scattered LED light from reaching the multichannel plate detector and to block the very weak chromophore fluorescence. A WG320 cut-off filter from Schott selected the tryptophan emission above 320 nm and blocked scattered excitation light. Data analysis was performed by SVD analysis as described5 and the DOS version of the fitting program Globals Unlimited. Radiationless energy transfer is well understood theoretically and well established experimentally. The rate of energy transfer kT is given by

kT ) 8 . 71 × 1023

J κ2 ΦD -1 s 6 τD n4 RDA

(2)

∑ ci(t)(τi - τP)

where J is the spectral overlap integral between the emission spectrum of the donor and the absorption spectrum of the acceptor, in units of cm3 M-1. κ2 is the angular factor of the interaction between the transition dipole moments of donor (µbD) and acceptor (µbA). Using the angles R, β, and δ defined in Figure 3 we have

c0τP

κ2 ) [cos β - 3 cos R cos δ]2

i

ci (t) is the concentration of molecules in intermediate i at time t. c0 is the initial concentration of molecules in P at t ) 0

(3)

Depending on the values of R, β, and δ, varies between 0 and 4. n is the index of refraction. RDA is the distance between κ2

Transient Tryptophan Fluorescence in PYP

Figure 4. (A) Transient tryptophan fluorescence at pH 9.9 at 0 (dotted line) and 500 mM KCl (solid line), 10 mM Tris buffer, filters WG320 and UG11, temperature 23 °C. Average of 30 flashes. (B) Transient absorption data at 360, 390, 420, 440, and 490 nm for the same sample (500 mM KCl) as in A. The common vertical lines in A and B indicate the time constants for a joint fit of the fluorescence and absorption data with a sum of three exponentials. The fit curves are superimposed on the data. (C) Transient fluorescence at pH 7.8, 9.2, and 10.0. Conditions:100 mM KCl, 10 mM Tris, filters WG305 and UG11, temperature 23 °C.

the transition dipole moments of donor and acceptor in angstroms (see Figure 3). τD is the fluorescence lifetime of the donor in the absence of the acceptor in sec. ΦD is the fluorescence quantum yield of the donor in the absence of the acceptor. According to eq 2, the rate constant for energy transfer is proportional to the product of J and κ2. 3. Results Photocycle Kinetics Monitored by the Transient Fluorescence of W119. The transient fluorescence ∆F/F0 of W119 was measured as described in Materials and Methods using the setup of Figure 2. Representative data at pH 9.9 are shown in Figure 4A. For comparison, the corresponding time courses of the absorbance changes ∆A of the same sample, at various detection wavelengths, are shown in Figure 4B. The fluorescence and absorbance data could be fitted well by a common set of three exponentials. The fit curves are also shown in Figure 4A and B. The global time constants are indicated by the common set of vertical lines. These time constants of 162 µs, 1.3 ms, and

J. Phys. Chem. B, Vol. 112, No. 30, 2008 9121

Figure 5. (A) Fluorescence decay of W119 in the dark and in the light, in the presence of background illumination from an LED emitting at 470 nm. Excitation wavelength 298 nm. Filters: UG11 and WG320(2x). (B) Light minus dark difference decay curves at pH 9.7 and 7.9. The arrow marks the longest 1.3 ns decay time required in the three exponential fit of the data. (C) pH dependence of the contributions of the species with the fluorescence lifetimes 0.05 ns (I2′) and 1.3 ns (I1′) to the photostationary equilibrium. The solid curves show a simultaneous fit of the data to the Henderson-Hasselbalch equation with a pKa of 9.95.

660 ms correspond to the transitions between the I1 and I1′ intermediate, the I1/ I1′ equilibrium to I1/ I1′/ I2′ equilibrium and the recovery, respectively (see scheme of Figure 1). These assignments were made previously.17 Whereas the absorption signal monitors the photocycle via the transient changes of the chromophore, the fluorescence signal monitors the cycle via the fluorescence changes of W119, i.e. at a site well removed from the chromophore (∼16 Å) at the interface between the central β-sheet and the N-terminal domain. From the excellent agreement between the kinetic time constants we conclude that at both sites the same transitions are sensed (observed). To obtain the kinetic constants, the fluorescence seems to be superior, since the amplitude of the I1 to I1′ and I1′ to I2′ transitions are much larger in the fluorescence than in the absorbance signal, where these transitions are only weakly apparent in the time trace at 390 nm. The information content of these two methods differs furthermore in that the changes are due to differences in the

9122 J. Phys. Chem. B, Vol. 112, No. 30, 2008 extinction coefficients and fluorescence lifetimes of the intermediates, respectively. The initial value of ∆F/F0 around 1 µs is small and positive. It can be calculated accurately from the sum of the three exponential fit amplitudes and has the value 0.075. This sum a1+a2+a3 is indicated by the horizontal line in Figure 4A. At these early times the only intermediate in the cycle is I1. We can then calculate the fluorescence lifetime of I1 from eq 1. The fraction cycling was determined as 0.4 in this experiment by comparison of our time-resolved absorption curves with the data from ref 17. For the fluorescence lifetime of P and apoPYP we used the previously determined values of 0.18 and 4.8 ns, respectively.5 Depending on the fraction of apo-PYP in the sample, we find from eq 1 for the fluorescence lifetime of I1 values between 0.25 ns (4% apo) and 0.28 ns (7% apo).The small value of ∆F/F0 leads as expected to a fluorescence lifetime for I1 that is only slightly larger than that of P. With a risetime of 162 µs ∆F/F0 increases to about 0.32 in the I1 to I1/I1′ transition (Figure 4A). In this transition a fraction of I1 decays to I1′, and these two intermediates are in equilibrium.17 The large increase in fluorescence means that W119 has a much larger fluorescence lifetime in the newly formed intermediate I1′ than in I1. With a time constant of 1.3 ms I1 and I1′ decay into an equilibrium of the three intermediates I1, I1′, and I2′, and this is accompanied by a large decrease in fluorescence. This means that the newly formed I2′ intermediate has a shorter fluorescence lifetime than the preceding I1/I1′ mixture. This finding is consistent with our earlier studies of the tryptophan fluorescence in the acid pH range, where we obtained a fluorescence lifetime of 0.04 ns for the I2′ intermediate.5 The data we discussed so far were in the presence of 500 mM KCl. Corresponding data in the presence of 0 mM KCl are also presented in Figure 4A. The initial value of ∆F/F0 is the same in 0 and 500 mM KCl indicating that the amount of I1 cycling and its lifetime are unaffected by the salt. The salt effect on the time constants is quite small and mainly limited to the recovery kinetics which are slowed at 500 mM KCl. This effect is well-known.20,21 There is however a very large effect on the amplitudes of the I1 to I1′ and I1′ to I2′ transitions, which both increase significantly. The maximal fluorescence at 500 µs increases from 0.24 to 0.32. This effect suggests that in the presence of 500 mM KCl the I1/I1′ equilibrium is shifted in the direction of I1′, as if the I1′ intermediate is stabilized by the presence of salt. It is not possible to determine the fluorescence lifetimes of I1′ and I2′ from the transient fluorescence time traces using eq 1 alone, since this equation contains too many unknowns. We can make reasonable estimates however by using additional information about the intermediate time courses from experiments performed under similar conditions (pH 10 and 50 mM KCl).17 In this way we find from the concentrations of I1 and I1′ around 500 µs (the maximum of ∆F/F0) and the value of ∆F/F0 at that time, that the lifetime of I1′ varies between 1.0 and 1.8 ns, depending on the percentage apo-PYP and the KCl concentration. In the next section we will determine these lifetimes directly and accurately from measurements of the fluorescence decay curves of a photostationary mixture of these intermediates. In Figure 4C the transient fluorescence changes are presented as a function of pH. There is clearly a strong pH dependence in the alkaline pH range. The time courses for the three pH values (at 100 mM KCl) could be fitted by common time constants of 120 µs and 1 ms for the first two transitions and a

Hoersch et al. pH dependent recovery rate. The pH effect mainly affects the amplitudes. The largest effects are on the amplitude of the I1 to I1′ transition and on the fluorescence of the I1/I1′/I2′ equilibrium in the late part of the cycle (>20 ms). At pH 7.8 apparently no I1′ is formed. Moreover the fluorescence change is negative in the microsecond region. This means that at pH 7.8 an intermediate dominates in the microsecond-equilibrium with a fluorescence lifetime much smaller than P. From our previous work we know that this is due to I2′ with a lifetime of 0.04 ns.5 From corresponding transient absorption measurements it is known that with increasing pH more I1′ is formed and less I2′ with a pKa of about 9.9.17 Since, as shown above, I1′ has a much larger fluorescence lifetime than P, this explains the strong increase of the fluorescence with pH in the first transition. The change in sign of the ms plateau, to positive values above pH 9 is likewise due to the decrease in the amount of I2′ (shorter lifetime than P) and the simultaneous increase in the amounts of I1 and I1′ (longer lifetime than P). Determination of the Tryptophan Fluorescence Lifetimes in the I1′ and I2′ Intermediates. In the previous section we were able to obtain the fluorescence lifetime of the I1 intermediate from the initial value of the transient fluorescence change, since no other intermediates contributed. The values of the life times in the other intermediates were determined from the fluorescence decay curves of a photostationary equilibrium of all intermediates generated by the presence of background illumination from an LED emitting at 470 nm. Because of the long recovery time of 660 ms these intermediates accumulate in sufficient amounts in the presence of the intense background illumination provided by this source. The time-resolved tryptophan fluorescence decay in the dark (P, black line) and in a photostationary equilibrium (red line) are presented in Figure 5A. The two decay curves cross. Below 0.4 ns the signal from the photostationary equilibrium is smaller than that of P. This means that a species contributes in the equilibrium with a lifetime smaller than that of P. For times above 0.4 ns the decay curve in the light is above the dark curve, indicating the presence of an intermediate with a lifetime longer than that of P. Because of the customary logarithmic intensity scale this crossover is difficult to observe in the data of Figure 5A. We therefore present the same data on a linear intensity scale in Figure 5B (red curve). Moreover we show here the light minus dark difference intensity. This has the added advantage that contributions that are the same in the two states cancel. The longest living component with a lifetime of about 4.8 ns is due to a small amount of apo-PYP which lacks the chromophore.5 The tryptophan fluorescence is thus not quenched by energy transfer, leading to the long 4.8 ns lifetime. The contribution from this population, which does not have a photocycle, is of no interest. Difference decay curves like those of Figure 5B were measured at 10 different pH values between pH 8 and 11 (only data for pH 7.9 and 9.7 are shown). A joint SVD decomposition of this data set was performed as described.5 The two significant components could be fitted with a sum of three exponentials with fluorescence lifetimes of 0.05, 0.17, and 1.3 ns. The 1.3 ns lifetime is marked by the arrow in Figure 5B. Since this is a light minus dark difference decay, the rising component with 0.17 ns is due to P, whereas the two decaying components are due to intermediates. This value for P is in excellent agreement with our previous determination of 0.18 ns.5 The value of 0.05 ns is in very good agreement with our previous determination of 0.04 ns for I2′.5 For I1 we obtained a lifetime of 0.25 to 0.28 ns, close to that of P. Therefore this contribution cannot be distinguished from that of P. The longest

Transient Tryptophan Fluorescence in PYP lifetime of 1.3 ns must thus be assigned to I1′. This is consistent with the results of the previous section where we found that the tryptophan fluorescence lifetime in I1′ was estimated to lie between 1.0 and 1.8 ns. Further support for this assignment comes from an analysis of the pH dependence of the decay amplitudes of the two species with decaying components (0.05 and 1.3 ns). The pH dependence of the amplitudes of these species is evident from the traces for pH 7.9 and 9.7 in Figure 5B. Their amplitudes were normalized by dividing them by their sum. The resulting pH dependence of the 0.05 and 1.3 ns components is plotted in Figure 5C. A joint fit of these data to the Henderson-Hasselbalch equation with a pKa of 9.95 is shown in Figure 5C. The pH dependence of the composition of the long living equilibrium of photointermediates has been investigated previously by flash photolysis.17 It was found that the contribution of I2′ decreases and of I1′ increases with pH with a common pKa of 9.9.17,33 The 0.05 ns species should thus be assigned to I2′ and the 1.3 ns species to I1′. Calculating the Tryptophan Fluorescence Life Times from the High Resolution Intermediate Structures and Structural Predictions. Information about the chromophore structure may be obtained from a comparison of the measured fluorescence lifetimes with those calculated on the basis of the available high resolution X-ray structures. This method worked well for P and the I2 intermediate in our previous study in the pH range 4 to 8, where we found excellent agreement between the measured and calculated structure-based lifetimes.5 There we measured life times of 0.18, 0.8, and 0.04 ns for P, I2, and I2′, respectively.5 In this investigation we obtained in addition the life times of I1 from transient fluorescence (0.26 ns) and of I1′ from the fluorescence decay (1.3 ns). For I1 two different structures were obtained from timeresolved X-ray diffraction at room temperature,26 which are shown in Figure 6B and 6C. In the photocycle model derived from the data these I1 intermediates coexist in a branched parallel part of the photocycle.26 In the first structure (PDB code 1TS7, Figure 6B) the two hydrogen bonds to E46 and Y42 are preserved and the chromophore structure is similar to that of P (Figure 6A). The isomerization around the 7-8 double bond involves the thioester part of the chromophore leading to a reorientation of the carbonyl group. The κ2 value of 2.91 which we calculate from this structure using eq 3 is thus close to that of the dark state P (2.65). In the second structure the hydrogen bond with E46 is broken, the chromophore ring has moved, and the transition dipole moment changed its orientation reducing the κ2 value to 1.18 (Figure 6C). These values together with the angle input values, the spectral overlap integrals, the Fo¨rster radius, the donor-acceptor distance and the calculated fluorescence life times are listed in Table 1. The overlap integral J was calculated using the I1 and I1′ spectra from17 with a spectral resolution of 10 nm. For the wavelength points 300, 310, and 320 nm we extrapolated the I1 and I1′ spectra, making the reasonable assumption that they have the same spectral shape as for P. For the other parameters required for the calculation of the Fo¨rster transfer rate and the fluorescence lifetime according to eq 2, i.e. the index of refraction, the fluorescence quantum yield, and the fluorescence lifetime of W119 in the absence of energy transfer, we took the same values as those in ref 5. For the structure with the two hydrogen bonds (Figure 6B) we calculated a lifetime of 0.22 ns in excellent agreement with our observed lifetime between 0.25 and 0.28 ns. For the structure with only one hydrogen bond (Figure 6C) our calculation results in a lifetime of 0.71 ns which is due to the

J. Phys. Chem. B, Vol. 112, No. 30, 2008 9123

Figure 6. X-ray structures of the chromophore region for the groundstate P(A), the photointermediate I2(D) and the two different coexisting structures for the I1 intermediate (B and C) from ref 26. Hydrogen bonds to neighboring amino acid side chains are indicated by red dashed lines (red: E46; dark blue: Y42; magenta: R52).

much smaller κ2 value of 1.18. Since there is almost perfect agreement with one structure and a major discrepancy with the other, we conclude that under the conditions of our experiment (solution, alkaline pH) only the structure with the two hydrogen bonds (Figure 6B) is present in the photocycle. To get an idea of the sensitivity of the calculated fluorescence lifetimes in the two I1 structures on the angles and distances of donor and acceptor, we presumed a structural uncertainty of (1° for the angles R, β, δ and ( 0.2 Å for RDA. This estimate is based on the differences in the three structure models for P, listed in Table 1 in ref 5. Calculation of the lifetimes for every possible donor-acceptor configuration leads to standard deviations of (0.01 ns and (0.05 ns for the two I1 lifetimes of 0.22 ns and 0.71 ns. For the I1′ intermediate no X-ray structure exists. So we screened the set of known intermediate structures for one whose

9124 J. Phys. Chem. B, Vol. 112, No. 30, 2008 κ2 value in combination with the I1′ overlap integral leads to a calculated fluorescence lifetime that matches the measured one (1.3 ns). Such an agreement was obtained when we gave the chromophore in the I1′ intermediate the structure observed in I2. The result of this hybrid calculation is shown in the last two lines of Table 1. The lifetimes calculated in this way for structure 1TS6 (PDB) varies between 1.45 and 2.41 ns and between 1.30 and 2.10 ns for structure 1S4S. This range is due to the uncertainty in the choice of the origin of the acceptor transition dipole moment, as illustrated in Figure 3. The direction of the transition dipole moment is given by the line connecting the two ends of the conjugated system.5,34 This leads to an apparent increase of the donor-acceptor distance from 16.3 to 18.7 Å and an associated increase in the lifetime. The value calculated in this way we define as τmax. Alternatively we could displace the transition dipole moment to the origin before isomerization in P, as shown in Figure 3. This leads to the second value we term τmin. Our measured value of 1.3 ns is close to the lower bound τmin. At acid and neutral pH, the I2′ intermediate is preceded by the I2 intermediate, both of which have protonated chromophores. At alkaline pH, I2′ is preceded by I1′, which has presumably a deprotonated chromophore, based on its λmax value of 425 nm.17,35 Since the structure of I1′ is consistent with that of I2 it is appropriate to call I1′ the deprotonated or alkaline form of I2. 4. Discussion The kinetics of the photocycle of PYP have been investigated by time-resolved absorption spectroscopy in the UV/vis 6,7,15–19 and in the IR,12,13 by time-resolved resonance Raman spectroscopy,23 and by time-resolved X-ray diffraction.24–29 Here we introduced a novel method and monitored the photocycle kinetics by the transient fluorescence of W119. Fluorescence is a highly sensitive method that responds to changes in fluorescence lifetime occurring during the photocycle. From a comparison between the time traces from the fluorescence (Figure 4A) and absorbance (Figure 4B), we concluded that the kinetic time constants derived from these methods are in good agreement, as expected. Using eq 1 we found that the fluorescence lifetime of tryptophan-119 lies between 0.25 and 0.28 ns in the I1 intermediate. Although we could not determine the fluorescence lifetime of the I1′ intermediate in the same way as for I1 from eq 1 alone, our observations allowed us to estimate its lifetime by using additional information on its time course from previous photocycle experiments at pH 10.17 This procedure led us to an estimated lifetime between 1 and 1.8 ns. Whereas the difference in absorption spectra between I1 and I1′ is small, the change in fluorescence lifetime is large. The I1 to I1′ transition was thus easy to detect in the fluorescence (Figure 4A), but was barely detectable in the absorbance signal (Figure 4B). To obtain the fluorescence lifetimes of the I1′ and I2′ intermediates with high precision we measured the fluorescence decay curves after ps-excitation in a photostationary mixture of P, I1, I1′, and I2′ and in the dark (P). In the presence of background illumination from an LED emitting at 470 nm a sufficient amount of these intermediates could be accumulated in a photostationary state for analysis. From an analysis of the light minus dark difference decay curves we found fluorescence life times of 0.17, 0.05, and 1.3 ns for P, I2′, and I1′, respectively. The first two values are in excellent agreement with the values of 0.18 and 0.04 ns which we determined in our previous study.5 Further support for the assignment of the 1.3 ns component to I1′ was obtained from the pH dependence of the amplitude of

Hoersch et al. this species in the mixture. We found that this amplitude increased with pH according to the Henderson-Hasselbalch equation with a pKa of 9.95 (Figure 5C). From transient absorption measurements it is known that the I1′ concentration in the cycle increases with pH in the same way with a pKa of 9.9.17 The 1.3 ns species should thus be identified with I1′. Having obtained the lifetime of W119 in the I1 and I1′ intermediates, we used their values to make predictions for the chromophore structures of these intermediates. The fluorescence of W119 differs in each intermediate since the rate of energy transfer between the donor W119 and the 4-hydroxycinnamoyl acceptor changes. According to the available X-ray structures from time-resolved and intermediate trapping studies,24–29 the distance between donor and acceptor does not change during the cycle.26 The changes in the rate of energy transfer are thus due to differences in the spectral overlap integral and the κ2 factor (see eq (3)). From the coordinates of the chromopore in the intermediates, we derived the relevant angles R, β, and δ (Figure 3), which allowed us to compute κ2. The spectral overlap integral J was calculated as described. Using J and κ2 a unique prediction was made for the fluorescence lifetime of W119 in each intermediate. These structure-derived values were compared with the measured fluorescence life times. In our previous study we measured the fluorescence life times of P, I2, and I2′.5 For P and I2 there was excellent agreement between the measured (0.18 and 0.82 ns) and the structure-derived (0.15 and 0.85 ns) fluorescence life times.5 Our analysis of the experimental data was based on the assumption that each photointermediate has a single lifetime. It is well-known, however, that a single tryptophan may have multiple fluorescence lifetimes. In our system the assignment of one decay time to each conformational species is validated by the excellent agreement for P and I2, as discussed further in ref 5. For I2′ there was no agreement. The most likely reason for this discrepancy is that the protein state that was assigned to I2′ in the structural work differs from the I2′ intermediate in solution. In solution the protein has undergone a major structural change in I2′ and is partially unfolded, causing for example additional quenching of the Trp fluorescence by nearby amino acids as discussed in ref 5. This transition is likely blocked or occurs with reduced extent in crystals. We applied the same procedure here to the I1 and I1′ intermediates. The I1 intermediate is present at all pH values but only accumulates significantly with background illumination at alkaline pH. The two coexisting I1 intermediates observed in recent time-resolved X-ray diffraction work at room temperature26 differ significantly in their chromophore structures, as illustrated in Figure 6B and 6C. As a consequence of the different chromophore orientations the transition dipole moments and the corresponding κ2 values differ by a factor of 2.5 (Table 1), leading to predicted life times of 0.22 ns (Figure 6B) and 0.71 ns (Figure 6C). The structure of Figure 6B in which the chromophore is hydrogen bonded to both Y42 and E46 is in very good agreement with our measured lifetime of 0.25 to 0.28 ns. The structure of Figure 6C with only one hydrogen bond to Y42 is clearly not compatible. We conclude that the structure of Figure 6C does not occur in solution. Our experiments confirm that the structure of Figure 6B is the solution structure from the excellent agreement between measured and calculated lifetimes. Time-resolved and low temperature steady state FTIR experiments showed that the carboxyl group of E46 remains protonated in I1, suggesting that the hydrogen bond between this group and the O- of the chromophore is preserved in this intermediate.13,4 Our experiments provide structural support for

Transient Tryptophan Fluorescence in PYP this suggestion. Time-resolved resonance Raman experiments demonstrated the presence of two chromophore conformations in the I1 intermediate, which are in a pH dependent equilibrium with a pKa of 6.36 At the high pH of our experiments we would detect only the high pH species. Since the structure of I1′ is unknown we did not have a structure-derived κ2 factor to calculate the fluorescence lifetime in this intermediate. Instead we searched for an intermediate with a chromophore structure and associated κ2 factor that together with the I1′ overlap integral would lead to a lifetime close to the observed 1.3 ns. In the choice of intermediate we were guided by the idea that at acid and neutral pH there are three sequential intermediates I1, I2, and I2′ absorbing at 460, 370, and 350 nm, respectively, whereas at alkaline pH the role of I2 seems to be taken over by I1′: with I1, I1′, and I2′ absorbing at 460, 425, and 350 nm. The chromophore is deprotonated in I1 and I1′, and protonated in I2 and I2′. From the structural work it is known that the chromophore is more exposed to the aqueous medium in I2 than in I1.24 We therefore selected as a structural model for I1′ the I2 structure, reasoning that I1′ may be just the deprotonated form of I2 at alkaline pH. In the last two lines of Table 1 two different I2 structures are listed with κ2 values of 0.22 and 0.26. In the second column we call this intermediate I1′ since we used the I1′ spectral overlap integral together with the I2 chromophore structure in the calculation of the fluorescence lifetime. The measured lifetime of I1′ of 1.3 ns is close to the calculated lower bound listed in the last two columns of Table 1. We conclude that the hypothesis that in I1′ the chromophore structure is the same or very similar as in I2 is consistent with the measured fluorescence lifetime of I1′. I1′ may thus be called the “alkaline form” of I2. Our observation that the equilibrium between I1 and I1′ shifts toward I1′ at high salt (Figure 4A) is consistent with an exposed/accessible and negatively charged chromophore in I1′. At high ionic strength the exposed negative charge in I1′ would be more effectively screened than in I1, where the charge is buried, stabilizing the I1′ form. At very high pH a new ground-state species, the so-called “alkaline form”, is formed in which E46 is deprotonated.37 Since the pKa for this transition is above 11.5,37 the contribution of this species in our experiments was negligible. 5. Conclusion We introduced a novel method, transient fluorescence, to monitor the photocycle kinetics of PYP. Using this solution method we determined the fluorescence lifetime of the unique tryptophan-119 in the I1 intermediate of 0.26 ns. This value is in excellent agreement with the lifetime calculated on the basis of the proposed crystal structure of I126 with both hydrogen bonds of the chromophore’s O- to E46 and Y42 preserved but excludes the proposed coexisting chromophore structure with only one hydrogen bond to Y42.26 By measuring the picosecond-fluorescence decay of W119 in a photostationary mixture of intermediates we obtained the fluorescence lifetime of the I1′ intermediate. Using its value of 1.3 ns we predict that the chromophore in I1′ has the same structure as in I2. Acknowledgment. We thank Chandra Joshi, Berthold Borucki, and Terry Meyer for helpful discussions. This work was supported by NIH grant GM 66146 (to M.A.C.) and DFG grant He 1382/14-2 (to M.P.H. and H.O.) References and Notes (1) Cusanovich, M. A.; Meyer, T. E. Biochemistry 2003, 42, 4759.

J. Phys. Chem. B, Vol. 112, No. 30, 2008 9125 (2) Groot, M. L.; van Wilderen, L. J. G. W.; Larsen, D. S.; van der Horst, M. A.; van Stokkum, I. H. M.; Hellingwerf, K. J.; van Grondelle, R. Biochemistry 2003, 42, 10054. (3) Heyne, K.; Mohammed, O. F.; Usman, A.; Dreyer, J.; Nibbering, E. T. J.; Cusanovich, M. A. J. Am. Chem. Soc. 2005, 127, 18100. (4) Xie, A.; Hoff, W. D.; Kroon, A. R.; Hellingwerf, K. J. Biochemistry 1996, 35, 14671. (5) Otto, H.; Hoersch, D.; Meyer, T. E.; Cusanovich, M. A.; Heyn, M. P. Biochemistry 2005, 44, 16804. (6) Borucki, B.; Joshi, C. P.; Otto, H.; Cusanovich, M. A.; Heyn, M. P. Biophys. J. 2006, 91, 2991. (7) Shimizu, N.; Imamoto, Y.; Harigai, M.; Kamikubo, H.; Yamazaki, Y.; Kataoka, M. J. Biol. Chem. 2006, 281, 4318. (8) Rubinstenn, G.; Vuister, G. W.; Mulder, F. A.; Du¨x, P. E.; Boelens, R.; Hellingwerf, K. J.; Kaptein, R. Nat. Struct. Biol. 1998, 5, 568. (9) Bernard, C.; Houben, K.; Derix, N. M.; Marks, D.; van der Horst, M. A.; Hellingwerf, K. J.; Boelens, R.; Kaptein, R.; van Nuland, N. A. Structure 2005, 13, 953. (10) Harigai, M.; Imamoto, Y.; Kamikubo, H.; Yamazaki, Y.; Kataoka, M. Biochemistry 2003, 42, 13893. (11) Imamoto, Y.; Kamikubo, H.; Harigai, M.; Shimizu, N.; Kataoka, M. Biochemistry 2002, 41, 13595. (12) Xie, A.; Kelemen, L.; Hendriks, J.; White, B. J.; Hellingwerf, K. J.; Hoff, W. D. Biochemistry 2001, 40, 1510. (13) Brudler, R.; Rammelsberg, R.; Woo, T. T.; Getzoff, E. D.; Gerwert, K. Nat. Struct. Biol. 2001, 8, 265. (14) Meyer, T. E.; Tollin, G.; Hazzard, J. H.; Cusanovich, M. A. Biophys. J. 1989, 56, 559. (15) Borucki, B.; Devanathan, S.; Otto, H.; Cusanovich, M. A.; Tollin, G.; Heyn, M. P. Biochemistry 2002, 41, 10026. (16) Joshi, C. P.; Borucki, B.; Otto, H.; Meyer, T. E.; Cusanovich, M. A.; Heyn, M. P. Biochemistry 2005, 44, 656. (17) Joshi, C. P.; Borucki, B.; Otto, H.; Meyer, T. E.; Cusanovich, M. A.; Heyn, M. P. Biochemistry 2006, 45, 7057. (18) Hoff, W. D.; van Stokkum, I. H. M.; van Ramesdonk, H. J.; van Brederode, M. E.; Brouwer, A. M.; Fitch, J. C.; Meyer, T. E.; van Grondelle, R.; Hellingwerf, K. J. Biophys. J. 1994, 67, 1691. (19) Hendriks, J.; van Stokkum, I. H. M.; Hellingwerf, K. J. Biophys. J. 2003, 84, 1180. (20) Borucki, B.; Kyndt, J. A.; Joshi, C. P.; Otto, H.; Meyer, T. E.; Cusanovich, M. A.; Heyn, M. P. Biochemistry 2005, 44, 13650. (21) Hoersch, D.; Otto, H.; Joshi, C. P.; Borucki, B.; Cusanovich, M. A.; Heyn, M. P. Biophys. J. 2007, 93, 1687. (22) Borucki, B.; Otto, H.; Heyn, M. P. J. Phys. Chem. B 1999, 103, 6371. (23) Pan, D. H.; Philip, A.; Hoff, W. D.; Mathies, R. A. Biophys. J. 2004, 86, 2374. (24) Genick, U. K.; Borgstahl, G. E.; Ng, K.; Ren, Z.; Pradervand, C.; Burke, P. M.; Srajer, V.; Teng, T. Y.; Schildkamp, W.; McRee, D. E.; Moffat, K.; Getzoff, E. D. Science 1997, 275, 1471. (25) Genick, U. K.; Soltis, S. M.; Kuhn, P.; Canestrelli, I. L.; Getzoff, E. D. Nature 1998, 392, 206. (26) Ihee, H.; Rajagopal, S.; Srajer, V.; Pahl, R.; Anderson, S.; Schmidt, M.; Schotte, F.; Anfinrud, P. A.; Wulff, M.; Moffat, K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 7145. (27) Perman, B.; Srajer, V.; Ren, Z.; Teng, T.; Pradervand, C.; Ursby, T.; Bourgeois, D.; Schotte, F.; Wulff, M.; Kort, R.; Hellingwerf, K.; Moffat, K. Science 1998, 279, 1946. (28) Ren, Z.; Perman, B.; Srajer, V.; Teng, T. Y.; Pradervand, C.; Bourgeois, D.; Schotte, F.; Ursby, T.; Kort, R.; Wulff, M.; Moffat, K. Biochemistry 2001, 40, 13788. (29) Schmidt, M.; Pahl, R.; Srajer, V.; Anderson, S.; Ren, Z.; Ihee, H.; Rajagopal, S.; Moffat, K. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 4799. (30) Gensch, T.; Hendriks, J.; Hellingwerf, K. J. Photochem. Photobiol. Sci. 2004, 3, 531. (31) Kyndt, J. A.; Vanrobaeys, F.; Fitch, J. C.; Devreese, B. V.; Meyer, T. E.; Cusanovich, M. A.; Van Beeumen, J. J. Biochemistry 2003, 42, 965. (32) Otto, H.; Lamparter, T.; Borucki, B.; Hughes, J.; Heyn, M. P. Biochemistry 2003, 42, 5885. (33) Imamoto, Y.; Harigai, M.; Kataoka, M. FEBS Lett. 2004, 557, 75. (34) Heyn, M. P.; Borucki, B.; Otto, H. Biochim. Biophys. Acta 2000, 1460, 60. (35) Hendriks, J.; Hoff, W. D.; Crielaard, W.; Hellingwerf, K. J. J. Biol. Chem. 1999, 274, 17655. (36) Unno, M.; Kumauchi, M.; Hamada, N.; Tokunaga, F.; Yamauchi, S. J. Biol. Chem. 2004, 279, 23855. (37) Meyer, T. E.; Devanathan, S.; Woo, T.; Getzoff, E. D.; Tollin, G.; Cusanovich, M. A. Biochemistry 2003, 42, 3319.

JP801174Z