REVIEW ARTICLE

Ultrafast fluorescence upconversion technique and its applications to proteins Haik Chosrowjan1, Seiji Taniguchi1 and Fumio Tanaka1,2 1 Division of Laser Biochemistry, Institute for Laser Technology, Utsubo-Honmachi, Nishiku, Osaka, Japan 2 Department of Chemistry, Faculty of Science, Chulalongkorn University, Bangkok, Thailand

Keywords charge transfer; flavoproteins; fluorescence upconversion; fluorescence upconversion microscope; FMN-binding protein; intraprotein electrostatic interactions; photoactove yellow protein; photoisomerization; point mutants; ultrafast spectroscopy Correspondence H. Chosrowjan, Division of Laser Biochemistry, Institute for Laser Technology, Utsubo-Honmachi, 1-8-4, Nishiku, Osaka 550-0004, Japan Fax: +81 6 6878 1568 Tel: +81 6 6879 8756 E-mail: [email protected] (Received 28 October 2014, revised 15 December 2014, accepted 17 December 2014) doi:10.1111/febs.13180

The basic principles and main characteristics of the ultrafast time-resolved fluorescence upconversion technique (conventional and space-resolved), including requirements for nonlinear crystals, mixing spectral bandwidth, acceptance angle, etc., are presented. Applications to flavoproteins [wild-type (WT) FMN-binding protein and its W32Y, W32A, E13R, E13K, E13Q and E13T mutants] and photoresponsive proteins [WT photoactive yellow protein and its R52Q mutant in solution and as single crystals] are demonstrated. For flavoproteins, investigations elucidating the effects of ionic charges on ultrafast electron transfer (ET) dynamics are summarized. It is shown that replacement of the ionic amino acid Glu13 and the resulting modification of the electrostatic charge distribution in the protein chromphorebinding pocket substantially alters the ultrafast fluorescence quenching dynamics and ET rate in FMN-binding protein. It is concluded that, together with donor–acceptor distances, electrostatic interactions between ionic photoproducts and other ionic groups in the proteins are important factors influencing the ET rates. In WT photoactive yellow protein and the R52Q mutant, ultrafast photoisomerization dynamics of the chromophore (deprotonated trans-p-coumaric acid) in liquid and crystal phases are investigated. It is shown that the primary dynamics in solution and single-crystal phases are quite similar; hence, the photocycle dynamics and structural differences observed at longer time scales arise mostly from the structural restraints imposed by the crystal lattice rigidity versus the flexibility in solution.

Introduction Time-resolved fluorescence spectroscopy is one of the most widely used techniques for studying the structure, function and reaction dynamics of macromolecules in chemistry and biology [1–3]. Fluorescence is often sensitive to small environmental changes of chromophores – small molecules embedded in proteins and absorbing light, and hence responsible for the protein’s color. Thus, fluorescence measurements can reveal ligand-induced conformational changes in proteins, the

origins of charge transfer reactions, solvent relaxation phenomena, and local conformational changes in and around the chromophore in proteins. Most fluorescence decays occur in the time window of ~ 100 fs to nanoseconds, so measurements require short light pulses and high temporal resolution instrumentation. Many different techniques have been developed to obtain time resolution in fluorescence spectroscopy. For instance, photocathode-based techniques such as time-correlated

Abbreviations BBO, b-barium borate; ES, electrostatic; ET, electron transfer; FBP, FMN-binding protein; FWHM, full width at half maximum; Iso, isoalloxazine; NLO, nonlinear optical; PYP, photoactive yellow protein; WT, wild-type.

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

3003

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

single-photon counting [4] and streak camera [5], allow the detection of weak, time-resolved fluorescence signals; however, very careful deconvolution procedures are needed to obtain picosecond time resolution. At the subpicosecond resolution level, these techniques become both expensive and unstable, and system maintenance for day-to-day reproducibility of sensitive sample measurements is challenging and often impossible. In recent years, a number of advanced and stable ultrafast lasers (primarily Ti:sapphire-based lasers with ~ 100-fs or shorter pulse widths) and associated optoelectronic instruments have emerged and become commercially available. Hence, in practice, to obtain time resolution comparable to the excitation laser pulse width, nonlinear optical (NLO) laser sampling techniques could be the best choice. The use of an optical Kerr effect as an optical shutter was first proposed in 1968 [6]. It makes use of the transient birefringence (third-order nonlinear effect) induced in a medium with high nonlinear susceptibility v3 by an intense laser pulse to create an ultrafast shutter. Liquids (CS2, benzene, and toluene) or solidstate materials (glasses, fused silica doped with gold nanoparticles, etc.) have been used as a gate, and an instrument response function as fast as ~ 200 fs was demonstrated [7]. However, the Kerr shutter contrast is inherently poor, owing to the nuclear motion-induced slow birefringence recovery component. Furthermore, low sensitivity and spectral restriction to a visible range limit the applications of this technique. Another nonlinear sampling technique is based on the phenomenon of sum or difference frequency generation of light (second-order nonlinear effect) in a nonlinear crystal [potassium dihydrogen phosphate or KH2PO4, LiNbO3, b-barium borate (BBO) etc.], thus being an intrinsically high-resolution spectroscopic technique [8,9]. Because the signals are generated at the sum or difference frequencies of the emission and the gate pulse (higher or lower photon energies), this technique is called fluorescence ‘upconversion’ or ‘downconversion’ for sum or difference frequency generation, respectively. Although fluorescence downconversion is less common and deserves to be developed and applied more systematically [10], here we address only the fluorescence upconversion technique and its applications to proteins. With regard to time resolution, measurement sensitivity, and accuracy, the fluorescence upconversion technique is ultimately the most competitive one. This technique has been adopted for investigations in the UV [11], visible [12] and near-IR [13] spectral regions, and has been used to study many diverse phenomena, such as solvation dynamics [14–16], intramolecular coherent vibrations [17,18], ultrafast photoisomeriza3004

tion reaction dynamics [19,20], charge transfer reactions [21,22], fluorescence properties of DNA nucleosides and nucleotides [23], solvation responses at multiple sites in a globular protein [24], and water dynamics in biological recognition [25], in condensed matter. Several excellent review articles on the fluorescence upconversion technique and its applications have already been published [26–28], so in the section ‘Fluorescence upconversion technique’, the basic principles of the technique, requirements for nonlinear crystals, mixing spectral bandwidth, acceptance angle, etc. have been only briefly summarized. In the section ‘Application to flavoproteins’, as an application example of the conventional fluorescence upconversion technique, we present and discuss the effects of electrostatic (ES) charge variations in the proximity of donors and acceptors on ultrafast electron transfer (ET) dynamics of FMN-binding protein (FBP). In our laboratory, alongside the conventional (only time-resolved) fluorescence upconversion apparatus that is mostly used for studies in the liquid phase, we have also developed a microscope-integrated fluorescence upconversion system for time-resolved and space-resolved studies on protein reaction dynamics and other ultrafast phenomena in the solid-state phase. The space-resolved fluorescence upconversion technique is described in the subsection ‘Microscope-integrated (space-resolved and timeresolved) fluorescence upconversion apparatus’, and an example of its application to single crystals of photoactive yellow protein (PYP) is presented in the section ‘Ultrafast photoisomerization reaction dynamics of PYP in crystal and liquid phases’.

Fluorescence upconversion technique Basic principles The time resolution mechanism underlying the fluorescence upconversion technique is illustrated in Fig. 1A. The upconversion is actually a cross-correlation between the fluorescence and a probe laser pulse. At time t = 0, the sample is electronically excited by second harmonics (in UV spectral range applications – third harmonics) of an ultrafast laser pulse with frequency xp. The collected incoherent fluorescence (xf) and the probe laser pulse xp arriving at time t = s are cofocused in an NLO crystal oriented at an appropriate angle with respect to the fluorescence and laser beams. Sum frequency photons are generated only during the time when the probe laser pulse is present in the crystal, acting as a ‘gate’, thus keeping the time resolution within the laser pulse width. The time evolution of the fluorescence may then be traced by varying FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

the delay s of the probe laser beam. It is easy to show that the intensity of the signal beam at sum frequency and at a given delay time s is proportional to the correlation function of the fluorescence with the probe laser pulse. The main features for successful upconversion measurements are briefly summarized below. Additional details of these estimations, including corresponding formulas, can be found in books dealing with the principles of nonlinear optics, e.g. [29].

example is the O + O ? E conversion in type I NLO crystals, where O and E stand for ordinary (parallel polarization for x) and extraordinary (perpendicular polarization for 2x) beams propagating in the NLO medium. Hence, for the collection of different polarized emission components, rotation of excitation pulse polarization with a thin half-wave plate is required. Phase-matching condition The upconversion process is efficient only when the condition for phase matching is satisfied. This happens for a narrow band of wavelengths centered at a wavelength determined by the phase-matching angle h of

Polarization The upconversion process (sum frequency generation) is intrinsically a polarization selection process. An A

Phase matching angle θ

Fluorescence ω f

Upconversion signal ωS

NLO crystal

Laser pulse ωp

Excitation laser pulse ωExc.

Fluorescence ωf

Delayed probe laser pulse ωp

Delay τ 0

B

PMT Mono chromator

Nd:YAG Laser (SHG)

Photon Counter

τ

t

532 nm

Ti:Sapphire Laser

Delay Stage

Pulse Compressor ~70 fs, 800 mW BBO (Type I)

Computer

ω

Fig. 1. (A) Schematic diagram showing the basic principles of the fluorescence upconversion technique. (B) Schematic diagram of the conventional fluorescence upconversion apparatus. BS, beam splitter; PMT, photo-multiplier tube; SHG, second harmonic generation.

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

800 - 880 nm, 120 fs

BS 2ω

BBO

Sample λ/2 Plate

3005

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

the NLO crystal. The conditions for the involved frequencies x and corresponding wave vectors k (2pn/k) are xS = xf + xp and kS = kf + kp, respectively. Here the subscripts S, f and p denote the upconverted signal, fluorescence and probe laser beams, respectively. In practice, the angle between kf and kp is kept constant (~ 15° in our apparatus). For BBO crystals, which are well suited for the upconversion, the phasematching angle (the angle between the crystal optical axis and the vertical axis) is ~ 30° for 800-nm and 400nm mixing (signal at 266.7 nm). For other monitoring wavelengths, the crystal angle can be gently tuned. For the probe wavelength at ~ 800 nm, the BBO crystal can cover a 250–2000-nm broad mixing spectral range. Acceptance angle The acceptance angle is the angle where the phase mismatch is < p/2. This is another important factor in upconversion experiments. As the fluorescence is spontaneous, it is emitted in all directions from the excited spot of the sample, and then collected and refocused onto the NLO crystal in a broad cone. Hence, the larger the acceptance angle that can be phase-matched by the crystal, the larger the upconversion efficiency. Roughly, the acceptance angle / increases inversely with the crystal length L. Thus, for thinner crystals, the focus can be tightened, owing to a larger acceptance angle. As a result, the total upconversion signal will stay relatively constant for thinner crystals, in addition to providing better time resolution. For an L = 0.4 mm BBO crystal, the acceptance angle is estimated to be ~ 8°. Spectral bandwidth The upconversion spectral bandwidth is estimated from the spectral position difference when the quantum efficiency drops to 50%. Additionally, there is a complicated interplay between the NLO crystal thickness and fluorescence spectral bandwidth. For an L = 0.4 mm BBO crystal at 820 nm/500 nm mixing, the spectral bandwidth is < 1 nm. Upconversion is an intrinsically high-resolution spectroscopy technique. Quantum efficiency of upconversion The quantum efficiency of upconversion can be estimated for the ‘small signal’ condition, i.e. no depletion of the fluorescence and probe pulse powers. For example, for a 100-fs probe pulse, with 0.5-W average power at 820 nm, a spot diameter of 0.1 mm, and a 76-MHz repetition rate, it is ~ 0.001% for an 3006

L = 0.4 mm BBO crystal. This is, however, more than enough for monitoring with average UV photomultipliers. The obtained signal/noise ratio is routinely ≥ 10. Group velocity mismatch In nonlinear processes such as sum frequency generation, the mismatch between the group velocity of ultrafast probe pulse and fluorescence may lead to temporal broadening of the generated upconversion signal. This restriction is more severe than the one imposed by phase mismatching discussed above. This broadening effect can be precompensated by a pulse compressor and/or minimized by a proper choice of optical elements by the use of refractive optics and thinner NLO crystals. Conventional (only time-resolved) fluorescence upconversion apparatus As mention above, the ultrafast fluorescence upconversion technique has many peculiarities, which is probably why instruments with higher time resolution (~ 100 fs or higher) are not available commercially. We have built a reliable system in our laboratory (Fig. 1B) with high temporal resolution and sensitivity; this is briefly described below. A Ti:sapphire laser system (Verdi-V8 pumped Mira 900; Coherent, Santa Clara, CA, USA) is used as a light source (120 fs, 76 MHz, 800 mW at 820 nm). The pulses are further compressed up to ~ 70-fs full width at half maximum (FWHM) with a prism pair compressor. The second harmonic (~ 20 mW at 410 nm) is generated in a 0.1-mm thin BBO crystal, and focused onto the sample circulating in a flow cell (50 mL
min 1) with 0.50-mm or 1-mm light path lengths to generate the fluorescence. It is then collected with a pair of parabolic mirrors and focused, together with the residual fundamental laser pulse, on a 0.4-mm (or 0.2-mm) BBO type I crystal to generate the upconverted signal at the sum frequency. After passing through a grating monochromator (1200 g
mm 1; Acton Research Corp., Trenton, NJ, USA), the fluorescence is detected with a photomultiplier (R1527P) coupled to a photon counter (C5410) system (both from Hamamatsu Photonics K. K., Hamamatsu City, Japan). The fluorescence decay curves can be obtained by varying the optical path length of the computercontrolled delay stage for the fundamental laser pulse. Ten scans (~ 6.67-fs or 20-fs steps) in alternate directions are usually accumulated to obtain a single transient with an acceptable signal/noise ratio. As an FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

instrument response function, the cross-correlation signal between the fundamental and its second harmonic pulse is used (the FWHM is typically kept between 90 fs and 200 fs, depending on the application). All measurements are usually carried out at ~ 20 °C. In the experiments described below, the optical density per 1-cm path length was ~ 3 at 410 nm. Microscope-integrated (space-resolved and timeresolved) fluorescence upconversion apparatus For specific applications to solid-state materials, especially protein single crystals, we have developed a femtosecond time-resolved and micrometer space-resolved fluorescence measurement system. Generally, measuring ultrafast dynamics of light-sensitive solid-state materials is a tremendous challenge. The simple reason is that they cannot be ‘flown’ like liquid samples to minimize accumulative effects and thermal damage. This is especially true for solid protein systems. To overcome this problem, ultraweak excitation intensities below the accumulative and/or damage thresholds of a given sample have to be used; in most cases, these did not yield a measurable signal. After several ‘trial and error’ attempts, we overcame these problems and successfully integrated a microscope into the fluorescence upconversion system, as shown schematically in Fig. 2.

The same Ti:sapphire laser and instruments as described in the subsection ‘Conventional (only timeresolved) fluorescence upconversion apparatus’ were used in this system. The second harmonic (~ 1 mW) was generated in a 0.1-mm thin BBO crystal, attenuated with a dark filter, and introduced into an inverted confocal microscope (IX71; Olympus, Shinjuku, Tokyo, Japan) from the back port. The fluorescence from the sample, positioned on the stage center plate, was collected in the backscattering geometry and guided to the outside of the microscope from the right side port. Then, the fluorescence was focused onto a nonlinear BBO crystal (0.4 mm), where it was mixed with the time-delayed fundamental laser pulse and upconverted at the sum frequency. After being passed through a grating monochromator, it was detected by a photomultiplier coupled with a photon counter system. The fluorescence decay curves were obtained in the same manner as described above. We used a Schwartzshield refractive objective lens (9 40) and a special design of the microscope’s inner light-guiding optics, achieving a time resolution of ~ 200 fs. The spatial resolution was estimated to be ~ 2 lm. In this configuration, ultraweak excitation conditions could be realized (~ 60 fJ per pulse at 410 nm). In all experiments on PYP single crystals presented in the section ‘Ultrafast photoisomerization reaction dynamics of PYP in crys-

Excitation laser spot

~ 50 µm x40 reflective objective lens

Sample (protein single crystal, etc.)

Probe pulse at 820 nm (~ 70 fs fwhm) Fig. 2. Schematic diagram of the femtosecond time-resolved and micrometer space-resolved fluorescence upconversion system. Insert (top): WT PYP single crystal appearance with (right) and without (left) laser excitation.

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

Excitation pulse at 410 nm (~100 fs fwhm)

Signal Z

Fluorescence BBO NLO Crystal

Microscope

Y X

3007

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

tal and liquid phases’, the excitation beam was perpendicular to the longitudinal C-axis of the crystal.

A

Trp32

FMN

Application to flavoproteins Flavoproteins contain FAD or FMN as a cofactor, and play an important role in oxidation, oxygenation and ET reactions [30]. Applications of the ultrafast fluorescence upconversion technique to various ‘nonfluorescent’ flavoproteins – riboflavin-binding protein, medium-chain acyl-CoA dehydrogenase, flavodoxin, D-amino acid oxidase, etc. – have been performed in our laboratory [31–33] and by other groups [34,35]. Briefly, the flavin enzymes are especially interesting model systems for elucidating ultra-fast ET reactions taking place in the chromophore’s binding pockets. When the photoexcited flavin chromophore is in the oxidized form, it acts as a strong electron acceptor; therefore, if such aromatic residues as tryptophan and/ or tyrosine are in close proximity to flavin, strong fluorescence quenching via ET on an ultrafast time scale (10 14–10 10 s) occurs. As an application example, we present here our recent studies on wild-type (WT) FBP and its six point mutants W32Y, W32A, E13R, E13K, E13Q, and E13T. Generally, proteins, including FBP, contain many ionic groups that may influence the intraprotein ET rates. The specific purpose of this study was to understand and elucidate the role and influence of ES charges and corresponding energies on the ET rate. The calculations of ES energies were based on Coulomb interactions between charged residues, FMN cofactor and Trp32 in the protein [for example, WT FBP has, overall, 13 negatively charged residues (eight glutamates and five aspartates) and 13 positively charged residues (four lysines and nine arginines)]. The corresponding ion-pair distances between FMN, Trp32 and those residues present in each system were derived from the known protein structures. Additional details of this study can be found elsewhere [36]. Briefly, FBP from Disulfovibrio vulgaris (Miyazaki F) is one of the smallest flavoproteins (122 residues, 13 kDa), and contains FMN as a cofactor [37]. It is thought to play an important role in ET processes in the bacterium; however, the whole picture of the electron flow and coupling to the redox proteins is not yet clear. According to the three-dimensional structures determined by X-ray crystallography [38], there are three potential quenchers located close to the isoalloxazine (Iso) rings of the FMN (Fig. 3A): Trp32, Tyr35, and Trp106. Corresponding center-to-center distances from Iso to Trp32, Tyr35 and Trp106 were 0.71 nm, 0.77 nm and 0.85 nm, respectively. First, to determine which residue 3008

Glu13

Tyr35 Trp106 B

Fig. 3. (A) Chromophore-binding pocket of the WT FBP enzyme. (B) Fluorescence decay curves of WT FBP and W32Y and W32A point mutants at steady-state fluorescence maxima (~ 530 nm) with average lifetimes of ~ 170 fs, ~ 11 ps, and ~ 40 ps, respectively.

is the dominant quencher in this protein, the fluorescence dynamics of two point mutants, W32Y and W32A, were studied and compared with the dynamics of WT FBP (Fig. 3B). It is important to note that the fluorescence upconversion technique can experimentally determine whether there is a substantial amount of water in the chromophore environment or whether the protein chromophore-binding pocket structure is flexible. Such information can be obtained by spectral measurements and wavelength-dependent emission dynamics analysis. For example, emission decays at shorter spectral regions and increases at longer spectral regions FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

H. Chosrowjan et al.

(dynamic Stokes shift) are characteristic of the solvation dynamics phenomenon, and such dynamic behavior, if observed for a chromophore buried in a protein binding pocket, would indicate that a substantial amount of water is present around the chromophore. On the other hand, emission anisotropy dynamics, if any, would indicate more flexibility of the chromophore-binding pocket structure. In the present study, transient fluorescence decays were measured at several wavelengths between 480 nm and 600 nm, and no dynamic Stokes shift or other marked wavelength dependencies were observed in all seven proteins examined (Fig. 4A). Furthermore, the time-resolved anisotropy was ~ 0.4 and constant for all systems (Fig. 4B), confirming that neither rotational motions nor changes in the electronic state of Iso take place. Hence, we can safely conclude that the protein chromophore-binding pocket structures are sufficiently rigid in all systems, and further discussions concern the single-wavelength measurements at the emission maxima (~ 530 nm) of these proteins. For comparison, the ET rate in the A

B

Fig. 4. (A) Example of fluorescence wavelength dependence: E13T mutant. No wavelength dependence was observed in six FBP proteins examined. (B) Example of fluorescence anisotropy: E13Q mutant. The time-resolved anisotropy was ~ 0.4 and constant for all six FBP systems examined.

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

Fluorescence upconversion in liquids and solids

W32A mutant, where tryptophan is replaced with neutral alanine possessing no quenching capability, is > 200 times lower than the ET rate in WT FBP, indicating that the ET contributions from two other quenchers, Tyr35 and Trp106, are negligible. Even when Trp32 is replaced by the potential quencher tyrosine in the W32Y mutant, the ET rate is ~ 65 times lower than that of WT FBP. Hence, Fig. 3B clearly shows that the main fluorescence quencher in FBP is Trp32, owing to its close proximity to Iso and higher (~ 8 eV) ionization potential. Next, to elucidate the effect of the ES charge distribution in the protein on the ET rate, we investigated four single-substitution isomorphs: E13K, E13R, E13T and E13Q. In all of these systems, the negatively charged Glu13 was replaced by a positively charged lysine (E13K) or arginine (E13R), and a neutral threonine (E13T) or glutamine (E13G). Interestingly, according to the X-ray crystallographic structures of these mutants (the Protein Data Bank codes for WT FBP and the E13K, E13R, E13T and E13Q mutants are 1FLM, 3AWH, 3AMF, 3A6Q, and 3A6R, respectively), the average distances between Iso and all three quenchers, Trp32, Tyr35, and Trp106, were almost unchanged. Moreover, the hydrogen bonding network between Iso and Gly49, Pro47 and Thr31 in WT FBP was unaltered in all other systems as well. Hence, one would expect the same ET rate as in WT FBP for all four mutants. However, as shown in Fig. 5A, the experimentally obtained ET rates varied for these systems. The observed fine tuning of fluorescence decay curves describing the ET rate changes could be ascribed to the ES charge redistribution in the protein chromophore-binding pocket. We used Kakitani–Mataga ET model theory [36,39] to calculate ET rates. Briefly, the Kakitani–Mataga ET model is based on Marcus’s original theory [40], but extends it and can be applied to both adiabatic and nonadiabatic ET reactions. Indeed, calculated ET rates confirmed that the main reason for the observed rate changes is the change in net ES energies of Trp32 in each system. For reference, the net ES energy of Trp32 in WT FBP was ~ 0.026 eV; however, in the E13K and E13R mutants, it increased to ~ 0.29 eV, and in the E13T and E13Q mutants it was ~ 0.4 eV. Calculations of the total free energy gap ( ΔG°T) showed that its value was smallest for the E13T and E13Q mutants (~ 0.065 eV), followed by the E13K and E13R mutants, and then by WT FBP, which has the largest value, at ~ 0.45 eV. Interestingly, the ET rates from Trp32 to excited Iso* were fastest in the E13K and E13R mutants, followed by WT FBP, the E13T mutant, and the E13Q mutant, respectively. This

3009

Fluorescence upconversion in liquids and solids

A

E13Q E13T WT FBP E13K E13R

B

Fig. 5. (A) Fluorescence decay curves of WT FBP and the E13R, E13K, E13T and E13Q mutants at steady-state fluorescence maxima (~ 530 nm). The dotted line represents the instrumental response function. The curves for fitting to experimentally obtained data have been omitted for clarity. (B) Trp32 to excited Iso* ET rate dependence on the total free energy gap, ΔG°T. Donor– acceptor distances were similar in all five FBP isomorphs. The main reason for the observed ET rate changes is the different net ES energy of Trp32 in each system.

seemingly inconsistent result is clarified in Fig. 5B, where the ET rate (kET) dependence on the total free energy gap ( ΔG°T) is plotted. The observed bellshaped ET rate dependence explains why, for WT FBP, the ET rate decreases while the total energy gap increases. This is one of the rarely observed cases when the ET process in a native protein occurs in the inverted region of the energy gap law. Ultrafast photoisomerization reaction dynamics of PYP in crystal and liquid phases PYP is a water-soluble photoreceptor first isolated from the phototropic purple bacterium Ectothiorhodospira halophila [41]. PYP is presumably responsible for 3010

H. Chosrowjan et al.

the negative phototaxis response of this microorganism. Studying PYP provides a renewed opportunity to better understand light transduction phenomena at the molecular level, because, in contrast to rhodopsins, it is small (125 residues, 14 kDa), very stable, and undergoes a fully reversible photocycle. For PYP, as for rhodopsins, photoisomerization of its chromophore (deprotonated trans-p-coumaric acid) buried in the protein has been identified as the primary photoreaction triggering the photocycle. The latter occurs not only in solution, but also in protein monocrystals (P65 and P63 symmetries), as shown by time-resolved absorption and X-ray crystallography studies [42,43]. The kinetics (nanosecond–millisecond time scale) and the nature of protein conformational changes in monocrystals, however, differ remarkably from those in solution [42,44]. Namely: (a) the overall photocycle rate is highly accelerated in PYP crystals versus solution [43]; (b) the blue-shifted intermediate PYPM is partially unfolded [45], whereas in crystals the conformational changes are small and limited to the immediate environment of the chromophore [46]; and (c) the solution kinetic model of the photocycle is not directly applicable to crystals [47]. It is also worth mentioning that high-resolution structural information on proteins comes almost exclusively from X-ray diffraction studies [48], whereas most functional and kinetic studies are carried out in solution or under solvent conditions very different from those used for crystallization. Nevertheless, in the interpretation of those observations and the development of kinetic models for PYP, it is usually assumed that, in solution and crystal phases, the initial structures of PYP are essentially the same and that the observed discrepancies in photocycle dynamics for solution versus crystal conditions arise solely from the structural restraints imposed by the crystal lattice rigidity (versus flexibility in solution). Although this interpretation is plausible for the slower (nanosecond–millisecond) dynamics of intermediate states, the similarity of initial pG states (also denoted as dark states) is not necessarily given, and has yet to be confirmed experimentally. There are clues in the literature on PYP indirectly questioning the similarity of initial dark states of PYP in solution and crystals. For instance, the structural information obtained from Xray crystallography (PYP crystals) and NMR investigations (PYP solution), although similar in general (a, b-folding), predict different hydrogen bonding networks and strengths around the chromophore. Moreover, the steady-state absorption and emission spectra, although similar in shape, slightly (2–3 nm) shift to the longer wavelengths for PYP crystals FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

versus those in solution. Calculations [49] of the partial atomic driving force of the positively charged Arg52 in the photoisomerization reaction predict a negative value (preventing the chromophore isomerization), i.e. a higher reaction rate, if Arg52 is replaced by a neutral amino acid. This result evidently contradicts the experimental observations with the R52Q mutant, where slower dynamics were observed [50]. In an attempt to clarify these problems, we comparatively studied the ultrafast dynamic properties of PYP

single crystals and PYP in solution by using the fluorescence upconversion technique. As discussed in the subsection ‘Microscope-integrated (space-resolved and time-resolved) fluorescence upconversion apparatus’, low excitation energy conditions allowed measurement of fluorescence decays of PYP at the same crystal spot without thermal, mechanical or chemical damage being caused. Up to 10 scans in alternate directions have been accumulated for one transient. We also measured decays at several

A

B

PYP crystal

C

PYP solution

D

R52Q crystal

R52Q solution

Fig. 6. Fluorescence decays of WT PYP single crystal (A), WT PYP in solution (B), R52Q mutant single crystal (C) and R52Q mutant in solution (D) excited at 410 nm and monitored at three different wavelengths. The decays were fitted with a sum of three exponential functions, and the best-fit parameters are listed in Tables 1 and 2, respectively.

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

3011

H. Chosrowjan et al.

Fluorescence upconversion in liquids and solids

different crystal spots; however, no inhomogeneous or other effects affecting the dynamics could be observed. Obviously, the crystals were of good quality and suitable for the measurements. In Fig. 6A, the ultrafast fluorescence dynamics of a WT PYP single crystal (P65 symmetry) measured at three different wavelengths are shown. The decays represent parallel components of the fluorescence when the polarization of the excitation pulse is perpendicular to the crystal’s longitudinal (C) axis. We also measured the perpendicular component of the fluorescence under the same excitation conditions. The only difference observed was the signal ratio between the perpendicular and parallel components of the fluorescence (~ 6.5 times the expected value of 7.27), with no detectable changes in the anisotropy’s dynamic behavior. This clearly showed that the polarization ratio depends solely on the orientation of the transition dipole moment of the chromophore with respect to the crystal axes, the absorption/emission transition moments are parallel, and no degenerated electronic states are involved. For comparison, decays of WT PYP in solution measured with the conventional fluorescence upconversion system with the same time resolution as in the case of PYP single crystals (~ 200 fs) are shown in Fig. 6B. The best-fit parameters are listed in Table 1. From Fig. 6A,B, it can be concluded that the fluorescence dynamics of WT PYP single crystals are quite close to those measured in solution. A spectral sharpening effect, similar to PYP solution dynamics, was also observed in PYP crystals. These results show that the ultrafast fluorescence dynamics of PYP in crystal and solution conditions are essentially the same. Hence, we conclude that the initial dark structures, and the intraprotein structural changes and their

temporal evolution on the femtosecond–picosecond time scale, are similar in liquid and crystal phases. In Fig. 6C,D, the fluorescence dynamics of the R52Q mutant in crystal and solution phases at different observation wavelengths are shown; the best-fit parameters are listed in Table 2. From Fig. 6C,D, it can be concluded that the fluorescence dynamics of the R52Q mutant in single-crystal form and solution are similar as well, although they are slower than the WT PYP dynamics. Hence, we conclude that there are no essential differences between R52Q mutant initial dark-structure and excited state dynamics in solution versus monocrystals. Another aim of these experiments was to address the role of the positively charged Arg52 in the stabilization of chromophore’s negative charge in the protein binding pocket. Briefly, in WT PYP monocrystals, arginine is linked by two hydrogen bonds to the carbonyl oxygen atoms of Thr50 and Tyr98 [48]. In solution, these two hydrogen bonds do not exist any more, but the main hydrophobic core containing the chromophore remains rather rigid and solvent-inaccessible [51]. The role of the positive charge near the phenolate group of the chromophore can be probed by replacing arginine with neutral glutamine. This approach is also justified by the X-ray crystallographic data on PYP and the R52Q mutant showing that the protein chormophorebinding pocket structure is not much altered [52]. From Fig. 6, it can be seen that the primary dynamics of the R52Q mutant in both crystal and liquid phases, although slightly slower, resemble those of WT PYP, and hence loss of positive charge does not affect qualitatively the primary dynamics of PYP. More discussions on this subject can be found in [53]. We have also performed nonresonance Raman scattering experiments

Table 1. Three exponential fitting parameters for WT PYP single crystals and PYP solution at different emission wavelengths. s2 (a2)

s1 (a1)

s3 (a3)

Wavelength (nm)

Crystal

Solution

Crystal

Solution

Crystal

Solution

480 500 550

320 fs (0.32) 470 fs (0.18) 290 fs (0.42)

325 fs (0.32) 450 fs (0.28) 295 fs (0.44)

2.1 ps (0.37) 2.1 ps (0.54) 2.1 ps (0.43)

1.5 ps (0.44) 1.8 ps (0.45) 1.6 ps (0.36)

17 ps (0.31) 19 ps (0.28) 20 ps (0.15)

13 ps (0.24) 14 ps (0.27) 16 ps (0.2)

Table 2. Three exponential fitting parameters for R52Q mutant single crystals and R52Q solution at different emission wavelengths. s1 (a1)

s2 (a2)

s3 (a3)

Wavelength (nm)

Crystal

Solution

Crystal

Solution

Crystal

Solution

480 500 550

825 fs (0.33) 850 fs (0.33) 800 fs (0.46)

825 fs (0.35) 1 ps (0.4) 815 fs (0.42)

10 ps (0.54) 8.6 ps (0.53) 8.8 ps (0.36)

11 ps (0.44) 12 ps (0.43) 13 ps (0.38)

140 ps (0.13) 150 ps (0.14) 130 ps (0.18)

130 ps (0.21) 140 ps (0.17) 160 ps (0.2)

3012

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

H. Chosrowjan et al.

on PYP solutions and mico-Raman measurements on PYP single crystals [54]. The observed differences were related mostly to the relative intensities of Raman bands attributable to nonrandom orientation of proteins in the crystal samples. In particular, the ratio between Raman intensities for perpendicular and parallel excitations relative to the crystal’s longitudinal axis was ~ 10, supporting the results obtained by fluorescence upconversion experiments.

Summary We have briefly presented the potential of ultrafast time-resolved fluorescence spectroscopy. The time-resolution characteristics of existing transient fluorescence spectroscopic techniques have been discussed, and the basic principles of the ultrafast fluorescence upconversion method, including requirements for nonlinear crystals, mixing spectral bandwidth, acceptance angle, etc., have been presented. Details of our home-made (conventional and space-resolved) femtosecond timeresolved fluorescence upconversion apparatus have been presented, and applications to two classes of protein, i.e. flavoproteins (FBP and its mutants) and photoresponsive proteins (PYP and its R52Q mutant in solution and as single crystals), have been demonstrated. Investigations on the influence of ionic charges on ultrafast fluorescence dynamics in FBP have been summarized. We have shown that replacement of Glu13 and the resulting modification of the electrostatic charge distribution in the protein chromophorebinding pocket fine tunes the ultrafast fluorescence quenching dynamics and ET rate in FBP. We have concluded that, together with donor–acceptor distances, electrostatic interactions between ionic photoproducts and other ionic groups in the proteins are important factors influencing the ET rates in proteins. For PYP, we have shown that the ultrafast photoisomerization dynamics in liquid and crystal phases are quite similar. We conclude that the photocycle dynamics and structural differences observed in solution versus crystals at longer time scales arise mostly from the structural restraints imposed by the crystal lattice rigidity, and do not originate from the intrinsic dark-structure differences in the chromophore-binding site, potential surface distortions, or the primary photodynamics of PYP’s excited state.

Acknowledgements We thank Dr Y. Imamoto from Kyoto University and Dr M. Kitamura from Osaka City University for providing protein samples and helpful discussions. We FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

Fluorescence upconversion in liquids and solids

thank also the Japan Society for the Promotion of Science (JSPS) for financial support (Grant-in-Aid for Scientific Research No. 26410029).

Author contributions FT: Calculations; HC: Experiments, manuscript preparation; ST: Experiments.

References 1 Lakowicz JR (2006) Principles of Fluorescence Spectroscopy, 3rd Ed, Springer Science+Business Media, LLC, NY. 2 Wolfbeis OS, Hof M, Huterer R & Fidler V (2005) Springer Series on Fluorescence: Fluorescence Spectroscopy in Biology; Advanced Methods and Their Applications to Membranes, Proteins, DNA and Cells. Springer-Verlag, Heidelberg, Berlin. 3 Saha R, Verma PK, Rakshit S, Saha S, Mayor S & Pal SK (2013) Light driven ultrafast electron transfer in oxidative redding of Green Fluorescent Proteins. Sci Rep 3, 1–7. 4 Duncan RR, Bergmann A, Cousin AM, Apps DK & Shipston MJ (2004) Multi-dimensional time-correlated single photon counting (TCSPC) fluorescence lifetime imaging microscopy (FLIM) to detect FRET in cells. J Microsc 215, 1–12. 5 Saha S, Mandal PK & Samanta A (2004) Solvation dynamics of Nile Red in a room temperature ionic liquid using streak camera. Phys Chem Chem Phys 6, 3106–3110. 6 Duguay MA & Hansen JW (1968) Optical sampling of subnanosecond light pulses. Appl Phys Lett 13, 178–180. 7 Gu JL, Shi JL, You GJ, Xiong LM, Qian SX, Hua ZL & Chen HR (2005) Incorporation of highly dispersed gold nanoparticles into the pore channels of mesoporous silica thin films and their ultrafast nonlinear optical response. Adv Mater 17, 557–560. 8 Mahr H & Hirsch MD (1975) Optical up-conversion light gate with picosecond resolution. Opt Commun 13, 96–99. 9 Gelot T, Touron-Touceda P, Cregut O, Leonard J & Haacke S (2012) Ultrafast site-specific fluorescence quenching of 2-aminopurine in a DNA hairpin studied by femtosecond down-conversion. J Phys Chem A 116, 2819–2825. 10 Sajadi M, Quick M & Ernsting NP (2013) Femtosecond broadband fluorescence spectroscopy by down- and upconversion in b-barium borate crystals. Appl Phys Lett 103, 173514. 11 Zhang L, Kao Y, Qiu W, Wang L & Zhong D (2006) Femtosecond studies of tryptophan

3013

Fluorescence upconversion in liquids and solids

12

13

14

15

16

17

18

19

20

21

22

fluorescence dynamics in proteins: local solvation and electronic quenching. J Phys Chem Lett B 110, 18097–18103. Jimenez R, Fleming GR, Kumar PV & Maroncelli M (1994) Femtosecond solvation dynamics of water. Nature 369, 471–473. Tomi T, Shibata Y, Ikeda Y, Taniguchi S, Chosrowjan H, Mataga N, Shimada K & Itoh S (2007) Energy and electron transfer in the photosynthetic reaction center complex of Acidiphilium rubrum containing Zn-bacteriochlorophyll a studied by femtosecond up-conversion spectroscopy. Biochemica et Biophysica Acta (BBA) –. Bioenergetics 1767, 22–30. Jarzeba W, Walker C, Johnson AE & Barbara PF (1991) Nonexponential solvation dynamics of simple liquids and mixtures. Chem Phys 152, 57–68. Glasbeek M & Zhang H (2004) Femtosecond studies of solvation and intramolecular configurational dynamics of fluorophores in liquid solution. Chem Rev 104, 1929–1954. Gustavsson T, Baldacchino G, Mialocq JC & Pommeret S (1995) A femtosecond fluorescence upconversion study of the dynamic Stokes shift of the DCM dye molecule in polar and non-polar solvents. Chem Phys Lett 236, 587–594. Rubtsov IV & Yoshihara K (1999) Vibrational coherence in electron donor-acceptor complexes. J Phys Chem A 103, 10202–10212. Chosrowjan H, Taniguchi S, Mataga N, Unno M, Yamauchi S, Hamada N, Kumauchi M & Tokunaga F (2004) Low-frequency vibrations and their role in ultrafast photoisomerization reaction dynamics of photoactive yellow protein. J Phys Chem B 108, 2686–2698. Chosrowjan H, Mataga N, Shibata Y, Tachibanaki S, Kandori H, Shichida Y, Okada T & Kouyama T (1998) Rhodopsin emission in real time: a new aspect of the primary event in vision. J Am Chem Soc 120, 9706–9707. Du M & Fleming GR (1993) Femtosecond timeresolved fluorescence spectroscopy of bacteriorhodopsin: direct observation of excited state dynamics in the primary step of the proton pump cycle. Biophys Chem 48, 101–111. Mataga N, Chosrowjan H, Shibata Y, Yoshida N, Osuka A, Kikuzawa T & Okada T (2001) First unequivocal observation of the whole bell-shaped energy gap low in intramolecular charge separation from S2 excited state of directly linked porphyrin-imide dyads and its solvent-polarity dependencies. J Am Chem Soc 123, 12422–12423. Wan C, Xia T, Becker HC & Zewail AH (2005) Ultrafast unequilibrated charge transfer: a new channel in the quenching of fluorescent biological probes. Chem Phys Lett 412, 158–163.

3014

H. Chosrowjan et al.

23 Onidas D, Markovitsi D, Marguet S, Sharonov A & Gustavsson T (2002) Fluorescence properties of DNA nucleosides and nucleotides: a refined steady-state and femtosecond investigation. J Phys Chem B 106, 11367–11374. 24 Abbyad P, Shi X, Childs W, McAnaney TB, Cohen BE & Boxer S (2007) Measurement of solvation responses at multiple sites in a globular protein. J Phys Chem B 111, 8269–8276. 25 Pal SK & Zewail AH (2004) Dynamics of water in biological recognition. Chem Rev 104, 2099–2123. 26 Fleming GR (1986) Chemical Application of Ultrafast Spectroscopy. Oxford University Press, New York. 27 Xu J & Knutson JR (2008) Ultrafast fluorescence spectroscopy via upconversion: applications to biophysics. Methods Enzymol 450, 159–183. 28 Zhang XX, Wuerth C, Zhao L, Resch-Genger U, Ernsting NP & Sajadi M (2011) Femtosecond broadband fluorescence upconversion spectroscopy: improved setup and photometric correction. Rev Sci Instrum 82, 063108. 29 Shen YR (1984) The Principles of Nonlinear Optics. Wiley-Interscience, New York. 30 Lehninger AL, Nelson DL & Cox MM (2000) Lehninger Principles of Biochemistry. Worth Publisher Inc, New York. 31 Mataga N, Chosrowjan H, Shibata Y & Tanaka F (1998) Ultrafast fluorescence quenching dynamics of flavin chromophores in protein nanospace. J Phys Chem B 102, 7081–7084. 32 Mataga N, Chosrowjan H, Shibata Y, Tanaka F, Nishina Y & Shiga K (2000) Dynamics and mechanisms of ultrafast fluorescence quenching reactions of flavin chromophores in protein nanospace. J Phys Chem B 104, 10667–10677. 33 Mataga N, Chosrowjan H, Taniguchi S, Tanaka F, Kido N & Kitamura M (2002) Femtosecond fluorescence dynamics of flavoproteins: comparative studies on Flavodoxin, its site-directed mutants, and riboflavin binding protein regarding ultrafast electron transfer in protein nanospaces. J Phys Chem B 106, 8917–8920. 34 Zhong D & Zewail AH (2001) Femtosecond dynamics of flavoproteins: charge separation and recombination in riboflavin (vitamin B2)-binding protein and in glucose oxidase enzyme. Proc Natl Acad Sci USA 98, 11867–11872. 35 Zhong D & Zewail AH (2001) Femtosecond dynamics of a drug-protein complex: daunomycin with Apo riboflavin-binding protein. Proc Natl Acad Sci USA 98, 11873–11878. 36 Taniguchi S, Chosrowjan H, Tanaka F, Nakanishi T, Sato S, Haruyama Y & Kitamura M (2013) A key factor for ultrafast rates of photoinduced electron transfer among five flavin mononucleotide binding proteins: effect of negative, positive, and neutral

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

H. Chosrowjan et al.

37

38

39

40

41

42

43

44

45

charges at residue 13 on the rate. Bull Chem Soc Jpn 86, 339–350. Kitamura M, Kojima S, Ogasawara K, Nakaya T, Sagara T, Niki K, Miura K, Akutsu H & Kumagai I (1994) Novel FMN-binding protein from Disulfovibrio vulgaris (Miyazaki F). Cloning and expression of its gene in Escherichia coli. J Biol Chem 269, 5566–5573. Suto K, Kawagoe K, Shibata N, Morimoto Y, Higuchi Y, Kitamura M & Yasuoka N (2000) How do the Xray structure and the NMR structure of FMN-binding protein differ? Acta Crystallogr D Biol Crystallogr 56, 368–371. Kakitani T & Mataga N (1985) New energy gap laws for the charge separation process in the fluorescence quenching reaction and the charge recombination process of ion pairs produced in polar solvents. J Phys Chem 89, 8–10. Marcus RA (1956) On the theory of oxidationreduction reactions involving electron transfer. I. J Chem Phys 24, 966–978. Meyer TE (1985) Isolation and characterization of soluble cytochromes, ferredoxins and other chromophoric proteins from the halophilic phototrophic bacterium Ectothiorhodospira halophila. Biochim Biophys Acta 806, 175–183. Xie A, Keleman L, Hendriks J, White BJ, Hellingwerf KJ & Hoff WD (2001) Formation of a new buried charge drives a large-amplitude protein quake in photoreceptor activation. Biochemistry 40, 1510–1517. Kort R, Ravelli B, Schotte F, Bourgeois D, Crielaard W, Hellingwerf KJ & Wulff M (2003) Characterization of photocycle intermediates in crystalline photoactive yellow protein. Photochem Photobiol 78, 131–137. Mano E, Kamikubo H, Imamoto Y & Kataoka M (2003) Comparison of the photochemical reaction of photoactive yellow protein in crystal with reaction in solution. Spectroscopy 17, 345–353. Hoff WD, Xie A, van Stokkum IHM, Tang XJ, Gural J, Kroon AR & Hellingwerf KJ (1999) Global conformational changes apon receptor stimulation in photoactive yellow protein. Biochemistry 38, 1009–1017.

FEBS Journal 282 (2015) 3003–3015 ª 2014 FEBS

Fluorescence upconversion in liquids and solids

46 Genick UK, Soltis SM, Kuhn P, Canestrelli IL &  resolution of an Getzoff ED (1998) Structure at 0.85 A early protein photocycle intermediate. Nature 392, 206–209. 47 Ng K, Getzoff ED & Moffat K (1995) Optical studies of a bacterial photoreceptor protein, photoactive yellow protein, in single crystals. Biochemistry 34, 879–890. 48 Borgstahl GEO, Williams DR & Getzoff ED (1995)  structure of photoactive yellow protein, a The 1.4 A cytosolic photoreceptor: unusual fold, active site and chromophore. Biochemistry 34, 6278–6287. 49 Yamada A, Ishikura T & Yamato T (2004) Direct measure of functional importance visualized atom-byatom for photoactive yellow protein: application to photoisomerization reaction. Proteins: Struct, Funct, Bioinf 55, 1070–1077. 50 Chosrowjan H, Mataga N, Shibata Y, Imamoto Y & Tokunaga F (1998) Environmental effects on the femtosecond-picosecond fluorescence dynamics of photoactive yellow protein: chromophores in aqueous solutions and in protein nanospaces modified by sitedirected mutagenesis. J Phys Chem B 102, 7695–7698. 51 Dux PE, Rubinstenn G, Vuister GW, Boelens R, Mulder FAA, Hard K, Hoff WD, Kroon AR, Crielaard W, Hellingwerf KJ et al. (1998) Solution structure and backbone dynamics of the photoactive yellow protein. Biochemistry 37, 12689–12699. 52 Shimizu N, Kamikubo H, Yamazaki Y, Imamoto Y & Kataoka M (2006) The crystal structure of the R52Q mutant demonstrates a role for R52 in chromophore pKa regulation in photoactive yellow protein. Biochemistry 45, 3542–3547. 53 Changenet-Barret P, Plaza P, Martin MM, Chosrowjan H, Taniguchi S, Mataga N, Imamoto Y & Kataoka M (2007) Role of arginine 52 on the early steps of the PYP photocycle. Chem Phys Lett 434, 320–325. 54 Taniguchi S & Chosrowjan H (2011) Fluorescence dynamics of photoactive yellow protein micro-crystals using femtosecond time-resolved fluorescence microscope. Rev Laser Eng 39, 931–937.

3015