Conformational and Intermolecular Interaction Dynamics of Photolyase/Cryptochrome Proteins Monitored by the Time‐Resolved Diffusion Technique^†

Abstract

Cryptochrome (CRY), a blue light sensor protein, possesses a similar domain structure to photolyase (PHR) that, upon absorption of light, repairs DNA damage. In this review, we compare the reaction dynamics of these systems by monitoring the reaction kinetics of conformational change and intermolecular interaction change based on time‐dependent diffusion coefficient measurements obtained by using the pulsed laser‐induced transient grating technique. Using this method, time‐dependent biomolecular interactions, such as transient dissociation reactions in solution, have been successfully detected in real time. Conformational change in (6‐4) PHR has not been detected after the photoexcitation by monitoring the diffusion coefficient. However, the repaired DNA dissociates from PHR with a time constant of 50 μs, which must relate to a minor conformational change. However, CRY exhibits a considerable diffusion change with a time constant of 400 ms, which indicates that the protein–solvent interaction is changed by the conformational change. The C‐terminal domain of CRY is shown to be responsible for this change.

Introduction

Light is essential for life. To use light, there are a variety of light‐absorbing proteins, not only light sensors but also repair systems for damaged DNA. Interestingly, one photosensor protein, cryptochrome (CRY), possesses a similar structure to photolyase (PHR), which has a function of DNA repair. Proteins of the photolyase/cryptochrome family are found in many biological species and mediate a wide variety of physiological functions. Hence, this family of proteins has attracted a much attention. Proteins of the photolyase/cryptochrome family have been extensively studied from the structural, biochemical and spectroscopic approaches. There has been significant progress in studies on the static protein structures, functions and photochemistry of flavin adenine dinucleotide (FAD) cofactor. The structural and functional studies provide further understanding of the functional difference among the proteins in the family. In this review, we will describe the reaction dynamics of the PHR and CRY systems based mainly on a new technique for monitoring the time‐dependent diffusion change.

Photolyase is a DNA repair enzyme, which catalyzes light‐dependent repair of DNA damage, cross‐linkages between adjacent pyrimidine bases caused by ultraviolet (UV) irradiation 1-4. UV radiation mainly produces two types of pyrimidine dimers: cyclobutane pyrimidine dimers (CPDs) and pyrimidine (6‐4) pyrimidone photoproducts [(6‐4) PPs]. For repairing damage, two types of PHRs have been identified: one specific for CPDs (CPD PHR) and the other specific for (6‐4) PPs [(6‐4) PHR]. PHR absorbs light energy by the FAD cofactor. Using light energy, PHR donates an electron to UV‐damaged DNA, catalyzing the splitting of the mutagenic pyrimidine dimers.

The crystallographic structures of protein–substrate complexes, both for CPD PHR 5 and for (6‐4) PHR 6, 7, have been reported. These structures showed that the substrate binding site is close to the FAD cofactor both in CPD PHR and in (6‐4) PHR. In the DNA repair reactions, PHRs first form a protein‐damaged DNA complex in the dark 8, 9. Although there are some oxidative states of the chromophore, FAD, it is known that the fully reduced form (FADH^‐) is catalytically active for both CPD PHR and (6‐4) PHR. After photoexcitation of FADH^‐ in CPD PHR, an electron transfer occurs from the excited chromophore to the damaged site of DNA. This electron transfer causes a cleavage of the chemical bonds of the DNA dimer (DNA repair) and, at the same time, a transient radical FADH· is formed. After this DNA repair, FADH^‐ is restored by a back electron transfer from the DNA 1, 2. These forward and back electron transfers between FADH^‐ and CPDs have been investigated by an ultrafast absorption technique 10. It was reported that the formation and decaying processes of FADH· are completed within 1 ns after the photoexcitation, and hence, the restoration of CPDs occurs very fast at least within this time scale.

The repair mechanism for (6‐4) PHR is thought to be more complex because (6‐4) PP has a more complicated structure than that of CPD. So far, two models have been proposed. In one model, the transient water molecule formation model, the C‐O bond at the 5′ base is directly broken by the initial proton transfer 6. In the other model, the transient four‐membered ring formation model, a simple oxetane ring formation promotes to transfer an oxygen atom from the 5′ to 3′ base first and the C6‐C4 bond split 11, 12. A recent ultrafast spectroscopic experiment has revealed that the forward and back electron transfers between FADH^‐ and (6‐4) PPs are completed in less than 10 ns 11. For both CPD PHR and (6‐4) PHR, both the rate for the splitting of pyrimidine dimers and the mechanisms of the electron transfer between protein and substrate have been characterized. In contrast, little is known about the dynamics of PHR over a longer time range, including the dynamics of the PHR–DNA interaction of the PHR–DNA complex during and/or after the light‐dependent repair. What is the fate of the repaired DNA? Such slow dynamics should be important to understanding the complete reaction of the DNA repair as well as for discussing the turnover rate of the reaction.

Cryptochromes are blue light receptors found in many living organisms 13, 14. CRYs have a FAD‐containing photoreceptor domain whose structure is similar to that of PHR (PHR‐like domain) 15. Despite the structural similarity of the PHR‐like domain, CRYs show little or no DNA repair activity. Instead, CRYs exhibit diverse biological functions such as growth and development in plants 16, circadian clocks in animals 17 and magnetic sensors in birds 14. DASH (Drosophila, Arabidopsis, Synechocystis, Human)‐type cryptochromes (CRY‐DASHs) are a relatively new member of the photolyase/cryptochrome protein family 18. CRY‐DASH was initially considered a CRY because it exhibited little or no DNA repair activity. However, a subsequent study showed that CRY‐DASH proteins from bacterial, plant and animal sources could repair single‐stranded DNA (ssDNA) 19. Therefore, CRY‐DASH was reclassified as a PHR for ssDNA.

The structures of plant CRYs, animal CRYs and CRY‐DASHs have been investigated 15, 18, 20-22. In addition to the N‐terminal photoreceptor PHR‐like domain, most CRYs possess an additional characteristic C‐terminal extension (cryptochrome C‐terminal [CCT] domain) 16, 23. However, the crystal structure of a full‐length protein containing a CCT domain was obtained only for Drosophila CRY 22. On the other hand, although the crystal structure of the PHR‐like domain was obtained for a plant CRY, represented by Arabidopsis thaliana cryptochrome1 (AtCRY1) 15, the structure of the CCT domain has not been reported. Indeed, it was shown that the isolated CCT domain is unstructured 24. Interestingly, in spite of the unstructured character, this CCT domain of the plant CRYs has an important functional role, which was demonstrated by genetic and physiological studies of truncated forms of CRYs; that is, the CCT domain is a functional part for information transduction to protein partners and can function in isolation when fused to a reporter gene 25, 26. CRY‐DASH proteins have no CCT domain 18, so they are similar to PHRs not only in function but also in structure.

The initial light‐dependent reaction of AtCRY1 was studied by transient absorption (TA) 27 and Fourier transform infrared (FTIR) spectroscopy 28. They showed that the photoexcitation of oxidized FAD in the protein induces an intraprotein electron transfer to the FAD from an aromatic amino acid residue and formation of the FAD radical (FADH˙). This radical state has been considered to be the signaling state in plant CRYs 29, 30. As the chromophore FAD that absorbs light is contained in the PHR‐like domain and the electron transfer is presumed to take place within the PHR‐like domain of CRYs 31-33, the light signal should be transferred from the PHR‐like domain to the CCT domain. It is highly possible that conformational changes in the protein moiety involve in the signal transduction process. Indeed, conformation changes in the protein moiety have been detected by FTIR 28 and partial proteolysis 24 experiments after the photoreduction of FAD. However, the conformational changes determined by FTIR were relatively small. Because proteolysis analysis does not have time resolution, it provides no kinetic information. More recently, a time‐resolved FTIR experiment was reported and found a conformational change in the β‐sheet in the PHR‐like domain of plant CRY (from Chlamydomonas) in the late microsecond time range 34. However, no kinetic study has been performed on the potential rearrangement of the CCT domain, which has been suggested to play a key role in the CRY function.

To date, for studying kinetics of chemical reactions, UV/visible light detection methods, using the characteristic absorption of cofactors, have been one of the main techniques. These methods are sensitive to reactions in the vicinity of the cofactors, and have been applied to processes such as photoinduced electron transfer between FADH^‐ and DNA during the DNA repair by PHRs and photoreduction of the FAD cofactor during the CRY signaling. However, it is difficult to obtain information on global protein dynamics or intermolecular interactions apart from the cofactor using these methods, because the whole protein size is very large compared with that of the cofactor. Such dynamics, however, are essential for understanding the kinetics and mechanisms of the photolyase/cryptochrome family protein functions. For example, at the initial stage of the DNA repair reaction by PHR, PHR has to bind with the target DNA and electrons have to be transferred to the pyrimidine dimers buried inside the DNA duplex. After the repair, PHR has to release the restored DNA product to start a new repair reaction cycle with another damaged DNA. Thus, to understand the DNA repair reaction, it is important to capture intermolecular interactions between PHR and DNA. Time‐resolved information on the conformational changes in the protein moiety is also essential for a better understanding of the signaling mechanisms of CRY. To observe time‐resolved conformational dynamics throughout the protein, we have developed a method based on time‐resolved detection of the diffusion change by the pulsed laser‐induced transient grating (TG) technique 35-42.

In this paper, we review the conformational and intermolecular interaction dynamics of (6‐4) PHR 43 and AtCRY1 44 based mostly on our time‐resolved diffusion technique. This review is organized as follows: The principle of time‐resolved diffusion technique is introduced in the Materials and Methods section; studies on the light‐induced conformational change and protein‐DNA interaction dynamics during the DNA repair reaction by PHR are described in the section titled Conformation Change and DNA Repair Reaction of (6–4) Photolyase; studies on the photo‐induced conformational dynamics of AtCRY1 are presented in the Detection of Conformational Change in AtCRY1 section; and, finally, the similarities and differences in the photolyase/cryptochrome families are discussed in the section titled PHR and CRY Proteins.

Materials and Methods

Principle of the TG method

In this section, the principle of the time‐resolved TG technique is described, in particular, for the measurement of the diffusion coefficient (D). In the TG method, an interference pattern of light intensity with a wavenumber q is produced by crossing two laser beams within the coherence time inside a sample solution. By this light of which intensity is spatially modulated, a chemical reaction is started. As the number of reaction depends on the light intensity, this chemical reaction produces a spatial modulation of concentrations of chemical species, that is reactant, (possible) intermediates and products. Because different chemical species possess different polarizabilities, the refractive index of the solution is changed depending on the concentrations of the chemical species. As a result, a spatial refractive index modulation is created. This modulation acts as a grating, so that a probe light passing through the interference region is diffracted as the TG signal. The efficiency of the diffraction is determined by many factors. Using approximations such that the absorption of the probe light is negligible, the sample thickness is longer than the fringe length and the diffraction efficiency is small, one may find that the TG intensity (I_TG) is proportional to the square of the refractive index difference (δn) between the peak null of the grating pattern:

(1)

where α is a constant representing the sensitivity of the experimental system.

There are several origins of the refractive index change. In the time region, we are interested in (μs–s), and at a rather weak excitation light intensity, there are three main contributions. First, the temperature change, which is induced by the thermal energy released from the excited states and from the enthalpy change of the reaction, results in a refractive index change [the thermal grating (δn_th)]. Second, any changes in the absorption spectrum, even far from the probe wavelength, change the refractive index at the probe wavelength (population grating). Third, changes in molecular volume also contribute to the signal (volume grating). The sum of the population grating and volume grating terms is called the species grating (δn_spe). The contribution of created species to the species grating term is positive and that of the depleted reactant is negative. Hence, the refractive index change of the species grating term is given as the difference between the refractive index change arising from the reactant (δn_R) and the product (δn_P). Here, it should be noted that the “product” in this case does not necessarily mean the final product, but can be any molecule produced from the reactant at the time of observation, including the intermediates. Hence, the observed TG signal (I_TG(t)) is expressed as:

(2)

Temporal profile of the thermal grating signal is determined by the heat releasing process and thermal diffusion process. Upon very fast heat releasing compared with the thermal diffusion, the thermal grating signal decays single exponentially with a rate constant of D_thq² (D_th: the thermal diffusivity). Similarly, the temporal profile of the species grating signal is determined by the kinetics of the chemical reactions and molecular diffusion processes. If there is no chemical reaction in the observation time window and the molecular diffusion coefficient (D) is time independent, the profile can be calculated by a simple molecular diffusion equation, which shows that the q‐Fourier component of the concentration modulation decays with a rate constant of Dq² for both the reactant and the product. Hence, using δn_R (>0) and δn_P (>0), which are, respectively, the initial refractive index changes arising from changes in the reactant and the product concentrations, the time development of the TG signal may be expressed by:

(3)

where D_R and D_P are the diffusion coefficients of the reactant and the product, respectively. The TG signal expressed by this equation gives rise to a characteristic rise–decay profile (e.g. Fig. 1), so that one can notice whether or not D is changed by the reaction. When the reaction does not change D, the first and the second terms of Eq. 3 are canceled and a weak single exponential decay is observed. Apparently, the characteristic rise–decay profile (diffusion peak) intensity increases with increasing difference between D_R and D_P.

Figure 1

(A) Simulated TG signals using Eq. 3 with δn_P:δn_R = 1:0.95, D_R = 5.0 × 10⁻¹¹ m²s⁻¹, q² = 3.0 × 10¹¹ m⁻² and various D_P‐values: (black) 5.0 × 10⁻¹¹ m²s⁻¹, (blue) 4.5 × 10⁻¹¹ m²s⁻¹, (green) 4.0 × 10⁻¹¹ m²s⁻¹, (red) 3.5 × 10⁻¹¹ m²s⁻¹. The signal intensity increases with an increasing difference in D. (B) Simulated TG signals using Eq. 4 with δn_P:δn_I:δn_R = 0.95:0.95:1, k = 50 s⁻¹, D_R = D_I = 5.0 × 10⁻¹¹ m²s⁻¹, D_P = 3.5 × 10⁻¹¹ m²s⁻¹ at various q²‐values: (black) 1.0 × 10¹³ m⁻², (blue) 5.0 × 10¹² m⁻², (green) 1.0 × 10¹² m⁻², (orange) 5 × 10¹¹ m⁻², (red) 1 × 10¹¹ m⁻². These simulated signals show that the intensity of the rise–decay profile increases with increasing the observation time range (lower q²‐value).

When D depends on time, the temporal profile of the TG signal depends on the observation time range (i.e. q²‐value) and thus cannot be fitted using Eq. 3. To obtain kinetic information from the signal, the TG signal representing the diffusion (diffusion signal) should be analyzed based on a proper model. One typical case is the following scheme:

(Scheme 1)

Here, we assume that an intermediate I is created very fast after the photoexcitation of the reactant R and the product P is produced from I with a rate constant k. After solving diffusion equation coupled with the reaction kinetics, one may find that δn_R(t) and δn_P(t) are given by:

(4)

where δn_I and D_I are the initial refractive index change and D of the intermediate species, respectively. In many cases, conformation changes in a two‐state manner. Typical TG signals for D_R = D_I > D_P are shown in Fig. 1. Because there are multiple parameters in Eq. 4, it is generally very difficult to determine k from the signal at a single q²‐value. Hence, to obtain reliable parameters, it is necessary to fit the signals measured at various q².

There are many advantages in the TG method. One of them, particularly for the determination of D, is a high sensitivity of this method. This high sensitivity comes from two facts. First, as shown in Eq. 3 and Fig. 1, the TG signal always comes from the contribution of the reactant and product. As demonstrated by the simulation curves (Fig. 1), the diffusion signal is stronger, when the difference between D_R and D_P becomes larger. Even a small change in D gives rise to a relatively strong characteristic TG signal. In other words, the TG technique can determine the difference in D between the reactant and the product (or intermediates) sensitively. The other source of the high sensitivity comes from the fact that only the reacting molecules contribute to the signal. This point is in contrast with other techniques such as light scattering or nuclear magnetic resonance (NMR) methods, which usually measure diffusions of all molecules present in the solution. This point becomes clear, when the quantum yield of the reaction is low. In this case, a dominate signal of the light scattering or NMR comes from the nonreacting molecules (i.e. reactant) and signal for D_P (or D_I) may be masked by the strong contribution of the reactant. The time‐resolved detection of D on a fast time scale is, of course, a unique advantage of the TG method.

The diffusion coefficient D is an important physical property. The value depends on various factors of the solution such as the temperature and viscosity. The value also depends on the size of the diffusing molecule. Apparently, when the size of the molecule changes due to an association or dissociation reaction, the D‐value changes. Sometimes, the relationship between D and the molecular size is described by the Stokes–Einstein equation that is expressed by:

(5)

where k_B, T, η, a and r are the Boltzmann constant, temperature, viscosity of the solution, a constant representing the boundary condition between the diffusing molecule and the solvent and the radius of the molecule, respectively. Hence, D decreases with increasing molecular size. The D‐change detection has been applied to studies on the dynamics for the homogenous oligomerizations 45.

Interestingly, not only the size of the diffusing molecule, D also depends on the conformation. For example, D of a protein in the native state was reported to be much larger than that in the unfolded state even if the molecular volume is the same 35, 42, 46. This fact is a basis of detection of conformational change by using the time‐resolved D method. Although the exact origin of the D‐change by the conformational change remains unknown, it has been suggested that the intermolecular interaction between the protein and water may be a dominant factor.

Photoreaction dynamics of (6‐4) photolyase

Conformational change in the vicinity of the chromophore

Photoreaction of (6‐4) PHR in the vicinity of the FAD chromophore has been studied by the light absorption detection technique. In particular, the absorption spectrum change on the ultrafast time scale has been reported to detect the electron transfer reaction 11. Here, the reaction in a relatively slower time region (>μs) is described.

Typical absorption spectrum changes (the TA spectra) after photoexcitation of full‐length (6‐4) PHR possessing reduced FAD without the substrate (damaged DNA) are shown in Fig. 2. The assignments of the spectrum changes were made as follows. First, the enhanced absorption peak in the wavelength range from 500 nm to 700 nm agrees well with the reported absorption spectrum for the neutral flavin radical FADH·(Fig. 2A) 47. Hence, the enhanced absorption in this wavelength range was attributed to the formation of a radical of PHR. The absorption change near 449 nm was assigned to the light‐induced oxidation of the reduced PHR. The temporal profile of the absorption change monitored at 633 nm is shown in Fig. 2B, which corresponds to the FADH· absorption change. The signal decayed monotonically to the baseline with two time constants of 20 ms and 2.5 s.

Figure 2

Photoinduced redox reactions in (6‐4) PHR monitored by the TA method. (A) TA spectra observed after 0.2 ms (red), 1 ms (green) and 20 ms (blue) after photoexcitation of the reduced form of (6‐4) PHR at 355 nm. Pale blue and pink curves on the figure show the absorption spectra of FADH^‐ and FADH·, respectively. (B) Temporal profiles of the TA signal of (6‐4) PHR with damaged DNA sample (orange) monitored at 633 nm under exactly the same conditions as the measurement of (6‐4) PHR (red). Concentrations of (6‐4) PHR and DNA were 120 and 110 μm, respectively.

The effect of the DNA repair to the kinetics monitored by the TA method was obtained by comparing the temporal profiles at 633 nm with and without the substrate (Fig. 2B). Although the temporal profile of (6‐4) PHR with the substrate in a long time range (<1 ms) was similar to that without the substrate, the signal intensity was slightly weaker (~ 15%). Interestingly, this value is close to the reported quantum yield of the light‐induced repair reaction (about 11%) 8. Considering that light‐induced conversion of reduced PHR to the PHR radical occurs by an electron transfer to the substrate, one may understand that the electron is reversed by back electron transfer from the restored DNA after the photorestoration. Therefore, it was suggested that the decrease in the absorption intensity reflects the decrease in PHR radical by the repair reaction. An additional fast decay was observed in a short time range (<100 μs). This TA signal for PHR with substrate was reproduced well by the sum of three exponential functions with time constants of 50 μs, 20 ms and 2.5 s. As described above, the time constants of 20 ms and 2.5 s agree with those obtained for PHR without the substrate. The origin of the 50 μs phase clarified by the TG method is described in the next section.

Conformational change and DNA repair reaction of (6‐4) photolyase

After photoexcitation of (6‐4) PHR by a pulsed laser beam, the TG signal appeared (<20 ns). The signal decayed to the baseline once followed by a rise and decay profile (Fig. 3). As the time constant of the initial decay phase depended on the q²‐value and agreed well with D_thq², this component was attributed to the thermal grating. The next decay phase was expressed by a sum of q²‐independent and q²‐dependent phases. The decay rate of the signal in 10 μs–10 ms did not depend on q². Hence, this phase should represent the reaction dynamics of (6‐4) PHR possibly including the kinetics observed by the TA signal. Because the decay rate of the last phase was clearly dependent on q², this phase was attributed to the diffusion process of (6‐4) PHR. If the D‐value of the protein changes by the reaction, a characteristic rise–decay profile should be observed as described in the principle section. However, as the last decay of this signal was a single diffusion component, it is certain that the D‐value of the protein does not change before (reactant) and after (product) light excitation (D_P~D_R). This is in contrast to many photoreceptor proteins including CRY, which is described later in this paper. Because the D‐change occurs by a conformational change and/or a change in oligomeric state, this observation signifies that the conformational change in (6‐4) PHR is small. The D‐value of (6‐4) PHR was determined to be 4.6 × 10⁻¹¹ m²s⁻¹.

Figure 3

Photoreaction and diffusion dynamics of (6‐4) PHR determined by the TG method. The TG signals were measured at q² of (red) 2.86 × 10¹² m⁻², (orange) 1.33 × 10¹² m⁻², (green) 2.67 × 10¹¹ m⁻², (blue) 1.24 × 10¹¹ m⁻² and (dark blue) 6.68 × 10¹⁰ m⁻². The best fit curves are shown by the black lines. The phases of the q‐independent reaction and q‐dependent diffusion dynamics are shown in the figure. The weak decay feature of the diffusion signal (q‐dependent rate) indicates that there is no D‐change in the photoreaction of (6‐4) PHR.

The DNA repair reaction of (6‐4) PHR was determined by comparing the TG signals for (6‐4) PHR alone to that in the presence of a DNA oligonucleotide with or without a single (6‐4) PP under the same conditions. It was found that adding undamaged DNA did not change the overall TG profile of (6‐4) PHR alone (Fig. 4A). Hence, the undamaged DNA does not interact with (6‐4) PHR in the ground state as well as the light state. Interestingly, the TG profile for (6‐4) PHR changed drastically upon the addition of damaged DNA carrying (6‐4) PP. The initial thermal grating signal decayed to the baseline, then showed a very small rise–decay feature followed by a strong rise–decay peak (Fig. 4A).

Figure 4

Observation of the light‐induced DNA repair of (6‐4) PHR by the TG method. (A) TG signals of (6‐4) PHR (red), with intact DNA (blue) and with damaged DNA (black), were measured at q² = 3.20 × 10¹¹ m⁻². (inset) Magnification of the longitudinal axis range. (B) TG signals of (6‐4) PHR + damaged DNA sample observed at various q². The q²‐values are (red) 1.03 × 10¹² m⁻², (green) 1.11 × 10¹² m⁻², (blue) 2.45 × 10¹¹ m⁻² and (dark blue) 4.80 × 10¹⁰ m⁻². The best fit curves based on the product dissociation model are shown by the black lines. Concentrations of (6‐4) PHR and intact/damaged DNA were 80 μm and 70 μm, respectively.

From the q² dependence of the TG signal (Fig. 4B), the rise–decay signal was assigned to the diffusion signal of (6‐4) PHR. The rise–decay feature of the diffusion signal indicated that there are two different species having different D. From the signs of δn for the rise and decay components and Eq. 2, the rise and decay components were assigned to the diffusions of the product and reactant, respectively. According to the theoretical expression described in Section Materials and methods, the kinetics of the D‐change can be determined from the q²‐dependent feature. After normalization of the signal by the thermal grating signal intensity, which reflects the number of photoexcited protein, the intensities of rise–decay diffusion signals were constant over the relatively small q² region (Fig. 4B). This feature indicates the D‐change reaction is completed at this longer time. Therefore, the rise–decay profile should be reproduced by a biexponential function in this time region [Eq. 3] and the diffusion coefficients for the product (D_P = 1.1 × 10⁻¹⁰ m²s⁻¹) and the reactant (D_R = 4.9 × 10⁻¹¹ m²s⁻¹) were determined. This D_R‐value for (6‐4) PHR during the repair is almost identical to the experimentally determined D‐value for (6‐4) PHR alone.

As the D‐change was not observed in the protein alone or with undamaged DNA, the D‐change results from the DNA repair reaction. From the magnitude of D, D_P was attributed to the D of freely diffusing DNA. Therefore, the observed D‐change in the DNA repair reaction is a clear indication of the dissociation process of the restored DNA from PHR. This dissociation model is consistent with the low (insignificant) binding affinity for product DNA (undamaged DNA) with PHR.

The observed TG signals at various q² were analyzed based on the product dissociation model:

(Scheme 2)

where daDNA and unDNA are damaged DNA and undamaged DNA, respectively. k is the rate constant of the dissociation. In this model, four different species (PHR:daDNA, PHR:unDNA, PHR and unDNA) are considered to contribute to the TG signals, but the D‐values for three of the four, PHR (60.5 kDa), PHR:daDNA (69 kDa) and PHR:unDNA (69 kDa), are predicted to be nearly identical (D_R). Using a fitting function obtained by solving the diffusion equation based on the scheme, the TG signals were consistently reproduced by D_R = 4.3 × 10⁻¹¹ m²s⁻¹, D_P = 1.0 × 10⁻¹⁰ m²s⁻¹ and k⁻¹ = 50 μs.

Further support of the assignments was obtained by an experiment under conditions with excess damaged DNA as follows. When the substrate concentration ([DNA]) is smaller than PHR (such as [PHR]/[DNA] = 80 μm/70 μm), the diffusion signal was observed immediately after the photoexcitation by the laser pulse (Fig. 4). On the other hand, when the substrate concentration was larger than PHR (such as [PHR]/[DNA] = 80 μm/90 μm), the diffusion peak did not appear initially (for the initial several pulses of the excitation light) and the signal was similar to that of protein alone. The diffusion peak gradually appeared after successive laser irradiation if the solution was not stirred. This observation was explained in terms of the product dissociation as follows. Under the conditions of [PHR] < [DNA], the free damaged DNA (i.e. without binding to PHR) exists in the solution. Hence, as soon as the light‐dependent repair reaction takes place, the repaired DNA is released and new damaged DNA will bind to PHR:

(Scheme 3)

Therefore, the overall reaction in the system may be

(Scheme 4)

Because the diffusion coefficients of the reactants (PHR:daDNA and daDNA) should be very similar to those of the products (PHR:daDNA and unDNA), the grating signals cancel each other and the diffusion signal should not be observed. Upon continuously irradiation of the solution by the pulsed laser light, however, the repair reaction successively takes place in the light‐illuminated area, and free damaged DNA is locally depleted. By sufficient light irradiation, a condition of [PHR] > [unDNA] will be satisfied in the light illumination area, so that the diffusion signal appears.

Recently, the possibility of a two‐photon process for the DNA repair by (6‐4) PHR has been suggested 48 based on an experiment that monitored the DNA repair by the absorption changes at 265 nm (intact thymines) and at 325 nm [(6‐4) PP] after successive pulsed light excitations of (6‐4) PHR. It was observed that although the reactant (6‐4) PP indeed decreased after one pulse excitation, the intact thymines mostly increased after several shots. This observation indicated that absorption of the first photon induced depletion of the (6‐4) PP, but not formation of the intact thymines. However, according to the above TG experiment, the diffusion signal intensity reflecting the dissociation of repaired DNA increased almost linear with the excitation laser power. Furthermore, the signal diminished monotonically after successive pulse irradiation. All these observations support that the DNA repair catalyzed by (6‐4) PHR is a one‐photon process. Further examination and careful discussion on the number of required photons for the DNA repair are needed.

The rate for the formation of the proposed transient oxetane intermediate was reported to be fast (hundreds of picoseconds) 11. Thus, the repair reaction by (6‐4) PHR is likely to be completed on the nanosecond time scale, like that by CPD PHR. If this estimation is correct, why is the repaired DNA releasing rate as slow as 50 μs? It is highly possible that the slow rate is determined by the conformational change in (6‐4) PHR during the DNA repair. The conformation at the DNA binding site controls the selective binding to daDNA and releases the repaired DNA. Indeed, the conformational change was suggested by the crystal structures of the (6‐4) PHR:DNA complex before and after repair 6; that is, the DNA duplex was fully opened at the damaged site with the (6‐4) lesion flipped out almost 180° into the PHR active site and the hydrogen bonding of the Gln residue to the repaired DNA product was smaller than that to the damaged DNA substrate, favoring the release of the repaired DNA. It is natural to expect that the conformational rearrangements are much slower than the rate for the electron transfer‐induced bond cleavage reaction of DNA. Hence, the complete repair process involves the local conformational changes at the DNA binding site and the rate of this conformation change was observed as the releasing of the restored DNA. PHR has to seek and bind a new substrate to start a new restoration cycle after product dissociation (Fig. 5). Therefore, the dissociation rate obtained here determines the upper limit for the DNA repair reaction.

Figure 5

Schematic illustration of the protein–DNA interaction during DNA repair by (6‐4) PHR. After the repair reaction, which completes within the nanosecond time scale, (6‐4) PHR exhibits a conformational change near the DNA binding site in the vicinity of the FAD cofactor, which is represented by yellow spheres, with a time constant of 50 μs. This conformational change triggers the dissociation of repaired DNA product from (6‐4) PHR.

Detection of conformational change in AtCRY1

Photolyase/cryptochrome proteins contain FAD cofactor in any one of the three redox states: oxidized, radical or fully reduced. In PHRs, protein with fully reduced FAD is catalytically active, while those with the other FAD states are completely inactive. The catalytically inactive form of PHR can be photoreduced to restore the catalytically active form. This reduction is called “photoactivation” of FAD. This photoreduction is also considered important for CRY signaling. The photoreduction following excitation of oxidized FAD in AtCRY1 has been studied by the transient absorption technique over a wide time range 34. Upon photoexcitation, the oxidized FAD is reduced to anion radical FAD^‐ and subsequently protonated within a few microseconds to form neutral radical FADH· 49, 50. This neutral radical has been considered to represent the signaling state in AtCRY1. Transient absorption spectra over a longer time range detected a contribution from the transient amino acid radicals formed concomitantly with FADH· 27. The spectra contained three recovery phases. The amplitude spectrum for a 1 ms recovery phase deviated from the difference spectrum between FADH· and oxidized FAD. The deviation was attributed to an electron transfer from Tyr to Trp·. The amplitude spectra for the other 5 ms and >100 ms recovery phases closely follow the difference spectrum between FADH· and oxidized FAD. They were attributed to electron back transfer from FADH· to TyrO· to return the original oxidized state. The photoreduction is considered to be essential for CRY signaling, because an AtCRY1 mutant lacking the Trp residue for the proposed electron transfer pathway resulted in a virtual loss of biological function in vivo 51. It was suggested that the photoreduction leads to conformational changes in the PHR‐like and CCT domains, but time‐resolved detection of the conformational change was not successful. The TG method is a powerful way to detect the conformational change.

The TG signal of AtCRY1 was reported (Fig. 6A). After the thermal grating signal (which is not apparent in Fig. 6A, but is apparent in the magnified signal in Fig. 6B), a rise–decay signal was observed. Because the time scales of the rise–decay components were q² dependent, this phase indicates a protein diffusion process (diffusion signal). The rise–decay feature clearly indicates that AtCRY1 shows a D‐change during the photoreaction, in contrast to (6‐4) PHR shown in Section Conformational change and DNA repair reaction of (6‐4) photolyase. Because the signs of δn for the rise and decay components were, respectively, determined to be positive and negative, the rise and decay components were attributed to the diffusions of the product and the reactant, respectively. The faster rate of the rising component indicates that D of the reactant is larger than that of the product.

Figure 6

(A) TG signals of the full‐length AtCRY1 at various q²‐values; (red) 4.1 × 10¹¹ m⁻², (orange) 1.9 × 10¹¹ m⁻², (green) 7.3 × 10¹⁰ m⁻² and (blue) 3.5 × 10¹⁰ m⁻². Signals were normalized by the thermal grating intensity. (B) Magnification of the Figure (A) in the time region of 5 μs–200 ms. The q‐independent decay components were observed in the milliseconds time range (arrows). The black dashed curves in these figures show the best fit curves based on the two‐state model.

The kinetics of the D‐change was obtained from the q² dependence of the signal. At a large q², the diffusion signal appeared at a fast time scale and its intensity was weak. The intensity, however, dramatically increased at a smaller q², where the diffusion signal appeared at a longer time (Fig. 6A). This time dependence of the diffusion signal intensity implies that D of AtCRY1 changed within this observation time window. D_R and D_P were determined from the diffusion signal in the smallest q², that is in the long time scale where the D‐change reaction should be almost completed as D_R = 1.46 × 10⁻¹¹ m⁻²s⁻¹ and D_P = 0.62 × 10⁻¹¹ m⁻²s⁻¹. Using these values and D_I = D_R, the observed TG signals at all q² were reproduced well based on the Eqs. 3 and 4. The time constant (k⁻¹) for the D‐change was determined to be k⁻¹ = 400 ms.

Between the time scales for the thermal grating and diffusion signals, another weak species grating signal was observed (Fig. 6B). The time constants for this weak species signal were q² independent and they were 1 ms and 7.5 ms. Previous TA studies showed rapid absorption changes due to the creation of the radical form of AtCRY1 followed by a 1 ms kinetics, which was attributed to the intraprotein electron transfer from Trp· to Tyr 27. On the basis of this observation, the 1 ms dynamics observed by the TG method was attributed to the population grating signal reflecting the absorption change due to this electron transfer reaction. The flavin radical detected by the TA method disappeared with time constants of 7.2 ms and >140 ms. The 7.5 ms dynamics observed in the TG signal was also attributed to this dark reversion process of the FAD cofactor from the radical form to the ground state. If the system is completely recovered with these time constants, the TG signal should disappear with these time constants, too. However, interestingly, the strong diffusion signal still observed even after this 7.5 ms decay. This fact indicates that the light‐induced conformational changes in the protein part should be retained even after the recovery of the cofactor. It was concluded that the origin of the D‐change is due to the significant and large conformational change in the AtCRY1 protein.

Interestingly, it was reported that this diffusion signal did not appear for a sample of an AtCRY1 mutant (W324F) lacking the terminal surface‐exposed Trp residue of the proposed electron transfer pathway (Fig. 7). The TA 27 and FTIR 28 studies on AtCRY1 suggested that an aromatic residue (Trp or Tyr) was involved in the photoreaction. Indeed, the W324F mutant shows almost the same absorption spectrum as the wild‐type protein, but shows no light‐induced absorption change, indicating the mutant has no photoreduction reactions 51. On the basis of these results, it is certain that the D‐change occurs by the electron transfer reaction. Therefore, the conformational change in the full‐length wild‐type protein is triggered in the N‐terminal PHR‐like domain.

Figure 7

Transient grating signal of the wild‐type full‐length (black), the CCT domain truncated (blue) and W324F mutant (red) of the full‐length AtCRY1 at q² = 2.3 × 10¹² m⁻² under the same experimental conditions.

Which part of the protein is changed? CRYs are generally larger than PHRs, because of the C‐terminal extension of their sequences. Plant CRYs, particularly, have an additional characteristic C‐terminal domain (CCT domain). The plant CCT domains have been implicated in signal transduction and shown to function as a scaffold for other proteins. NMR studies combined with proteolysis indirectly suggested that some conformational change occurs in the CCT domain upon light irradiation. To directly examine and distinguish light‐induced conformational changes in the N‐ and C‐terminal domains of AtCRY1, the TG signals of truncated AtCRY1 protein lacking the CCT domain (trAtCRY1) were investigated under the same experimental conditions used for full‐length AtCRY1 (Fig. 7, blue). In contrast to the full‐length protein, which shows the characteristic large diffusion signal, trAtCRY1 does not show this component—only a very weak diffusion signal. These observations implicate that the major component of the D‐change‐sensitive conformational change in AtCRY1 is derived from the CCT domain.

A previous experiment using a binding assay coupled with proteolysis revealed that the CCT domain has a stable tertiary structure owing to the interaction with the PHR‐like domain in the dark 24. In the light state, however, this interaction was weakened due to a conformational change in the CCT domain. Considering these results and the conformational change directly observed by the TG method, we may obtain following picture of the reaction. The conformation of AtCRY1 in the dark state is relatively packed due to the interdomain interactions between the PHR‐like and CCT domains. After the light‐induced reduction of the FAD cofactor in the PHR‐like domain, the electron transfer takes place and it changes the interaction between the PHR‐like and CCT domains. This change results in the dissociation of the CCT domain from the PHR domain. By this dissociation, the surface area of AtCRY1 is exposed to the solvent, and it causes the D‐value reduction. This conformational change would also expose interaction sites for other molecules, so that it activates signals in the photomorphogenesis pathways.

The TG signal of trAtCRY1 without the CCT domain showed a very weak diffusion signal reflecting a small D‐change (Fig. 7, blue). This small change in D is consistent with a small conformational change in the vicinity of FAD, as observed by the previous FTIR study 28. The TG study suggests that PHRs exhibit relatively small conformational changes upon light illumination, whereas CRYs show conformational changes by involving their evolved C‐terminal extensions to increase dynamics for signaling. The question may arise, how is the interdomain interaction between the PHR‐like domain and the CCT domain regulated with such a small conformation change? Similar features have sometimes been observed for other photosensor proteins, such as the LOV (light–oxygen–voltage) and BLUF (photosensor of blue light using FAD) proteins. For example, clear conformational changes have not been observed after the photoexcitation of the BLUF domain. The circular dichroism (CD) spectra of BLUF proteins AppA and TePixD did not change upon illumination 52, 53, and NMR studies of AppA and BlrB showed only small chemical shifts upon excitation 54. The diffusion coefficient of the BLUF domain of YcgF was not changed by photoexcitation 55, although D is sensitive to conformational changes. Despite these minor structural changes in the BLUF domain, the interprotein and interdomain interaction is controlled by photoexcitation. As the driving force of the reactions, it was recently proposed that the random structural fluctuation is a key factor 56. When the BLUF domain of TePixD is photoexcited, the fluctuation is enhanced without changing the averaged conformation. It may be reasonable to speculate that the fluctuation at the PHR‐like domain is changed by the photoexcitation and controls the interdomain interaction with the CCT domain. The schematic illustrations of the possible reactions are shown in Fig. 8.

Figure 8

Schematic illustration of the proposed conformational change in AtCRY1. The intraprotein electron transfer from FAD (yellow spheres) to Trp residues (purple spheres) induces a change in the PHR‐like and CCT domains. This change results in the dissociation of the CCT domain from the PHR‐like domain with a time constant of 0.4 s, and this dissociation increases the friction of the protein diffusion and reduces the diffusion coefficient.

PHR and CRY proteins

The photoreactions of both (6‐4) PHR and AtCRY1 are initiated by the light‐induced redox reaction of the FAD cofactor. However, these proteins were found to show different conformational dynamics. (6‐4) PHR, which functions as a DNA repair enzyme, showed no D‐change, while the AtCRY1, which functions as a photoreceptor protein, showed a substantial D‐change during photoreactions. Detailed analysis revealed that the observed D‐change in the photoreaction of the AtCRY1 reflects the protein conformational change related to the C‐terminal extension (CCT domain). This observation is consistent with a structural feature in the protein family where PHR lacks the CCT domain. The different conformational dynamics of PHR and CRY can be understood by the difference in the function of these proteins. For the DNA repair reaction by PHR, changing the global protein conformation is not so important, because PHR catalyzes the reaction by donating an electron to the DNA bound to it in the vicinity of the FAD cofactor that absorbs light. In contrast, CRY needs to transmit the light‐induced signal to the protein partner to regulate biological functions. Thus, conformational rearrangements of the CCT domain play an important role in the CRY signaling. PHR and CRY proteins likely mediate different biological functions by fully using their structural differences and by showing the different conformational dynamics.

However, PHR and CRY proteins are implied to share common “dissociation” properties. The PHR dissociates the repaired DNA by an interaction change between protein and DNA, while the CRY also likely dissociates the CCT domain from the PHR‐like domain by an interdomain interaction change. After receiving the light using the common PHR‐like domains, PHR and CRY both show an interaction change on the molecular surface triggered by the redox reaction of the FAD cofactor. This interaction change regulates the dissociations of DNA or the C‐terminal domain, resulting in the different biological functions. This dissociation property might be important for the functions of the other CRY proteins such as mammalian clock CRY, which has a strikingly high similarity to the (6‐4) PHR. Although the qualitative “dissociation” properties are similar to each other, the rate constants are very different; that is, while the rate of the DNA dissociation from (6‐4) PHR is 50 μs, that of dissociation of the CCT domain from the PHR domain of CRY is 400 ms, which is approximately 10⁴ times slower.

We believe the knowledge obtained here provides essential information for the understanding of the reactions and functions of the proteins of the photolyase/cryptochrome family.

Acknowledgements

We are deeply indebted to all authors of the papers referenced in this article. Part of this study was supported by the Grant‐in‐Aid for Scientific Research (25288005) and the Grant‐in‐Aid for Scientific Research on Innovative Areas (Research in a proposed research area) (20107003, 25102004) from the Ministry of Education, Science, Sports and Culture in Japan (to M.T.).

Biographies

• 

  Masato Kondoh obtained his B.Sc., M.Sc. and Ph.D. (in Chemistry) from Kyoto University. His research interest is conformational and intermolecular reaction dynamics of protein. He is currently trying to detect the dynamics of membrane proteins at interface.

• 

  Masahide Terazima is a professor at the Graduate School of Science, Kyoto University, Kyoto, Japan. He obtained his B.Sc. from Kyoto University in 1982 and completed his Ph.D. at the Graduate School of Science, Kyoto University, in 1987. He became an assistant professor at the Faculty of Science, Tohoku University, at Sendai in 1986. He joined the Faculty of Science, Kyoto University, in 1990 and was promoted to be a full professor in 2001. His current research interests include development of new methods to study reaction mechanisms of biological proteins, such as time‐resolved thermodynamics and time‐resolved diffusion methods.