Dynamics and mechanism of DNA repair by a bifunctional cryptochrome Authors: Luyao Yan, Xiaodan Cao, Lijuan Wang, Jie Chen, Aziz Sancar, Dongping Zhong Published: 2024 ABSTRACT SignificanceThe discovery of some cryptochromes possessing photolyase function is exciting and such bifunctional cryptochromes may play significant roles in vivo. Little is known about the molecular mechanisms of bifunctional cryptochromes (CRYs) and their photocycles, and the active state of the critical cofactor flavin in vivo for the dual functions. Here, we focused on the photolyase function of the bifunctional Chlamydomonas reinhardtii CRY (CraCRY) on repairing ultraviolet light induced (6-4) photoproduct. We elucidated the dynamics and mechanism of repair by CraCRY at the most fundamental level by integrating femtosecond spectroscopy with molecular biology. Our findings may unify the active state of the cofactor in vivo of photolyase/cryptochrome family and provide insights into the dual functions of the members of this family. The photolyase/cryptochrome proteins belong to a flavoprotein superfamily, acting as photoenzyme/photoreceptor under blue-light illumination (1–5). The two kinds of photoproteins share high sequence and structural similarities, but perform completely different functions in organisms (6–8). Photolyases (PLs) use blue-light energy to repair UV-induced DNA damage, including cyclobutane pyrimidine dimer (CPD) and pyrimidine (6–4) pyrimidone photoproduct (6-4PP) (9–12). Cryptochromes (CRYs) are involved in various blue-light regulatory mechanisms (13–16), such as photoperiodic flowering in plants and circadian rhythms in animals. Recently, several proteins have been suggested to have dual functions for both DNA repair and signal transduction (17–22). One of these bifunctional proteins, an animal-like cryptochrome (namely aCRY) from the green algae Chlamydomonas reinhardtii (CraCRY), has been extensively studied recently (23–38), showing both 6-4PP repair and signaling properties upon blue-light excitation (38). But the molecular mechanisms for its dual functions still remain unclear. Here, we focus on its repair dynamics to elucidate the molecular mechanism and catalytic photocycle by mapping out the dynamic evolution of the repair process. The complex structure of CraCRY has been recently solved using X-ray crystallography, including its cofactor FAD and substrate 6-4PP (Fig. 1), but without the disordered C-terminal extension. This extension contains approximately 100 amino acids, mainly for the signal transduction function. Here, the 6-4PP repair occurs only at the active site in the photolyase homology region even in the absence of the C-terminal extension. In the active site, the crystal structure shows two conserved histidine residues, H357 and H361 (Fig. 1), corresponding to H364 and H368 in Arabidopsis thaliana 6-4PL (At6-4PL), surrounding the substrate 6-4PP. The At6-4PL is so far the only 6-4PL for which the repair dynamics have been resolved (39); the derived reaction scheme of repair is also shown in Fig. 1. We observed the critical role of the first H364 as the key proton donor for repair in At6-4PL. Here, in CraCRY, the second histidine H361, not first H357, is at a short distance of 3.2 Å with the oxygen atom of the hydroxyl group at the 5′ side of 6-4PP (Fig. 1) and can be a potential proton donor. Using site-directed mutagenesis, we prepared two single mutants, H357A and H361A, and one double mutant, H357F/H361F to determine the role of both histidines in the repair reaction. X-ray structure of Chlamydomonas reinhardtii CraCRY (in blue) bound to double-stranded DNA (in orange) containing a 6-4PP lesion (PDB: 5ZM0) with a close view of the local structure at the active site, showing the relative positions in Å of the catalytic cofactor FADH‾ (including LfH‾ and Ade), histidine residues H357 and H361, and the 6-4PP substrate. The 6-4PP lesion is flipped out of the DNA and inserted into the active site. The two most conserved histidine residues, H357 and H361 are close to the 6-4PP. The repair scheme involves both electron and proton transfer, including the cyclic intramolecular electron transfer between LfH‾ and Ade (FET1, BET1a, and BET1b), the forward intermolecular electron transfer to the substrate (FET2 and FET3), the futile backward electron transfer (BET2), and the repair channel, including a critical PT, subsequent bond breaking and making, and final proton and electron return (PR and ER). We first conducted titration experiments to obtain CraCRY dissociation constant (Kd) of the complex and measured its repair quantum yield (QY). The obtained dissociation constant of the wide type (WT) with 8-mer single-stranded DNA containing 6-4PP is 7.01 × 10−6 M, indicating strong binding interactions (SI Appendix, Fig. S1). Using the EcPL repair as a control (40), the measurements of QY for CraCRY, i.e., the increase of the repaired base thymine absorption at 266 nm and the decrease of 6-4PP absorption at 325 nm, are shown in SI Appendix, Fig. S2 A and B. The repair QY is (6.6 ± 0.04)% for the WT in H2O and (3.7 ± 0.03)% in D2O, smaller than those of At6-4PL (39). Surprisingly, both mutations of H357A and H361A show negligible repair, with QY of 0.014% and 0.025%, respectively (SI Appendix, Fig. S2C). For the double mutation H357F/H361F, no trace of repair was observed, indicating that the histidine residue must participate in repair. To understand these repair processes at the most fundamental level, we then employed femtosecond spectroscopy to dissect the enzymic repair reaction into a series of elementary steps and characterize their detailed dynamics. We followed the repair evolution, captured the important intermediates, and finally elucidated the dynamic picture of the repair process. Results and Discussion Electron Bifurcation and Dominant Intermolecular Tunneling. CraCRY, like all other photolyases, contains a fully reduced flavin adenine dinucleotide (FADH‾) as a catalytic cofactor, noncovalently situated at the bottom of the active site in an unconventional U-shaped conformation with a short distance between the lumiflavin (LfH‾) and adenine (Ade) moieties (9, 11). As reported before (9, 41–44), an intramolecular electron transfer (ET) usually occurs between LfH‾* and Ade, i.e., through forward ET (FET1) and back ET (BET1), with formation of intermediates LfH• and Ade‾. Fig. 2A shows the transient signals of the WT and three mutants probed at 800 nm upon 400 nm excitation. All transient signals display a stretched exponential decay, Aexp(−t/τ)β (A, amplitude, τ, lifetime and β, stretched parameter), in nanoseconds and this stretched behavior results from coupling between the ET and the active-site solvation processes on a similar timescale (45–47). The transients of the mutants all become faster while the dynamics of the WT in D2O buffer surprisingly slows down. Using our data analysis model (SI Appendix), the derived values for FET1 are τ = 1.5 to 14.6 ns and β = 0.76 to 0.85 (the structure Inset in Fig. 2B). The resulting average τ values are 1.8 to 15.9 ns, as listed in SI Appendix, Table S1. For the WT, we obtained FET1 in 8.6 ns, which is significantly longer as compared with the FET1 in EcPL and At6-4PL (9, 11). These stretched results seemingly indicate that the active-state solvation extends to the nanosecond timescale, affecting the slow FET1 process. Intramolecular electron transfer in CraCRY upon excitation at 400 nm. (A) Absorption transients of the excited-state flavin (LfH‾*) of the WT in H2O and D2O and three mutants probed at 800 nm (λpr). (B, Upper Panel) The transient of WT in H2O, probed at 650 nm, from two contributions of the excited-state LfH‾* (blue solid line) and the intermediate semiquinone state LfH• (yellow solid line). The semiquinone LfH• signal includes two components: one cyclic ET pathway (yellow dashed line) and one autoionization channel (yellow dotted line). (B, Lower Panel) The transient of the double mutant H357F/H361F, probed at 650 nm, from two contributions of the excited-state LfH‾* (blue solid line) and the intermediate semiquinone state LfH• (yellow solid line). The semiquinone LfH• signal includes two components: one cyclic ET pathway (yellow dashed line) and one autoionization channel (yellow dotted line). We further examined the back ET (BET1 in Fig. 2B) by detecting the dynamics of the radical intermediate LfH• at 650 nm. The transients are shown in Fig. 2B for the WT and double mutant H357F/H361F. Besides the excited-state LfH‾* signals which were detected at 800 nm in Fig. 2A, we obtained the radical LfH• signals which follow a reverse kinetics with a slow formation (FET1) and a fast decay (BET1b in Fig. 1). The BET1b dynamics, returning the original ground state and finishing the redox cycle, is fast in 134 ps for the WT and 43 ps for the mutant H357F/H361F. The other two mutants of H357A and H361A, which initial signals are enlarged in SI Appendix, Fig. S3, clearly show the LfH• contributions. We also need to include a minor autoionization channel with a formation time in tens to hundreds of nanoseconds, as observed before (9); this electron must return to the LfH• to form the original LfH at a longer time scale since we did not observe the steady-state LfH• formation after prolonged blue-light illumination. All BET1b times are listed in SI Appendix, Table S1. As discussed below, the BET process in fact can also lead the system to return the original excited state, i.e., BET1a in Fig. 1, and the derived timescales (see below) are in 500 to 1,400 ps, also listed in SI Appendix, Table S1, much slower than the BET1b by a factor of 5 to 30. This channel is very minor for CraCRY and makes a negligible contribution to the transients shown in Fig. 2B. Fig. 3A shows the transients in the presence of the substrate 6-4PP. Strikingly, all the dynamics in the complex become much faster than those in the proteins alone, from several nanoseconds to hundreds of picoseconds, indicating a dominant intermolecular electron tunneling process from the LfH‾* to the substrate 6-4PP (FET2 in Fig. 1), mediated by the intervening adenine moiety as observed in the At6-4PL (9) and EcPL (43) before. These dynamics still follow a stretched exponential behavior with a larger β = 0.88 to 0.98 and a faster τ = 229 to 325 ps. The average FET2 are in 281 to 534 ps, much faster than the FET1, indicating that the electron bifurcation from LfH‾* greatly favors the direct tunneling to the substrate 6-4PP rather than the two-step hopping to Ade then to the substate as widely observed in the repair of CPD in class-II photolyases (48). For the WT, we obtained the FET2 being about 300 ps in both H2O and D2O. For the three mutants, the electron tunneling between the cofactor and the substrate still occurs as the WT even though the mutants have nearly no repair function. Intermolecular electron transfer between the cofactor and the 6-4PP and subsequently PT for DNA repair by CraCRY. (A) Absorption transients of the enzyme alone (dashed line) and enzyme–substrate complex (solid line) probed at 800 nm. In the presence of 6-4PP, the excited-state dynamics of LfH‾* become much faster due to direct electron tunneling to 6-4PP. (B) Absorption transients probed at 650 nm for the WT (red in H2O and blue in D2O) and double mutant H357F/H361F (green). The two single mutants are shown in the Inset. Clearly, all the transients show a bump around 200 ps and all the three mutants decay to zero. Note the autoionization signals were removed. (C) The transient of the WT probed at 650 nm shows the two components of LfH‾* and LfH•. The LfH• signal also has two components: One (LfH•1) is the dynamics of the futile BET2 branching and the other (LfH•2) is the portion for DNA repair. Capturing Intermediates and Critical Proton Transfer (PT). We tuned the probing wavelength at 650 nm to catch the intermediate LfH•. The detected signals only contain two contributions, the excited-state LfH‾* and the intermediate LfH•. Fig. 3B shows the transients of the WT in H2O and D2O and the three mutants. All the transients show an obvious bump around a few hundreds of picoseconds (the Inset in Fig. 3C and SI Appendix, Fig. S4), indicating a reverse kinetics as observed in Fig. 2B. Significantly, all transients of the three mutants decay to nearly zero while for the WT in H2O and D2O the transients exhibit a plateau at a longer time, indicating that a certain fraction of LfH• does not return the original ground-state LfH‾ yet and the repair has not been completed. Since the three mutants have negligible repair activities, after the direct tunneling, an ultrafast BET occurs to close the ET redox cycle and restore the original state. Thus, the two long components of LfH• in H2O and D2O are the portions for the repair channel, directly proportional to their repair quantum yields, as determined above (6.6%:3.7%). Fig. 3C shows the transient of the WT in H2O with deconvolution into the excited state LfH‾* and the intermediate LfH•. Using the kinetic model (SI Appendix) based on the reaction scheme in Fig. 1, after the initial electron injection the reaction bifurcates again and the LfH• signal consists of two components: One represents the futile channel BET2 to return the original state without repair, showing a reverse kinetics with a slow formation FET2 and a fast decay (BET2+PT), and the other one is a repair channel, proceeding to a proton-transfer (PT) process with the long component LfH• (Fig. 3C). Knowing the total decay rate (BET2+PT) and the amplitude ratio of two channels, we determined the BET2 and PT in 151 ps and 2,029 ps, respectively. Similar to the At6-4PL (9, 49), the reaction dominantly channeled into the BET2 pathway and thus the repair quantum yield is low. The bifurcation results in the repair channel with quantum yield of only about 6.9%. In D2O, the BET2 is 183 ps, similar to that in H2O, while the PT dynamics slows down to 4,543 ps, exhibiting a clear isotope effect and resulting in even lower quantum yield of about 3.9%. As reported before (39), the PT is the critical step for successful repair as the reaction passes over the transition state to form products. The observation of the PT process here also indicates the active-site histidine residues being protonated that can provide a proton for the repair. At the active site, the two histidine residues are conserved for most 6-4 photolyases except that the prokaryotic 6-4PL has the second histidine being leucine (50). The three solved X-ray structures of Dm6-4PL (51), At6-4PL (49), and CraCRY show their two histidine residues nearly in the same positions and the active-site configurations are highly conserved (SI Appendix, Fig. S5). In the presence of the 6-4PP substrate, the two known complex structures of Dm6-4PL and CraCRY show different configurations of the two histidine positions. For Dm6-4PL, the two histidine residues (H365 and H369) remain the similar configuration to that in absence of the substrate and the first histidine at a close distance of 2.8 Å to the substrate acts as the proton donor, while for CraCRY, the first histidine (H357) flips out with a distance of 4.9 Å to the substrate and the second histidine (H361) gets closer to the substrate at 3.2 Å (SI Appendix, Fig. S5). Thus, in CraCRY it is probably the second H361, not the first H357, to donate a proton to the substrate. Furthermore, either histidine mutant causes nearly no repair, indicating that the two histidine residues must change their local configurations by mutations. The two conserved histidine residues seem to structurally couple together and also have a role of double protection to ensure a proton source as seen in Dm6-4PL and CraCRY. For the prokaryotic 6-4 photolyases, they only have the first histidine at the active site and this histidine must maintain a unique configuration to successfully donate a proton for repair, as also observed in Dm6-4PL (51). We further tuned our wavelengths to the UV region to capture substrate-related species.‾ Fig. 4 shows two typical transients probed at 323 nm and 315 nm. Through systematic analyses with all determined timescales from the visible detection, we decomposed each transient signal into substrate-related components of 6-4PPH• and 6-4PP, and the total flavin cofactor contributions of LfH‾*, LfH• and LfH‾. The signal of 6-4PPH• was relatively small, due to the slower PT (2,029 ps) and the much faster BET2 (151 ps), resulting in less population of the 6-4PPH• intermediate. At 323 nm, the signal is mostly positive and the flavin contributions are dominant. At 315 nm, the signal initially is positive and then quickly turns negative, mainly resulting from two ground-state bleaching contributions of LfH‾ and 6-4PP. Femtosecond absorption transients probed in UV region. (A) The transients of the WT in H2O and D2O probed at 323 nm (dark red and blue lines) and H357F/H361F mutant (green line). The signal for the WT includes contributions from three forms of flavin, excited state LfH‾*, ground state LfH‾, and intermediate state LfH• (blue line), as well as 6-4PP (yellow line) and substrate-related intermediate 6-4PPH• (purple line). (B) The transients of the WT probed at 315 nm (dark red and blue lines) and H357F/H361F mutant (green line). At 315 nm, the contribution of flavin mainly comes from the ground-state bleaching of LfH‾, resulting in an obvious negative signal (blue line). The signal of H357F/H361F mutant decays to zero through a futile ET redox cycle. Quantum Yields, Electron Transfer Properties, and Repair Photocycle. We have resolved the key elementary reactions during the repair and obtained their timescales. From electron transfer energetics (see below), we can estimate the ET reaction time of FET3, i.e., one electron hopping from Ade‾ to 6-4PP, in a few picoseconds by knowing the driving force (−ΔG0), the donor-acceptor distance (r), a range of reorganization energy (λ), as well as the total repair quantum yield determined above. The observed dynamics include six ET reactions and one PT and their timescales cover about six orders of magnitudes from a few picoseconds of FET3 to about one microsecond of PT in the mutants (the Upper panel in Fig. 5A). With all those times, we can calculate the quantum yield for each reaction pathway. According to the reaction scheme in Fig. 1, the repair quantum yield after the PT step should be Φ = (ΦFET1 × ΦFET3 + ΦFET2) × ΦPT if we ignore the BET1a channel. The resulting values are shown in the Lower panel of Fig. 5A, also listed in SI Appendix, Table S2. Again, the intramolecular FET1 branching is minor and the direct tunneling FET2 is dominant, similar to the observation of class-I CPD photolyases such as in EcPL (43–45). For the WT in H2O, the Φ value is 0.0656, which is equal to the total repair quantum yield of 6.6% determined from the steady-state measurement, i.e., Φ = ΦTotal. This result is critical and shows that after the PT process the repair reaction proceeds in down-hill manner and never goes back till the repair is completed. For the very small repair of the two mutants H357A (0.014%) and H361A (0.025%), the derived PT process could be as slow as about 1 μs, indicating an extremely slow PT process. Thus, the histidine residue must be in a very unfavorable configuration after mutation of the other histidine. For the double mutation H357F/H361F, due to the lack of a proton donor at the active site, repair cannot occur even though the initial direct electron tunneling takes place in hundreds of picoseconds, as fast as the WT, but subsequently an ultrafast back electron transfer follows and hence closes a futile redox cycle without repair. The reaction times, quantum yield branchings, and ET energy properties. (A) The Upper panel shows the reaction times of each elementary reaction involved in DNA repair. LT, lifetime; FET, forward electron transfer; BET, backward electron transfer; PT, proton transfer. Gray dashed lines connect the two or three competing channels that determine the repair efficiency. The Lower figure shows the branch quantum yields of FET1, FET2, FET3, and PT and total repair quantum yield, following the equation Φtotal = (ΦFET2 + ΦFET1 × ΦFET3) × ΦPT. (B) Two-dimensional plot of the Franck–Condon part of electron transfer of CraCRY. ΔG0, free energy; λ, reorganization energy. The charge recombination step (BET1b) falls in the Marcus inverted ET region (−ΔG0 ≥ λ), while all other ET reactions fall in the Marcus normal region (−ΔG0 ≤ λ). The typical range of λ here is in the region of about 1.0 to 1.3 eV. Using the Marcus ET theory (52–54) with the obtained ET timescales, we can estimate the ET properties such as driving forces (−ΔG0) and reorganization energies (λ). Knowing the complex structures, i.e., the donor-acceptor distances, we can estimate the coupling constants (J) according to our previous determination of those values for EcPL complex (9, 41, 43). For the ET pair of LfH‾*(LfH‾)/Ade with a mixing hopping distance at around 4.6 Å, the J is about 12 meV; for the LfH‾*/6-4PP pair with a long intervening tunneling distance at about 7.8 Å, the J is around 3.0 meV; and for the Ade‾/6-4PP pair with a shorter distance at 2.8 Å, the J can be as large as 25 meV. As detailed in SI Appendix, we figured out all energetics of five ET processes of FET1, BET1b, FET2, FET3, and BET2, as shown in Fig. 5B and SI Appendix, Table S3; among the five ET dynamics, FET1, FET2, FET3, and BET2 are in the normal region and the BET1b is in the inverted region. It is noted that in order to have reasonably low reorganization energies (λ) for FET3, we estimated the FET3 reaction times of about 3 ps. With all these determined values, we can easily calculate the BET1a times, which are also listed in SI Appendix, Table S1 and shown in Fig. 3A (the Upper panel), and the derived ET energetics are also in the normal region of Fig. 5B. The BET1a process is much slower than the BET1b by about one order of magnitude, thus channeling all the population into BET1b and FET3 after the first ET bifurcation. Most reorganization energies are in the range of about 1.1 to 1.3 eV, consistent with our previous observations for photolyases and cryptochromes (9, 41, 43, 44), in which contribution is mostly from the flavin cofactor due to its butterfly bending motion, resulting in the configuration changes during reduction (55–57). Notice that the reorganization energies for the mutants become smaller, compared with the values of the WT, suggesting that the mutations of the protonated histidine into hydrophobic alanine or phenylalanine reduced local polarity and increased local rigidity. Fig. 6 shows the repair photocycle of 6-4PP with all determined timescales. The repair dominantly takes the direct tunneling pathway, still using the intervening adenine as a mediator to enhance the coupling between the flavin cofactor and the substrate. This strategy is the most efficient and also used in the repair of CPD by class-I EcPL and AnPL (9, 41, 45). The distant class-II CPD photolyases such as AtPL and DmPL mainly take a less efficient route mostly with a two-step hopping pathway (48) here using the adenine as an intermediate. Again, the observation further validates the functional role of the folded configuration of the flavin cofactor. After the critical PT, as examined above through the quantum yield calculations, the processes go to the complete repair. During the remaining processes, a series of atomic rearrangements, such as a proton relay, bond cleavages, and formations, occurred. Finally, one proton returned the original histidine and one electron got back to the flavin, completing the 6-4PP repair photocycle in a time of at least tens of nanoseconds. These longtime dynamics could be further studied using the infrared spectroscopy or the direct structural imaging by femtosecond X-ray free electron laser (XFEL). Recently, the CPD repair by class-II MmPL has been studied by solving a series of structural snapshots from femtoseconds to 200 μs (58–60), covering 11 orders of timescales. One difficulty in solving various transient structures for the enzyme catalysis is the mixture of various populations. Fig. 6B shows the population changes with time for various flavins and substrate-related species. Given a snapshot during the enzyme reaction, the average of diffracted structures can be a mixture of various configurations. In a sense, it seems that we need to resolve the reaction dynamics and know the detailed timescales using femtosecond spectroscopy, and then we can carefully choose the specific moments to take diffracted imagines, preferably of one dominant species. In Fig. 6C, we illustrated a molecular “movie” of structural changes during the repair reaction. With the increased availability of XFEL, more and more biological complex reactions can be revealed. The repair photocycle of CraCRY, population changes, and different substrate intermediates. (A) Photocycle of CraCRY repairing 6-4 thymine photoproducts. The elementary steps include initial forward electron transfer (FET1 and FET2), futile backward electron transfer (BET1b and BET2), subsequent PT, and final proton and electron return after repair. The PT rate is the rate-determining step of the DNA repair efficiency, controlling the overall repair quantum yield. (B) Evolution of flavin-associated and substrate-associated population changes. After 1 ns, there is almost no population of 6-4PP•‾. (C) A molecular “movie” shows the key snapshots (I-IV) for the initial electron transfer to the substrate to form 6-4PP•‾, the critical PT from the nearby histidine to form 6-4PPH• and a later time to finally form the products of two normal bases. Conclusion We have used femtosecond spectroscopy and site-directed mutations to study the biological dynamics of a dual-function CraCRY and here focused to reveal its DNA repair function. This enzyme is the second system reported so far, for which dynamic evolution was mapped out and thus the molecular mechanism and repair photocycle were elucidated. Similar to the previously studied At6-4PL and class-I CPDPL, the catalytic reaction of CraCRY mainly adopts the direct tunneling pathway in 300 ps and avoids the two-step hopping channel in about 8 ns to minimize the ET bifurcation. The subsequent second reaction bifurcation is critical and decides the total repair quantum yield. The essential PT in 2 ns for a successful repair competes with the intrinsic back ET in 151 ps to return the original ground state. Obviously, this PT process is challenging and needs a perfect configuration with a short distance to effectively transfer the proton. From the two studied systems of CraCRY and At6-4PL, this competition results in at most 10% for the repair channel. Evolution seems not to have achieved the optimization of the active-site environment yet to favor the critical PT, probably due to other constraints. The PT step is necessity like a dividing line in the transition state. Our isotope effect further confirmed this step, resulting from electron-coupled PT. After the proton shifts from the histidine residue to the oxygen atom of the hydroxyl group on the 5′ base of 6-4PP, a proton relay occurs and another proton moves from the hydroxyl group to the N3 of the 3′ base. Subsequently, a transient oxetane structure is formed (Fig. 6C) and the bond breaking and forming immediately take place. Finally, after the electron returns to flavin to recover the original hydroquinone form and the proton returns the original histidine residue, the 6-4PP is restored into the two normal bases. All these remaining steps maintain 100% quantum yield, without any branching or going back to cause the permanent DNA damage. CraCRY possesses a conserved HH1LARH2 motif in the substrate binding pocket, identical to that in At6-4PL and Dm6-4PL. In the prokaryotic 6-4PL (such as PhrB), this motif is not conserved and contains HH1IQRL (50). Thus, to secure the repair of 6-4PP, at least one histidine must be near the oxygen atom of the hydroxyl group on the 5′ base of 6-4PP substrate to provide a kind of proton shuttling with the substrate during repair while two histidine residues near the substrate can double protect such a proton source to assure the repair happening. One question is still debatable: Which active redox state of flavin is in vivo for the dual functions? We believe that with all these detailed results we obtained for the 6-4PP repair the DNA repair function must start from the initial anionic hydroquinone flavin in vivo. It is hard to imagine that the two functions would use two diffident redox states in vivo considering the reduction potential in cells. The elucidation of the DNA repair using the anionic hydroquinone form of flavin may give us a solution, i.e., whether the bifunctional cryptochromes, or even cryptochromes in general, would use the same redox form of hydroquinone flavin for both functions in the environment of cell or one of the dual functions is redundant. Materials and Methods Preparation of CraCRY and Mutants. We transformed the DNA plasmid containing the CraCRY gene into Escherichia coli cell BL21(DE3). Then, the cells were grown in LB medium at 37 °C. Isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to induce protein expression when the optical density (OD) reached 0.9, and the cells were further grown at 15 °C for 20 to 22 h. The cells were harvested by centrifugation and the supernatant was discarded. Next, the pellets were lysed using an ultrasonic cell disruptor to release the CraCRY protein. After centrifugation, the supernatant was collected for further purification. The supernatant was applied to a Ni-NTA column for nickel affinity chromatography, selectively binding proteins with a His tag. The target protein was then eluted using imidazole, and the yellow CraCRY protein was collected. The eluted protein was further purified using an ion exchange column. The sample was loaded onto a heparin column and separated based on surface charge using a sodium chloride gradient to isolate the CraCRY protein. Finally, the purified CraCRY protein was stored in a buffer containing 50% glycerol to maintain protein stability and frozen at −80 °C for further measurement. In the study of mutants, we designed key histidine-related mutations (H357A, H361A, and H357F/H361F) at the active site. All mutant plasmids were constructed using the QuickMutation™ kit (Beyotime) based on the wild-type CraCRY plasmid, and PCR was performed using the ProFlex™ PCR system (Thermo Fisher Scientific). DNA sequencing was conducted on all mutated DNA plasmids to ensure accurate results. Preparation and Purification of 6-4PP. We prepared a DNA solution (GTATTATG) at a concentration of 0.5 mg/mL in ultrapure water. We loaded 5 mL of the DNA solution into a dish with a diameter of 6 cm, and placed three dishes into a sandwich bag. Then we used high-purity argon to purge the sandwich bag for 5 min to remove oxygen and reduce by-product production, and sealed the bag. Next, we irradiated the DNA solution with two 254 nm lights for 1 h, and concentrated the solution and stored it in the dark at −80 °C. The distance between the dishes and the 254-nm lamps was about 2 cm. Next, we purified the UV-irradiated DNA sample using a C18 column on high-performance liquid chromatography (HPLC). Buffer A is TEAA buffer (20 mL acetic acid and 55.8 mL triethylamine are mixed in 4 L TEAA buffer to a final volume of approximately 4 L, and then acetic acid is slowly added to adjust the pH to 7.0), and buffer B is acetonitrile. The purification method for 6-4PP includes approximately 35 min of 8.1% buffer B (separation), 3 min of 30% buffer B (cleaning), and 7 min of 8.1% buffer B (re-equilibration). The single injection volume is 100 μL, and the flow rate is 4.5 mL/min. We collected the 325 nm absorption peak (6-4PP), and the small peak after 6-4PP (its Dewar valence isomer), and the strong peak eluted by 30% B (mainly undamaged DNA). Finally, we concentrated and dried the separated 6-4PP product and stored it at −80 °C in the dark to prevent the transformation of 6-4PP into Dewar valence isomer. Measurement of Repair Activity. To quantify the enzymatic activity of CraCRY, we used EcPL as a control. The procedure to determine the relative quantum yield and dissociation constant has been described earlier (61). First, to determine the dissociation constant of wild-type CraCRY, we prepared a mixture of 2 μM concentration of CraCRY and an 8-mer single-stranded DNA containing 6-4PP at different concentrations (10 μM, 20 μM, 50 μM, 100 μM, 150 μM, and 200 μM). We placed the mixture of CraCRY and 6-4PP in a cuvette with an optical path of 1 mm and added 20 mM dithiothreitol (DTT) as a reducing agent. We used a white light to irradiate the samples at the same distance and recorded the absorbance spectra of each sample at a series of illumination times, taking the average absorption differential spectra at 325 nm and 266 nm, respectively. The typical results are given in SI Appendix, Figs. S1 and S2. Femtosecond-Resolved Transient Absorption Spectroscopy. After purification, CraCRY mainly exists in its oxidized form. We used a glove box to deoxygenate CraCRY. The protein was subjected to repeated cycles of high-purity argon gas purging and vacuum pumping inside the glove box, with three cycles performed to thoroughly remove oxygen from the solution. Then, in the glove box filled with argon, we mixed the protein with 20 mM DTT and transferred the mixture into a customized sealed reaction cell, ensuring that both the gas and the CraCRY protein solution in the reaction cell are anaerobic. Next, the protein was illuminated with white light, converting it to the pure reduced state FADH‾, which is the catalytic state of the enzyme. The protein concentration used in femtosecond transient absorption studies was approximately 100 μM. For deoxygenation of the CraCRY complex with 6-4PP, the procedure was similar to the above steps. After preparing the anaerobic CraCRY, pre-deoxygenated 6-4PP was added into the reaction cell. All femtosecond-resolved measurements were performed using the transient absorption method, which has been described in detail previously (42). In brief, for all measurements, a pump pulse at 400 nm was used. The pump pulse energy was typically attenuated to approximately 50 nJ/pulse before focused on the sample. We tested different pump energies and found that high pump energy would generate some nonlinear effects that overlapped with the repair dynamic signal. Therefore, it is necessary to ensure that the pump energy is as low as possible. Probe pulses were set at wavelengths of 800 nm, 650 nm, 323 nm, and 315 nm to detect different species. The corresponding molecular reaction kinetics model is described in detail in SI Appendix. Supplementary Material Appendix 01 (PDF)