Introduction

Cryptochromes regulate the circadian clock, growth and development of organisms, and are homologs of photolyases1,2. Photolyases obligately require a flavin adenine dinucleotide (FAD) cofactor in the anionic hydroquinone (hq, FADH−) state to repair UV-induced DNA lesions, such as cyclobutane pyrimidine dimers or 6-4 photoproducts, via external light energy3,4,5. To repair a lesion in DNA, a photolyase first binds to a DNA segment containing the lesion and then flips out the lesion into a pocket just above the FAD cofactor. The hq state FAD cofactor is excited by light and sends an electron to the lesion, forming a neutral semiquinone (nsq, FADH●): anionic lesion radical pair. The anionic lesion radical is not stable and quickly splits into two pyrimidines. Finally, the excess electron on one pyrimidine returns to nsq to reform the hq state FAD, and the repaired DNA segment disassociates from the photolyase3,4,5.

Many cryptochromes also contain an FAD cofactor, which is used to detect light cues. However, the photoreactions of cryptochromes are generally different from those of photolyases1,2. Drosophila cryptochrome (dCry) is a type I animal cryptochrome that is purified with an oxidized (ox) state FAD cofactor. Upon blue light illumination, the FAD cofactor rapidly converts into the anionic semiquinone (asq, FAD●−) state, which is the possible signaling state of dCry6,7,8. A cysteine residue (Cys416) facing N5 of FAD, instead of a conserved asparagine residue in many photolyases, is considered to be responsible for asq formation in dCry6,7. Intriguingly, dCry contains a C-terminal tail (CTT) that mimics lesion-containing DNA and binds to a homologous position of the substrate-binding pocket in photolyases (Fig. 1A). There is a conserved FFW motif (534-536) in CTT, with Phe534 located at the center of the binding pocket in the resting-state structure of dCry9,10. His378 is a remarkable residue adjacent to Phe534 in the binding pocket. The homologous histidine residue in 6-4 photolyases is known to be crucial for the repair of 6-4 photoproducts11,12,13,14,15.

Fig. 1: Proteolysis assay to detect CTT release of dCry.
figure 1

A The structure of dCry. The CTT, which is located at a binding pocket above the FAD cofactor (yellow stick), is shown in magenta. Trypsin cleavage sites (Lys289, Arg298, Arg430, and Arg513) are shown in green stick representation. Among them, Arg430 and Arg513 are light-accessible sites in the vicinity of CTT. B Comparison of two different proteolysis conditions to detect light-induced structural changes in dCry. Lane M was the marker. Lane U represents the untreated dCry protein. Lanes D-L and L-L, a trypsin:dCry molar ratio of 1:50 was used to digest dCry in the dark and light, respectively. The formation of fragments b and c was light independent and was determined to be 1-289 and 299-542 of dCry by mass spectrometry (Supplementary Figs. 1 and 2). Fragments a (1-513) and a’ (1-503) were detected only under light conditions. Lanes D-H and L-H, a trypsin:dCry molar ratio of 1:1 was used to digest dCry in the dark and light, respectively. Compared with those of lanes D-L and L-L, the bands of fragment b were weakened both in the dark and light. In some experiments, fragment b could not be detected, which did not interfere with the analyses. Under light conditions, fragment d of ~15 kDa was formed by trypsin cleavage at Arg298 and Arg430. Band t was trypsin applied for proteolysis. C Absorption spectra of the photoreduction of dCry under blue light (λmax = 440 nm, irradiance of ~9.5 W m−2) and oxidation. The pH value of the protein solution was 7.5. D Proteolysis of dCry using a trypsin:dCry molar ratio of 1:1 to detect CTT release during photoreduction and redocking after oxidation for 100 min. E Kinetics of the fractions of various redox states (▪, the ox state; ●, the asq state; ▲, the nsq state; ▼, the total of the redox states; the hq state was not detected; one representative result is shown), and CTT release of dCry (orange stars, represented by the intensity ratio of fragment d to the sum of fragments c and d) during photoreduction. The red line represents the fitted curve of asq formation during photoreduction (kpr1 ~ 0.06 s−1, Supplementary Table 1). The degree of CTT release corresponded well with the asq formation.

Full size image

The Drosophila circadian clock is driven by four core components. The transcriptional activators Clock and Cycle activate the transcription of several clock genes, including period (Per) and timeless (tim). Per and Tim are translated and act as transcriptional repressors to downregulate the transcription of their own genes. Then, a transcriptional-translational feedback loop is formed to produce an ~24 h clock cycle16,17. The major physiological function of dCry is to entrain the circadian clock with light17,18,19. Under blue light, dCry releases its CTT, interacts with Tim, and recruits the E3 ubiquitin ligase Jetlag (Jet) to promote the degradation of Tim20,21,22,23. Prolonged illumination also promotes the degradation of dCry by Jet or BRWD321,24. However, the factors that trigger CTT release have not been fully elucidated. Both photo- and chemical reduction of FAD to the asq state have been shown to induce CTT release8. Nevertheless, a dCry mutant (L405E/C416N) that forms nsq by photoreduction has been regarded as a functional mimic of the dark state that does not release CTT25. It was proposed that His378 protonation at the proper position is important for CTT release26. However, another study reported that His378 does not affect the transduction of the structural signal to the CTT but stabilizes CTT in the resting-state conformation27. It was declared that the signaling-state formation of dCry was pH dependent that was suppressed under basic conditions26,27,28. And it was also observed that higher fractions of nsq (which is insufficient to trigger CTT release) formed at lower pH values during the photoreduction of dCry28. These conflicting results need to be reconciled for a full understanding of dCry.

In this work, an improved limited proteolysis method was developed, by which trypsin cleavage at a site adjacent to CTT (Arg430) in dCry was readily detected, which corresponded well with asq formation by photoreduction. Using this method combined with spectroscopy analyses and light-induced degradation experiments in Drosophila S2 cells, dCry and its typical mutants were systematically investigated to elucidate the effects of the ionic and protonation states of flavin on its functions. Similar to a previous study25, the nsq-forming mutants (C416N and L405E/C416N) presented weak CTT release in the nsq state. However, we found that if the mutants were photoreduced to the hq state, which is also anionic, CTT release was promoted. Correspondingly, we found that decreasing the pH led to more nsq formation during photoreduction of dCry, as well as the inhibition of CTT release, and its oxidation after photoreduction was faster. These results emphasize that the negative charge on flavin is crucial for CTT release. The pH-dependent nsq formation of dCry and the protonation of the C416N mutant during photoreduction suggested the existence of an alternative proton transfer pathway in dCry. A side tunnel was found by examination of the dCry structure, which may transfer protons from the solvent to the FAD cofactor. We found that mutants in CTT [including the mutant that deleted CTT (dCT), F534A, F535A, and W536A] formed asq predominantly during photoreduction, accompanied with lower oxidation rates. Molecular dynamics simulations revealed that the mutants in CTT have a narrower and less curved entrance of the side tunnel, which could suppress proton transfer and the formation of nsq. This might be responsible for the previously observed enhanced light sensitivity of the Trp536 mutants8. The end of the side tunnel joins the binding pocket, where His378 is located. Mutation of His378 to a neutral alanine or a positively charged lysine residue resulted in the formation of a greater nsq fraction during photoreduction. In contrast, the substitution of a negatively charged aspartic acid residue for His378 suppressed nsq formation. On the basis of these results, a role of His378 was proposed that serves as a proton interceptor to adjust the protonation of the flavin cofactor. A proper fraction of nsq formed during photoreduction of dCry is important for fast recovery when it returns to the dark.

Results

Trypsin cleavage at Arg430 is a good indicator of CTT release

Limited proteolysis is a powerful method that has already been used to investigate the light-induced conformational changes of dCry8,22. In previous works, low concentrations of trypsin (1:800 or 1:100, trypsin:dCry weight ratios; molar ratios of ~1:300 or 1:40) and long incubation time (10 min)22, or high concentrations of trypsin (1:1 molar ratio) and short incubation time (30 s)8 were used. Upon blue light illumination, dCry was photoreduced to the asq state, CTT proteolytic susceptibility increased, and fragments with CTT removed (1-494/1-503/1-513) were observed after trypsin digestion8,22. The positions of the light-accessible site Arg513 and the CTT in dCry are shown in Fig. 1A. We reproduced the results using a trypsin:dCry molar ratio of 1:50 with an incubation time of 1 min. Light-induced fragments of (1-513) and a’ (1-503) were observed in lane L-L but not in lane D-L (Fig. 1B and Supplementary Fig. 1). However, in practice, the protocol is difficult to perform, and false-positive or false-negative results are often obtained. This was due to the strong susceptibilities of Lys289 and Arg298 both in the dark and light conditions, as most full-length dCry was cleaved at these positions to form fragments b (1-289) and c (299-542) before the formation of the light-induced fragments a and a’, making the fragments a and a’ too faint to be easily detected.

In contrast, when a high concentration of trypsin (1:1 molar ratio) with incubation time of 1 min was applied, cleavage at Arg430 was observed upon light exposure, resulting in a ~15 kDa fragment d (299-430), accompanied with a decrease in fragment c (299-542) (Fig. 1B, lane L-H, compared with lane D-H; and Supplementary Figs. 1 and 2). A survey of the dCry structure revealed that Arg430 is located in the vicinity of CTT, and is accessible by trypsin only after CTT is released from its binding pocket (Fig. 1A). Notably, the light susceptibility of Arg430 has already been reported in previous works8,22, but cleavage at this site had not been used as a main criterion for CTT release.

The relationship between the extent of photoreduction and the cleavage of Arg430, i.e., CTT release, was investigated. Blue light illumination of dCry at pH 7.5 quickly converted the ox state of FAD into a large fraction of the asq state and a small fraction of the nsq state (Fig. 1C and E, Supplementary Table 1). Under the same conditions, trypsin cleavage of dCry resulted in increasing intensity ratios of the fragment d to the sum of fragments c and d, which represented CTT release and corresponded well with the asq formation (Fig. 1D, E, and Supplementary Fig. 3). Nevertheless, incubation in the dark for 100 min under aerobic conditions led to full reoxidation of the FAD cofactor, and the trypsin susceptibility of Arg430 was also lost (Fig. 1D; Supplementary Fig. 4). These results indicate that the photoreduction and CTT release of dCry are correlated and fundamentally reversible.

CTT release was suppressed in the nsq state but promoted in both the asq and hq states

Phylogenetic analyses revealed that animal cryptochromes, including dCry, are derived from eukaryotic 6-4 photolyases1,2,3. The structures of dCry and Drosophila 6-4 photolyase (Dm6-4) share many features but also present some important differences (Fig. 2A)9,10,29. Both dCry and Dm6-4 contain four conserved tryptophan residues, through which electrons can be transferred from the solvent to the FAD cofactor during photoreduction30,31. While photolyases can be photoreduced and protonated to nsq and/or hq states30,32, dCry can only be photoreduced into a large fraction of the asq state6,7,8 and a small fraction of the nsq state. This discrepancy has been attributed to the different residues facing N5 of FAD in each protein: a cysteine in dCry (Cys416) but an asparagine in Dm6-4 (Asn403) (Fig. 2A)6,7. Nevertheless, Leu401 and Leu405 in the vicinity of Cys416 provide the solvent barrier that suppresses protonation of the FAD cofactor in dCry25,33, but the homologous Glu392 in Dm6-4 may transfer protons from the solvent to FAD through Asn403 (Fig. 2A). According to the structural comparison of dCry and Dm6-4, two mutants, C416N and L405E/C416N, were constructed.

Fig. 2: The photoreduction, oxidation, and CTT release properties of the C416N mutant.
figure 2

A Comparison of the structures of dCry (left panel) and Drosophila 6-4 photolyase (Dm6-4, right panel). Both dCry and Dm6-4 contain four conserved tryptophan residues (gray sticks), through which electrons can be transferred from the solvent to the FAD cofactor (yellow stick) during photoreduction. The residue facing N5 of FAD is a cysteine (Cys416) in dCry but is an asparagine (Asn403) in Dm6-4. Nevertheless, Leu401 and Leu405 provide solvent barriers that suppress the protonation of FAD in dCry. The homologous positions in Dm6-4 are Gln388 and Glu392, through which protons may be transferred to FAD to form nsq and/or hq states. Some important sites in dCry and Dm6-4 are depicted in stick representations. A possible alternative proton transfer pathway in dCry is also shown (top left corner in the left panel). B Short-term illumination of the C416N mutant converted the oxidized FAD cofactor into the asq state (left panel). The asq state FAD in C416N was quickly oxidized to the ox state (kasqox ~ 0.21 min−1) and protonated to the nsq state (kpt ~ 0.41 min−1, Supplementary Table 1) (middle and right panels. ▪, the ox state; ●, the asq state; ▲, the nsq state; ▼, the hq state; and ♦, the total of the redox states; one representative result is shown). The red line in the right panel represents the fitted curve of overall asq decomposition (~0.62 min−1). C Illumination of the C416N mutant for several minutes converted most of the FAD cofactor into the nsq state (left panel). The nsq state of FAD in C416N was very stable toward oxidation, which needs more than 1000 min to be fully oxidized (middle and right panels. The symbols are the same as above). The red line in the right panel represents the fitted curve of nsq oxidation (knsqox ~ 0.001 min−1, Supplementary Table 1). D Long-term illumination led to hq state formation in the C416N mutant (left panel). The oxidation of hq-state FAD in C416N was relatively fast that was completed within 80 min (middle and right panels. The symbols are the same as above). The red line in the right panel represents the fitted curve of hq oxidation (khqox ~ 0.03 min−1, Supplementary Table 1). E Proteolysis assay of C416N to detect CTT release during photoreduction (for 3600 s) and redocking after oxidation for 1000 min. F Kinetics of the fractions of various redox states (the symbols are the same as above), and CTT release of C416N (orange stars, represented by the intensity ratio of fragment d to the sum of fragments c and d) during photoreduction. The red lines represent the fitted curves of asq protonation (kpt ~ 6.5  × 10−3 s−1, i.e., ~0.39 min−1) and hq formation (kpr2 ~ 1.3 × 10-4 s−1, Supplementary Table 1). The degree of CTT release was in line with asq and hq formation.

Full size image

Consistent with previous studies34, the asq state formed in the C416N mutant after short-term illumination (Fig. 2B, left panel). The decomposition of the asq state was very fast, a fraction (~34%) of which was fully oxidized, and the remaining fraction (~66%) was converted into the nsq state (Fig. 2B, middle and right panels). By multiple linear regression and exponential fitting, it was calculated that the overall decomposition rate of asq was ~0.62 min−1 (at pH 7.5). In particular, the oxidation rate of asq (kasqox) was ~0.21 min−1, which was comparable to that of wild-type dCry at pH 7.5 (~0.17 min−1, Supplementary Fig. 4C and Table 1). The protonation rate of asq (kpt) was calculated to be ~0.41 min−1 (Supplementary Table 1). The fast oxidation rate suggested that the CTT of the C416N mutant would be released as well as that of wild-type dCry in the asq state. Indeed, limited trypsin proteolysis of C416N after short-term light exposure resulted in a band corresponding to fragment d (Fig. 2E and Supplementary Fig. 5). However, the band was relatively faint, which might be due to the fast protonation of asq to nsq during operation.

Illumination of the C416N mutant for several minutes resulted in the transition of most FAD into the nsq state (Fig. 2C, left panel). The nsq state of C416N was extremely stable in the dark, the oxidation rate of which (knsqox) was only ~0.001 min−1 (Fig. 2C, middle and right panels; and Supplementary Table 1). Trypsin proteolysis of C416N in the nsq state produced only weak bands corresponding to fragment d (Fig. 2E and Supplementary Fig. 5), indicating that this state of C416N mimicked the dark state. From the photoreduction kinetics, it was calculated that the protonation rate kpt of asq in C416N under light was ~0.39 min−1 (Fig. 2F and Supplementary Table 1), which was approximately the same as the value determined in the dark. This indicated that the protonation of asq in C416N was light independent. Further extending the illumination time to hours led to the formation of some hq state in the C416N mutant (Fig. 2D, left panel; the molar extinction coefficient spectra of C416N in the ox, asq, nsq, and hq states are shown in Supplementary Fig. 6A). Intriguingly, the hq state C416N was oxidized much faster than the nsq state (khqox ~ 0.03 min−1; Fig. 2D, middle and right panels; Supplementary Table 1). Nevertheless, trypsin proteolysis of C416N resulted in intense fragment d bands and weakened fragment c bands, which were in line with hq formation (Fig. 2E, F, and Supplementary Fig. 5).

As previously reported25, short-term illumination of the L405E/C416N mutant nearly completely converted the FAD cofactor into the nsq state, confirming that protons could be efficiently transferred to FAD through the glutamic acid residue at site 405 during photoreduction (Fig. 3A, left panel). The oxidation of L405E/C416N in the nsq state was even slower (knsqox ~ 3.3 × 10-4 min−1; Fig. 3A, middle and right panels; and Supplementary Table 1). Trypsin proteolysis provided only very faint bands corresponding to fragment d in this state, corroborating that L405E/C416N in the nsq state was an indeed dark-state mimic25 (Fig. 3C and Supplementary Fig. 7). Long-term illumination of L405E/C416N for hours also produced some hq state (Fig. 3B, left panel; the molar extinction coefficient spectra of L405E/C416N in the ox, nsq, and hq states are shown in Supplementary Fig. 6B). The oxidation rate of the hq state L405E/C416N was significantly elevated (khqox ~ 0.08 min−1; Fig. 3B, middle and right panels; and Supplementary Table 1). The L405E/C416N mutant in the hq state also formed intense fragment d bands in proteolysis experiments (Fig. 3C, D, and Supplementary Fig. 7).

Fig. 3: The photoreduction, oxidation, and CTT release properties of the L405E/C416N mutant.
figure 3

A Short-term illumination of the L405E/C416N mutant converted the oxidized FAD cofactor into the nsq state (left panel). The nsq state of FAD in L405E/C416N was extremely stable toward oxidation, which needs more than 2000 min to be fully oxidized (middle and right panels. ▪, the ox state; ●, the asq state; ▲, the nsq state; ▼, the hq state; and ♦, the total of the redox states; one representative result is shown). The red line in the right panel represents the fitted curve of nsq oxidation (knsqox ~ 3.3 × 10−4 min−1, Supplementary Table 1). B Long-term illumination led to hq state formation in the L405E/C416N mutant (left panel). The oxidation of hq state FAD in L405E/C416N was relatively fast that was completed within 60 min (middle and right panels. The symbols are the same as above). The red line in the right panel represents the fitted curve of hq oxidation (khqox ~ 0.08 min-1, Supplementary Table 1). C Proteolysis assay of L405E/C416N to detect CTT release during photoreduction (for 3600 s) and redocking after oxidation for 2000 min (D) Kinetics of the fractions of various redox states (the symbols are the same as above), and CTT release of L405E/C416N (orange stars, represented by the intensity ratio of fragment d to the sum of fragments c and d) during photoreduction. The red line represents the fitted curve of hq formation (kpr2 ~ 2.2 × 10−3 s−1, Supplementary Table 1). The degree of CTT release was in line with hq formation.

Full size image

Light-induced degradation experiments in Drosophila S2 cells revealed that more than half of the dCry protein was degraded upon 40 min of illumination (Supplementary Fig. 8A). Light also induced degradation of the C416N mutant but in a much slower manner (Supplementary Fig. 8B). Interestingly, the L405E/C416N mutant was degraded slightly faster than C416N upon illumination (Supplementary Fig. 8C), which might be due to a higher photoreduction rate to the hq state (kpr2) in the L405E/C416N mutant (~2.2 × 10−3 s−1, Fig. 3D and Supplementary Table 1) than in the C416N mutant (~1.3 × 10−4 s−1, Fig. 2F and Supplementary Table 1).

These results demonstrated that the hq state, which is also anionic as the asq state, could promote CTT release in dCry mutants (although wild-type dCry could not be photoreduced to the hq state). The asq state C416N mutant could also be protonated, albeit at a slower rate, suggesting the existence of an alternative proton transfer pathway other than site 405 in dCry (Fig. 2A, left panel).

pH-dependent nsq formation and its role in the activation of dCry

A unique property of dCry is the asq formation upon light illumination, which is proposed to be the signaling state of dCry6,7,8. Recently, it was reported that during photoreduction of dCry, a small fraction of nsq was formed26,28. This phenomenon was also observed in our studies. At pH 6.5, photoreduction of dCry exhibited prominent absorption peaks at 600 nm and 647 nm (Fig. 4A, left panel), which are characteristic of nsq. The nsq fractions were significantly greater than those formed during photoreduction at pH 7.5 (Fig. 1C). Assuming that the absorption spectrum of each state of FAD in dCry was pH independent, the molar extinction coefficient spectra of dCry in the asq and nsq states were obtained by manipulation of the photoreduction spectra at different pH values (Supplementary Fig. 9A). The photoreduced dCry protein at pH 6.5 resulted in weakened fragment d bands after trypsin proteolysis, accompanied with less asq formation (Fig. 4A, middle and right panels, and Supplementary Fig. 10), indicating that CTT release was inhibited at this pH.

Fig. 4: pH-dependent photoreduction, CTT release, and oxidation of wild-type dCry.
figure 4

A Illumination of dCry at pH 6.5 resulted in the formation of greater amounts of nsq (left panel), accompanied by lower fractions of asq and inhibition of CTT release (middle and right panels. ▪, the ox state; ●, the asq state; ▲, the nsq state; ▼, the total of the redox states; the hq state was not detected; one representative result is shown. Orange stars, the degree of CTT release represented by the intensity ratio of fragment d to the sum of fragments c and d). The red line represents the fitted curve of asq formation during photoreduction at pH 6.5 (kpr1 ~ 0.06 s−1, Supplementary Table 1). B Illumination of dCry at pH 9.0 led to the formation of nearly pure asq (left panel), and CTT release increased (middle and right panels. The symbols are the same as above). The red line represents the fitted curve of asq formation during photoreduction at pH 9.0 (kpr1 ~ 0.07 s−1, Supplementary Table 1). C pH-dependent nsq fractions of the total semiquinone (asq + nsq) formed during photoreduction. D Relationship between the photoreduction rate (kpr1) and pH. E pH-dependent oxidation rates of asq (kasqox) in dCry after photoreduction.

Full size image

In contrast, illumination of dCry at pH 9.0 produced a nearly pure asq state because protonation barely occurred at this pH (Fig. 4B, left panel). To our surprise, trypsin cleavage of photoreduced dCry at pH 9.0 resulted in heavier fragment d bands and mutually weakened fragment c bands (Fig. 4B, middle and right panels, and Supplementary Fig. 11), revealing that CTT release of dCry was enhanced at this pH. This finding challenged previous statements that higher pH inhibited the signaling-state formation of dCry26,27,28, and ruled out the possibility that the protonation of His378 is necessary for CTT release26. Nevertheless, it unified the conclusion from the nsq-forming mutants that nsq formation inhibited CTT release due to the lack of a negative charge, and that asq formation promotes CTT release25. Examination of the photoreduction data at different pH values revealed that nsq formation in dCry was strongly pH dependent (Fig. 4C). Nevertheless, the nsq fractions of the total semiquinone (asq + nsq) were approximately constant during photoreduction at each pH (Supplementary Table 2), implying that there was fast equilibration between the asq and nsq states. The protons needed to form nsq must be transferred through an unidentified proton transfer pathway in dCry. In contrast to previous studies26,28, we did not observe a significant pH dependence of the photoreduction rates in dCry (Fig. 4D). Instead, the oxidation rates of photoreduced dCry were found to be strongly pH dependent: lower pH values led to higher oxidation rates accompanied with faster CTT redocking, and vice versa (Fig. 4E, Supplementary Fig. 4, Supplementary Fig. 9B–H and Supplementary Table 1). This may be due to the instability of the nsq state in the wild-type dCry milieu. Considering the fast equilibration between nsq and asq in dCry, the formation of a proper fraction of nsq would increase the oxidation rates of the total semiquinone, resulting in faster recovery when dCry is returned to the dark.

The protonation of FAD was suppressed in the mutants in CTT

Another unique feature of dCry is that it contains CTT located at the binding pocket in the resting state9,10, and CTT release upon illumination activates dCry20,21,22,23. CTT has a conserved FFW motif (534-536), with Phe534 at the center of the binding pocket9,10. It is well known that CTT plays key roles in activation of dCry35,36,37 and an important work on the activation mechanism of dCry has actually been performed with CTT mutants (W536A and W536F)8. However, systematic spectroscopic and structural studies on the mutants in CTT are lacking. To gain a deeper understanding of the roles of CTT in dCry activation, the mutant that deleted CTT (dCT, 1-516), and F534A, F535A, and W536A were constructed and investigated.

The dCT mutant could be readily photoreduced into nearly pure asq state similar to that of wild-type dCry at pH 9.0 (Fig. 5A, left panel). It was not surprising that the dCT mutant presented constant trypsin susceptibility at Arg430, which produced fragment d both in the dark and under light (Fig. 5A, middle and right panels, and Supplementary Fig. 12). The dCT mutant was unstable in S2 cells irrespective of light presence (Supplementary Fig. 13A). However, the photoreduction of dCT was essentially pH independent, and the formation of nsq was minimized at all pH values tested (pH 6.5–9.0, Supplementary Fig. 14). Nevertheless, the oxidation rates of dCT after photoreduction were also pH independent (~0.06 min−1, Supplementary Fig. 14 and Supplementary Table 1), which was equal to that of wild-type dCry at pH 9.0 (~0.06 min−1, Fig. 4E, Supplementary Fig. 9, and Supplementary Table 1).

Fig. 5: The photoreduction and CTT release properties of the mutants in CTT.
figure 5

A Illumination of the mutant that deleted CTT (dCT, 1-516) quickly photoreduced the oxidized FAD cofactor into a nearly pure asq state (left panel). Owing to the lack of CTT, the dCT mutant showed constant cleavage at Arg430 [middle and right panels. ▪, the ox state; ●, the asq state; ▲, the nsq state; ▼, the total of the redox states; the hq state was not detected; one representative result is shown. Orange stars, the intensity ratio of fragment d to the sum of fragments c’ (299-516) and d]. The red line represents the fitted curve of asq formation in dCT during photoreduction (kpr1 ~ 0.03 s−1, Supplementary Table 1). B Illumination of F534A also produced nearly pure asq state (left panel). The CTT release of F534A was impaired because of the central position of Phe534 (middle and right panels. The symbols are the same as above, except that orange stars represent the degree of CTT release according to the intensity ratio of fragment d to the sum of fragments c and d). The red line represents the fitted curve of asq formation in F534A during photoreduction (kpr1 ~ 0.06 s−1, Supplementary Table 1). C Illumination of W536A gave similar spectral results to those of dCT and F534A (left panel). However, the CTT release of W536A was enhanced (middle and right panels. The symbols are the same as above). The red line represents the fitted curve of asq formation at W536A during photoreduction (kpr1 ~ 0.07 s−1, Supplementary Table 1). D Structural representations of the middle tunnel that would form after CTT release (left panel), and the side tunnel through which protons might be transferred to FAD (middle and right panels). In the right panel, the side tunnel was depicted by CAVER 3.035, the entrance of which is formed by the phosphate binding loop8,12 and α-helix 8 (α8). The side tunnel deeply accesses the FAD cofactor, and joins the middle tunnel (with a narrow bottleneck near the docked CTT) in a highly curved manner. E Molecular dynamics simulations revealed that α8 of the mutants in CTT (dCT, F534A, F535A, and W536A) shifted inward, making the entrances of the side tunnels narrower. The tunnels of the CTT mutants have lower curvatures than that of wild-type dCry, with different exits that correspond well with their mutation sites.

Full size image

The photoreduction behaviors of F534A, F535A, and W536A were similar to that of the dCT mutant (Fig. 5B, left panel; Supplementary Fig. 15A; and Fig. 5C, left panel), revealing that the protonation of asq was suppressed in all the mutants in CTT. Owing to the central position and potentially important interactions of Phe534, CTT release of the F534A mutant upon illumination was impaired compared with that of wild-type dCry (Fig. 5B, middle and right panels, and Supplementary Fig. 16). However, CTT release was enhanced in the photoreduced F535A and W536A mutants (Supplementary Fig. 15B and 15C; and Fig. 5C, middle and right panels, and Supplementary Fig. 17). The enhanced light-sensing ability of W536A was in line with the previous work, in which a W536 mutant was used in kinetic experiments because it produced sharper light/dark band differences in the trypsin proteolysis assay8. These phenomena could be easily explained by the fact that the asq fractions are predominant in the CTT mutants during photoreduction, which is more competent to induce CTT release. The light-induced degradation of F534A in S2 cells was comparable to that of wild-type dCry (Supplementary Fig. 13B), and that of F535A and W536A was slightly faster than that of wild-type dCry (Supplementary Fig. 13C, D). The oxidation rates of all these mutants after photoreduction were lower than that of wild-type dCry under neutral conditions (Supplementary Fig. 15D, Supplementary Fig. 18, and Supplementary Table 1).

CTT release upon illumination in dCry and its derived proteins, or CTT deletion in the dCT mutant, would result in opening of the binding pocket and forming a middle tunnel (Fig. 5D, left panel), through which electron acceptors such as oxygen could be accessible by the FAD cofactor. However, the data of the mutants in CTT revealed that protons could not be transferred to FAD through this tunnel. After careful examination of the structure of dCry, a side tunnel was found, the entrance of which is formed by the phosphate binding loop8,12 and α-helix 8 (α8) (Fig. 5D, middle and right panels). The side tunnel of wild-type dCry was depicted by CAVER 3.038, which accesses the FAD cofactor deeply, and joins the middle tunnel in a highly curved manner (Fig. 5D, right panel). Protons might be transferred through the side tunnel to the isoalloxazine ring of FAD to protonate it. Molecular dynamics simulations revealed that α8 in all of the mutants in CTT (dCT, F534A, F535A, and W536A) shifted inward, making the entrances of the side tunnels to be narrower (Fig. 5E, Supplementary Fig. 19, and Supplementary Table 3). And the throughputs of the tunnels were predicted to be lower than that of wild-type dCry (Supplementary Table 3). These factors could suppress proton transfer and the protonation of FAD. Nevertheless, the curvatures of the tunnels of the CTT mutants were lower than that of wild-type dCry (Fig. 5E and Supplementary Table 3). The protons might leak to solution through the less curved tunnels, which would also prevent the protonation of FAD.

His378 is a proton interceptor that adjusts the protonation of FAD in dCry

Dissection of the dCry structure revealed that the end of the side tunnel joins the binding pocket and accesses the FAD cofactor in a side direction (Fig. 6A). His378 is located at the end of the side tunnel as well as at the center of the binding pocket. The homologous residue in 6-4 photolyases is also vital for their repair activity11,12,13,14,15. Owing to the central position and potentially important roles of His378, this residue has been intensively studied, although the results are somewhat controversial26,27,28. To investigate the roles of His378 in proton transfer and the activation of dCry, three His378 mutants (H378A, H378K, and H378D) were generated.

Fig. 6: His378 acts as a proton interceptor during photoreduction.
figure 6

A Dissection of the dCry structure shows the details of the side tunnel, which joins the binding pocket, and accesses the FAD cofactor in a side direction. His378 is located at the end of the side tunnel, in the vicinity of Phe534 and the FAD cofactor. B Illumination of the H378A mutant slowly converted the oxidized FAD cofactor into asq and nsq. The fractions of nsq were greater than those of wild-type dCry under neutral conditions. Prolonged illumination converted some nsq into hq (left panel). Proteolysis analysis of H378A showed that CTT binding was less tight in the dark and that CTT release was less sufficient in light than in wild-type dCry, but still increased in line with asq and hq formation (middle and right panels. ▪, the ox state; ●, the asq state; ▲, the nsq state; ▼, the hq state; and ♦, the total of the redox states; one representative result is shown. Orange stars, the degree of CTT release represented by the intensity ratio of fragment d to the sum of fragments c and d). C Photoreduction of the H378K mutant was even slower, but significant amounts of nsq still formed (left panel). The proteolysis of H378K was similar to that of H378A (middle and right panels. The symbols are the same as above). D Illumination of the H378D mutant resulted in rapid formation of nearly pure asq (left panel). Proteolysis of H378D showed that the CTT remained in a half-release state both in the dark and in the light (middle and right panels. The symbols are the same as above). The red line represents the fitted curve of asq formation in H378D during photoreduction (kpr1 ~ 0.05 s−1, Supplementary Table 1).

Full size image

The photoreduction of the H378A mutant was relatively slow, which is consistent with the findings of a previous study28. Nevertheless, it was clearly observed that nsq was formed together with asq during photoreduction, and the nsq fractions of the total semiquinone of the initially photoreduced H378A were even greater than those of wild-type dCry (Supplementary Table 4). Prolonged illumination of H378A converted some nsq into hq (Fig. 6B, left panel). These results ruled out the possibility that His378 is a proton donor of FAD because the alanine residue cannot provide a proton28. Proteolysis experiments of H378A showed that CTT binding was less tight in the dark and that CTT release was less sufficient in light than that in wild-type dCry (Fig. 6B, middle and right panels, and Supplementary Fig. 20). These results confirmed that His378 is important for both CTT docking in the dark and CTT release in light27. The photoreduction of H378K was even slower than that of H378A, but the formation of nsq still occurred (Fig. 6C, left panel; and Supplementary Table 4). Trypsin proteolysis of H378K gave similar results to those of H378A (Fig. 6C, middle and right panels, and Supplementary Fig. 21).

The situation of the H378D mutant was different and had never been reported previously. The photoreduction of H378D was as fast as that of wild-type dCry, and a nearly pure asq state was formed (Fig. 6D, left panel; Supplementary Table 4). Protons passing through the side tunnel might be attracted by the negatively charged aspartic acid residue before accessing the FAD cofactor; therefore, nsq formation was severely inhibited in the H378D mutant. The imidazole side chain of the histidine residue could also attract protons at neutral pH. Thus, only a small fraction of nsq is formed during photoreduction of wild-type dCry. Without a proton interceptor such as a histidine or an aspartic acid residue at the end of the side tunnel, a greater nsq fraction would be formed during photoreduction, just as the cases of the H378A and H378K mutants. Trypsin proteolysis of H378D showed that its CTT was maintained in a half-release state both in the dark and light (Fig. 6D, middle and right panels, and Supplementary Fig. 22). The negatively charged aspartic acid residue might mimic the anionic asq or hq state of FAD, leading to worse docking of CTT in the H378D mutant in the dark.

The light-induced degradation of H378A in S2 cells was similar to that of wild-type dCry (Supplementary Fig. 23A). However, the degradation of H378K and H378D was impaired compared with that of wild-type dCry (Supplementary Fig. 23B,C), which was in line with the previous study26. This finding suggested that His378 might also participate in the recognition of the E3 ubiquitin ligases that regulate dCry degradation.

Discussion

In the circadian clock system of Drosophila, dCry plays an important role in entraining the clock with light17,18,19. Phylogenetic analyses revealed that dCry was derived from eukaryotic 6-4 photolyases1,2,3. In contrast to photolyases, dCry contains the unique CTT that docks in the substrate binding-like pocket in the dark9,10. CTT release is a crucial step for dCry to transduce the light signal to its interaction partners, such as Tim and Jet20,21,22,23. However, the mechanism of CTT release is not fully understood. Another difference between dCry and photolyases is that photolyases can be photoreduced to the nsq and hq states30, but most FAD in dCry is only able to be photoreduced to the asq state6,7,8.

Cys416 in dCry faces N5 of FAD, the steric hindrance of which hampers the protonation of asq to produce nsq. The homologous position in photolyases is generally an asparagine residue, which can form a hydrogen bond with nsq and hq to stabilize them (Fig. 2A)39,40,41. The C416N mutant can be photoreduced into three states: asq, nsq, and hq. Like wild-type dCry, short-term illumination converted the oxidized FAD of C416N into the asq state. The oxidation rate of asq in C416N was as fast as that of wild-type dCry (~0.21 min−1 vs. ~0.17 min−1, Supplementary Table 1), implying that CTT could also be released in this state. The asq in C416N also underwent light-independent protonation to produce the nsq state. The nsq in C416N was quite stable, which was oxidized ~170-fold slower than wild-type dCry in the asq state (~0.001 min−1 vs. ~0.17 min−1, Supplementary Table 1), indicating that the electron acceptor-accessibility of the nsq state FAD in C416N was suppressed. Proteolysis assays also revealed that CTT was barely released in this state, confirming that the nsq state is a dark-state mimic25. The long-term illumination of C416N resulted in the formation of some hq state. The oxidation of hq in C416N was relatively fast (~0.03 min−1, Supplementary Table 1), and proteolysis assays and light-induced degradation experiments in Drosophila S2 cells revealed that CTT was efficiently released in this state. The results demonstrated that the asq and hq states, both of which contain a negative charge, are capable of promoting CTT release, whereas the ox and nsq states, both of which are neutral, cannot trigger CTT release. Therefore, the negative charge on the FAD cofactor is crucial for CTT release. The center residue in CTT, Phe534, contains an electron-rich phenyl side chain, which is docked in the binding pocket in the resting state. When the FAD cofactor becomes negatively charged during photoreduction, it creates an electrostatic repelling force on the phenyl side chain of Phe534, inducing CTT ejection from the binding pocket (Fig. 7A). Nevertheless, the ox and nsq states of FAD are in planar conformations, and the asq and hq states of FAD are in bending conformations42. These conformational changes in different states of FAD may also contribute to triggering CTT release (Fig. 7A).

Fig. 7: The working models of dCry.
figure 7

A The negative charge and conformational changes of the FAD cofactor regulate CTT release. Wild-type dCry can be photoreduced to a large fraction of the asq state (FAD●−) and a small fraction of the nsq state (FADH●) under neutral conditions. The hq state (FADH−) can only be obtained in some dCry mutants (such as C416N and L405E/C416N) after prolonged illumination. The asq state and the hq state FAD trigger CTT release. B His378 of dCry adjusts the protonation degree of FAD during photoreduction to achieve a balance between light sensitivity and dark recovery. In the L405E/C416N mutant, the FAD cofactor can be efficiently protonated through the Glu405 site during short-term photoreduction, resulting in a nearly pure nsq state. Both of the light sensitivity and dark recovery of L405E/C416N are inhibited (in the Asn416 background). In the F535A or W536A mutant, the FAD protonation is suppressed during photoreduction, resulting in a nearly pure asq state. The light sensitivity of F535A or W536A is increased, but their dark recovery is impaired.

Full size image

In addition to Cys416, Leu401 and Leu405 establish solvent barriers that also ensure the arrest of the asq state of dCry during photoreduction (Fig. 2A)25,33. The L405E/C416N mutant provides a proton transferer at site 405, which aids in rapid protonation of the asq state to the nsq state25. In the nsq state, CTT release in the L405E/C416N mutant was also suppressed. Nevertheless, long-term illumination of L405E/C416N produced the hq state, which also promoted CTT release. However, how did the C416N mutant obtain protons without the proton transferer at site 405? A counterexample was the D393N mutant of Chlamydomonas reinhardtii cryptochrome CPH1 (Asp393 is the homologous position of Cys416 in dCry), which could hardly be photoreduced and protonated to the nsq state43 because solvent barriers such as Leu378 and Trp382 also exist in CPH1. These findings imply that an alternative but less efficient proton transfer pathway is present in dCry.

Recent studies revealed that a small fraction of nsq forms together with asq during the photoreduction of dCry and that nsq formation is pH dependent26,28. This phenomenon also suggested an alternative proton transfer pathway in dCry. Interestingly, CTT release was suppressed during photoreduction at lower pH, accompanied with more nsq formation. In contrast, CTT release was promoted at higher pH values, which was in line with greater asq formation (Fig. 4). Although these results are compatible with the findings from the C416N and L405E/C416N mutants, they contrast with previous studies that reported that protonation of a key residue (His378) was important for CTT release26 and that the formation of the signaling state required the uptake of a proton27. Nevertheless, if nsq formation does not benefit CTT release, why does it form during photoreduction of dCry under neutral conditions? In-depth analysis revealed that the nsq fractions of the total semiquinone (asq + nsq) were nearly constant during photoreduction at each pH value, indicating that the two semiquinone states had a fast equilibration between them. And it was observed that the greater fractions of nsq formed, the faster oxidation of total semiquinone was. At pH 9.0, the oxidation rate of photoreduced dCry was only approximately one third of that at pH 7.5 (~0.06 min−1 vs. ~0.17 min−1, Supplementary Table 1). Therefore, the formation of a small fraction of nsq during photoreduction may aid in the fast recovery of dCry if it returns to the dark (Fig. 7B). The Cys416 residue should be responsible for the destabilization of nsq in wild-type dCry. Moreover, how does wild-type dCry in the nsq state flavin access electron acceptors if CTT release has been inhibited? A plausible interpretation is that the fast equilibration between nsq and asq would lead to oscillation of CTT between the docking and release states, which would suppress trypsin digestion but would not prevent electron acceptors from entering.

The mutants in CTT (dCT, F534A, F535A, and W536A) formed a nearly pure asq state during photoreduction. Nevertheless, all the mutants in CTT presented lower oxidation rates than wild-type dCry did under neutral conditions. This phenomenon seems to be counterintuitive for the dCT mutant because it had generally been considered that without the CTT, the mutant should be oxidized faster. However, this is reasonable because the asq state is relatively more stable than nsq in the dCry milieu, regardless of whether the CTT is released or deleted. The CTT release or deletion opens the binding pocket to form the middle tunnel. Our results suggested that electron acceptors could be accessed by FAD through the middle tunnel, but the protons could not. Examination of the dCry structure revealed a side tunnel structure, which is not present in eukaryotic 6-4 photolyases (Figs. 5D and 6A). Nevertheless, the entrance of the side tunnel was predicted to be narrower and less curved in the mutants in CTT by molecular dynamics simulations (Fig. 5E and Supplementary Table 3). This might be due to the lack of proper support by CTT, causing an inward shift in α8. Protons might be transferred to FAD through the side tunnel in wild-type dCry and the CTT-intact mutants but were blocked in the mutants in CTT. Owing to more fractions of asq formed in the F535A and W536A mutants, they exhibited increased CTT release upon illumination. However, these mutants could not recover quickly to the dark state due to their lower oxidation rates. This might be the reason that natural selection did not favor these mutants as photoreceptors.

His378 is located at the center of the binding pocket, as well as the end of the side tunnel (Fig. 6A). The H378A and H378K mutants produced greater fractions of nsq than wild-type dCry did during photoreduction (Fig. 6B, C, and Supplementary Table 4). Nevertheless, the H378D mutant was converted into a nearly pure asq state under illumination (Fig. 6D). The methyl side chain of the alanine residue cannot accept protons. The side chain of the lysine residue is already protonated under neutral conditions. These two residues had little effect on proton transfer through the side tunnel to FAD; therefore, higher fractions of nsq formed during photoreduction of the H378A and H378K mutants. The carboxyl side chain of aspartic acid and the imidazole side chain of histidine may attract protons before being transferred to FAD and suppress the protonation of large amounts of asq in the H378D mutant and wild-type dCry. In fact, this interpretation involves minor issues that the aspartic acid residue would not accept more protons than would the histidine residue under neutral conditions, considering the pKa values of the side chains (~3.9 for aspartic acid vs. ~6.0 for histidine); however, it was actually observed that the H378D mutant had lower fractions of nsq than wild-type dCry did during photoreduction. Perhaps the negatively charged aspartic acid residue might change the paths of the transferred protons via electrostatic forces, preventing them from reaching the FAD cofactor but not needing to associate with the protons. Interestingly, the CTT of H378D remained in a half-release state both in the dark and in the light. The negatively charged aspartic acid residue at site 378 might also trigger CTT release, as the anionic asq- or hq-state FAD does, although less efficiently. These results revealed that the residue at site 378 could adjust the protonation degree of FAD during photoreduction (Fig. 7B).

In summary, the negative charge on FAD is crucial for CTT release. The side tunnel was found in dCry, through which protons could be transferred to a fraction of asq, protonating it into nsq. Proper fractions of nsq formed in dCry are important for fast recovery when it is returned to the dark (Fig. 7). Further efforts are required to fully elucidate the activation mechanism of dCry, which will not only help elucidate the circadian clock system, but also provide ideas for the design of light-gated molecular machines.

Methods

Gene cloning, protein expression and purification

The cloning, expression and purification procedures for dCry and its mutants were adapted from a previous report44. The gene encoding dCry was chemically synthesized and cloned into pET28a plasmid. The point mutants of dCry were obtained by site-directed PCR mutagenesis. The dCT mutant was constructed with the CTT (517-542) deleted. The recombinant plasmids of dCry and its mutants were transformed into competent Rosetta(DE3) cells. The cells were cultured in LB medium containing 50 μg ml−1 kanamycin at 37 °C until A600 = 0.6 ~ 0.8. Protein expression was induced with 0.5 mM Isopropyl β-D-thiogalactoside and incubating at 20 °C for 20 h in the dark. The cells were harvested and disrupted by sonication. The proteins were purified by using a Ni-IDA Sepharose 6FF column (BBI) and a Superdex 200 10/300 GL column (GE Healthcare). The purified proteins were stored in protein buffer containing 50 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM EDTA and 10% glycerol unless otherwise specified. For the pH-dependent experiments, buffer with Tris-HCl was used between pH 7.0 and pH 9.0, and buffer with MES was used when the pH was less than 7.0.

Absorption spectroscopic analyses

The absorption spectra of the protein samples were recorded on a UV-1800 spectrophotometer equipped with a TCC-240A temperature controller (Shimadzu). A semi-micro quartz cuvette of 200 μl was used. The scanning range was 200–750 nm. The temperature was maintained at 16 ± 0.5 °C during recording.

To determine the photoreduction processes, the protein samples were supplemented with 10 mM DTT and injected into the cuvette under aerobic conditions. The samples were kept in an ice water jacket and illuminated with a blue LED (λmax ~ 440 nm). The irradiance of the LED was measured by an OHSP-350S spectral irradiance colorimeter (Hopoocolor). The absorption spectra of the samples were recorded quickly at intervals. To monitor the oxidation processes, the cuvette containing the photoreduced sample was exposed to air and placed into the spectrophotometer in the dark. The spectra were automatically recorded at set intervals. Each kinetics experiment was performed at least three times.

The molar extinction coefficient spectra of dCry and its mutants were obtained via a method adapted from previously described approaches32,43. The proteins of wild-type dCry and some mutants were purified in a pure ox state. The molar extinction coefficient spectra of these proteins in the ox state were obtained by recording the spectra before and after addition of SDS (0.5%). The free FADox concentration was determined by using the extinction coefficient ε450 = 11300 M−1 cm−1. A two-state transition stage was judged by existing isosbestic points and reciprocal changes at the characteristic absorption peaks of the two states. To obtain the molar extinction coefficient spectrum of one state, at least two spectra during the two-state transition stage were selected. Then a fraction (z) of one spectrum was subtracted from another spectrum, so that the characteristic fine structures of the other state were no longer visible. The spectrum was then divided by (1 − z) and the molar concentration to obtain the final spectrum. If the transitions occurred in more than two states, more spectra were selected. The fine structures of one state were eliminated by the former method, and then another state, until the molar extinction coefficient spectrum of the last pure state was obtained. The photoreduction and oxidation of wild-type dCry at a certain pH value were pseudo-two-state transitions, because the asq and nsq states changed proportionally (Supplementary Table 2). The molar extinction coefficient spectra of dCry in the asq or the nsq state could only be obtained from the photoreduction and oxidation spectra at different pH values via the former method. To obtain the kinetics of various redox states during photoreduction or oxidation of dCry and its mutants, spectra between 350 and 750 nm were fitted by the multiple linear regression method using the molar extinction coefficient spectra as independent data. The kinetics data were fitted by an exponential model or sequential model to obtain the reaction rate constants.

Trypsin digestion assays

The protein samples were kept in the dark, illuminated, or illuminated and oxidized as described above. To perform the limited trypsin proteolysis assays, the protein samples were withdrawn and mixed immediately with a 1:50 or 1:1 molar ratio of recombinant trypsin (Yaxin Biotechnology) and incubated at 25 °C for 1 min. The reactions were terminated by mixing with 2xSDS-PAGE loading buffer and heating at 100 °C for 10 minutes. After centrifugation, the samples were analyzed via 12.5% SDS-PAGE. The gels were stained with Coomassie brilliant blue R-250. The densities of the bands were analyzed via ImageJ (Fuji). The degree of CTT release was represented by the intensity ratio of fragment d (299-430) to the sum of fragments c (299-542) and d. For each protein, at least three independent experiments were performed.

Mass spectrometric analyses

To determine the trypsin cleavage sites, the bands corresponding to fragments b and c in the dark and light samples and those corresponding to fragment d in the light samples were cut from Coomassie brilliant blue-stained gels. The gel strips were decolored and washed, then reduced with 10 mM DTT and alkylated with 10 mM iodoacetamide (IAA). The samples were further digested with trypsin, and the resulting peptides were extracted and analyzed via LC‒MS/MS via a Q Exactive mass spectrometer coupled with an Easy nLC 1200 nanoflow UHPLC system (Thermo Fisher).

Cell culture stability assays

The genes encoding wild-type dCry and its mutants with an N-terminal 3xV5 tag were subcloned and inserted into the pAc5.1 vector (Invitrogen). Drosophila S2 cells were cultured at 27 °C in Sf-900 II insect cell culture medium (Thermo Fisher) supplemented with 10% heat-inactivated FBS (Sigma) and penicillin/streptomycin. The plasmids were electrotransfected into S2 cells via a Gene Pulser X-cell electroporation system (Bio-Rad). The transfected S2 cells were cultured in the dark for 48 h. Then, the cells were divided into aliquots, kept in the dark or illuminated with a 440 nm LED for different time periods. After that, the cells were harvested (~2.5 × 104 cells per lane), and the protein levels were determined by Western blotting using an anti-V5 antibody (HT401, ProteinFind). The actin levels in the samples were measured as internal controls by using an anti-actin antibody (AC038, Abclonal). The densities of the bands were analyzed via ImageJ (Fuji).

Molecular dynamics simulations

The initial structure of dCry was obtained from the Protein Data Bank (PDB ID: 4GU5)10. The dCT mutant was generated with the CTT (517-542) deleted. The mutants in CTT (F534A, F535A, and W536A) were generated by mutating the target amino acid residues via the ChimeraX software (UCSF). From the multiple rotamer conformations generated, the optimal mutant conformation exhibiting no steric clashes with adjacent residues was selected. All MD simulations were performed via the GPU-accelerated version of GROMACS 2024.345. The interatomic interactions were modeled via the Amber force field. Atom charges for the FAD cofactor were obtained through RESP charge fitting on the basis of electrostatic potentials calculated at the B3LYP/6-311 G (d, p) level of theory. The protein-FAD complex was solvated in a cubic water box containing TIP3P water molecules, with periodic boundary conditions applied in three dimensions. Sodium and chloride ions were added to neutralize the system and achieve a physiological NaCl concentration of 150 mM. The system was subjected to initial energy minimization to relieve any steric clashes or unfavorable interactions. Energy minimization followed by equilibration in the NVT and NPT ensembles for 100 ps each was performed for each system. Subsequently, production MD simulations were carried out at a constant temperature of 300 K in the NPT ensemble via Langevin dynamics with a collision frequency of 2.0 ps−1. The pressure was maintained at 1 atm using the Parrinello-Rahman barostat with a time constant of 2.0 ps. The time step for the simulations was set to 2 fs, and Lincs constraints were applied to the hydrogen bonds. Electrostatic interactions were calculated via the particle mesh Ewald (PME) method with a cutoff distance of 1.0 nm. Each simulation was run for a total duration of 200 ns. Root-mean-square deviation (RMSD) values of the simulated structures with respect to the initial structures were used to validate the simulations that should reach a steady phase in the last 20 ns (Supplementary Fig. 19B). Root-mean-square fluctuation (RMSF) values were calculated to identify the significant structural changes in the simulated structures (Supplementary Fig. 19C). The distances between Arg237 (CZ) in α8 and Lys264 (CE) in the phosphate binding loop of the simulated structures were measured to detect the movements of α8 (Supplementary Fig. 19D).

The existence of possible tunnels for proton transfer in dCRY and its mutants was determined by analyzing the crystal structure (PDB ID: 4GU5), as well as the snapshots of the MD simulations during the last 20 ns via CAVER 3.038. The starting points were set as the center coordinates of three residues in the phosphate binding loop and α8 (Val240, Gln254, and Asp261). All other parameters were set as default values.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.