Role of Two Histidines in the (6-4) Photolyase Reaction*

The reaction mechanism of Xenopus (6-4) photolyase was investigated using several mutant enzymes. In the active site, which is homologous between the cis,syn cyclobutane pyrimidine dimer and (6-4) photolyases, four amino acid residues that are specific to (6-4) photolyase, Gln 288 , His 354 , Leu 355 , and His 358 , and two conserved tryptophans, Trp 291 and Trp 398 , were substituted with alanine. Only the L355A mutant had a lower affinity for the substrate, which suggested a hydrophobic interaction with the (6-4) photoproduct. Both the H354A and H358A mutations resulted in an almost complete loss of the repair activity, although the Trp 291 and Trp 398 mutants retained some activity. Taking the pH profile of the (6-4) photolyase reaction into consideration with this observation, we propose a mechanism in which these histidines catalyze the formation of the four-mem-bered ring intermediate in the repair process of this enzyme. When deuterium oxide was used as a solvent, the repair activity was decreased. The proton

of the 3Ј-base is transferred to the 5Ј-pyrimidine ring in the  photoproduct. Although both the CPD and the (6-4) photoproduct are formed by UV at adjacent pyrimidine sites in DNA strands, the structural features of the  photoproduct are quite different from those of the CPD. Therefore, the influence of these two photolesions on replication also differs (4).
Photoproducts must be repaired in cells to maintain genetic integrity. Photolyase is a unique DNA repair enzyme that eliminates UV-derived photoproducts by electron transfer from the catalytic cofactor, FAD, using light in the near UV/blue region (5,6). Thus far, two types of photolyases have been isolated. One is a DNA photolyase specific for the CPD (referred to as CPD photolyase in this report) and the other is a (6-4) photolyase specific for the (6-4) photoproduct (7,8). Photoreactivation of the CPD has been found in a wide range of organisms. The Escherichia coli and Anacystis nidulans enzymes were characterized in detail, and their structural information is available (9,10). The homologous genes have been isolated from many sources (5,7,11,12). In contrast, the (6-4) photolyase activity has been detected in some higher eukaryotes. The cDNAs of (6-4) photolyase have been cloned from Drosophila melanogaster (13), Xenopus laevis (14), Danio rerio (15), and Arabidopsis thaliana (16). Interestingly, the  photolyases are quite similar to the CPD photolyases, although (6-4) photolyase does not bind or repair the CPD (13). The  photolyase has been considered to have evolutionarily diverted from the CPD photolyase (17). In fact, we have shown that the (6-4) photolyases possess features quite similar with those of the CPD photolyases (14,18). On the other hand, unlike the CPD, Kim et al. (19) rejected the possibility that a single electron transfer to the  photoproduct could result in the restoration of the original DNA bases. It was suggested that a function additional to the common photolyase activities is required for the  photolyase to restore the original bases.
It was proposed that the (6-4) photolyase repairs the (6-4) photoproduct first by converting it to a four-membered ring and then by splitting the two pyrimidines, as does the CPD photolyase (oxetane intermediate model, see Ref. 19). In previous reports, we showed by high pressure liquid chromatography analyses that both the T(6-4)T and T(6-4)C photoproducts are restored completely to the original bases, despite the difference in the functional group at the C-4 of the 3Ј-base (18,20). These results strongly suggested that the transfer of the -NH 2 or -OH group of the  photoproduct is an intramolecular reaction within the substrate. Additionally, Zhao et al. (21) supported the oxetane intermediate model by using chemically synthesized analogues of the (6-4) photoproduct as the substrate. However, the (6-4) photoproducts neither form this intermediate nor turn into the original bases spontaneously. Theoretical calculations on the base portion of T(6-4)T and its oxetane isomer estimated the gap to be about 14.5-16.5 kcal/ mol, suggesting that perturbation of the (6-4)/oxetane equilibrium is unlikely to be a feature of the photoenzymic repair mechanism as the computed value exceeds the likely difference in binding energy between the two species (22). Although the cycloreversion process by electron transfer has been studied intensively (18,19,21,23,24), the mechanism of this intermediate formation by the (6-4) photolyase is still unclear.
In this paper, we identified the residues crucial for the photoreactivation by the  photolyase. Due to the high similarity in the primary structures at the active sites, the reason for the marked difference in the substrate specificity of each type of photolyase has been unclear. A structural model of the  photolyase in complex with the (6-4) photoproduct was constructed to discuss the substrate recognition and the reaction mechanism. We propose that the two conserved histidine residues, which are not present in the CPD photolyases, act as an acid and a base to catalyze the formation of the four-membered ring intermediate.

EXPERIMENTAL PROCEDURES
Binding and Repair Assays-Xenopus (6-4) photolyase and its substrate, the 49-mer oligonucleotide containing a single (6-4) photoproduct, were prepared for in vitro assays, as described previously (14,18,25). The sequence of the oligonucleotide was 5Ј-d(AGCTACCATGCCT-GCACGAATTAAGCAATTCGTAATCATGGTCATAGCT)-3Ј. The (6-4) photoproduct derived from two thymidines was incorporated into the TTAA sequence context, which is the recognition site of the MseI restriction endonuclease. The oligonucleotide was labeled isotopically at the 5Ј terminus, annealed with the complementary strand, and then used as the substrate.
To detect the photolyase activity in vitro, restriction site restoration assays were performed, as described previously (18). The substrate (1 nM) was treated with the photolyases in 30 l of a 50 mM buffered solution containing 1 mM dithiothreitol under daylight irradiation by a fluorescent lamp. The mixture was diluted to 150 l with sterile water, treated with the same volume of phenol/chloroform (50:50, v/v), and precipitated with ethanol. The pellet was dissolved in 10 l of sterile water, digested with MseI (New England Biolabs, Inc.) overnight, subjected to electrophoresis on a 10% polyacrylamide gel containing 7 M urea, and visualized by autoradiography. The cleaved and uncleaved fractions of the DNA fragment were quantified with a Fuji bioimaging analyzer (BAS-2000, Fuji Photo Film). For the isotope effect, D 2 O, KOD, and (DOCD 2 ) 2 CDOD (glycerol-d 8 ), which were 99 atom %, were purchased from Cambridge Isotopes. The salt used in the preparation of buffers was initially dissolved in D 2 O and was concentrated in several freeze-dry cycles prior to the addition of glycerol-d 8 . The substrate was diluted in buffer prepared with D 2 O and glycerol-d 8 . The enzymes were dissolved in buffer containing D 2 O and glycerol-d 8 and were concentrated with a Centriprep-50 apparatus (Millipore Corporation), and their absorption and fluorescence spectra were found to be virtually identical to those of the enzymes dissolved in H 2 O.
Substrate binding was detected with electrophoretic mobility shift assays (EMSAs), as described previously (18). The described substrate (50 pM) was incubated with 0.05-1 nM of the photolyase in 10 l of 100 mM Tris-Cl (pH 8.0) on ice for 30 min in the dark, and the resultant mixture was applied to a 5% nondenaturing polyacrylamide gel. The electrophoresis was performed under yellow light to prevent photoreactivation, and then the gel was autoradiographed for visual inspection. The bound and unbound fractions of the DNA fragment were quantified on the bioimaging analyzer.
The in vivo effect of the photolyase was detected by the survival rate of UV-irradiated E. coli strain SY2 (uvrA Ϫ , recA Ϫ , phr Ϫ ) containing the plasmid pRT2, which carries the wild type E. coli CPD photolyase gene, as described previously (14). E. coli SY2 (pRT2) was transformed with a series of mutant constructs derived from pGEX-Xl64PR. The transformed cells were grown in LB medium containing 150 mg/liter ampicillin at 37°C overnight. The expression of the photolyase gene was induced by adding isopropyl-␤-D-thiogalactopyranoside to the medium to a final concentration of 0.1 mM and shaking for an additional 1 h at 37°C. Aliquots of cells were plated on LB agar and irradiated first by UV at an intensity of either 0.3 or 0.6 J/m 2 and subsequently with daylight fluorescence lamps for 1 h. The plates were incubated in the dark at 37°C overnight, and surviving colonies were counted the next day. All experiments, except the photoreactivation treatment, were performed under yellow light.
Construction of Xenopus  Photolyase Mutants-Alanine-substituted mutations were introduced at Gln 288 , Trp 291 , His 354 , Leu 355 , His 358 , and Trp 398 in Xenopus (6-4) photolyase, respectively. Based on the described plasmid, pGEX-Xl64PR (14), we constructed mutated plasmids using the Site-directed Mutagenesis System Mutan®-Super Express Km (Takara). The BglII/SphI fragment digested from pGEX-Xl64PR was subcloned into the BamHI/SphI sites of pKF18. The resultant plasmid was named pKF-Xl64 and was used as the template. The underlines indicate the modified sequences. According to the manufacturer's instructions, mutations were introduced by polymerase chain reaction using the described oligonucleotides and pKF-Xl64. Mutated fragments derived from pKF-Xl64 were inserted at the proper position of pGEX-Xl64PR with the appropriate restriction endonucleases, NdeI and MunI. In all cases, the nucleotide sequence was determined to ensure that no additional mutation was introduced. The commercially available E. coli strain JM 109 (Takara) and the strain SY2 (pRT2) were transformed with the resultant pGEX-XL64PR plasmids carrying a single mutation for the protein production and the in vivo assays, respectively. Glutathione S-transferase (GST)-fused proteins were used for all experiments, except that the GST was removed for the substrate binding assays. The wild type enzyme and the mutants were purified as described previously (14,18). The production of the photolyases was checked by both SDS-polyacrylamide gel electrophoresis (4 -20% gradient) and an enzymatic assay using the GST Detection Module (Amersham Pharmacia Biotech), which is based on the detection of the fused GST activity. According to the manufacturer's instructions, 2 l of the cell extracts from the transformed E. coli were added to 150 l of a solution containing 1 mM 1-chloro-2,4-dinitrobenzene and 1 mM glutathione, and the change in absorbance at 340 nm was monitored on a UV spectrophotometer.
Protein Modeling-The amino acid sequences of E. coli CPD photolyase and Xenopus (6-4) photolyase were aligned by using the commercially available software, GENETYX-MAC Version 8.0 (Software Development Co., Ltd.). A tertiary model of Xenopus (6-4) photolyase from Glu 221 to Pro 420 was constructed by comparative modeling based on the structure of E. coli CPD photolyase (Protein Data Bank code, 1dnp) (9), using our original programs as follows: a loop search method for the backbone structure (26), a dead-end elimination method for the side chain structure (27), and a conformation energy minimization method for structure refinement (28), using the AMBER force field (29). The coordinates of the free (6-4) photoproduct of thymidylyl(3Ј 3 5Ј)-thymidine, derived from an NMR study (30), were kindly provided by Dr. Byong-Seok Choi. The (6-4) photoproduct was manually docked to the active site of Xenopus (6-4) photolyase, using the molecular graphics program Insight II (Molecular Simulation Inc.), and the energy of the complex conformation was further minimized (28).

RESULTS
The pH Profile of Xenopus  Photolyase-The substrate specificity and the apparent reaction of the (6-4) photolyases are quite different from those of the CPD photolyases, although both photolyases repair the damage by photoinduced electron transfer from reduced FAD to the substrate (18). Fig. 2 shows the pH profile of Xenopus (6-4) photolyase. A single T(6-4)T photoproduct in a 49-mer oligonucleotide was photoreactivated with Xenopus (6-4) photolyase at various pH values. The activity was detected with the described coupled enzyme assay (18), which is based on the restoration of the restriction enzyme sensitivity of a recognition sequence containing the photoproduct by photoreactivation. Repair by Xenopus (6-4) photolyase was most efficient at high pH (pH 8.5-9.0), although flavin is most efficiently excited at a lower pH. The maximal activity occurred at pH 8.5 and was 2.4-fold higher than that at pH 6.5, whereas a remarkable decrease of catalysis was observed upon approaching pH 6.0.
Ala Substitution-To determine the catalytic residues essential for the formation of the proposed four-membered ring intermediate, a site-specific mutation analysis was carried out. Although the crystal structures of the CPD photolyases did not contain the substrate (9, 10), the mutation analysis and other experiments located the interaction sites with DNA in the C-terminal helical domain (31)(32)(33)(34)(35)(36). The C-terminal halves of the (6-4) and CPD photolyases are highly conserved. The homology of Xenopus (6-4) photolyase to E. coli CPD photolyase increases to 35.8% in this region, whereas the (6-4) photolyases share 26.5-28.5% homology with E. coli CPD photolyase in the full-length comparison (Fig. 3A). This region of E. coli CPD photolyase also contains most of the amino acids involved in the binding of the flavin cofactor, and the FAD-binding sites are well conserved in the (6-4) photolyase (Fig. 3B). Eleven of 14 amino acids involved in FAD binding in Xenopus (6-4) photolyase are homologous, and 8 of them are identical in Xenopus (6-4) photolyase. The flavin chromophore is deeply buried in the center of the C-terminal helical domain (9, 10). This conservation indicates the structural similarity between the CPD and (6-4) photolyases. Although the proposed DNA binding residues are not very well conserved, most of the aromatic residues are conserved (Fig. 3C). Particularly, Trp 291 and Trp 398 of the Xenopus enzyme, equivalent to Trp 277 and Trp 384 of E. coli photolyase, respectively, are conserved in all of the (6-4) photolyases. Trp 277 and Trp 384 of E. coli CPD photolyase are supposed to be located between the cofactor and the sub-strate (31,32,34) and to mediate the electron transfer (31,34). The conservation of the FAD-binding sites and the -systems of the aromatic residues is consistent with the fact that the (6-4) photolyase transmits a photoinduced electron from the reduced FAD to the substrate, as the CPD photolyase does. In contrast, the marked difference between the CPD and (6-4) photolyases lies in the hydrophilic residues. The substitutions from E. coli CPD photolyase to Xenopus (6-4) photolyase include Glu 274 3 Gln 288 , Asn 341 3 His 354 , Arg 342 3 Leu 355 , and Met 345 3 His 358 . Remarkably, the hydrophilicity is inverted at Leu 355 and His 358 in the Xenopus enzyme from the counterparts in the bacterial CPD photolyase, although these residues are conserved well in the (6-4) photolyases. We thought that these residues could confer the substrate specificity to the (6-4) photolyases. Based upon the observation described above, we selected Gln 288 , Trp 291 , His 354 , Leu 355 , His 358 , and Trp 398 of Xenopus (6-4) photolyase as targets for alanine mutagenesis.
Mutation Effects on Substrate and Cofactor Binding-The wild type and alanine-substituted (6-4) photolyases were prepared as fusion proteins with GST, as reported previously (14), and were characterized by comparison of the absorption spectra. Production of the photolyases was monitored by the GST activity, as described under "Experimental Procedures." The alanine-substituted photolyases were produced at a level sim- The 49-mer oligonucleotide duplex (0.5 nM) was incubated on ice with the (6-4) photolyase (10 nM) at various pH values with irradiation from a fluorescent lamp for 1 h. The enzyme was then extracted with organic solvents, and the duplex was digested with the MseI restriction endonuclease. The (6-4) photoproduct was within the TTAA sequence context, and the repair of the photoproduct enabled MseI to cut the duplex between the two thymidines, generating the 21-mer product. The optimal pH range for the electron donation by FAD is shaded. ilar to that of the wild type. For all photolyases, the flavin chromophore was identified by the absorption spectrum. The enzymes exhibited a prominent peak at 450 nm with a shoulder at 470 nm, which is typical of the fully oxidized forms of flavin. These results suggested that the overall structures of the fused enzymes were intact and that the flavin binding was not perturbed in the mutants. Due to their availability, GST-fused (6-4) photolyases were used for the repair assays.
The (6-4) photolyase binds the substrate independent of light and initiates the repair only upon absorbing a near UV-visible photon, as CPD photolyase does. To characterize the substrate binding of the (6-4) photolyase mutants, EMSA was performed in the dark by using the 49-mer oligonucleotide containing a single T(6-4)T photoproduct as a substrate. All of the obtained photolyases displayed an affinity for the substrate. We generated the binding isotherms of the mutants under the conditions of constant substrate and various enzyme concentrations (Fig.  4). All mutants except Leu 355 showed the affinity that is comparable to wild type. In contrast, apparently, the mutation of Leu 355 resulted in a large decrease in the affinity to the substrate.
Analysis of DNA Repair by Ala-substituted Mutants-To study the catalysis by the mutants, we used two assays. First, we tested the photoreactivating activity of the photolyases in vitro. Fig. 5A shows the results of the restriction site susceptibility assays. The repair activity was not reduced by the mutation at either Gln 288 or Leu 355 , whereas H358A showed a decrease in the cleaved fraction (lane 7). In addition, as revealed in the figure, the H354A mutant lost the catalytic activity (lane 5). The ratio of DNA repaired by the bulky residue mutants, histidines and tryptophans, was quantified at different times (Fig. 5B). Removal of the -system of Trp 291 or Trp 398 resulted in a reduction of the activity to 2(6-4)0% of the wild type, whereas the mutation of the histidine residues had a particularly large influence on the catalysis. The H354A mutant lost the repair activity completely, and the rate of repair by the H358A mutant was only 1.5% that by the wild type enzyme.
Since E. coli does not photoreverse the (6-4) photoproduct, the production of the (6-4) photolyase in cells would result in an increased resistance to UV light in the presence of the photoreactivating light (14). By using this feature, we investigated the in vivo activity of the photolyases (Fig. 5C). E. coli SY2 (uvrA Ϫ , recA Ϫ , phr Ϫ ) cells containing pRT2, which carries the wild type E. coli CPD photolyase gene, were transformed with a series of mutant vectors containing the (6-4) photolyase gene and were irradiated with UV light at either 0.3 or 0.6 J/m 2 . At these UV doses, the main photoproduct is the CPD (70 -80%), whereas the (6-4) photoproduct forms 20 -30% of the total photoproducts. Although irradiation at 0.3 J/m 2 did not cause any apparent difference between the mutants (data not shown), the results at 0.6 J/m 2 reflected those of the in vitro photoreactivation experiments, as shown in Fig. 5C. At 0.6 J/m 2 , the survival rate of control cells transformed with the pGEX-4T-2 construct was 2.1 ϫ 10 Ϫ2 . The W291A and W398A mutants exhibited rather weak resistance as compared with the wild type. The increase from the negative control was 5.0 Ϯ 0.4-and 5.4 Ϯ 1.0-fold, respectively, whereas the cells transformed with the wild type enzyme increased the UV resistance by 10-fold. In   FIG. 4. Binding characteristics of the (6-4) photolyase mutants. The 49-mer oligonucleotide duplex containing the (6-4) photoproduct (50 pM) was incubated with the photolyases (25-400 pM) on ice for 30 min. The DNA-protein complexes were separated on a 5% nondenaturing polyacrylamide gel and were quantified on a bioimaging analyzer.

FIG. 5. Repair assays for the (6-4) photolyase mutants. A and B,
repair activity of the (6-4) photolyase mutants in vitro. The repair activity was detected by the restriction site restoration assay. The 49-mer oligonucleotide duplex (0.5 nM) was incubated on ice with the (6-4) photolyase (10 nM) in a solution (30 l) containing 50 mM Tris-HCl (pH 8.0) and 1 mM dithiothreitol with irradiation by a fluorescent lamp, and was treated with the MseI restriction endonuclease at 37°C overnight. The cleaved 21-mer oligonucleotide was separated from the original oligonucleotide on a 10% polyacrylamide gel containing 7 M urea and was visualized by autoradiography. The wild type (6-4) photolyase was diluted 10-fold in A. C, photoreactivating activity of Xenopus (6-4) photolyase mutants produced in E. coli. The in vivo effect on photoreactivation of the photolyase mutants was determined by the survival rate of UV-irradiated E. coli SY32 (pRT2) cells transformed with a series of vectors carrying the mutant photolyase gene. For the wild type enzyme and several mutants, an increase of UV resistance was detected by comparison of the survival rate at 0.6 J/m 2 with that of the control cells. The survival rate of the control cells transformed with the pGEX-4T-2 construct was 2.1 ϫ 10 Ϫ2 at this UV dose.
contrast, the survival rates of the cells transformed with pGEX-H354A and pGEX-H358A were 0.9 -3.5 ϫ 10 Ϫ2 and 1.8 -3.3 ϫ 10 Ϫ2 , respectively. The H354A and H358A mutants showed no increase in the UV resistance by photoreactivation (1.1 Ϯ 0.6-and 1.2 Ϯ 0.4-fold against the control cells transformed with pGEX-4T-2). Apparently, the phenotype of the mutants that lack the histidine residue in the active site was different from that of W291A or W398A. From these results, we concluded that both His 354 and His 358 are essential for the catalytic activity of (6-4) photolyase.
Isotope Effect-The results of the repair assays revealed an unusual pH dependence of the (6-4) photolyase reaction (Fig.   2). Unlike other amino acids, the pK a value of histidine is variable due to the properties of the imidazole. Hence, we reasoned that these properties of the histidine could be used for the formation of the oxetane or azetidine intermediate, and we assumed that either His 354 or His 358 could act as a general acid, and the other as a general base, to facilitate the formation of the proposed four-membered ring. To obtain data supporting this proposition, the isotope effect was tested. We photoreactivated the T(6-4)T-containing 49-mer with the photolyase in heavy water. If proton transfer is involved in the (6-4) photolyase reaction, then the isotope effect by deuterium would cause a difference in the rate of the photoreactivation. As shown in Fig. 6, the deuterated reaction resulted in a 44% decrease in the rate of repair. DISCUSSION To gain insight into the interactions between the (6-4) photolyase and its substrate, we constructed a structural model of Xenopus (6-4) photolyase over the region comprising amino acids 221-420 by using the crystal structure of E. coli photolyase (9) as a starting point (Fig. 7). Since the C-terminal halves of the proteins, which interact with the FAD and contain the putative substrate-binding site, are highly conserved between the (6-4) and CPD photolyases, as shown in Fig. 3A, this model can be used reliably to discuss the substrate recognition and the catalytic reaction. The cavity leading to the chromophore in the E. coli CPD photolyase is conserved in the (6-4) photolyase (Fig. 7). Compared with the cavity in the CPD photolyase, the path to the chromophore in the (6-4) photolyase is somewhat narrow due to the bulky residues of His 354 and His 358 , as shown in Fig. 7A. All of the residues shown in Fig. 3C are located on From the crystal structure of E. coli CPD photolyase, Park et al. (9) predicted that the interaction for its substrate recognition occurs near the rim of the cavity. At this site, the residues are hydrophobic on one side and polar on the other, and this asymmetry fits well with the hydrophobic cyclobutane ring and the polar opposite edges. This substrate-binding site has been verified by site-directed mutagenesis (31,32), docking simulations (33)(34)(35), and atomic force microscopy (36). We put the substrate, i.e. a dinucleoside monophosphate containing the  photoproduct, at the corresponding site in the (6-4) photolyase with the same orientation of the 5Ј and 3Ј components. The internucleoside phosphate was located near the conserved basic residue (Lys 420 ). In the final model after energy minimization, His 354 , Leu 355 , and His 358 are close to the photoproduct, as shown in Fig. 7A. When the photoproduct is incorporated into a DNA strand, both the 5Ј and 3Ј extensions can interact with the enzyme along the trace of the positive electrostatic potential shown in Fig. 7B. In this study, we showed that Leu 355 is crucial for the substrate binding (Fig. 4). From these results and the docking model, it can be concluded that the hydrophobic interaction between the 3Ј-pyrimidone ring and Leu 355 plays an important role in substrate binding. It is noteworthy that Arg 342 of E. coli CPD photolyase, which is the counterpart of Leu 355 of Xenopus  photolyase, is also required for the substrate binding (32), although leucine is quite different from arginine in terms of polarity.
Since the FAD is buried deeply in the center of the C-terminal helical domain, the conservation of the FAD-binding site suggests a structural similarity between the CPD and  photolyases. In addition, not only the -systems of Trp 277 and Trp 384 at the proposed active site of E. coli CPD photolyase (equivalent to Trp 291 and Trp 398 in Xenopus (6-4) photolyase) but also most of the aromatic residues are well conserved at the region surrounding the FAD, as shown in Fig. 3A. Considering the similarity in the electron transfer between the CPD and (6-4) photolyases (18), the aromatic residues in the (6-4) photolyase probably play a role similar to that of the corresponding residues in the CPD photolyase. Like the CPD photolyase, which exhibits high activity over the pH range from 5.2 to 7.9 (37), the FAD in the (6-4) photolyase is surrounded by these aromatic residues and lies in the hydrophobic core (Fig. 7), which enables efficient electron transfer free from environmental influences. However, the (6-4) photolyase exhibited a remarkable pH dependence with maximal activity at pH 8.5, as shown in Fig. 2. The inhibitory effect at neutral pH observed for the (6-4) photolyase should result from a catalytic function of this enzyme, which is absent for CPD photolyases.
We noticed that protonation of histidine would agree well with the inhibitory effect observed for the activity of (6-4) photolyase. Since the function of histidine is uniquely influenced by pH, as compared with other amino acids, it is possible that the histidine(s) at the active site caused the unusual pH dependence of the (6-4) photolyase. Indeed, the mutants in this study showed that the two histidines at the active site are required for the catalysis to repair the (6-4) photoproduct, in contrast to the CPD photolyases (Fig. 5). We proposed previously that the (6-4) photolyase donates an electron in a manner similar to that of the CPD photolyase (18), and the oxetane intermediate model proposed by Kim et al. (19) is most likely. According to their model, the (6-4) photolyase repairs the (6-4) photoproduct first by converting it to an intermediate with the four-membered ring, oxetane for TT or azetidine for TC, which is also formed in the photochemical reaction to the (6-4) photoproduct, as shown in Fig. 1 (3). Zhao et al. (21) supported this model by using synthetic analogs of the (6-4) photoproduct as substrates. Laser flash photolysis, fluorescence quenching, and product analysis experiments also supported the oxetane reversal mechanism (24). On the other hand, Heelis and Liu (22) suggested that perturbation of the T(6-4)T/oxetane equilibrium is unlikely to be a feature of the photoenzymic repair mechanism as the value estimated for the free energy difference between the hydrated T(6-4)T and oxetane species (ϳ14.5-16.5 kcal/mol) exceeds the likely difference in binding energy between the two species. To repair (6-4) photoproducts efficiently, the (6-4) photolyases should enhance the rate of formation of the four-membered ring intermediates. Zhao et al. Asp 388 , respectively, in Xenopus (6-4) photolyase) as candidates for the catalytic residues. However, two Asps are also conserved in the CPD photolyase, and in our model, these Asps are located on the opposite side to Gln 288 in Xenopus (6-4) photolyase. In addition, the alanine substitution at Gln 288 had no effect on the (6-4) photolyase activity, as shown in Fig. 5, A and C. Therefore, we propose that His 354 and His 358 , which are located at the active site and are crucial for the activity, catalyze the intermediate formation. Fig. 8 shows a possible mechanism for the formation of the four-membered ring catalyzed by two histidines. In the docking model, the locations of His 354 and His 358 would allow hydrogen bonding to the N-3 of the 3Јpyrimidone and the hydroxyl group on the 5Ј-pyrimidine, respectively. Therefore, His 358 abstracts a proton from the hydroxyl group or the protonated amino group at C-5 of the 5Ј-base, and at the same time, His 354 protonates the N-3 of the 3Ј-base to generate a highly electrophilic iminium ion. A nucleophilic attack to the cationic 3Ј-C-4 by the oxygen anion or the nitrogen lone pair results in the formation of the oxetane or azetidine intermediate. The inhibition at neutral pH and the isotope effect observed in this study can be explained by this mechanism. At neutral pH, His 358 , which triggers the repair reaction of the (6-4) photolyase, is protonated and fails to abstract the proton. The isotope effect is attributed to the difference in the atomic weights of hydrogen and deuterium in this proton transfer process. The observation of this isotope effect shows that the formation of the oxetane intermediate is the rate-limiting step in the reaction mechanism shown in Fig. 8.
Our novel finding is that the two histidines, which are not present in CPD photolyases, are required for the photoreversal of the (6-4) photoproduct. From the pH profile, the mutation analyses, and the structure model, we propose that His 354 and His 358 not only give the substrate specificity for the (6-4) photolyases but also act as an acid and a base, respectively, to form the four-membered ring intermediate.