Investigation of the Cyclobutane Pyrimidine Dimer (CPD) Photolyase DNA Recognition Mechanism by NMR Analyses*

The cyclobutane pyrimidine dimer (CPD) is one of the major forms of DNA damage caused by irradiation with ultraviolet (UV) light. CPD photolyases recognize and repair UV-damaged DNA. The DNA recognition mechanism of the CPD photolyase has remained obscure because of a lack of structural information about DNA-CPD photolyase complexes. In order to elucidate the CPD photolyase DNA binding mode, we performed NMR analyses of the DNA-CPD photolyase complex. Based upon results from 31 P NMR measurements, in combination with site-di-rected mutagenesis, we have demonstrated the orientation way as previously reported (17). The CPD photolyases from T. thermophilus HB27 and its mutants were prepared, according to an established procedure (18). The enzymes labeled with 2 H were also obtained in the same way, except that the E. coli strains expressing them were cultured for 24 h in M9 medium containing D -glucose- 2 H 7 and 2 H 2 O instead of D -glucose and H 2 O, respectively. Surface Plasmon Resonance (SPR) Measurements— The binding constants of the CPD photolyase and its mutants to d(GTAT[CPD]TATG),

P NMR measurements, in combination with site-directed mutagenesis, we have demonstrated the orientation of CPD-containing single-stranded DNA (ssDNA) on the CPD photolyase. In addition, chemical shift perturbation analyses, using stable isotope-labeled DNA, revealed that the CPD is buried in a cavity within CPD photolyase. Finally, NMR analyses of a double-stranded DNA (dsDNA)-CPD photolyase complex indicated that the CPD is flipped out of the dsDNA by the enzyme, to gain access to the active site.
Irradiation of DNA with ultraviolet (UV) light produces various damaged bases, leading to cellular transformation and cell death (1)(2)(3). One of the major products formed by UV irradiation is the cyclobutane pyrimidine dimer (CPD) 1 (Fig. 1), from the photo [2 ϩ 2] cycloaddition of the 5,6-double bond of two adjacent pyrimidine nucleotides. Organisms have a variety of enzymes playing crucial roles in repair-systems for damaged DNA, including nucleoside excision repair systems and photoreactivation (4). CPD photolyases, which function as members of the DNA repair systems, restore the CPD to normal pyrimidines by photoreactivation. To elucidate the mechanism of DNA repair by CPD photolyase, various biological and spectroscopic experiments have been performed. This enzyme has a flavin adenine dinucleotide (FAD) as an essential cofactor (5).
The FAD is excited by light, and then transfers one electron to the CPD bases (6 -8). After the electron transfer, the cyclobutane ring splits and then one electron is transferred back to the FAD (9,10). In order to understand the mechanism based on structural analyses, the crystal structures of the CPD photolyases from Escherichia coli, Anacystis nidulans, and Thermus thermophilus have been solved without the substrates, i.e. CPD-containing DNAs (11)(12)(13). The x-ray studies revealed that the enzymes share a similar global fold, which consists of an ␣/␤ domain and a helical domain. The helical domain is composed of clusters I and II, and a cavity is formed between the clusters, where the FAD is deeply buried. It has been suggested that the cavity is used for the CPD binding, because the asymmetric polarity of the cavity fits well with that of the CPD (11)(12)(13)(14)(15). Based upon the crystal structures without the substrates, two research groups have proposed computer models of the DNA-CPD photolyase complex, in which the relative orientations of the DNA chain are different from each other (11)(12)(13)(14)(15). However, no crystallographic or NMR structure information on the complexes is presently available.
Here, we report NMR analyses of the DNA recognition mechanism by T. thermophilus CPD photolyase with a molecular weight of 48 kDa. Initially, we analyzed DNA complexed with the CPD photolyase and its mutants by 31 P NMR. These analyses revealed the orientation of the bound DNA relative to the enzyme. Next, more detailed analyses of the CPD lesion, utilizing stable isotope-labeled DNA, revealed that the CPD base is buried in the cavity in the DNA-CPD photolyase complex, and that the CPD is flipped out of the DNA helix in the complex. We discuss the correlation between the structural information and activity on the basis of the present NMR data.

EXPERIMENTAL PROCEDURES
Preparation of Oligonucleotides and CPD Photolyase from T. thermophilus-Oligonucleotides labeled with 13 C and 15 N were prepared according to a published method (16). The other oligonucleotides without the CPD lesion were purchased from ESPEC Oligo Service Corp. d(CGCAAT[CPD]TAAGCCG) was chemically synthesized as previously reported (17). The other oligonucleotides containing CPD lesions were prepared by UV irradiation (0.9 J/cm 2 ) with a FUNA-UV-LINKER FS-800 (Funakoshi) and were purified in the same way as previously reported (17). The CPD photolyases from T. thermophilus HB27 and its mutants were prepared, according to an established procedure (18). The enzymes labeled with 2 H were also obtained in the same way, except that the E. coli strains expressing them were cultured for 24 h in M9 medium containing D-glucose-2 H 7 and 2 H 2 O instead of D-glucose and H 2 O, respectively.
Surface Plasmon Resonance (SPR) Measurements-The binding constants of the CPD photolyase and its mutants to d(GTAT[CPD]TATG), * This work was supported by grants from the Japan New Energy and Industrial Technology Development Organization (NEDO) and the Ministry of Economy, Trade, and Industry (METI) and by a Japan Society for the Promotion of Science fellowship (to T. T.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
containing a biotin at the 5Ј-end, were determined using a BIAcore 1000 instrument (BIAcore AB) in the same way as reported previously (17), except that the enzyme concentration range was 10 -100 nM.
NMR Measurements-Four different buffers were used for the NMR measurements. Buffer I: 5 mM sodium phosphate buffer, pH 7.3, containing 200 mM NaCl, and 3 mM NaN 3 in 2 H 2 O. Buffer II: Buffer I plus 10 mM dithiothreitol, 1 mM EDTA, and 5% ethylene glycol, with the 2 H 2 O reduced to 10%. Buffer III: Buffer I plus 10 mM dithiothreitol-2 H 10 and 5% ethylene glycol-2 H 6 with a 100 mM concentration of NaCl. Buffer IV: the pH of Buffer III was changed to 6.5, and the 2 H 2 O was reduced to 10%. The experiments to assign the 31 P resonances of the oligodeoxynucleotides in the absence of the enzymes were carried out at a concentration of 0.5-2 mM in 420 l of Buffer I, in the same way as reported previously (19). Buffer II was used for one-dimensional 31 P NMR measurements in the absence and presence of the enzymes and for 31 P-31 P exchange spectroscopy (EXSY) experiments. The experiments described above were performed on a Bruker DRX400 spectrometer, and those described below were monitored on a Bruker DRX600 spectrometer. One-dimensional 1 H, 1 H-13 C heteronuclear single quantum coherence (HSQC) (20), HCCH type-(21), HCN (22), HCP (23), and 13 C heteronuclear single quantum coherence-nuclear Overhauser effect spectroscopy (HSQC-NOESY) (20) spectra were used for the detection and assignments of the resonances from CPD labeled with stable isotopes. In these experiments, the NMR samples were dissolved in Buffer III. The measurements described above were performed at 37°C. The other experiments described below were carried out at 30°C, and the samples were dissolved in Buffer IV. One-dimensional 1 H, 1 H-15 N HSQC, and 15 N NOESY-HSQC (20) were used for the detection and assignments of the imino resonances of dsDNA.

Binding Constants of the Wild-type CPD Photolyase and Its
Mutants for ssDNA Containing CPD-We determined the binding constants of the CPD photolyase from T. thermophilus for d(GTAT[CPD]TATG) by SPR measurements. The binding constants of the enzyme were measured under conditions with various salt concentrations (Table I). Higher NaCl concentrations resulted in significant reductions of the affinity, suggesting that electrostatic interactions contribute to the CPD binding.
In order to identify the residues that are responsible for electrostatic interactions, we determined the binding constants of mutants with the replacement of basic residues, which are widely distributed on the surface of the CPD photolyase, and the Trp residues, which are important for enzyme activity (11)(12)(13)(14)(15); R134A, R141A, R142A, R196A, R201A, R203A, R232A, R233A, R234A, K240A, W247A, W353A, R311A, R366A, L372A, R376A, R382A, K397A, R407A, and R419A. (e.g. R134A is a mutant, with the replacement of Arg with Ala at position 134.) It was confirmed that the global folds and the structures around the FAD for all mutants were not affected by the introduction of the mutations, by their 1 H NMR spectra, and UV/visible absorption spectra, respectively (data not shown). The binding constants of the mutants are summarized in Table II. The binding of R311A, R366A, and W353A for the single-stranded DNA (ssDNA) was not detected under our experimental conditions. The binding constants of K240A, R141A, R201A, R376A, and W247A were 7.4 ϫ 10 6 , 1.1 ϫ 10 7 , 9.4 ϫ 10 6 , 9.8 ϫ 10 6 , and 8.0 ϫ 10 6 M Ϫ1 , respectively; which are all less than half of that of the wild type. The binding constants of the other mutants were almost identical to that of the wild type. 31 P NMR analyses of ssDNA-CPD Photolyase Complexes-In order to determine the binding site of the CPD photolyase on DNA, we measured the 31 P NMR spectra of d(TAT[CPD]TATG) (ss7mer) in the absence and presence of the CPD photolyase ( Fig. 2, a-c). Assignments of 31 P resonances derived from the ss7mer in the free form were accomplished through the analyses of 1 H-31 P HSQC spectra, along with total correlation spectroscopy (TOCSY) and rotational Overhauser effect spectroscopy (ROESY) spectra, via a sequential assignment methodology (spectra not shown) (19). The established assign-

d(GTAT(CPD)TATG)] under various salt concentrations
The binding constants were determined by SPR with 10 -100 nM protein concentrations and running buffer (10 mM HEPES, pH 7.3, 3.5 mM EDTA, 0.005% Tween-20, 10 mM 2-mercaptoethanol, and NaCl). a Binding between the CPD photolyase and non-damaged DNA was also observed under these conditions. b Binding between the CPD photolyase and the CPD-containing ssDNA was not observed under these conditions.
b No significant chemical shift changes were observed for these mutants.
c Responses were not observed for these mutants. d Bound resonances were not obtained because of low affinities.
ments of the 31 P resonances are indicated in Fig. 2a.
To assign the resonances from the bound form, the 31 P-31 P EXSY spectrum was measured in the presence of a 0.5 molar excess of DNA. All of the free resonances could be connected with the bound ones, by observations of the exchange peaks (Fig. 2b). A comparison between the 31 P NMR spectra of DNA in the free and bound forms (Fig. 2, a and c) revealed that the P -1 , P 0 , P 1 , and P 2 resonances in the bound form were significantly broad and showed drastic chemical shift changes, upon the addition of the CPD photolyase. In a similar way, the 31 P NMR spectra of d(GTAT[CPD]TATG) and d(GTAT[CPD-]TATGC) were measured in the absence and presence of the CPD photolyase, and all of the resonances were assigned. In the spectra of d(GTAT[CPD]TATG) and d(GTAT[CPD-]TATGC), the P Ϫ1 , P 0 , P 1 , and P 2 resonances were also significantly perturbed, in terms of both the chemical shift and the line width, upon complex formation with the CPD photolyase (spectra not shown).
To determine which residues of the CPD photolyase interact with each phosphate of the DNA, we also measured the 31 P NMR spectra of the ss7mer complexed with the CPD photolyases mutants. Fig. 2d shows a 31 P NMR spectrum of the ss7mer in the presence of 1.5 molar equivalents of R201A. Compared with the spectrum of the wild-type complex (Fig. 2c), the P 0 and P Ϫ1 resonances have different chemical shifts, whereas the other resonances remained the same. In a similar manner, we determined the positions of the phosphate groups with resonances that were perturbed by the mutations in the ss7mer-CPD photolyase complex, as follows: P Ϫ2 , P Ϫ1 , and P 0 by K240A, P Ϫ2 by R141A, P 1 , P 2 , and P 3 by R376A, P 2 and P 3 by R407A, P Ϫ2 and P Ϫ1 by E244A, P 0 and P 1 by W247A, and P 2 by L372A. On the other hand, the 31 P spectra of the ss7mer in the presence of K397A, R134A, R142A, R196A, R203A, R232A, R233A, R239A, R382A, and R419A were identical to that in the presence of the wild type. These results are summarized in Table II.
Detection of Resonances Derived from CPD-To detect the resonances derived from the CPD lesion, we prepared d(AT[CPD]TAC), in which the CPD lesion was labeled with 13 C and 15 N (ss5mer), and measured the 1 H-13 C HSQC spec-tra (spectra not shown). In the spectrum, only the resonances originating from the CPD lesion were selectively observed. The resonances were assigned by using HCCH-type experiments (21), in which the 1 H-13 C cross-peaks from the deoxyribose rings and the cyclobutane ring were connected within each ring. HCN spectra (22) were useful for the connection of the resonances from the 1Ј-positions of the deoxyriboses and the resonances from the 6-positions of the bases via the 1-positions, which were labeled with 15 N. The 5Ј to 3Ј orientation was determined by the HCP spectra (23). Thus, all resonances involving the damaged bases were unambiguously assigned, by using the throughbond connectivities. Fig. 3, a-c show panels of the 1 H-13 C HSQC spectrum of the ss5mer upon the addition of a substoichiometric amount of the CPD photolyase labeled with 2 H. Both of the resonances originating from the free and bound forms were observed in the spectrum. The established assignments of the resonances from the free form are indicated in Fig. 3a. Assignments of the resonances derived from the methyl groups of the CPD bases in the complex (Fig. 3c) were performed by observations of the exchange-peaks between the resonances from the free and bound forms in the 13 C HSQC-NOESY spectra. The assignments of the other bound resonances were carried out, based on their chemical shifts (Fig. 3, a and b). The resonances derived from the 2Ј-position of the deoxyriboses could not be identified, due to line broadening. The assignments revealed that the chemical shifts of the CPD resonances were shifted upon complexation of the CPD photolyase (Fig. 3, a-c). Among them, the chemical shifts of the resonances from the bases, i.e. the methyl groups and the 6-position, were markedly shifted upfield along the 1 H dimension upon complexation of the CPD photolyase (Fig. 3, b and c). There are two possibilities for these significant chemical shift changes. One is the ring current effect of the aromatic groups of the CPD photolyase in the binding site, and the other is that of the nucleotides adjacent to the CPD.
To examine the effect of the aromatic groups in the CPD binding site, the 1 H-13 C HSQC spectra of the ss5mer in the presence of the CPD photolyase mutants were measured. In the spectrum of the ss5mer-W247A complex, the chemical shifts of the resonances from the CPD, including the methyl resonances ( Fig. 3d), were shifted downfield in comparison with those of the ss5mer-wild type complex. The methyl resonance of the 5Ј-side showed an especially large chemical shift difference, and its chemical shift was similar to that of the free form (Fig. 3d).
We measured the 1 H NMR spectra of pT[CPD]Tp, which is a CPD analog bearing a phosphate group, but lacking nucleotides on either side of CPD, and the ss5mer complexed with the wild type (Fig. 4, a and b). In the spectrum of the pT[CPD]Tp-wildtype complex, two characteristic resonances (ϳϪ0.8 ppm and Ϫ1.3 ppm) were observed (Fig. 4b), with chemical shifts quite similar to those of the methyl proton resonances from the ss5mer-wild type complex (Fig. 4a).
NMR Analyses of CPD-containing dsDNA Bound to the CPD Photolyase-To detect the base pair resonances of a DNA duplex, we prepared d(CGGCTTAATTGCG) labeled with 15 N, which is complementary to an unlabeled CPD-containing ssDNA, i.e. d(CGCAAT[CPD]TAAGCCG). After the formation of doublestranded DNA (dsDNA), the 1 H-15 N HSQC spectra of the dsDNA in the free and bound forms were recorded at 30°C (Fig. 5). In these spectra, the cross-peaks from the imino protons of guanines and thymines from the non-damaged strand were selectively observed. The imino resonances in the free form were assigned, using the NOEs between imino protons observed in the 15 N NOESY-HSQC and 1 H-1 H NOESY spectra. Assignments of the bound resonances were carried out, using the 1 H NOE difference spectra. The 15 N chemical shifts that were characteristic of each guanine and thymine imino groups were helpful for the assignments. The cross-peaks from Nos. 3,4,9,11, and 12 were clearly observed in the spectra of the bound form. A weak cross-peak from either No. 5 or 8 was also detected.
To investigate the environment of the CPD lesion in the dsDNA-CPD photolyase complex, we measured the 1 H NMR spectrum of the dsDNA complexed with the deuterated wildtype CPD photolyase (Fig. 6a). Two resonances with chemical shifts almost identical to the methyl group resonances of the ss5mer-wild type complex (Figs. 3c and 4a, indicated by arrows in Fig. 6a), were observed in the high field region of the spectrum. Fig. 6b shows the 1 H NMR spectrum of the dsDNA-W247A complex. A resonance with a chemical shift identical to that of the 3Ј-side methyl group in W247A-ss5mer CPD (Figs. 3d and 4b, indicated by an arrow in Fig. 6b) was observed in the high field region of the spectrum.  (Table I) suggested that electrostatic interactions, between phosphate groups in DNA and basic residues in CPD photolyase, significantly contribute to the DNA binding of the enzyme. The 31 P NMR analysis showed that P Ϫ1 , P 0 , P 1 , and P 2 were perturbed, in terms of both the chemical shift and the line broadening, upon the addition of d(TAT[CPD]TATG) (Fig. 2, a and b). These results indicate that a phosphate group of the CPD, and one phosphate group flanking at the 5Ј-side and two flanking at the 3Ј-side of CPD, are responsible for binding to CPD photolyase.
In order to investigate the basic residues that are involved in the electrostatic interaction, we determined the affinities of the mutants with substitutions of the basic residues, especially the Arg residues, to Ala (Table II). Fig. 7 shows the mapping of the affected residues identified in the mutant experiments on the crystal structure of the CPD photolyase from T. thermophilus without the substrates (13). The mapping indicates that the residues that participate in the binding are located in the region surrounding the cavity formed by clusters I and II, which could be a possible binding pocket for the CPD (13,15).
Orientation of DNA Relative to the DNA Binding Site-The 31 P NMR spectra of the CPD-containing ssDNA complexed with a variety of the mutants were measured. Some of the resonances from the ssDNA-mutant complexes have different chemical shifts from those from the ssDNA-wild type complex, indicating that the phosphate groups of the ssDNA with different chemical shifts are spatially close to the mutation position. Fig. 2d is a representative spectrum, in which the chemical shifts of P Ϫ1 and P 0 of the ssDNA-R201A complex were different from those of the ssDNA-wild type complex, suggesting that the P Ϫ1 and P 0 sites are located in the vicinity of Arg 201 . In a similar way, we obtained the spatial relationship between the residues of the CPD photolyase and the phosphate groups of  (Table II). The results obtained from the present experiments are also summarized in Fig. 7. As shown in Fig. 7, the mapping reveals that the 5Ј-side of the DNA lies on cluster I, the 3Ј-side is on cluster II, and the CPD resides at the cavity formed by both of the clusters.
On the basis of molecular dynamics simulations, the binding modes of the complex have been proposed (11)(12)(13)(14)(15)). In the model proposed by Sanders and Wiest, the orientation of the DNA is opposite to that in our results. On the other hand, the model by Vande Berg and Sancar has the same DNA direction of DNA as in our model.
The CPD Bases Are Buried in the Cavity on the DNA Binding Surface-To investigate the environment of the CPD bound to the enzyme, we measured the resonances derived from a CPD lesion labeled with 13 C and 15 N in a single-stranded pentamer (ss5mer), in both the free and bound forms. Interestingly, the resonances from the 6-position and the methyl groups of the base showed remarkable chemical shift changes along the 1 H dimension, upon complexation with the CPD photolyase (Fig. 3,  b and c).
In the spectrum of the pT[CPD]Tp-CPD photolyase complex (Fig. 4b), two methyl resonances were observed at the same positions as those from the ss5mer-CPD photolyase complex (Fig. 4a). This result indicates that the drastic chemical shift changes are induced, not by the flanking nucleotides, but by the CPD photolyase.
Considering the facts that CPD exists in the cavity in the bound form, as described in the previous section, and that the cavity is composed of some aromatic groups, such as Trp 247 , Trp 353 , and FAD (Fig. 7), it is most likely that the drastic chemical shift change is caused by the ring current shift of the aromatic groups. In order to define the position of the CPD relative to the cavity, we performed NMR analyses, using W247A.
As a result, the chemical shifts of the CPD methyl resonances were significantly different between the wild-type complex (Fig. 3c) and the W247A complex (Fig. 3d). Among them, the resonance from the 5Ј-methyl group of CPD showed a remarkable chemical shift difference between the wild-type complex and the W247A complex. These results suggest that  (Table II), the residues subjected to the mutations are classified into three types: the binding constants of the mutants are nearly equal to that of the wild type (cyan), are less than half of that of the wild type (yellow), and the responses were not observed for the mutants (red). FAD, which is inside the cavity, is shown in green. The phosphorus positions relative to the residues, which were obtained from a 31 P NMR analysis of the mutants (Table II), are also mapped.
Trp 247 mainly and partially causes the chemical shift change of the methyl resonances of the 5Ј-and 3Ј-sides, respectively. Therefore, we concluded that the bases of CPD are buried in the cavity and the side chain of Trp 247 is in the closest vicinity of the 5Ј-side methyl group of the bases in the ss5mer-CPD photolyase complex. This is consistent with the conclusion from the 31 P NMR analyses described above: the 5Ј-side of DNA lies on cluster I, where Trp 247 exists.
Flipping of CPD from the DNA Duplex in the Complex-The CPD photolyase is reportedly able to recognize and repair a CPD lesion embedded in dsDNA as well as that in ssDNA (24 -30), although McAteer et al. (30) reported that a CPD in dsDNA exists within the DNA helix and forms Watson-Crick type hydrogen-bonds with the complementary strand. One possible recognition mode for the buried CPD in dsDNA by the CPD photolyase is that the enzyme unwinds the dsDNA and then recognizes the CPD in dsDNA.
On the other hand, it has been established that one of the important recognition modes for a damaged base is that repair enzymes recognize the flipped out damaged base. Several groups of crystallographers have revealed the structures of DNA protein complexes with the base flipping (31)(32)(33)(34)(35)(36)(37)(38). However, in the case of the CPD photolyase, no structural evidence of CPD-flipping has been shown.
In order to investigate the mode of either CPD-flipping or DNA-unwinding, we performed a structural analysis of the dsDNA-CPD photolyase complex. We measured the 1 H-15 N HSQC spectrum of dsDNA, consisting of a non-labeled strand containing the CPD lesion and its complementary strand labeled with 15 N (Fig. 5). It should be noted that the observation of the imino resonances provides proof of the formation of the base pairs and the duplex. As shown in Fig. 5b, the resonances originating from the base pair at position Nos. 3, 4, 9, 11, and 12 are observed in the dsDNA-CPD photolyase complex, indicating that the CPD photolyase does not induce large dsDNAunwinding that is required for the recognition of the CPD bases within the DNA helix.
We also measured the 1 H NMR spectrum of the dsDNA-CPD photolyase complex, and compared it with that of the ss5mer-CPD photolyase complex (Figs. 4, a and c; 6, a and b). As shown in Fig. 6a, the spectrum of the dsDNA-CPD photolyase complex displayed two resonances, with chemical shifts identical to those of the bound ssDNA (Figs. 3c, and 4a). In addition, in the spectrum of the dsDNA-W247A complex (Fig. 6b), a high fieldshifted methyl proton resonance was observed at the position identical to that of the methyl group at the 3Ј-side of the CPD in the ss5mer-W247A complex (Figs. 3d and 4c). These results support the conclusion that the CPD photolyases binding modes are the same for a CPD embedded in dsDNA and a CPD in ssDNA. In other words, the bases of the CPD in dsDNA complexed with the CPD photolyase are also inserted into the cavity of the CPD photolyase. Taken together with the results of the retention of the duplex form and the insertion of the bases into the cavity, we conclude that the CPD is flipped out of the duplex in the complex.
Process Leading to CPD-flipping-In the previous section, we concluded that the CPD is flipped out in the bound form. However, as described above, the bases of the CPD exist inside the CPD-containing dsDNA in the free form (30). The difference between the DNA structures in the free and bound forms raises the question of how the CPD photolyase recognizes the CPD in dsDNA.
In addition to the chemical structure of the CPD base, the B II conformation of the backbone flanking the CPD on the 3Ј-side is the most pronounced characteristic of the CPD-containing dsDNA structure (30). Thus, the CPD photolyase may recog-nize the B II conformation of the backbone in the first step of DNA recognition. Vande Berg and Sancar demonstrated that Arg 452 in the CPD photolyase from yeast, which corresponds to Arg 311 from T. thermophilus, is essential for the enzyme to distinguish CPD from other normal bases (15). The present study revealed that Arg311 is located on the 3Ј-side of CPD in the complex (Fig. 7). Therefore, we suggest that the region containing Arg 311 is necessary for the CPD photolyase to search for the distortion of the backbone, and as described above, to form an ion pair with the 3Ј-phosphate group adjacent to the CPD upon binding.
Biological Significance of CPD-flipping-The CPD photolyases repair CPD lesions by an electron transfer between FAD and CPD. The efficiency of the electron transfer depends on the distance between the FAD and the CPD. If the CPD bases are inside the DNA helix in the complex, then the CPD should be more than 20 Å away from the FAD, which would make it difficult to transfer the electron from FAD to CPD with high efficiency. Therefore, CPD-flipping is required for the efficient CPD repair reaction by the enzyme.
The CPD photolyases repair both active and inactive genes (39). Therefore, the enzyme should search for and repair CPD lesions without interfering with the transcriptional regulation. In the present study, we found that the stable duplex exists even in the vicinity of the CPD in the dsDNA-CPD photolyase complex. This base-flipping mechanism is also required for minimizing the structural changes in the dsDNA and avoiding perturbations of the transcriptional regulation. Base-flipping may be the most efficient and universal repair system by the CPD photolyases.