Characterization of (6-4) Photoproduct DNA Photolyase*

Pyrimidine (6-4) pyrimidone photoproduct is the second most abundant W photoproduct in DNA. Recently, it was reported that ilrosophila melunogaster cell-free extracts restored the biological activity of (6-4) photo- product-containing DNA in a light-dependent reaction (Todo, T., Takemori, H., Ryo, H., Ihara, M., T., O., K., and Nomura, T. (1993) Nature 361, 372-374) concomitant with the loss of (6-4) photoproduct antigenic sites and (6-4) photoproduct-caused alkali-la-bile sites. In the present study we show that the (6-4) photoproduct but not its Dewar isomer is the substrate for the enzyme, that the enzyme has an action spectrum peak at 400 nm, and that the efficiency of repair per incident photon is very low compared with cyclobutane pyrimidine dimer photolyases. Furthermore, we pro- vide evidence that the (6-4) photoproduct photolyase converts the photoproduct to unmodified bases prob- ably through an oxetane intermediate. bovine serum albumin, 2 mMATP. The incubation for 30 min at 30 "C. The reaction products analyzed either on 12% polyacrylamide sequencing gel (photoreacti-vated DNA) or 0.8% agarose gel (plasmid DNA). Primer Elongation Assay-To find out if (6-4) photolyase restored the photoproduct to two thymines we tested the photoreactivated DNA for template function. A 10- and a 27-nucleotide primer complementary to the 3' part of the 49-mer (T[6-4]T) were synthesized. The 49-mer was incubated with 5 pg of cell-free extract in 100 pl of reaction buffer for IO min at 15 "C, either in the dark or under 380 nm photoreactivating light. The DNA was then extracted with phenoLkhloroform, precipitated with ethanol, and denatured by incubating in a IO-pl solution contain- ing 0.2 M NaOH, 0.2 mM EDTA for 30 min at 37 "C. (This does not appreciably cleave at 6-4 photoproducts that are alkali-labile at higher pH and temperature.) The DNA was neutralized by adding 3 p1 of 3 M sodium acetate, pH 5.2, precipitated with ethanol, and redissolved in 10 pl of 20 mM Tris-HC1, pH 7.5, 50 mM NaCl, and 2 mM MgCl,; the terminally labeled 10-nucleotide primer was added to 100 nM, or the 27-nucleotide primer was added to 5 nM, and the mixture was incubated at 70 "C for 2 min and slowly cooled to 30 "C for annealing the oli- gomers. For the experiments designed to measure primer elongation past the repaired lesion, to the primer-template (10 pl), 1 pl of 20 mM dithiothreitol, 10 mM ATP, 100 mM Tris-HC1, pH 7.4, 50 p~ MgCl,, a 5 mM concentration of each dNTP, and 2 units of DNA polymerase I were added. The reaction mixture incubated

DNA photolyase (deoxyribocylcobutadipyrimidine pyrimidine-lyase, EC 4.1.99.3, photoreactivating enzyme) is specific for cyclobutane type dimers; it does not repair (6-4) photoproducts (Brash et al., 1985). As a consequence, it has become common practice to expose UV-irradiated cells to photoreactivating light (350-450 nm) to study the effect of (6-4) photoproducts. Any residual mutagenic or cytotoxic effects remaining following photoreactivation are ascribed to (6-4) photoproducts (see Mitchell and Nairn, 1989). This is a justifiable approach as i t has been shown that in many organisms including man and Escherichia coli photoreactivating light does not eliminate  * This work was supported by National Institutes of Health Grants GM31082, CA40463, and ES5557. 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. photoproducts from DNA, although at high doses of 300-350 nm wavelengths the photoproduct is converted to its Dewar isomer photochemically (Taylor and Cohrs, 1987;Taylor et al., 1988). This photoisomerization was suggested to explain the nonenzymatic loss of (6-4) photoproducts, that was correlated with type I11 photoreactivation (Ikenaga et al., 1970;1971;Patrick, 1970), a process that is maximal at 313 nm and found only in certain bacteria (Ikenaga et al., 1970;Jagger et al., 1970). However, two recent papers have provided strong evidence that, in some organisms at least, there might be a (6-4) photoreactivating enzyme (Todo and Ryo, 1992;Todo et al., 1993). It was found that cell-free extract from Drosophila melanogaster contained two factors with specific affinity for UV-irradiated oligomer, as determined by band shift assay. One of these (factor 2 ) was identified as the cyclobutane pyrimidine dimer DNA photolyase because the band was absent in a mutant lacking the enzyme (Todo and Ryo, 1992) or when the experiment was conducted with UV-irradiated DNA photoreactivated with E. coli photolyase prior to mixing with the Drosophila cell-free extract. Thus it was concluded that factor 1 bound to (6-4) photoproduct. Surprisingly, however, exposure of the factor 1-DNA complex to high intensity light from a fluorescent lamp resulted in the disappearance of the retarded band. Similarly, when the DNA in factor 1 complex was exposed to light and then tested for (6-4) photoproducts by alkali hydrolysis and by disappearance of binding sites for Pyr[6,4lPyr-specific monoclonal antibodies, all Pyr[6,4lPyr were eliminated with the possible exception of C[6,4lC. Perhaps most significantly, this photoreactivation activity restored the biological function of a plasmid inactivated by Pyr[6,4lPyr (Todo et al., 1993). It was proposed that Drosophila contained a photolyase specific for (6-4) photoproducts.
This finding was quite surprising because formation of the (6-4) photoproduct involves the transfer of the group at C4 (-NH, or = OH) of the 3' base of the dinucleotide to the C-5 position of the 5' base concomitant with the formation of the sigma bond between the C-6 of the 5' base and the C-4 of the 3' base. Thus even if a n enzyme breaks the 6-4 C-C bond the bases would not be restored to their original forms. To address this and many other related questions stemming from these exciting findings we decided to investigate this activity further. We used a defined substrate containing a single T[6,4lT in a 49-base pair duplex  and nuclear extracts from a Drosophila cell line for these studies. Our work confirms the findings of Todo et al. (1993) and reveals that the enzyme may have a chromophore with A, , , = 400 nm and that photoreactivation with this enzyme, most likely, restores the dinucleotides making up the (6-4) photoproduct to unmodified form.

EXPERIMENTAL PROCEDURES
Materials-MseI restriction endonuclease and polynucleotide kinase were purchased from New England Biolabs. DNA polymerase I was

8535
This is an Open Access article under the CC BY license.
(6-4) Photoproduc obtained from Promega. E. coli endonuclease 111 was a gift from Dr. R. P. Cunningham (State University of New York, Albany), Enzyme and Substrate-Nuclear extract of the D. melanogaster Kc cell line was used as (6-4) photolyase. The extract was a kind gift of Dr. L. Searles (University of North Carolina). The extract was stored at -80 "C, and it was found that three or four freeze-thaw cycles did not affect the (6-4) photolyase activity. The substrate was a 49-base pair duplex containing either a (6-4) photoproduct or its Dewar isomer at the center (the two Ts that make up the photodimer are in bold): d(AGC-TACCATGCCTGCACGAATTAAGCAATTCGTAATCATGGTCATAG- CT).
The preparation and characterization of the oligomer containing either the T[6,4lT or the T[Dew]T forms of the (6-4) photoproduct have been described previously Svoboda et al., 1993). The oligomer was 5' end labeled with [y32P]ATP (7,000 Ci/mmol) and T4 polynucleotide kinase, annealed to the complementary strand, and the duplex was purified by gel electrophoresis as described previously .
Photoreactivation Assay and Action Spectrum-The photodimer in the 49-base pair duplex is a t a n MseI site (which incises between the two Ts in the TTAA sequence) and thus is resistant to cleavage by MseI. The assay measures the restoration of the susceptibility of the duplex to MseI cleavage. The reaction mixture (100 pl) contained 50 mM Tris-HC1, pH 7.4, 50 mM NaC1,5 mM dithiothreitol, 1 mM EDTA, 5% (v/v) glycerol, 10 pg of cell-free extract, and about 0.1 nM substrate. The reaction mixture was placed in an anaerobic cuvette with a rubber septum. The solution in the cuvette was deoxygenated for 10 min by a gentle stream of oxygen-free argon gas a t 15 "C. Following deoxygenation, the cuvette was placed in the irradiation chamber of a Quanta count monochromator (Photon Technology International) and exposed to photoreactivating light. The band width was * 5 nm. After irradiation, the sample was removed from the cuvette, placed in a microcentrifuge tube, and 3 p1 of 10% sodium dodecyl sulfate and 2 pl of proteinase K (10 mg/ml) were added and incubated at 55 "C for 60 min. The DNA was then extracted with phenollchloroform and precipitated with ethanol. The DNA was dissolved in 10 pl of 10 mM Tris-HC1, pH 7.9.50 mM NaC1,lO mM MgCl,, and 1 mM dithiothreitol. Two units of MseI were added, and the mixture was incubated a t 37 "C for 90 min. The digested DNA was separated on 12% polyacrylamide sequencing gels. The 49-mer band corresponding to unrepaired DNA and the 21-mer resulting from MseI incision of repaired DNA were excised and quantified by Cerenkov counting. The action spectrum was generated by conducting the photoreactivation reactions with wavelengths ranging from 300 to 500 nm (Payne and Sancar, 1990).
Endonuclease 111 Assay-To test if photoreversal created a saturated thymine ring we tested the photoreactivated DNA with endonuclease 111. Either the photoreactivated DNA (following phenokhloroform extraction) or 0.1 pg of pBR322 treated with OsO, to produce an average of four thymine glycols/plasmid as described by Katcher and Wallace (1983) was incubated with 20 nM endonuclease 111 in 10 p1 of reaction buffer containing 50 mM Tris-HC1, pH 7.5, 100 mM KC1, 10 mM dithiothreitol, and 10 m M MgCl,, 100 pg/ml bovine serum albumin, 2 mMATP. The incubation was for 30 min a t 30 "C. The reaction products were analyzed either on 12% polyacrylamide sequencing gel (photoreactivated DNA) or 0.8% agarose gel (plasmid DNA).
Primer Elongation Assay-To find out if (6-4) photolyase restored the photoproduct to two thymines we tested the photoreactivated DNA for template function. A 10-and a 27-nucleotide primer complementary to the 3' part of the 49-mer (T[6-4]T) were synthesized. The 49-mer was incubated with 5 pg of cell-free extract in 100 pl of reaction buffer for IO min a t 15 "C, either in the dark or under 380 nm photoreactivating light. The DNA was then extracted with phenoLkhloroform, precipitated with ethanol, and denatured by incubating in a IO-pl solution containing 0.2 M NaOH, 0.2 mM EDTA for 30 min a t 37 "C. (This treatment does not appreciably cleave at 6-4 photoproducts that are alkali-labile at higher pH and temperature.) The DNA was neutralized by adding 3 p1 of 3 M sodium acetate, pH 5.2, precipitated with ethanol, and redissolved in 10 pl of 20 mM Tris-HC1, pH 7.5, 50 mM NaCl, and 2 mM MgCl,; the terminally labeled 10-nucleotide primer was added to 100 nM, or the 27-nucleotide primer was added to 5 nM, and the mixture was incubated a t 70 "C for 2 min and slowly cooled to 30 "C for annealing the oligomers. For the experiments designed to measure primer elongation past the repaired lesion, to the primer-template (10 pl), 1 pl of 20 mM dithiothreitol, 10 mM ATP, 100 mM Tris-HC1, pH 7.4, 50 p~ MgCl,, a 5 mM concentration of each dNTP, and 2 units of DNA polymerase I were added. The reaction mixture was incubated at 37 "C for 90 min, and the reaction products were analyzed on 12% polyacrylamide sequencing gels. For determining the base inserted across from the repaired lesion, to the primer-template mixture (10 pl) in 100 m~ Tris-HC1, pH 7.4,5 mM MgCl,, T4 polymerase (2 units), and one dNTP at 10 p~ were added. The reaction was incubated a t 37 "C for 30 min, and the reaction products were precipitated with ethanol and analyzed on 15% polyacrylamide sequencing gels.

RESULTS
The (6-4) Photolyase Is Specific for (6-4) Photoproduct-The (6-4) photoproduct is almost quantitatively converted to its Dewar isomer (Fig. l ) by irradiation with 300-350 nm (Taylor and Cohrs, 1987, and thus under natural conditions a significant fraction of the (6-4) photoproduct is expected to be in its Dewar isomer form. Thus, we first wished to know if the (6-4) photolyase discovered by Todo et al. (1993) was active on one or both forms of the photoproducts. Substrates containing either T16-41T or T[DewlT were incubated with CFE from Drosophila, exposed to photoreactivating light, and then the repair of the photoproduct was quantified by measuring the restoration of the MseI susceptibility to the T[6-4lTAA or T[Dew]TAA site. The results shown in Fig. 2 indicate that the (6-4) photolyase is specific for the Kekule form and is essentially inactive on the Dewar isomer. The 1-2% activity observed with the Dewar isomer is probably caused by the low level of Kekule isomer contaminant in this T[DewlT preparation.
Action Spectrum of (6-4) Photolyase-Since the (6-4) photoproduct but not its Dewar isomer absorbs in the near UV (310-330 nm) and since the photoreactivating activity appears to be specific for the Kekule form we considered the possibility that the photoreactivation resulted from direct excitation of the photoproduct to initiate a photochemical reaction facilitated by a chromophoreless (6-4) photoproduct-binding protein.
This question is best answered by action spectrum measurements. Fig. 3 shows the absorption spectrum of T[6-4lT taken from the literature (Franklin et al., 1982) and the relative action spectrum of the (6-4) photolyase determined in this study. Clearly, the action spectrum is distinct from the T[6-4]T absorption spectrum. In fact, at the wavelengths (320-330 nm) corresponding to the T[6-4]T absorption peak the action spectrum has a valley, partly because at these wavelengths absorption by T[6-4]T leads to formation of nonsubstrate T[Dew]T. Thus, we tentatively conclude that the (6-4) photolyase has an intrinsic chromophore with an absorption peak at 400 nm.
Photolytic cross-section is the product of the molar absorption coefficient (E) of the photosensitizer and the quantum yield (6) of the photochemical reaction, and i t is a measure of the probability of an incident photon causing the photochemical reaction to occur. When the E of the chromophore is known €6 and therefore 4 can be measured easily. In a crude system, such as cell free extract, €6 can be measured by conducting the photoreactivation experiment under conditions in which all of the substrate is bound. Under such conditions, repair becomes a first-order function of the light dose, and from the first-order  (1982). Curve R represents the relative action spectrum of (6-4) photolyase, normalized to 400 nm, which shows the maximum activity. Photoreactivation was not detectable a t wavelengths greater than 520 nm.
plot of the repair reaction as a function of dose €4 can be obtained from the slope of the first-order plot by the following equation (Rupert, 1962) In ( where A is the irradiation wavelength in nm and L is the light dose in ergs.mm-'. To generate such a plot for the (6-4) photolyase we titrated the substrate for complete binding and then conducted the photoreactivation experiments with 360 nm. Fig.  4 shows the result of the dose dependence of photoreactivation.
Reaction Product of (6-4) Photolyase-Because of migration of the substituent at the C-4 position ofone base to the C-5 position of the other during formation of the (6-4) photoproduct, cleavage of the C-6-C-4 single bond is expected to generate a dihydropyrimidine and a pyrimidine ring that are likely to be mutagenic and cytotoxic. Therefore, we tested the photoreactivated DNA with two probes for unusual bases. E. coli endonuclease I11 is a multipurpose repair enzyme that has glycosylase/AP lyase activity on most if not all pyrimidine hydrates, glycols, and fragmentation products (see Wallace, 1988;Sancar and Sancar, 1988). We treated the photoreactivated DNA with endonuclease I11 and analyzed the reaction product on a sequencing gel. Fig.   5A shows that in this particular experiment about 30% of the (6-4) photoproduct has been eliminated as evidenced by sensitivity to MseI restriction endonuclease (lane 2 1. However, when this DNA was treated with excess endonuclease I11 no cleavage occurred (lane 3 ) under conditions in which endonuclease I11 cleaved at all thymine glycols in a plasmid substrate (data not shown). Thus, it appears that photolyase does not generate a saturated pyrimidine ring as a product. To test for the formation of some other pyrimidine derivative not recognizable by endonuclease I11 we also tested the repaired DNA for template activity. Fig. 5B shows that T16-41T is a block for polymerase I (lane 1 ) but that after photoreactivation a fraction of the primer proportional to the fraction ofT[6-4]T repaired is elongated past the photoproduct region (lane 2 ) . Thus, we conclude that the photoreactivated DNA is no longer a block for DNA polymerase I. This is in agreement with the results ofTodoet al. (1993), who showed restoration of transforming (replication) activity of a plasmid following photoreactivation, and provides further evidence that, most likely, (6-4) photolyase restores the pyrimidines to their normal form. A third line of evidence for this conclusion, of course, comes from the fact that the (6-4) photolyase renders the substrate susceptible to the MseI restriction endonuclease. It is well known that restriction endonucleases are quite sensitive to base modifications at the incision site, and MseI hydrolyzes the phosphodiester bond between the two thymines involved in T[6-4lT; it is unlikely that the enzyme would recognize and incise a t this sequence if following photoreacti- A final piece of evidence for restoration of the T[6-4]T to normal bases was provided by conducting chain elongation with single dNTPs using a primer terminating at the nucleotide preceding the lesion site. The results shown in Fig. 5C reveal that dA is incorporated very efficiently across the repaired lesion and that dG and dC are not incorporated at the detection limit of our assay. With dT as the sole nucleotide we do observe some (-1%) chain elongation past the repaired lesion. However, we observe the same level of synthesis with undamaged DNA as well, and therefore we ascribe this synthesis to "slippage" of the template a t this A-T-rich sequence. All of these data combined lead us to conclude that the (6-4) photolyase does restore T[6-4lT to canonical bases.

DISCUSSION
The work presented here confirms the findings of Todo et al.
(1993) that Drosophila contains a protein that mediates the light-dependent disappearance of the (6-4) photoproduct from DNA. We have shown further that the action spectrum peak for this activity is a t 400 nm and that this activity is specific for the Kekule form of the photoproduct, being completely inactive on the Dewar form. We have obtained preliminary evidence that the photoreaction restores the bases to their canonical (unmodified) forms. Our findings raise two questions regarding the mechanism and significance of this activity. The oxetane intermediate is considered essential for regenerating canonical bases. Mechanisms I and I1 involve energy transfer or direct excitation of the (6-4) product. Mechanism I11 involves electron transfer to/from the (6-4) product and Mechanisms IV and V electron transfer to/from oxetane intermediate. e Mech IV e Regarding mechanism, it is most likely that reversal of the plex (Mechanism I) or excitation transfer from a chromophore (6-4) product t o 2 thymidines proceeds via the same oxetane (Mechanism 11) may lead to oxetane formation. Alternatively, it intermediate that led to its formation. The presumed oxetane is possible that an excited state chromophore abstracts an elecphotoproduct of thymine has been shown to be unstable above tron from the pyrimidinone ring which is then attacked at C-4 -80 "C and thermally decompose to the (6-4) product (Rahn and by the hydroxyl group together with proton transfer and sub-Hosszu, 1969). It is not known, however, to what extent the sequent back electron transfer (Mechanism 111). oxetane intermediate also thermally decomposes back to the In another possible pathway, the enzyme first thermally original dipyrimidine. The instability of the oxetane interme-catalyzes the formation of the oxetane intermediate which then diate relative to the (6-4) product is probably caused by the is converted to two thymines by a photochemical mechanism. inherent ring strain of the oxetane and the loss of aromatic The oxetane intermediate is expected to have a A, , ,