Human DNA topoisomerase I-mediated cleavages stimulated by ultraviolet light-induced DNA damage.

DNA topoisomerases have been proposed as the proteins involved in the formation of the DNA-protein cross-links detected after ultraviolet light (UV) irradiation of cellular DNA. This possibility has been investigated by studying the effects of UV-induced DNA damage on human DNA topoisomerase I action. UV lesions impaired the enzyme's ability to relax negatively supercoiled DNA. Decreased relaxation activity correlated with the stimulation of cleavable complexes. Accumulation of cleavable complexes resulted from blockage of the rejoining step of the cleavage-religation reaction. Mapping of cleavage sites on the pAT153 genome indicated UV-induced cleavage at discrete positions corresponding to sites stimulated also by the topoisomerase I inhibitor camptothecin, except for one. Subsequent analysis at nucleotide level within the sequence encompassing the UV-specific cleavage site revealed the precise positions of sites stimulated by camptothecin with respect to those specific for UV irradiation. Interestingly, one of the UV-stimulated cleavage sites was formed within a sequence that did not contain dimerized pyrimidines, suggesting transmission of the distortion, caused by photodamage to DNA, into the neighboring sequences. These results support the proposal that DNA structural alterations induced by UV lesions can be sufficient stimulus to induce cross-linking of topoisomerase I to cellular DNA.

DNA topoisomerases have been proposed as the proteins involved in the formation of the DNA-protein cross-links detected after ultraviolet light (UV) irradiation of cellular DNA. This possibility has been investigated by studying the effects of UV-induced DNA damage on human DNA topoisomerase I action. UV lesions impaired the enzyme's ability to relax negatively supercoiled DNA. Decreased relaxation activity correlated with the stimulation of cleavable complexes. Accumulation of cleavable complexes resulted from blockage of the rejoining step of the cleavage-religation reaction. Mapping of cleavage sites on the pAT153 genome indicated UV-induced cleavage at discrete positions corresponding to sites stimulated also by the topoisomerase I inhibitor camptothecin, except for one. Subsequent analysis at nucleotide level within the sequence encompassing the UV-specific cleavage site revealed the precise positions of sites stimulated by camptothecin with respect to those specific for UV irradiation. Interestingly, one of the UV-stimulated cleavage sites was formed within a sequence that did not contain dimerized pyrimidines, suggesting transmission of the distortion, caused by photodamage to DNA, into the neighboring sequences. These results support the proposal that DNA structural alterations induced by UV lesions can be sufficient stimulus to induce cross-linking of topoisomerase I to cellular DNA.
Cyclobutane pyrimidine dimers (CPDs) 1 and (6 -4) photoproducts are the most prevalent lesions produced in DNA by ultraviolet (UV) light (1). However, pyrimidine dimers are not the only photochemical effect of UV light on cellular DNA. It has been shown that UV radiation induces also the formation of DNA-protein cross-links (2)(3)(4). Proteinase K treatment abolishes the cross-linking effect and reveals the presence of cryptic DNA strand breaks. Since this cross-linking is partially repaired, it has been suggested that this non-dimer DNA damage may play an important role in the biological effect of UV radiation (2,4). DNA topoisomerases have been proposed as possible candidates for the protein(s) involved in UV-induced DNA-protein cross-linking (4). This proposal is consistent with the formation of transient single and double strand breaks during DNA topoisomerase reactions, with covalent attachment of the enzymes to one terminus of the DNA nick (reviewed in Ref. 5).
DNA topoisomerases are ubiquitous enzymes involved in a number of crucial cellular processes, including replication, transcription, and recombination. A relationship between cellular responses to DNA damage and topoisomerases has been proposed (6 -9). The catalytic cycle of DNA topoisomerases can be divided into several steps: 1) enzyme-DNA binding; 2) DNA cleavage, resulting in a covalent attachment between the protein and one terminus of the DNA nick; 3) DNA strand passage; 4) poststrand passage DNA religation concerted with the enzyme turnover (reviewed in Ref. 10). Treatment with strong protein denaturants arrests the catalytic cycle after the cleavage event by trapping the transient DNA-protein intermediate of step 2, termed "cleavable complex." Several topoisomerase inhibitors stabilize the cleavable complex by interfering with the strand passage cleavage/religation equilibrium. Treatment of cellular DNA with these inhibitors results in cryptic single and double strand breaks associated with enzyme's covalent attachment to DNA (reviewed in Ref. 11).
UV photoproducts cause alterations of the DNA conformation that can affect the activity of DNA processing enzymes. It has been shown that UV irradiation of the substrate inhibits digestion of recognition sequences containing dimerizable pyrimidines by restriction enzymes (12). Moreover, the catalytic reaction of bacterial DNA topoisomerase I (13,14) and Drosophila melanogaster DNA topoisomerase II (15) is inhibited by UV damage in the target DNA. The molecular mechanism by which damage can affect the enzyme's DNA strand passage step is unknown. However, it has been proposed that the inhibitory lesions may be present in the enzyme's active site at the time of strand passage (15) or that the helical distortion induced by photodamage to DNA can slow down the diffusion of the helix through the DNA-protein bridge (13).
In an effort to verify whether eukaryotic DNA topoisomerases might be the proteins involved in the formation of UVinduced DNA-protein cross-links, the effects of short wave UVinduced photoproducts on the enzymatic activity of human DNA topoisomerase I were investigated. Our results indicate that the enzyme's relaxation reaction is inhibited by the presence of UV damage in the substrate. Reduced relaxation activity correlated with alteration of the cleavage/religation equilibrium of the reaction, resulting in the stimulation of cleavable complexes. This observation supports the notion that DNA topoisomerase I may be the protein involved in UV-induced DNA-protein cross-linking of cellular DNA.

EXPERIMENTAL PROCEDURES
Materials-Human DNA topoisomerase I, purified from HeLa cell nuclei according to Ishii et al. (16), was diluted to the appropriate concentration in DNA topoisomerase I diluent (40 mM Hepes, pH 8.0, 0.2 mg/ml bovine serum albumin, 2 mM dithiothreitol, 0.5 mM EDTA, pH 8.0, 40% glycerol, 6% poly(ethylene glycol)). One unit of enzyme is defined as the amount of topoisomerase I required to relax 50% of 0.25 g of pAT153 DNA after a 10-min incubation at 30°C. Micrococcus luteus PD-endonuclease was obtained by purification to the CM-cellulose step according to Grafstrom et al. (17).
Plasmid pAT153 DNA (3657 bp, Fig. 3B) was purified from stationary phase HB101 Escherichia coli cells grown in L-broth supplemented with 0.2% casamino acids and 10 g/ml tetracycline, by alkaline lysis method as described (18). Separation of the negatively supercoiled form DNA (RFI) from the nicked form DNA (RFII) was obtained on a NACS-37 column (Life Technologies, Inc.) as described previously (19). DNA was stored at Ϫ20°C in H 2 O at a concentration of 400 g/ml.
Preparation of UV-irradiated DNA-Ultraviolet light-irradiated DNA was prepared by exposure to 254-nm UV light from a germicidal lamp at a fluence rate of 1.89 J m Ϫ2 s Ϫ1 , as determined by a VLX ultraviolet intensity meter (Vilbert Lourmat, Marne-La-Vallé, France). The samples were kept on ice to avoid heating and evaporation. Under the conditions employed for short wave UV irradiation, approximately 80% of the photoproducts induced in DNA are CPDs. The remaining 20% are predominantly (6 -4) photoproducts (data not shown). The maximal UV dose employed in the present study (2200 J/m 2 ) is expected to generate approximately 100 CPDs/plasmid DNA molecule.
DNA Topoisomerase I-catalyzed Relaxation of Negatively Supercoiled DNA-The topoisomerase I-catalyzed DNA relaxation assay was carried out as follow: each reaction mixture (120 l total volume) contained 1.50 g of negatively supercoiled pAT153 DNA, nonirradiated or irradiated at the indicated doses, 50 mM Tris-HCl, pH 7.5, 120 mM KCl, 10 mM MgCl 2 , 0.5 mM dithiothreitol, 0.5 mM EDTA, pH 8.0, and 4.2 units of topoisomerase I. The reaction mixture was incubated at 30°C; 20-l aliquots were removed at the indicated times and added to 5 l of stop mix (final concentration 10% glycerol, 0.005% bromphenol blue, 0.1% SDS). DNA samples were analyzed by electrophoresis at 25 V for 16 h on a 1% agarose gel containing ethidium bromide in Tris-phosphate buffer (0.08 M Tris-phosphate, 0.8 mM EDTA, 0.5 g/ml ethidium bromide). After electrophoresis, the gel was subjected to photolysis by short wave UV light (254 nm) to nick all DNA samples. This was done to normalize the intensity of the observed bands, since supercoiled DNA and relaxed DNA are stained differently with ethidium bromide. After photonicking, the gel was destained and photographed under transilluminator at 300-nm UV light with Polaroid type 55 positive/negative film.
The negative of the gel photograph was scanned with Model GS-670 Densitometer (Bio-Rad) and peaks quantitated by integration with Bio-Rad Molecular Analyst software. The velocity of DNA relaxation was determined by quantitating the loss of supercoiled DNA as a function of time. The loss of supercoiled DNA was calculated as (A r / A r ϩA s ) ϫ 100, where A r is the area of the relaxed DNA band (RFIV) and A s is the area of the supercoiled DNA band (RFI). This automatically corrects for loading variations or for differences in transilluminator efficiency. Under the conditions employed, the band intensity in the negative of the gel photograph was directly proportional to the amount of DNA present. To examine the initial reaction velocity, the loss of supercoiled DNA was quantitated at the different time points of the kinetic reactions. Initial rates were calculated by a least squares fit analysis.
Preparation of End-labeled DNA Fragments and Mapping of DNA Breaks-To map topoisomerase I-mediated cleavage sites at genome level, plasmid pAT153 DNA (Fig. 3B) was first linearized with BamHI (position 375 of the genome) or ScaI (position 3140 of the genome) restriction enzymes; then the 5Ј DNA termini were dephosphorylated with calf thymus alkaline phosphatase and labeled with [␥-32 P]ATP and T4 polynucleotide kinase. When needed, the labeled DNAs were subjected to a second enzyme digestion with SalI (position 650 of the genome) or Asp700 (position 3257 of the genome), respectively. The longer DNA fragments were isolated and purified by low melting agarose gel and digestion with Gelase enzyme. DNA was purified by phe-nol-chloroform extraction and ethanol precipitation (18). The positions of cleavage sites, after DNA separation on alkaline agarose gels, were obtained by densitometric scanning of the autoradiographies and computer analysis of the data, as described above. DNA markers were run in the two outer lanes of all gels in order to check the uniformity of DNA migration throughout the gels. Regression curves of the logarithm of the fragment size (in base pairs), as a function of the migration distance of each fragment from a reference line, were determined for the DNA markers. Regression coefficients were consistently near 0.99. Each autoradiography lane was analyzed using the same reference line, and the size of each DNA fragment, induced by topoisomerase I cleavage, was computed. Fragment size determination was usually within 50 bp for a given fragment analyzed in different gels.
Localization at nucleotide level of topoisomerase I-mediated cleavage sites was obtained by linearization of pAT153 DNA with PvuI or Asp700 restriction enzymes. Protruding and blunt 3Ј-ends were labeled using terminal deoxynucleotidyl transferase and [␣-32 P]ddATP according to Brash (20) and then digested with SspI or EcoRI enzymes, respectively. The two 3Ј-end-labeled fragments (435 and 398 bp, respectively) were purified by 5% acrylamide gel electrophoresis, followed by electroelution and ethanol precipitation. Mapping of the phosphodiester bonds cleaved by topoisomerase I was obtained by electrophoresis of the reaction products on sequencing gel alongside with the Maxam and Gilbert sequence ladder of the same fragment (18).
DNA Topoisomerase I Cleavage Reactions-DNA fragments were incubated in 20 mM Tris-HCl, pH 7.5, 100 mM KCl, 10 mM MgCl 2 , 0.1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, pH 8.0, with appropriate amount of purified human topoisomerase I at 37°C for 5 min. Cleavable complexes were disrupted by addition of 1% SDS and proteinase K (100 g/ml), followed by incubation for 30 min at 37°C. DNA was then phenol extracted and ethanol-precipitated. When cleavage products were to be analyzed by alkaline agarose gel electrophoresis, equal counts from each samples were dissolved in 5 l of H 2 O, adjusted to 50 mM NaOH and 1 mM EDTA, and loaded into the wells of a 1% alkaline agarose gel. Alkaline agarose gels were run at 2 V/cm overnight and then dried. Dried gels were autoradiographed with Hyperfilm-MP (Amersham).
M. luteus PD-endonuclease Digestion and Piperidine Treatment-Analysis of UV photoproducts was essentially as described by Brash (20). To detect CPDs, the uniquely labeled DNA was treated with M. luteus PD-endonuclease, which cleaves the glycosidic bond of the 5Ј cyclobutane dimer and the phosphodiester bond 3Ј to the apyrimidinic site, leaving a free 3Ј hydroxyl end (21). The UV-irradiated DNA was incubated with saturating amount of M. luteus PD-endonuclease in 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 50 mM NaCl (final volume 20 l), at 37°C for 30 min. The reaction was stopped by phenol extraction, and DNA was subsequently purified by ethanol precipitation. To detect (6 -4) photoproducts, the labeled DNA was treated with hot alkali to break the labile glycosidic bond of the 3Ј pyrimidine photoproduct, subsequently leading to ␤-elimination at the apyrimidinic site (20). The UV-irradiated DNA was dissolved in 100 l of 1 M piperidine (Fluka) and heated for 30 min at 90°C. DNA fragments were precipitated by addition of sodium acetate, pH 5.2, to a final concentration of 0.3 M and 2.5 volumes of ethanol. The precipitates were washed twice with 75% ethanol and dried in a Speedvac concentrator to remove traces of remaining piperidine. The position of cleavages by M. luteus PD-endonuclease and by hot piperidine was estimated by running the treated samples on DNA sequencing gel, alongside with the Maxam and Gilbert sequence ladder of the same fragments. The relative frequency of UV photoproduct formation was calculated from the area of the peaks obtained by densitometric scans of the autoradiographies as above. DNA cleavage was expressed in arbitrary units relative to the most intense band. Since the relative rates of photodamage formation are not modified in the range of doses employed (21), the UV dose chosen for this analysis was calculated so to produce an average of 1 CPD per DNA strand.

RESULTS
Effect of UV-induced Photoproducts on DNA Topoisomerase I-catalyzed DNA Relaxation-A DNA relaxation assay was uti-lized to examine the effects of UV-induced damage in the substrate on the catalytic activity of human DNA topoisomerase I. This assay measured the conversion of naturally supercoiled pAT153 DNA (RFI) into relaxed covalently closed circular DNA (RFIV). RFI and RFIV DNAs were separated from the nicked molecules (RFII) by inclusion of ethidium bromide in the agarose gel. In this condition the relaxed topoisomers migrate as a single band ahead of the RFI band allowing an accurate quantitation of the reaction products by densitometric scanning of gel photographs. A time course of relaxation was carried out with catalytic amount of enzyme and pAT153 DNA molecules irradiated at increasing UV doses from 0 to 1750 J/m 2 (Fig. 1,  A and B). Under these conditions, the relaxation velocity of supercoiled DNA was linear for 20 min. Following a 20-min incubation, 0.7 unit of topoisomerase I relaxed about 65% of nonirradiated pAT153 DNA. As shown in Fig. 1C, the initial DNA relaxation rate linearly decreased as the level of UV-dose increased. At the highest UV dose, the initial velocity was reduced to about 20%. These results indicate a dose-dependent inhibition of topoisomerase I activity by UV-induced damage.
Effect of UV-induced Photoproducts on the DNA Cleavage/ Religation Equilibrium of Topoisomerase I-The catalytic cycle of topoisomerase I can be divided into several discrete steps involving DNA binding, DNA cleavage, DNA strand passage, religation of the DNA break, and enzyme turnover. The cleavage reaction involves a transient single-stranded break in the DNA backbone containing the enzyme covalently bound at the 3Ј side of the cleaved strand. This transient intermediate, termed cleavable complex, can be evidenced by freezing the cleavage/religation equilibrium with strong protein denaturants. In order to determine which specific reaction step(s) of the normal catalytic cycle is impaired by UV light-induced lesions in the target DNA, the effect of this damage on the DNA cleavage/religation equilibrium was examined.
Cleavage was studied by reacting 5Ј-end-labeled EcoRI-cut pAT153 DNA with topoisomerase I for 5 min at 37°C; the transiently nicked enzyme-DNA intermediate was trapped by addition of SDS and digestion with proteinase K followed by separation of reaction products on alkaline agarose gel and visualization by autoradiography. Cleavable complex stimulation was evaluated by measuring the decrease of full-length linear molecules as a function of UV dose. As shown in the gel photograph ( Fig. 2A), there is a continuous decrease of uncleaved molecules with the increase of UV irradiation. As summarized in the graph (Fig. 2B), where results were plotted as residual uncleaved DNA versus UV dose, there is linear dependence of global cleavage with increasing amount of photoproducts. This finding clearly indicates that the presence of UV-induced DNA lesions affects the DNA cleavage/religation equilibrium by causing the accumulation of cleavable complexes. This effect is analogous to that described for the specific topoisomerase I inhibitor camptothecin (CPT) (22), and for The results, obtained from data shown in A, were plotted as a percent relaxation activity versus the UV dose to which plasmid substrates were exposed. The relaxation activity of topoisomerase I in the absence of UV-induced photoproducts was set to 100%. The average standard errors were less than 5%. Assay conditions and densitometric analysis of reaction products on negatives of agarose gel photographs were as described under "Experimental Procedures." mismatches adjacent to a topoisomerase I cleavage site (23).
Genomic Localization of Topoisomerase I-mediated DNA Breaks Stimulated by UV Photoproducts in Plasmid pAT153 DNA-Eukaryotic topoisomerase I-mediated cleavages are nonrandomly introduced in the DNA helix. Cleavage sites are characterized by a weak sequence-specificity (24, 25) and a loose conformational consensus (26) with preference for curved DNA, be it stably (27,28) or dynamically bent (29). Although CPT has only a minimal effect on the sequence selectivity of the enzyme (reviewed in Refs. 30 and 31), the degree of stimulation differs among sites (32). Experimental and modeling studies have shown that UV damage is characterized by a small but significant deformation of the DNA double helix, which can affect protein-DNA interaction (12). Thus, it is possible to envisage that the modifications of the helical parameters produced by UV photoproducts may interfere with the sequence specificity of topoisomerase I-mediated cleavage. Therefore, it is of interest to study topoisomerase I cleavage specificity on UV-irradiated DNA and to compare this specificity with that described for CPT.
A uniquely end-labeled fragment (generated by secondary digestion of 5Ј-end-labeled BamHI-cut pAT153 DNA with SalI) was incubated with a concentration of DNA topoisomerase I that produced only limited DNA cleavage (Fig. 3A, lane 3), denatured in the presence of SDS and deproteinized with proteinase K. Localization of strand breakage was obtained by size analysis of the cleavage products separated by alkaline agarose Using two different uniquely 5Ј-end-labeled fragments, (5Ј-BamHI/SalI and 5Ј-ScaI/Asp700), genomic localization of cleavage sites was obtained. The position of these sites on the pAT153 map was determined by densitometric scanning of the autoradiograms and computer analysis of six independent experiments with an average standard error of Ϯ 25 nucleotides (Fig. 3B). The break site uniquely stimulated by UV damage (Fig. 3A, arrow) was at position 3134 (Ϯ11 bp) of the plasmid map. The same analysis performed on the complementary strand did not reveal any breakage site stimulated only by UV damage (data not shown).
Higher Resolution Analysis of Cleavage Site Specificity-The influence of UV damage on the cleavage properties of human topoisomerase I was further characterized by analysis, on denaturing acrylamide gel, of the break sites previously mapped within the region encompassing the site preferentially stimulated by UV photolesions. The position of UV-induced breakages was compared with the position of breaks stimulated by CPT and with CPDs distribution, obtained by running in parallel the same UV-irradiated fragment digested with the cyclobutane dimer-specific endonuclease from M. luteus (21).
The 752-bp EcoRI/PstI fragment (5Ј-end-labeled at the EcoRI restriction site) was incubated with topoisomerase I in the presence of increasing amount of CPT or after irradiation at increasing UV dose (Fig. 4). Cleavages with the enzyme alone are limited (lanes 2 and 7), while several additional cleavage sites were visible on UV-irradiated DNA (lanes 8 -11) or in the presence of CPT (from 0.75 to 5.0 M, lanes 3-6). Cleavage stimulation by CPT appeared concentration-dependent (lanes 3-6). No major change in band intensities and in cleavage pattern could be observed as a consequence of increase in UV irradiation (from 880 up to 2200 J/m 2 , lanes 8 -11).
Breakage sites within this fragment were tentatively classified into three categories according to the response elicited by UV or CPT. Topoisomerase I cleavages at a sites were stimulated by UV damage and by CPT, those at b sites only by CPT, and those at c sites specifically by UV irradiation of the target DNA. Among the c sites, the c1 site corresponded to the uniquely UV-stimulated site identified by alkaline agarose gel electrophoresis. Interestingly, c3 and c4 sites were formed at some distance from the major CPDs clusters (lane 12). To examine whether c sites were UV-specific, we increased the molar ratio of enzyme to DNA. We found that a 3-fold increase in the molar ratio of enzyme to DNA resulted in very weak breakage at c1 and c2 sites on nonirradiated DNA. These sites appeared slightly stimulated also by CPT. No effect was observed at the c3 and c4 sites even with more enzyme (data not shown). The cleavage sites stimulated by UV damage (c and a sites) were mapped at nucleotide level by running the reaction products on sequencing gels in parallel with Maxam-Gilbert chemical degradation reactions and PD-endonuclease digestion products of the same end-labeled fragment. The SspI-PvuI pAT153 restriction fragment (3Ј-end-labeled at the PvuI restriction site) was used to characterize c1 and c2 sites and the EcoRI-XmnI fragment (3Ј-end-labeled at the XmnI restriction site) to study c3 and c4 sites.
In Fig. 5, the portion of the gel containing the topoisomerase I cleavage products at the a1, a2, and c2 sites, obtained after stimulation with UV damage (lane 6), with CPT (lane 7) or with CPT on an irradiated substrate (lane 5), is shown. Control experiments, in which DNA was irradiated and topoisomerase I was omitted (lane 3), excluded the possibility that cleavages were caused by UV light. The two a sites, previously identified as single bands (Fig. 4, lanes 8 -11), appeared to be flanked by two b sites that were named b1 and b2. Interestingly, cleavage intensity at a1, a2, b1, and b2 sites was markedly suppressed, when CPT stimulation was performed on an irradiated substrate (compare lanes 5 and 7). Inspection of the nucleotide sequences at the 5Ј terminus of the topoisomerase I cleavages showed the presence of CPDs on the irradiated substrate (lane 1). This effect was not detected for other b sites that were instead located within sequences that did not contain photodamage in the scissible strand, suggesting that the presence of CPDs in the sequences adjacent to the break sites can interfere with CPT action (data not shown). Fig. 6 summarizes the nucleotide sequences containing the UV-stimulated sites, the positions of the UV-stimulated topoisomerase I-mediated cleavage sites with respect to the distribution of CPDs and (6 -4) photoproducts and the relative frequency of CPDs. Comparisons of the base sequences upstream and downstream from the UV-stimulated sites has not revealed any apparent specific elements that could explain the stabilization of cleavable complexes observed in the presence of UV damage. Analysis of the relative distribution of UV lesions on the scissible strand indicate a difference between a and c sites. All three a sites were positioned at the 5Ј side of pyrimidines runs that have an high probability to dimerize. With the exception of the c2 site, the c sites were formed within sequences that have a very low probability to contain damage in the scissible strand. This is particularly evident for the c3 site located at least 10 bases from two dimerized thymines.
Dissociation Kinetics of Enzyme-DNA Complexes by Heating-Cleavable complexes stimulation by camptothecin has been shown to be reversed by several treatments such as elevated temperature (33,34). To investigate the mechanism of topoisomerase I breakage stimulated by UV photodamage, the stability of preformed UV-or CPT-induced cleavable complexes was measured. The dissociation kinetics were followed by heating the reaction mixtures at 65°C before treating samples with SDS. The electrophoretic analysis of the resulting DNA fragments is shown in Fig. 7A. It is clear that the UV-and CPTstimulated cleavable complexes showed widely different sensitivities to heating. The enzyme-DNA complexes formed in the presence of CPT (a and b sites) dissociated very rapidly, while complexes formed after UV irradiation (a and c sites) decayed more slowly, with broken complexes still detectable after 15 min of heating. To quantify the differences, the residual cleavage, i.e. the cleavage frequency at a given time normalized to the cleavage frequency obtained in a sample from the same experimental series subjected to SDS immediately before heating, was plotted as a function of the incubation time at 65°C (Fig. 7B). The residual breakage at the CPT-stimulated a and b sites (shaded area) was reduced to less than 1% after 1 min at 65°C and cleavage was not detected at later times. In contrast, the complexes formed at UV-stimulated sites decayed FIG. 6. Relative frequency of CPDs formation and DNA sequences at sites of topoisomerase I cleavage stimulated by UV damage. Mapping of topoisomerase I cleavage sites within the amp r gene was performed using uniquely 3Ј-end-labeled restriction fragments essentially as described in Fig. 5 and under "Experimental Procedures." The position of the break sites is indicated by an arrow. The distribution and relative frequency of CPDs (closed bars) and the distribution of (6 -4) photoproducts, surrounding the topoisomerase I cleavage sites, were obtained as described under "Experimental Procedures"; the pyrimidines involved in the formation of (6 -4) photoproducts are underlined. more slowly and remained incomplete.
It is noteworthy that the UV-stimulated break sites appeared to have two dissociation rates; an initial rapid rate in the first min of incubation was followed by a very slow rate. The basis for the discontinuity in the decay curves is unclear but it might involve a rapid inactivation of the enzyme during the kinetic analysis or an inability to complete break religation. In the latter case, it might depend from the position of photodamage with respect to the break sites. In fact, the irradiated substrate is constituted by an heterogeneous population with respect to the position where damage is created. Thus, in the fraction of molecules, where damage was present at some distance from cleavage sites, resealing may occur at a much higher rate (initial rate), while in the fraction of molecules, where photodamage were formed close to the break sites, religation may be very slow or even prevented (slow rate). The latter explanation is supported by the observation that among the complexes stimulated by UV damage, there is a good correlation between the religation rate and the relative position with respect to damage. For example, the a1 and a2 sites, located in the scissible strand close to a run of thymines with high probability to dimerize, decayed very slowly. The c3 site, formed within a sequence without dimers, appeared instead to be reclosed more rapidly and to a greater extent. DISCUSSION Cyclobutane pyrimidine dimers and (6 -4) photoproducts, the most common DNA damage induced by exposing DNA to short wave UV radiation (1), produce a small but significant local distortion of the double helix that can interfere with the proper functioning of enzymes acting on DNA, such as restriction endonucleases (12) and DNA topoisomerases. We have reported previously the inhibition of prokaryotic DNA topoisomerase I catalysis by UV damage in the target substrate (13,14). Analogous effect has been described for D. melanogaster DNA topoisomerase II (15). In the present study we report the consequences of UV damage on the activity of human DNA topoisomerase I.
Our results indicate that the presence of UV photodamage in the irradiated substrate inhibited topoisomerase I action under steady state conditions. As determined by DNA relaxation assay, the initial rate of topoisomerase I catalysis decreased by approximately 50% when ϳ40 CPDs were present per plasmid pAT153 molecule. This level of inhibition was lower than that previously reported for the prokaryotic type I topoisomerase from M. luteus (13) and eukaryotic type II from D. melanogaster (15).
Decreased relaxation activity by UV photodamage correlated with an interference of the enzyme's cleavage/religation equilibrium resulting in the stabilization of the cleavable complex. Cleavable complexes stimulation, measured as induction of single strand breaks in linear substrates, increased linearly with the dose. UV-dependent enhancement of topoisomerase I-mediated breakages appeared to be due to a reduction in the closure rate of the broken complexes. Thus, as established for camptothecin, the molecular mechanism by which UV damage inhibited DNA topoisomerase I catalytic activity seems to depend on its effects on the religation step. However, we cannot exclude the possibility that the presence of photolesions in the passing strand at the time of strand passage may slow down strand diffusion through the DNA-protein bridge.
Several authors have demonstrated that DNA topoisomerase I acts at preferred sequences lacking a clear consensus and that camptothecin has only minor effects on the enzyme's nucleotide specificity (30,31). Initial low resolution mapping of UV-stimulated cleavage sites on the pAT153 genome did not reveal major differences with respect to those stimulated by camptothecin except for one position in the promoter of the amp r gene. Subsequent localization at nucleotide level of the breakage sites in the region encompassing the uniquely UV-stimulated site and in the neighboring sequences was carried out and a comparison between the UV-stimulated sites with those induced by camptothecin was done. Some sites were uniquely stimulated by UV damage (c sites) and appeared at some distance from CPT-stimulated sites except for one (c4 site). Most CPT-stimulated breakages were not stimulated also by the presence of photoproducts (b sites) except for few sites (a sites). The relative position of b sites with respect to CPDs distribution did not reveal any special features that could give some clues to understanding why they were not stimulated also by UV damage. One exception is offered by the two b sites flanking sites a1 and a2 where breakages at both sites were within a thymines run that dimerized with high efficiency. In this case, it is possible to speculate that the cyclobutane bonds between the two thymines flanking the break at the b sites may affect either the cleavage and/or the binding of the enzyme to the substrate. This possibility is supported by the observation that cleavage stimulation by CPT on an UV-irradiated substrate was severely reduced at these two sites, while it was not affected at b sites formed at sequences without dimers in the scissible strand. Also the neighbor a1 and a2 sites showed a marked reduction in cutting frequency when cleavage was examined in the above condition, suggesting that UV damage stimulation was predominant at these sites and that CPTstimulated cleavage was effective only on the fraction of molecules without damage in the flanking sequences. However, an alternative explanation to this latter observation can be formulated based on the model of the camptothecin-topoisomerase I-DNA ternary complexes proposed by Pommier (reviewed in Ref. 31). According to this model, the planar ring of camptothecin should stuck with the base at the 5Ј terminus of the DNA breaks within the cavity formed at the topoisomerase I cleavage sites. Thus, the presence of UV lesions on the 5Ј terminus of the breaks produces conformational deformations that may alter the camptothecin receptor sites and consequently affect the action of the drug.
Numerous studies have shown that steric factors determine the interaction of topoisomerase I active site with DNA (35)(36)(37). Cleaved sites are characterized by a set of distinct local helical parameters (twisting) (26,38), and cleavage efficiency is modulated by stable or dynamic curvature of the DNA molecule (writhing) (28,38,39). Evidences that photodimers cause alterations in DNA structure are offered by numerous physicochemical, biochemical, and modeling studies. Measurements of the shift of phased A-tract multimers containing site-specific CPDs (40) and of changes in band pattern of UV-irradiated topoisomers (41,42) have shown that CPDs cause a topological unwinding due to a combination of actual duplex unwinding (twisting) and negative supercoiling (writhing) resulting from bending of DNA (40). Thus, we hypothesize that the alteration in the twist and writhe consequent to damage formation can determine changes in the structural context that may either dislocate the enzyme-DNA interacting sites of few bases (c2 and c4 site) or drive into the optimal conformation for topoisomerase I activity sequences normally poorly (c1 site) or not recognized by the enzyme (c3 site). This effect may resemble that described in supercoiled DNA, where the activation of new sites, not detected in relaxed DNA, has been observed (29). In this respect the c3 site, formed in a sequence with very low content of pyrimidine dimers in the strands surrounding the cleavage site, is of special interest. In fact, if the above hypothesis is correct, at least in this case one has to assume that CPD conformational changes are not only localized at the damaged pyrimidines but may be propagated into neighboring sequences as postulated by Pearlman et al. (43).
It is well established in the literature (1) and visible in Fig.  6 that the frequency of dimer formation varies at different potential dimer sites. Therefore, at the UV dose employed in the mapping experiments (Fig. 5), the irradiated fragments consist of an heterogeneous population of molecules with respect to UV damage frequency and position. Because of that, it is quite difficult to assess which of the lesions, that can be formed in the sequences surrounding the UV-stimulated sites, is effective in stimulating cleavage. In this respect, it cannot be excluded the possibility that also UV photodamage present on the complementary strand can determine a structural variation that may be reflected on the scissible strand.
Although camptothecin does not seem to change significantly the breakage specificity of the enzyme in vitro, it has a greater stabilizing effect on some breakage sites than on others (reviewed in Refs. 30 and 31). The enhancement breakage factor inversely correlates with the reclosure rate (32). UV-induced cleavage appeared to have different characteristics. The enhancement factor did not seem to vary significantly from one site to another. In addition, cleavage frequency at the sites identified in the EcoRI/PstI pAT153 fragment did not seem to be UV dose-dependent. Nevertheless, from the dose-response curve of overall cleavages, measured with the full-length DNA, there was a linear increase of cleavable complexes stimulation up to a dose that introduced approximately 100 CPDs per pAT153 molecule. This contradiction may be explained by assuming that as the UV dose increases damage is formed also at less probable positions. This results in additional deformation of DNA structure that activates many (potential) cleavable sites. However, each of them, being formed in low amount, is cleaved with a very low frequency so that they cannot be seen as discrete bands but as an increased background (Fig. 3). Furthermore, the religation rate, reduced with respect to that measured with CPT, did not correlate with the extent of cleavage but rather with the position with respect to UV photodimers. When CPDs were present within the sequence surrounding the cleavage sites, the rejoining step afterwards was very slow. This case is particularly evident for the a1 and a2 sites that are flanked on the 5Ј side of the break site by runs of thymines with an high probability to dimerize. These sites showed a much lower religation rate when stimulated by UV with respect to CPT stimulation. This type of kinetic can be explained in terms of the extent of misalignment that the two ends at the site of the nick have acquired as result of the helical distortion imposed by the damage. Analogous explanation has been proposed for topoisomerase I cleavage at a mismatch when present in one enzyme's recognition sequence (23).
The UV doses employed in this study exceed the level of radiation to which cells commonly are exposed. However, the presence of UV lesions in critically located topoisomerase I recognition sites could have physiological consequences.
DNA topoisomerases have been proposed as the enzymes responsible for sister chromatid exchange (44,45) and chromosomal aberrations (46). Several lines of evidence suggest a role for eukaryotic DNA topoisomerases in mediating "illegitimate recombination" of genetic material (reviewed in Ref. 47). The mechanistic link derives from the capacity of topoisomerase interactive drugs to stimulate these forms of genotoxicity (Ref. 48 and reviewed in Refs. 5 and 46). The biochemical basis for such a function comes from the ability of topoisomerases to mediate cleavage and religation in two half-reactions separated by the cleavable complex that gives the enzymes the capacity to catalyze intra-and intermolecular DNA transfer reactions (36,49). Moreover, it has been speculated that also the generation of chromosomal rearrangements by DNA damage could derive from their effects on topoisomerases (44,50). Distortion of the DNA structure in the vicinity of unrepaired DNA damage might be sufficient stimulus to alter the correct function of these enzymes such that incorrect rejoining could happen. However, UV damage does not stimulate in vitro the topoisomerase II-mediated cleavable complex (15). This observation implies that type II enzymes are unlikely to be the protein that mediate UV-induced cross-links and consequently responsible for UV-induced chromosome rearrangements. Conversely, our finding that UV damage interfered with the activity of DNA topoisomerase I by stabilizing the DNA-protein intermediate necessary for strand exchange, supports the notion that UVstimulated DNA-protein cross-links may be mediated by topoisomerase I and that this cross-linking may account for at least part of the chromosomal rearrangements induced by UV light.