Binding of the Escherichia coli UvrAB proteins to the DNA mono- and diadducts of cis-[N-2-amino-N-2-methylamino-2,2,1-bicycloheptane]dichloroplatinum(II ) and cisplatin. Analysis of the factors controlling recognition and proof of monoadduct-mediated UvrB-DNA cross-linking.

The interactions of the Escherichia coli endonuclease UvrAB proteins with the DNA mono- and diadducts of both the cis-racemic exo-[N-2-amino-N-2-methylamino-2,2,1-bicycloheptane]dichloroplatinum(II) (complex 1) and cisplatin (cis-diamminedichloroplatinum(II) (cis-DDP)), have been studied. Complex 1 reacts faster with DNA than cis-DDP and gives monoadducts with a longer lifetime (8 h 20 min chelation t compared with 2 h 40 min for cis-DDP). Using pSP65 plasmid [3H]DNA, the filter binding assay was associated with the analysis of the nucleoprotein complexes to characterize the UvrAB recognition of the platinum adducts and to demonstrate the occurrence of platinum-mediated DNA-protein cross-linking. First, it is shown that the UvrAB proteins recognize the complex 1 mono- and diadducts with a higher affinity than those of cis-DDP. Fifteen times more cis-DDP adducts per plasmid are required than complex 1 adducts, to lead to similar UvrAB binding. However, the UvrAB proteins recognize monoadducts and diadducts of each complex with a similar affinity. Second, it is shown that UvrB is the protein involved in the nucleoprotein complexes formed from mono- and diadducts of complex 1 and cis-DDP. This protein is also partly cross-linked to DNA with a similar efficiency by monoadducts derived from complex 1 and cis-DDP. However, as UvrB has a greater affinity for the DNA adducts of complex 1 than for those of cis-DDP, more UvrB-platinum-DNA cross-links are formed with complex 1 than with cis-DDP. This study, using a bacterial repair system as a model, points to a possible strategy for making new cytotoxic platinum complexes for mammalian cells.

the platinum compounds for the DNA major groove and also to decrease the chelation rate of the platinum monoadducts. Preliminary studies revealed that the rate of monoadducts formation from these complexes was higher than that observed for cis-DDP and that their chelation was indeed very slow. Several of these compounds, like the norbornyl derivative of Fig. 1 (complex 1), were found to exhibit strong antitumor properties (10,49). Furthermore, these compounds were able to overcome cis-DDP resistance in L1210 cells. These observations suggest that their mechanism of action might be different from that of cis-DDP and might involve a biological processing of the monoadducts. Platinum mono and diadducts are repaired in prokaryotic and eukaryotic cells (2,(11)(12)(13)(14)(15)(16)(17)(18). Recent studies on trans-1,2-diaminocyclohexanedichloroplatinum(II), which is also characterized by a bulky nonexchangeable ligand, revealed that monoadducts were more efficiently repaired than diadducts by UvrABC excinuclease (14). The recognition of monoadducts still bearing a labile ligand, chloro or aqua, could lead to its substitution by a protein nucleophilic group. Therefore, such monoadducts might become lethal lesions because of their ability to cross-link repair proteins to DNA. Such DNAprotein cross-linking was not observed with (DACH) dichloroplatinum complexes; however, the authors did not study the interaction of UvrABC excinuclease with monoadducts which still had a labile aqua or chloro ligand.
In order to test this hypothesis, the UvrAB recognition (19,20) of complex 1 and cis-DDP DNA adducts was studied using the filter binding assay. The determination of the rate constants of the first platination and chelation steps, together with the use of a technique which rapidly eliminates unbound platinum allowed us to discriminate between the recognition of chloro monoadducts and that of the final diadducts. The crosslinking of UvrAB proteins to DNA via chloroplatinum adducts was determined. The results for complex 1 and cis-DDP are discussed.

Reagents, Enzymes, and pSP65 Plasmid [ 3 H]DNA-cis-DDP
was provided by Rhône-Poulenc Rorer. The platinum complex 1 (Fig. 1) was synthesized by Dr. M. Barreau and P. Mailliet (Rhône-Poulenc Rorer). The UvrA and UvrB proteins used in these studies were prepared and purified as described (21). The purity of the proteins (more than 95%) was determined by SDS-polyacrylamide gel electrophoresis. Nonradiolabeled and 3 H-labeled pSP65 DNA (3005 base pairs, from Promega) were prepared as described (22). [ 3 H]Thymidine (60 Ci/mmol) was from Amersham Corp. [ 14 C]Thiourea (53 mCi/mmol) was from DuPont NEN. Nitrocellulose filters (BA85) were from Schleicher and Schuell. Membrane for microdialysis (VSWP 2500) was from Millipore. Sephadex G10 was from Pharmacia LKB. Stock solutions of platinum derivatives were prepared in dimethyl formamide, and dilutions were performed just prior to the experiments in NaClO 4 10 mM adjusted to pH 5.5. Thiourea was from Sigma. The thiourea solution was made fresh in Tris-HCl 0.6 M, pH 7.5, and deionized on AG 501-X8 resin (Bio-Rad). Ultrogel AcA22 was from Biosepra.
DNA Platination Kinetics-pSP65 plasmid DNA (6.9 M nucleotide concentration) was incubated at 37°C with platinum derivatives (1.7 M) in 100 ml of NaClO 4 10 mM, pH 5.5. At the times indicated, 5-or 10-ml aliquots were taken, and the reaction was stopped by adding NaCl (0.5 M final concentration). Unbound platinum was eliminated by precipitation with ethanol. This technique allows the elimination of more than 95% of unbound platinum. After precipitation, aliquots were concentrated 10 -20 times in perchloric acid (0.5 M final concentration). Then aliquots were incubated 2 h at 70°C. DNA content of aliquots was determined at 260 nm. The actual DNA concentration was calculated using a calibration curve established with DNA solutions of known concentration treated with perchloric acid under the same experimental conditions. Platinum content was determined using a furnace-equipped atomic absorption spectrometer Perkin-Elmer 5100. A calibration curve was made between 0 and 500 g/liter.
Kinetics of Monoadducts to Diadducts Chelation-The method was described previously (14). Briefly, 0.5 mg/ml salmon sperm DNA was incubated with cis-DDP (510 M) and compound 1 (230 M) in TEN buffer (2 mM Tris-HCl, 0.2 mM EDTA and 25 mM NaCl, pH 7.4) for 30 min at 37°C. These concentrations and the time of incubation were chosen in order to obtain 1.5 platinum adducts/100 bases.
The reaction was stopped by increasing the NaCl concentration to 0.5 M and cooling at 0°C. Samples were dialyzed twice versus high-salt TEN buffer (TEN plus 0.5 M NaCl) at 0°C. The samples were precipitated with 0.8 volume of isopropanol, suspended at 1 mg/ml in high-salt TEN. Samples were then dialyzed versus two changes of NaClO 4 10 mM, pH 5.5, for 2 h at 0°C in order to remove chloride.
After aliquots were removed for determination of platinum content by AAS and of DNA concentration, the samples were divided into 200-l aliquots and either left at 4°C (controls) or incubated at 37°C. The amount of monoadduct present at each time point was determined by adding 20 l of 110 mM [ 14 C]thiourea (specific radioactivity of 5.23 Ci/mol) and incubating for 10 min at 37°C. The reaction was stopped by adding 2 ml of cold 5% trichloroacetic acid and cooling to 4°C. In order to reduce the background, samples were pelleted, the supernatant was removed by aspiration, and the pellet was redissolved in 0.2 ml of 0.1 N NaOH. After reprecipitation of the DNA with cold 5% trichloroacetic acid, the samples were collected by filtration on a glass fiber filter for further analysis.
Plasmid DNA Platination and Nucleoprotein Filter Binding Assay-Since nonlabeled DNA and [ 3 H]DNA were kept in TE buffer (10 mM Tris-HCl and 2 mM EDTA), samples (10 -50 l) of these DNA were microdialyzed against NaClO 4 10 mM, pH 5.5 at 0°C for 2 h prior to platinum treatment. [ 3 H]DNA (nucleotide concentration of 6.9 M) was treated with platinum compounds at a drug:nucleotide ratio of 1/4 and incubated in 10 mM NaClO 4 , pH 5.5, at 37°C. At the indicated time, the DNA platination was terminated by two consecutive 2-min centrifugations of 250-l aliquots on 2.5-ml Sephadex G10 columns that had been equilibrated previously with 10 mM NaClO 4 , pH 5.5. Control experiments revealed that this technique allows the elimination of more than 99% of unbound platinum complex. In addition, it was verified that no significant platination occurred during the centrifugation (data not shown).
The formation of stable nucleoprotein complexes between UvrAB and platinated DNA was measured by trapping the native protein-DNA complexes onto nitrocellulose filters as described (22). The 140-l reaction mixture consisted of 40 mM potassium morpholinopropane sulfonate buffer at pH 7.5, 85 mM KCl, 1 mM dithiothreitol, 15 mM MgSO 4 , 2 mM ATP (binding buffer), 200 fmol (each protein) of UvrA and UvrB, and 14 fmol of [ 3 H]DNA treated or not treated with platinum compounds. After 10-min incubation at 37°C, the reaction was stopped by addition of 5 ml of cold 2 ϫ SSC (0.3 M NaCl, 0.03 M trisodium citrate). After 5 min in ice in order to eliminate nonspecific binding, the nucleoprotein complexes were collected by trapping onto a BA85 nitrocellulose filter.
All the data were analyzed using nonlinear regression (GraphPAD program, San Diego, CA).
In the text, the results of the binding assay (Figs. 4A and 5A) are expressed as t1 ⁄2 values which correspond to the incubation time between DNA and the platinum complex required to obtain half the fraction of DNA retained on the filter at infinite time.
Since platinum derivatives interact preferentially with guanine that can be considered to be distributed at random, the location of platinum adducts results in a Poisson distribution. The UvrAB proteins binding reflects the amount of binding of platinum to DNA. Therefore, the number of proteins bound specifically to platinum adducts is also described by a Poisson distribution. The apparent average number of proteins bound to DNA (⑀u) can be expressed as follows, where ⑀ is the filter binding efficiency which is considered as the probability that a protein complexed to DNA at a particular site, is held on the filter, u is the average number of proteins bound to DNA, and f is the fraction of DNA retained on the filter (23,24). Plasmid DNA Platination and Isolation of DNA-Protein Complexes-An isotopic dilution of 3 H-labeled pSP65 (10,000 cpm/g) at a nucleotide concentration of 6.9 M was treated with platinum derivatives at a 1/4 drug:nucleotide ratio. Incubation was performed in 10 mM NaClO 4 , pH 5.5, at 37°C for 30 and 70 min, respectively, for complex 1 and cis-DDP. Such incubation times were chosen in order to obtain 30 and 40 platinum adducts, respectively, of complex 1 and cis-DDP (see Fig. 2). Reactions were stopped by addition of NaCl (0.5 M final concentration). Unbound platinum was eliminated by DNA precipitation with ethanol. DNA pellets were suspended in NaClO 4 10 mM, pH 5.5, and filtered through a Sephadex G50 column equilibrated with NaClO 4 . One part of the sample was used for the UvrAB binding, the other part was incubated 20 h at 37°C prior to the UvrAB binding, in order to allow the conversion of monoadducts into diadducts.
The UvrAB binding and the isolation of DNA-protein complexes were performed using a method described previously (25,26) after slight modifications. This method was as follows: 6.7 nM of 3 H-labeled pSP65, treated as above, was incubated with UvrA and UvrB proteins, both 94 nM, at 37°C in 80 or 140 l of binding buffer containing 20% glycerol. After 10 min of incubation at 37°C, samples were loaded onto an AcA22 gel exclusion column (1.0 ϫ 13 cm) equilibrated in the same binding buffer with 300 mM KCl at 25°C. The column was run at 4 ml/h, and 700-l fractions were collected. The DNA content of each fraction was quantified by scintillation counting of 50-l aliquots. For protein determination, aliquots corresponding to the peak fraction and equivalent either to 90 or 120 ng of DNA were analyzed by SDS-PAGE and silver staining.

RESULTS
DNA Platination-Plasmid DNA (6.9 M) was treated with the platinum complexes (r b ϭ 0.25, where r b ϭ adducts per base ratio) as described under "Materials and Methods." The quantification of platinum bound to DNA (i.e. DNA-platinum monoadducts and diadducts) was performed using AAS. The results presented in Fig. 2 show that DNA platination involves at least two consecutive steps. They show that, in our experimental conditions, complex 1 reacts faster with DNA than cis-DDP. Further investigation of the kinetic scheme is underway.
Kinetics of Monoadducts to Diadducts Chelation-The chelation reaction was monitored by trapping the monoadducts with [ 14 C]thiourea, as reported by Page et al. (14). In Fig. 3, the kinetics of chelation of the monoadducts derived from complex 1 is compared with that of the monoadducts from cis-DDP. The chelation t1 ⁄2 values are, respectively, 8 h 20 min and 2 h 40 min for complex 1 and cis-DDP.
UvrAB Binding to Platinated DNA in Which Most of the Adducts Are Monoadducts-DNA was treated with compound 1 and cis-DDP at a r b ϭ 0.25. At the times indicated in Fig. 4, aliquots were taken, and the small molecules in them were removed by gel filtration in order to eliminate unbound plati-num before incubation of the DNA fractions with the UvrAB proteins. The filter binding assay was then performed (see "Materials and Methods"). The DNA-UvrAB protein complex versus DNA-platination time relationship is a product of the platinum-adduct formation rate and the UvrAB binding efficiency to a given platinum-adduct. The results are presented in Fig. 4A. The amount of nucleoprotein complexes increases with the DNA-platination time. The formation of the nucleoprotein complexes is faster for complex 1 than for cis-DDP with respective t1 ⁄2 of 2.4 Ϯ 0.11 min and 39 Ϯ 0.9 min. Since in these experiments DNA was treated with the platinum complexes in the same experimental conditions as those used for the quan- FIG. 4. UvrAB binding to DNA treated with the platinum complexes to form mainly monoadducts. DNA was incubated with the platinum complexes at 37°C in NaClO 4 (10 mM, pH 5.5) as described under "Materials and Methods." At the times indicated, unbound platinum complexes were removed by two spun G10 chromatographies, and platinated DNA was incubated with UvrA and UvrB proteins. Then, the filter binding assay was performed (see "Materials and Methods"). Results are the mean of three independent experiments and are expressed with the standard error of the mean. A, UvrAB binding to platinated DNA as a function of the incubation time between DNA and the platinum complexes. q, complex 1; s; cis-DDP. The results were analyzed by nonlinear regression using GraphPad system (San Diego, CA). The best fits were obtained with exponential and sigmoid curves respectively for complex 1 (R 2 : 0.98) and cis-DDP (R 2 : 0.999). B, comparison of the relative UvrAB recognition efficiencies for DNA-monoadducts of complex 1 and cis-DDP. The apparent number of proteins bound per plasmid DNA (⑀u) was determined from the fitted curves of A according to Equation 1, where ⑀ is the filter retention efficiency (see text), and u is the average number of proteins bound to DNA. The amount of platinum per plasmid was computed from the fitted curves of tification of platinum bound to DNA, the apparent average number of proteins bound to platinated DNA (⑀u) as a function of the amount of platinum adducts can be derived (Fig. 4B). Linear regressions gave ⑀u ϭ 0.3 ϫ platinum adducts per plasmid and 0.02 ϫ platinum adducts per plasmid, respectively, for complex 1 and cis-DDP. One needs 15 times more adducts from cis-DDP than from complex 1 to obtain an identical apparent average number of proteins bound to DNA. The filter binding assay was performed with different levels of platination in the same experimental conditions for both platinum compounds (i.e. ⑀ can be assumed to be the same). This allows us to compare the relative UvrAB recognition efficiencies on platinated DNA obtained with complex 1 and cis-DDP. Therefore, UvrAB proteins recognize complex 1 monoadducts with a better efficiency than those from cis-DDP.
Recognition of the DNA Diadducts by UvrAB Proteins-For the complexes used in this study, the chelation step of the monoadducts to form the diadducts is much slower than the first platination step (Figs. 2 and 3). Therefore, one can compare the interaction of UvrAB proteins with both the mono-and diadducts formed with complexes 1 and cis-DDP. For compound 1, the slow chelation observed in Fig. 3 suggests that, in our conditions, even at a time of 120 min, the binding of the UvrAB proteins to the platinated DNA mainly corresponds to the formation of the nucleoprotein complexes with monoadducts.
In order to determine the UvrAB binding efficiency on DNA diadducts, we allowed the monoadducts free of unbound platinum compounds to undergo the chelation step before addition of the UvrAB proteins. At various platination times, samples were filtered on G10 columns and incubated for 20 h at 37°C. This length of incubation was shown to produce more than 80 and 99% diadducts, respectively, from complex 1 and cis-DDP (data extrapolated from Fig. 3). The filter binding assay yielded the following t1 ⁄2 of incubation to obtain half the maximum protein binding: 1.9 Ϯ 0.44 and 32.5 Ϯ 1.4 min, respectively, for complex 1 and cis-DDP (Fig. 5A). The apparent number of proteins bound to DNA as a function of platinum per plasmid was analyzed. Linear regression gave ⑀u ϭ 0.3 ϫ platinum adducts per plasmid and 0.03 ϫ platinum adducts per plasmid, respectively, for complex 1 and cis-DDP. About 10 times more cis-DDP diadducts than complex 1 diadducts are needed to obtain a similar apparent average number of proteins bound to DNA (Fig. 5B). Since the experimental conditions were identical to those described for monoadducts (similar ⑀), UvrAB proteins recognize complex 1 diadducts with a better efficiency than those formed from cis-DDP.
Determination of the Uvr Protein Involved in the Nucleoprotein Complexes-It has been shown that the UvrA protein delivers UvrB at the site of damage on DNA. However according to the UvrA-to-UvrB ratio in our reaction mixture, UvrA could also be involved in the nucleoprotein complex (25). The filter binding assay does not allow for the determination of which UvrA and/or UvrB protein(s) is/are involved in the nucleoprotein complexes. In order to address this issue, UvrA and UvrB proteins were incubated with nonplatinated DNA or platinated DNA containing either mono-or diadducts from each platinum compound. The DNA-protein complexes were isolated using a size exclusion chromatography, and their protein content was analyzed by SDS-PAGE and silver staining according to Orren and Sancar (25,26). Higher concentrations of pSP65 DNA and Uvr proteins than those used for the filter binding assay were required to allow the Uvr protein determination by silver staining. However, in order to allow a relevant comparison with the results obtained with the filter binding assay, the same UvrA-to-UvrB and Uvr protein-to-plasmid molecule ratios were kept identical in the reaction mixtures in both experiments. The results are presented in the Fig. 6. No UvrA and/or UvrB binding was observed with nonplatinated DNA. For platinated DNA, a major band corresponding to the UvrB protein was observed whatever the nature of the damage (mono-or diadducts) and whatever the platinum compound used (complex 1 or cis-DDP).
Determination of SDS/EDTA-resistant Cross-links between Platinated DNA and the UvrAB Proteins-According to the previous results, UvrAB binding to the platinum monoadducts occurs much faster than their chelation. Therefore, a reaction between these chloro monoadducts and a nucleophilic group of the UvrAB proteins might lead to covalent cross-linking. Such cross-linking cannot occur with diadducts. The UvrAB binding was performed on DNA mono-and diadducts. Protein-DNA cross-linking was evaluated using the resistance of the nucleoprotein complexes to a denaturing treatment by 0.2% SDS, 50 FIG. 5. UvrAB binding to DNA-platinum diadducts. DNA was treated as described in the legend to Fig. 4, but after filtration on G10 column, the samples were incubated 20 h at 37°C in NaClO 4 10 mM, pH 5.5, to allow the conversion of monoadducts to diadducts before incubation with UvrAB proteins. Then, the filter binding assay was performed. All the data were analyzed as described in the legend to Fig. 4. A, UvrAB binding to DNA platinum diadducts as a function of the ANA platination time. The best fits were obtained with exponential and sigmoid curves, respectively, for complex 1 ( mM EDTA. EDTA was used to chelate the Mg 2ϩ cation. Without ATP-Mg 2ϩ , the noncovalent UvrAB-DNA nucleoprotein complexes dissociate and cannot rebind (22). An incubation with SDS was added to the EDTA treatment in order to eliminate hydrophobic interactions between Uvr proteins and the DNA. In these conditions no nucleoprotein complexes were observed with a UV-treated DNA containing pyrimidine dimers (data not shown). The results are presented in Fig. 7. The amount of SDS/EDTA-resistant nucleoprotein complexes of platinated DNA with UvrAB increases with the amount of platinum adducts. Much more SDS/EDTA-resistant complexes are formed when the UvrAB proteins are added to the DNA immediately after the platination step than after the supplementary 20-h incubation period (Fig. 7A). When the apparent average number of proteins bound to DNA, obtained after the dissociating SDS/EDTA treatment, is expressed as a function of the number of platinum adducts per plasmid (Fig. 7B), the monoadducts formed from complex 1 appear to be more efficient than those formed from cis-DDP to induce cross-linking between the UvrAB proteins and platinated DNA.
Determination of the Uvr Protein Cross-linked to DNA via Platinum Monoadducts-The isolation and analysis of nucleoprotein complexes by SDS-PAGE and silver staining allowed the detection of the bound but noncross-linked UvrB protein. In these conditions, Uvr proteins cross-linked to DNA stay in the well resulting from the nonmigration of the pSP65 plasmid DNA in the SDS-polyacrylamide gel. Such cross-linked Uvr proteins could only migrate in the gel after a cyanide or a thiourea treatment which are known to dissociate platinum adducts from DNA (27,28). To identify the protein cross-linked to DNA, we used the following two steps procedure. First, nucleoprotein complexes were isolated and analyzed by SDS-PAGE and silver staining before and after SDS/EDTA treatment. The results are shown in Fig. 8. First, the SDS/EDTA treatment before gel exclusion column leads to the disappearance of the UvrB protein. Such a treatment allows the dissoci-ation of the noncross-linked UvrB protein from the platinated DNA, whatever the platinum derivative used. Second, SDS/ EDTA-treated pSP65 DNA was treated with thiourea, using conditions which dissociate the platinum adducts from the DNA (28). A cyanide treatment could not be used because of its interference with the silver staining step. The results are shown in Fig. 9. A band corresponding to the UvrB protein appears only for the pSP65 DNA which contains monoadducts of complex 1 and not for that containing either cis-DDP monoadducts or complex 1 diadducts. DISCUSSION The racemic complex, cis-[N-2-amino N-2-methylamino-2,2,1-bicycloheptane]dichloroplatinum(II) with exo configuration (complex 1) bears a bulky hydrophobic norbornyl substituent on the ethylenediamine ligand. It was designed to have a higher affinity than cis-DDP for DNA major groove and also to give monoadducts with longer lifetimes.
Two questions were addressed in this work: (i) do the UvrA and/or UvrB proteins of the Escherichia coli repair system exhibit different recognition efficiencies for the complex 1 and cis-DDP DNA adducts and for the mono-and diadducts of each complex and (ii) does complex 1 yield long-living monoadducts and are such adducts able to cross-link the UvrA and/or UvrB proteins to DNA?
First Platination and Chelation Steps-AAS monitoring of DNA platination by complex 1 shows that it is faster than that by cis-DDP (Fig. 2). The platination profiles of Fig. 2 are in agreement with a two-step process as previously observed for the reaction of DNA fragments with cis-DDP, the first one being the aquation of the chloro complex (29,30).
UvrAB Recognition of the DNA-Platinum Mono-and Diadducts-Using the filter binding assay, we compared the UvrAB recognition of platinum adducts formed with complex 1 and cis-DDP. The results show that the UvrAB proteins recognize, FIG. 7. Cross-linking between UvrAB proteins and platinated DNA. DNA samples containing increasing amounts of monoadducts or diadducts were prepared as described in the legends of Figs. 4 and 5. The platinated DNA was incubated with UvrAB proteins as described under "Materials and Methods." After a 10-min incubation with UvrAB proteins, samples (140 l) were divided into two parts. One part was used as a control for the filter binding assay (data not shown). To the second part, SDS and EDTA (0.2% SDS, 50 mM EDTA final concentration) were added, and the samples were incubated at 37°C for 30 min. After dilution by 5 ml of warmed 2 ϫ SSC buffer and an incubation at 37°C for 5 min, the nucleoprotein complexes resistant to SDS/EDTA treatment were collected onto nitrocellulose filter. respectively, the monoadducts and diadducts of complex 1 with a higher efficiency than the monoadducts and diadducts of cis-DDP. Furthermore, for complex 1, the UvrAB proteins recognize the mono-and diadducts with the same efficiency. For cis-DDP, a slightly better recognition is found for the diadducts. The nucleoprotein complexes formed, with an UvrAto-UvrB stoichiometric ratio, from the adducts of complex 1 and cis-DDP were isolated and analyzed by SDS-PAGE as such or after SDS/EDTA dissociation. Apart from tiny amounts of UvrA, UvrB was the major protein identified. The SDS/EDTA treatment led to the disappearance of stable nucleoprotein complexes with the filter binding assay (Fig. 7) together with the absence of UvrB on SDS-polyacrylamide gel (Fig. 8), suggesting that it is the only one involved in the stable recognition complex. Orren and Sancar (25) using the same UvrA-to-UvrB ratio found that UvrA was also involved in the nucleoprotein complexes in about twice the amount of UvrB. However, the UvrA concentration (Ͼ125 nM) was higher than ours (94 nM) with an UvrA-to-plasmid ratio higher than 14.
The UvrABC endonuclease is known to recognize a variety of DNA damages (19). This broad spectrum led the authors to postulate that the UvrABC endonuclease rather recognizes the DNA structural deformation induced by the various adducts than the characteristic features of each adduct. Despite the absence of structural data on the mono-and diadducts of complex 1 with DNA, it is likely that they are similar to those of the corresponding cis-DDP adducts. For the latter, it is known that the square planar coordination of platinum(II) is the major factor controlling the stereochemistry of the adducts and that it induces DNA bending for the intrastrand diadducts (31) but not for the monoadducts (32)(33)(34). Therefore one expected different UvrAB recognition of the mono-and diadducts for both complex 1 and cis-DDP. The UvrAB proteins recognize a variety of adducts even the O 6 -methylguanine ones which cause little distortion of the DNA helix (35). Our results showing a similar recognition of both the mono-and diadducts of complex 1 and cis-DDP suggest that simple structural determinants do not, by themselves, control the UvrAB recognition. Our UvrAB results for a population of lesions located randomly on a supercoiled DNA can be compared with the UvrABC repair results of defined adducts reported by Page et al. (14). They found a better UvrABC repair efficiency for the [G]Pt(dien) or [G]Pt-Cl(en) and [G]PtCl(DACH) monoadducts trapped with dithiothreitol than for the [GG] diadducts of cis-PtCl 2 (DACH). Visse et al. (36) concluded from footprinting experiments that the recognition complex involves binding of the proteins to the convex side of the bend formed by the platinum GG-chelate. Clearly, this does not explain the similar recognition of the mono-and diadducts of complex 1 and of cis-DDP. The norbornyl ethylenediamine ligand is bulkier than the two ammine ones, and molecular mechanics modeling of the mono-and diadducts shows that both remain in the major groove of the straight and bent structures (not shown). The more efficient binding of UvrB to both mono-and diadducts of complex 1 suggests that hydrophobic interactions might be favorable to recognition in the major groove area. Several results suggest that the UvrABC, or actually UvrAB recognition efficiency, is not directly related to the UvrABC incision efficiency. The latter depends upon the torsional constraint of the damaged DNA, whereas the former does not (22,37). Using supercoiled DNA, Beck et al. (12) showed that UvrABC incised cis-and trans-DDP-treated DNA with the same efficiency, whereas using linear DNA, they found that only the one treated with cis-DDP was incised. UvrABC incision efficiency is also dependent upon the sequence of the damaged DNA strand. It has been observed that the UvrABC incision efficiency of a DNA treated with acetylaminofluorene varied with the DNA sequence, whereas recognition did not (38). All these results suggest that the UvrAB recognition and UvrC incision steps might obey different binding criteria. The recognition would work for bent DNA structures as well as for any type of adducts, particularly with bulky hydrophobic groups, whereas incision would be more sensitive to the DNA sequence and the torsional constraint (39).
UvrB-platinum-DNA Cross-links-Thanks to the slow chelation reaction of our platinum monoadducts (Fig. 3), we studied the interaction of the UvrAB proteins with monoadducts still bearing an exchangeable (chloro or aqua) ligand on platinum. The results show that only UvrB is cross-linked to the DNA and that only the monoadducts of complex 1 are able to cross-link efficiently the protein. The mechanism of the cross-linking is unknown. However, it could result from a substitution of the labile ligand by a nucleophilic group of the protein such as a histidine imidazole or more likely a thioether group of a methionine or a thiol group of a cysteine, within the UvrB-DNA recognition complex. The substitution of the chloro ligand of a DNA platinum monoadduct by a sulfur nucleophile from the UvrB protein does not require an aquation step (40) and should be an efficient reaction, provided a favorable proximity would exist between the platinum and the sulfur nucleophile. Platinum derivatives react with thiol-containing peptides or proteins (41)(42)(43) and could react with the UvrB cysteine residues (44). That such a reaction requires UvrB binding to the platinated DNA is shown by the absence of UvrB cross-linking when the UvrAB proteins are incubated with the platinated DNA under SDS/EDTA dissociating conditions (data not shown). The UvrB cross-linking could be facilitated by its close contact with DNA when the preincision complex is formed (26,45). It is noteworthy that a small amount of UvrB-platinum-DNA crosslinked nucleoprotein complexes was also observed following the UvrAB treatment of DNA platinated with compound 1, after 20-h incubation to allow the chelation of the monoadducts. Such cross-linking probably results from the presence of remaining nonchelated monoadducts. According to the slow chelation kinetics of the complex 1-DNA monoadducts (Fig. 3), a 20-h incubation after the first platination is less than three times the 8 h 20 min chelation t1 ⁄2 and therefore too short to allow a complete conversion of the monoadducts (46). Fig. 7B shows that cross-linking between the UvrB proteins and platinated DNA occurs at very small platination ratios for complex 1, whereas it is not observed for cis-DDP until an average of 20 FIG. 9. Demonstration of the formation of cross-linked UvrBplatinum-DNA complexes using a thiourea treatment. Aliquots (120 ng of DNA) corresponding to SDS/EDTA-treated samples of Fig. 8  (lanes b and d) were incubated for 24 h at 37°C in the presence (ϩ) or in the absence (Ϫ) of 1 M thiourea, 0.3 M Tris-HCl, pH 7.5. In the absence of thiourea, the samples were incubated in 0.3 M Tris-HCl, pH 7.5. Then samples were analyzed by SDS-PAGE and silver staining. St, UvrA and UvrB standards; lanes a and b, DNA containing, respectively, 30 and 40 monoadducts of complex 1 and cis-DDP; lane c, DNA containing 30 diadducts of complex 1. ૺ, limit of the stacking gel. As already observed in the Fig. 6, platinated DNA does not enter the stacking gel until it was treated with thiourea treatment which dissociates platinum adducts from DNA. It has to be noted that the thiourea treatment does not interfere with silver staining. Furthermore, a silverstained smear is observed in all the wells even in the unloaded wells. Such a smear results from the heavy silver staining of the gel. adducts per plasmid (3005 base pairs) are formed. This can be correlated to the fact that 15 times more adducts per plasmid are required for cis-DDP than for complex 1 to lead to similar UvrB binding and explains the fact that no cross-linked UvrB protein was observed after the analysis by SDS/PAGE (Fig. 9). Actually, once the noncovalent nucleoprotein complex is formed, covalent cross-linking appears to be comparable for both the monoadducts of complex 1 and cis-DDP. However, since the UvrB protein has a greater affinity for complex 1-DNA adducts than for those of cis-DDP, more UvrB-platinum-DNA cross-links are formed with complex 1 than with cis-DDP. Therefore, complex 1-DNA monoadducts appear as a good tool to study the biological consequences of protein-DNA cross-linking.
In this work, bacterial UvrAB proteins were used for their ability to recognize DNA lesions induced by platinum monoadducts and diadducts. In mammalian cells, the specificity of the recognition and incision steps as well as the nature of the proteins are less characterized. Several proteins involved in repair or in transcription processes have been isolated for their ability to recognize platinum DNA adducts (2,47,48). Analogues of cis-DDP able to react quickly with DNA, to give long-lived monoadducts able to cross-link repair proteins may be good candidates to overcome the repair-mediated cis-DDP resistance (2).