Different recognition of DNA modified by aatitumor cisplatin and its clinically ineffective trans isomer by tumor suppressor protein p53.

The p53 gene encodes a nuclear phosphoprotein that is biologically activated in response to genotoxic stresses including treatment with anticancer platinum drugs. The DNA binding activity of p53 protein is crucial for its tumor suppressor function. DNA interactions of active wild-type human p53 protein with DNA fragments and oligodeoxyribonucleotide duplexes modified by antitumor cisplatin and its clinically ineffective trans isomer (transplatin) were investigated by using a gel mobility shift assay. It was found that DNA adducts of cisplatin reduced binding affinity of the consensus DNA sequence to p53, whereas transplatin adducts did not. This result was interpreted to mean that the precise steric fit required for the formation and stability of the tetrameric complex of p53 with the consensus sequence cannot be attained, as a consequence of severe conformational perturbations induced in DNA by cisplatin adducts. The results also demonstrate an increase of the binding affinity of p53 to DNA lacking the consensus sequence and modified by cisplatin but not by transplatin. In addition, only major 1,2-GG intrastrand cross-links of cisplatin are responsible for this enhanced binding affinity of p53. The data base on structures of various DNA adducts of cisplatin and transplatin reveals distinctive structural features of 1,2-intrastrand cross-links of cisplatin, suggesting a unique role for this adduct in the binding of p53 to DNA lacking the consensus sequence. The results support the hypothesis that the mechanism of antitumor activity of cisplatin may also be associated with its efficiency to affect the binding affinity of platinated DNA to active p53 protein.

The p53 gene encodes a nuclear phosphoprotein that is biologically activated in response to genotoxic stresses including treatment with anticancer platinum drugs. The DNA binding activity of p53 protein is crucial for its tumor suppressor function. DNA interactions of active wild-type human p53 protein with DNA fragments and oligodeoxyribonucleotide duplexes modified by antitumor cisplatin and its clinically ineffective trans isomer (transplatin) were investigated by using a gel mobility shift assay. It was found that DNA adducts of cisplatin reduced binding affinity of the consensus DNA sequence to p53, whereas transplatin adducts did not. This result was interpreted to mean that the precise steric fit required for the formation and stability of the tetrameric complex of p53 with the consensus sequence cannot be attained, as a consequence of severe conformational perturbations induced in DNA by cisplatin adducts. The results also demonstrate an increase of the binding affinity of p53 to DNA lacking the consensus sequence and modified by cisplatin but not by transplatin. In addition, only major 1,2-GG intrastrand crosslinks of cisplatin are responsible for this enhanced binding affinity of p53. The data base on structures of various DNA adducts of cisplatin and transplatin reveals distinctive structural features of 1,2-intrastrand cross-links of cisplatin, suggesting a unique role for this adduct in the binding of p53 to DNA lacking the consensus sequence. The results support the hypothesis that the mechanism of antitumor activity of cisplatin may also be associated with its efficiency to affect the binding affinity of platinated DNA to active p53 protein.
It is well established that platinum coordination complexes exhibit antitumor effects (1,2). The success of platinum complexes in killing tumor cells results from their ability to form on DNA various types of covalent adducts (3). The first platinum complex introduced in the clinic is cis-diamminedichloroplati-num(II) (cisplatin) 1 (1,4). Although the antitumor effects of cisplatin were discovered more than 30 years ago, the mechanism of its antitumor activity has not yet been fully understood. It has been shown (3) that this bifunctional platinum complex forms on DNA mainly intrastrand cross-links (CLs) between neighboring purine residues (ϳ90%). Other minor adducts are intrastrand CLs between two purine nucleotides separated by one or more nucleotides and interstrand CLs; a few adducts remain monofunctional. Transplatin (trans isomer of cisplatin) is clinically ineffective; thus both isomers have been widely used in studies of the structure-pharmacological activity relationship of platinum complexes (5). In these studies, one searches for differences between active and inactive compounds that may be responsible for the differences in the pharmacological effect. Transplatin-DNA adducts are interstrand CLs, and a relatively large portion of adducts remains monofunctional (6,7).
The adducts formed on DNA by bifunctional platinum compounds are capable of terminating DNA synthesis (1) and triggering several cellular processes such as apoptosis (8) or repair of lesions (9,10). It has become clear that under some circumstances p53 can play a major role in the activation of apoptosis (11,12) and actively participates in various processes of DNA repair via its ability to interact with components of the repair machinery and by its various biochemical activities (13).
The tumor suppressor protein p53 is a nuclear phosphoprotein consisting of 393 amino acids and containing four major functional domains (14). The transcriptional activation domain is located at the N terminus, whereas the sequence-specific DNA-binding domain is within the central part of p53. The C-terminal portion, which interacts with DNA in a nonspecific manner, contains an oligomerization and regulatory domains. p53 is a potent mediator of cellular responses against genotoxic insults (13) that exerts its effect through transcriptional regulation. Upon exposure to genotoxic compounds, p53 protein levels increase due to several post-transcriptional mechanisms. Cisplatin induces apoptosis in cells expressing either wild-type (wt) or mutant p53 so that a consensus on the significance of p53 for response to cisplatin has not been attainable (15). Nevertheless, on average, cells with mutant p53 are more resistant to the effect of cisplatin (16). Hence, it seems reasonable to conclude that p53 can control the processing of DNA adducts of platinum, depending on the cell type.
The tumor suppressor function of p53 protein is crucially related to its DNA binding activity. Active wt p53 binds as a tetramer to over 100 different response elements naturally occurring in the human genome. These response elements, which show functionality, differ in the details of their specific base sequence, but all contain two tandem consensus decamers, each a pentameric inverted repeat. Most consensus decamers, separated in the binding unit by 0 -21 base pairs (bp), follow the consensus sequence pattern (17) PuPuPuC(A/t)͉(T/ a)GPyPyPy, where Pu and Py are purines and pyrimidines, respectively, and the vertical bar denotes the center of pseudodyad symmetry. Four molecules of the DNA-binding domain of p53 bind the response elements with high cooperativity. They also bend DNA. It has been suggested (18) that this bending is localized mostly at the two pentamer CA͉TG junctions in the consensus DNA response element (CDRE) (by 25-28°at each junction) toward the major groove. Active wt p53 also over-twists the DNA response element by ϳ70°. This DNA twisting is uniformly distributed among the pentamers. It has also been suggested that due to many functions of p53 protein the demands for binding specificity and selectivity are necessarily extraordinary, which is accomplished through its tetrameric association with a repetitive binding site. Precise steric fit accommodated through both DNA bending and twisting appears extremely important in this binding site. Because DNA bending and twisting are coupled in the p53-DNA complex, the binding specificity of the p53 system as well as complex stability could be fine-tuned by agents that affect DNA bending and twisting. It has been shown (3) that DNA adducts of cisplatin and other platinum bifunctional compounds distort the conformation of DNA, including bending and changes in the twist angle. Thus, formation of the adducts by platinum compounds in the CDRE could affect its binding affinity to p53. No direct interaction between active wt p53 and platinum-modified DNA in cell-free media has yet been reported. Only binding of latent p53 protein, lacking sequence-specific DNA binding, to DNA modified by cisplatin has been reported (19). In the present work, we have used gel mobility shift assay methodology to investigate the binding affinity of the active human wt p53 system to the CDRE modified by either antitumor cisplatin or its clinically inefficient trans isomer in a cell-free medium. We have also examined binding of DNA lacking CDRE and modified by cisplatin or transplatin to reveal affinity of active wt p53 to platinated DNA containing no consensus nucleotide sequence. Thus, these studies could provide insight into the relative cytotoxicities of these two isomers, thereby potentially aiding in the rational design of new platinum drugs as well as illuminating aspects of the role of p53 in chemotherapy by platinum compounds.
Purification of the Active Wild-type Human p53 Protein-The human wt p53 protein was expressed in baculovirus-infected recombinant Sf9 insect cells. The details of the purification were described previously (22). The protein concentration was determined by the Bradford method. In the present paper the concentration of the p53 protein is related to tetrameric protein units.
Platination Reactions-Cisplatin and transplatin were purchased from Sigma. Monodentate diethylenetriaminechloroplatinum(II) chlo-ride ([Cl(dien)Pt]Cl) was kindly provided by Dr. G. Natile (University of Bari, Italy). Short PvuII fragments of pPGM1 and pBluescript II SK ϩ plasmid DNAs (474 and 448 bp, respectively) and an oligonucleotide duplex (oligo-CDRE) were incubated with cisplatin or transplatin in 10 mM NaClO 4 at 37°C for 48 h in the dark. The number of platinum atoms bound per nucleotide (r b value) was determined by flameless atomic absorption spectrophotometry or differential pulse polarography (23). The number of interstrand CLs formed by cisplatin in the oligo-CDRE duplex was determined using polyacrylamide gel electrophoresis (PAGE) under denaturing conditions in the same way as described in previous reports (24,25). The oligonucleotide duplexes containing single site-specific adducts of cisplatin were prepared and characterized as described previously (20, 26 -28).
Preparation of DNA-Protein Complexes-Formation of the complexes of the p53 protein with the 474-or 448-bp-long PvuII fragments of pPGM1 or pBluescript SK II ϩ , respectively, unplatinated or modified by cisplatin or transplatin was examined in a buffer containing 5 mM Tris-HCl, pH 7.6, 0.5 mM Na 3 EDTA, 50 mM KCl, 0.01% Triton X-100 in a total volume of 15 l. The nonmodified or platinated 474-or 448-bp fragment was mixed with a nonmodified 2513-bp-long fragment of pPGM1. The final amounts of the short and long fragments in the reactions were 150 and 850 ng, respectively (the molar ratio of these fragments was ϳ1). The molar ratio of p53 to the 474-or 448-bp fragment was 0 -6. Samples with p53 protein were incubated in ice for 30 min. After the incubation was completed, 3 l of the loading buffer (50% glycerol, 50 mM Na 3 EDTA, 2% bromphenol blue) was added, and the samples were loaded on the 1% agarose gel precooled to 4°C and electrophoresed in 0.5ϫ TBE buffer (0.09 M Tris borate, 2 mM Na 3 EDTA (pH 8.0)). The gel was finally stained by ethidium bromide.
Formation of the complexes of p53 with the oligonucleotide duplexes was examined in the same buffer as that used for analysis of the complexes of p53 with the plasmid fragments (see above) in a total volume of 12 l. The nonmodified or platinated duplexes were mixed with the nonmodified 2513-bp-long fragment of pPGM1. The final amounts of the duplexes and long fragment in the reactions were 20 and 120 ng, respectively. The molar ratio of p53 to duplex was 0 -3. Samples with p53 were incubated in ice for 30 min. After the incubation was completed, 3 l of the loading buffer (50% glycerol, 50 mM Na 3 EDTA, 2% bromphenol blue) was added, and the samples were loaded on the native 5% polyacrylamide gel (mono:bis(acrylamide) ratio, 29:1) precooled to 4°C in 0.5ϫ TBE buffer. The radioactivity associated with the bands was quantified by means of a Molecular Dynamics PhosphorImager (Storm 860 system with ImageQuant software).
The primary p53 monoclonal antibody (mAb) DO-1 (purified and characterized as described in Ref. 29) was also added to the p53-DNA complex (the molar ratio of mAb to p53 tetramer was 3), the mixture was incubated for an additional 30 min at 20°C, and the resulting p53-DNA-mAb complexes were loaded on the gels.
Other Chemicals-T4 polynucleotide kinase and [␥-32 P]ATP used for 5Ј-end radioactive labeling of the top strands of the oligonucleotide duplexes were purchased from New England Biolabs (Beverly, MA) and Amersham Pharmacia Biotech, respectively. Acrylamide, bis(acrylamide), agarose, and urea were from Merck KgaA (Darmstadt, Germany).

RESULTS
Binding of p53 Protein to Platinated DNA Containing the Consensus Response Element-pPGM1 plasmid was cleaved by PvuII (blunt end-forming enzyme, which cuts twice within the pPGM1). This cleavage produced 474-and 2513-bp fragments containing and lacking CDRE, respectively. The two fragments were separated on the agarose gel, extracted, and purified. The 474-bp fragment was further globally modified by cisplatin or transplatin at r b ϭ 0.02-0.1, and the unplatinated 2513-bp fragment was added as the nonspecific competitor. This mixture was incubated with various amounts of wt p53 (at molar ratios of p53 to 474-bp fragment in the range of 0 -6) and analyzed using agarose gel electrophoresis. The incubation of the unplatinated 474-bp fragment with increasing amounts of p53 resulted in the appearance of a new, more slowly migrating species, with a concomitant decrease of the intensity of the band corresponding to the 474-bp fragment incubated in the absence of p53 (shown for p53:474-bp fragment ratio of 0.95 in Fig. 2, lane 6). This result was in agreement with the previously published reports and demonstrated formation of a sequence-specific complex between DNA and p53 protein (30). Importantly, addition of DO-1 mAb (which maps to the Nterminal domain of p53) produced supershifted complexes that migrated still more slowly than the p53-474-bp complex (Fig.  2, lane 11), confirming the presence of p53 in the more slowly migrating species. In contrast, incubation of the 474-bp fragment modified by cisplatin at r b ϭ 0.02-0.06 with p53 (in the presence of the unplatinated 2513-bp fragment) considerably reduced the yield of the species migrating more slowly in the agarose gel (shown for a p53 to 474-bp fragment ratio of 0.95 in Fig. 2, lanes 7-9). This result is consistent with the idea that cisplatin adducts efficiently reduce binding affinity of the CDRE to active p53. The same experiments were performed with the 474-bp fragment globally modified by transplatin. No reduction in the intensity of the band corresponding to the sequence-specific p53-DNA complex as a consequence of the modification by transplatin was noticed even at r b as high as 0.1 (shown for r b ϭ 0.06 and a p53 to 474-bp fragment ratio of 0.95 in Fig. 2, lane 10). This result demonstrates the inefficiency of transplatin adducts to reduce the binding affinity of the CDRE to the active p53 protein.
Further studies were performed using a short (20 bp) oligodeoxyribonucleotide duplex, oligo-CDRE ( Fig. 1) whose sequence follows the consensus sequence pattern (17). The duplex was globally modified by cisplatin or transplatin to r b in the range of 0.02-0.06, and the unplatinated PvuII fragment of pPGM1, which was 2513 bp long and contained no CDRE, was added as the nonspecific competitor. These mixtures were incubated with p53 at various p53 to duplex molar ratios (0.1-3) and analyzed using native PAGE (Fig. 3). Incubation of the unplatinated oligo-CDRE with increasing amounts of p53 resulted in the appearance of the new, more slowly migrating species, with a concomitant decrease of the intensity of the band corresponding to the 20-bp duplex incubated in the absence of p53 (shown for a p53 to duplex ratio of 0.3 in Fig. 3, lane 2). This result confirmed the formation of the complex between oligo-CDRE and p53. In contrast, the incubation of oligo-CDRE modified by cisplatin at r b ϭ 0.02-0.06 with p53 reduced the yield of the species migrating more slowly in the gel. For instance, the modification of oligo-CDRE by cisplatin at r b ϭ 0.05 already completely inhibited formation of the complex between this duplex and p53 (shown for a p53 to duplex molar ratio of 0.3 in Fig. 3, lane 4). It was also verified using PAGE under denaturing conditions that the modification of oligo-CDRE by cisplatin at r b ϭ 0.05 also afforded duplexes interstrand-cross-linked by this drug (data not shown). The quantitative evaluation (24,25) of the radioactivities associated with the bands corresponding to interstrand cross-linked duplexes and duplexes containing no interstrand CL (24, 25) revealed ϳ20% of duplexes containing at least one interstrand CL of cisplatin.
Oligo-CDRE was also globally modified by transplatin and incubated with p53. No reduction of the intensity of the band corresponding to the p53-oligo-CDRE complex was noticed even at so high an r b as 0.05 (Fig. 3, lane 6), i.e. under conditions when cisplatin adducts inhibited formation of the complex between p53 and the duplex completely (Fig. 3, lane 4). Thus, these experiments confirmed that transplatin adducts do not affect the binding affinity of the CDRE to p53 protein.
Binding of Active p53 Protein to Platinated DNA Lacking the Consensus Response Element-We also investigated binding of p53 to the 448-bp fragment of the pBluescript II SK ϩ plasmid lacking CDRE but modified by cisplatin or transplatin. The plasmid was cleaved by PvuII, which yielded the 448-and 2513-bp fragments. The longer fragment was identical to that produced by the PvuII cleavage of pPGM1, whereas the shorter fragment only differed from the shorter fragment produced by PvuII cleavage of pPGM1 by lacking CDRE. The 448-bp fragment was globally modified by cisplatin or transplatin at r b ϭ 0.01-0.08. After the 448-bp fragment was platinated, the unplatinated 2513-bp fragment was added as the nonspecific competitor. These mixtures were incubated with p53 at various p53 to 448-bp fragment molar ratios (in the range of 0.5-6) and analyzed using agarose gel electrophoresis. Incubation of the unplatinated PvuII fragments with increasing amounts of p53 resulted in no changes in the migration of these fragments, demonstrating no effect on formation of the complex between p53 and DNA lacking CDRE (shown for a p53 to 448-bp fragment ratio of 6 in Fig. 4, lane 6). On the other hand, a new  5 and 10). A gel mobility retardation assay was performed in the presence of the unplatinated 2513-bp nonspecific competitor (PvuII fragment of pPMG1 lacking CDRE) in 1% agarose gel; concentrations of the 474-and 2513-bp fragments were 10 and 57 g/ml (3.3 and 3.5 ϫ 10 Ϫ8 M), respectively, and the concentration of the p53 protein was 0 (lanes 1-5) or 3.14 ϫ 10 Ϫ8 M (lanes 6 -11). r b values were as follows: 0 (lanes 1, 6, and 11); 0.02 (lanes 2 and 7); 0.04 (lanes 3 and 8); 0.06 (lanes 4, 5, 9, 10). Lane 11 was the same as lane 6, but mAb DO-1 was added at a molar ratio of mAb to p53 of 3. For other details, see "Experimental Procedures. "   FIG. 3. Binding of p53 to the 20-bp duplex containing CDRE (oligo-CDRE). The duplex was unplatinated (lanes 1 and 2) or globally modified to r b ϭ 0.05 by cisplatin (lanes 3 and 4) or transplatin (lanes 5  and 6). A gel mobility retardation assay was performed in the presence of the unplatinated 2513-bp nonspecific competitor (PvuII fragment of pPMG1 lacking CDRE) in 5% native polyacrylamide gel; concentrations of the oligo-CDRE and 2513-bp fragment were 1.6 and 10 g/ml (1.26 ϫ 10 Ϫ7 and 6 ϫ 10 Ϫ9 M), respectively, and the concentration of p53 was 0 (lanes 1, 3, and 5) or 3.9 ϫ 10 Ϫ8 M (lanes 2, 4, and 6). The oligo-CDRE was radioactively labeled at the 5Ј-end of the top strand. For other details, see "Experimental Procedures." species migrating in the gel considerably more slowly was observed if the 448-bp fragment modified by cisplatin at r b ϭ 0.01-0.08 was analyzed (shown for a p53 to 448-bp fragment ratio of 6 in Fig. 4, lanes 7-9). This result demonstrated formation of the complex between p53 and DNA lacking CDRE and modified by cisplatin (30). Importantly, supershifted complexes were noticed as a consequence of addition of mAb DO-1 to the complex of p53 with the 448-bp fragment modified by cisplatin (Fig. 4, lane 11). The same experiments were performed with the 448-bp fragment globally modified by clinically ineffective transplatin. No more slowly migrating species as a consequence of the modification by transplatin was noticed at r b ϭ 0.08 and a p53 to DNA ratio of 6 (shown for r b ϭ 0.08 and a p53 to 448-bp fragment ratio of 6 in Fig. 4, lane 10). Thus, these results indicate that the binding affinity of p53 to DNA lacking CDRE is enhanced selectively by the modification by cisplatin and not by transplatin.
Cisplatin forms several types of adducts, which occur in DNA with a different frequency and differently distort the conformation of DNA (3). To determine which specific adduct of cisplatin represents a structural motif responsible for recognition of cisplatin-modified DNA lacking CDRE by p53, we prepared a series of 20-bp duplexes with blunt ends. The nucleotide sequences of these oligonucleotides were designed (Fig. 1) so that they did not follow the consensus sequence pattern (17), but they allowed us to prepare the duplexes containing a single site-specific adduct of cisplatin, such as 1,2-GG or 1,3-GTG intrastrand CL in d(TGGT)/d(ACCA) or d(TGTGT)/d(ACACA), respectively, and 1,2-GG interstrand CL or a site-specific monofunctional adduct of the model platinum compound [Cl(dien)Pt]Cl (20) in the duplex d(TGCT)/d(AGCA). The duplexes containing the single adduct were prepared and purified, and the unplatinated PvuII fragment of pPGM1, which was 2513 bp long and lacked CDRE, was added as the nonspecific competitor. These mixtures were incubated with p53 at various p53 to duplex molar ratios (0.1-3) and analyzed using native PAGE. Incubation of the mixtures containing unplatinated oligonucleotide duplexes with p53 did not result in the appearance of any more slowly migrating species (shown for a p53 to duplex ratio of 3 in Fig. 5, lane 2). Similarly, no more slowly migrating species that would demonstrate formation of the complex between p53 and the platinated duplex was observed if p53 was incubated with the duplexes containing the single site-specific 1,3-GTG intrastrand, 1,2-GG interstrand CL of cisplatin, or monofunctional adduct of [Cl(dien)Pt]Cl (shown for a p53 to duplex ratio of 3 in Fig. 5, lanes 8, 10, 12). In contrast, incubation of the d(TGGT)d(ACCA) duplex containing 1,2-GG intrastrand CL of cisplatin with p53 did result in the occurrence of the species migrating more slowly in the gel (shown for a p53 to duplex ratio of 0.93-3 in Fig. 5, lanes  4 -6). These results demonstrate that among DNA adducts of cisplatin major intrastrand adducts between neighboring purine residues only produce a structural motif responsible for enhancement of the binding affinity of active p53 to cisplatinmodified DNA lacking CDRE.
Interestingly, the protein concentration at which half of the free d(TGGT)/d(ACCA) duplex containing the single 1,2-intrastrand CL was bound, was ϳ1.5 ϫ 10 Ϫ7 M (Fig. 5, lane 5), whereas that at which half of the unplatinated oligo-CDRE was bound was roughly 10 times lower. This rough comparison shows that the binding affinity of p53 to 1,2-intrastrand crosslink of cisplatin formed in DNA lacking CDRE is lower than the binding affinity of p53 to unplatinated CDRE. DISCUSSION

DNA Adducts of Antitumor Cisplatin Reduce the Binding Affinity of the Consensus Response Sequence to Active p53 Protein, whereas the Adducts of Clinically Ineffective Transplatin
Do Not-It has been shown that sequence-dependent conformational variability of response elements plays a critical role in the sequence-specific binding of p53 to DNA and the stability of the resulting complex. Extraordinary demands for this binding specificity and selectivity of p53 are closely related to its tetrameric association with CDRE, in which a precise steric fit is extremely important. It has been suggested (18) that steric fit is accommodated through strongly correlated DNA bending localized at the two highly bendable CA͉TG junctions and twisting uniformly distributed between the pentamers of the CDRE. The consensus sequences investigated in the present work contained several sites at which bifunctional adducts of cisplatin strongly distorting DNA conformation are formed. In particular, they contained the sites at which major intrastrand CLs between adjacent purines (at d(GG) or 5Ј-d(AG)) are formed. For instance, the structure of these adducts determined by NMR methods has revealed (31, 32) that these adducts induce the overall helix bend of 40 -78°toward the major groove, DNA unwinding of 25-27°, severe perturbation of hydrogen bonding within the 5Ј-coordinated GC bp, and distortion extended over at least 4 -5 bp at the site of the CL. Interestingly, the CDREs tested in the present work also contain the sites at which cisplatin forms less frequent interstrand CLs.  4. Binding of p53 to the PvuII fragment of pBluescript SK II ϩ DNA, which was 448-bp-long and lacked CDRE. The fragment was unplatinated (lanes 1 and 6) or globally modified by cisplatin ( lanes  2-4, 7-9, and 11) or transplatin (lanes 5 and 10). A gel mobility retardation assay was performed in the presence of the unplatinated 2513-bp nonspecific competitor (PvuII fragment of pPMG1 lacking CDRE) in 1% agarose gel; concentrations of the 448-and 2513-bp fragments were 10 and 57 g/ml (3.1 and 3.5 ϫ 10 Ϫ8 M), respectively, and the concentration of p53 was 0 (lanes 1-5) or 1.88 ϫ 10 Ϫ7 M (lanes 6 -11). r b values were as follows: 0 (lanes 1 and 6); 0.01 (lanes 2 and 7); 0.05 (lanes 3 and 8); 0.08 (lanes 4, 5, and 9 -11). Lane 11 was the same as lane 9, but mAb DO-1 was added at a molar ratio of mAb to p53 of 3. For other details, see "Experimental Procedures. "   FIG. 5. Binding of p53 protein to the 20-bp oligonucleotide duplexes lacking CDRE and containing no, or a single, sitespecific adduct of cisplatin. A gel mobility retardation assay was performed in the presence of the unplatinated 2513-bp nonspecific competitor (PvuII fragment of pPMG1 lacking CDRE) in 5% native polyacrylamide gel; concentrations of the oligonucleotide duplexes and 2513-bp fragments were 1.6 and 10 g/ml (1.26 ϫ 10 Ϫ7 and 6 ϫ 10 Ϫ9 M), respectively, and the concentration of p53 was 0 M (lanes 1, 3, 7, 9, and  11 Formation of these lesions in the CDREs investigated in the present work was confirmed by PAGE under denaturing conditions (see above). The interstrand CL, which is preferentially formed by cisplatin between opposite G residues in the 5Ј-GC/ 5Ј-GC sequence (24), induces several irregularities in the crosslinked base pairs and their immediately adjacent pairs (33). The cross-linked G residues are not paired with hydrogen bonds to the complementary cytosines, which are located outside the duplex and not stacked with other aromatic rings. All other base residues are paired, but distortion extends over at least 4 bp at the site of the CL. In addition, the cis-diammineplatinum(II) bridge resides in the minor groove, and the double helix is locally reversed to a left-handed, Z-DNA-like form. This adduct induces the helix unwinding by ϳ80°relative to B-DNA and also the bending of ϳ40°of the helix axis at the cross-linked site toward the minor groove. Thus, cisplatin formed in the CDREs investigated in the present work bifunctional adducts, which strongly disturb its secondary structure. The result of these perturbations is that the precise steric fit required for the formation and stability of the tetrameric complex of p53 with the consensus nucleotide sequence cannot be attained, so that p53 does not bind to its CDRE.
We demonstrated in the present work that 1,2-intrastrand CLs of cisplatin formed in the CDRE reduce its binding affinity to p53, whereas the same adducts formed in the sequences that do not follow the CDRE pattern afford DNA enhanced binding affinity to p53. Hence, the CDRE should have some intrinsic specific feature absent in the usual B-DNA that precludes p53 from binding to CDRE after its modification by cisplatin. It has been shown (18) that CDREs are already intrinsically curved in the flexible CATG tetramers, with the directionality close to that in the p53-DNA complexes. We suggest that unique alterations induced in the consensus sequence simultaneously by 1,2-intrastrand CLs of cisplatin and the intrinsic curvature of the CDRE are structural factors responsible for the reduced affinity of p53 to its consensus sequence modified by cisplatin.
Clinically ineffective transplatin also forms various types of adducts in DNA. It forms mainly monofunctional adducts at G residues and some amount of interstrand CLs between complementary G and C residues (6,7,25). Monofunctional adducts of transplatin (and cisplatin) affect DNA conformation only slightly, without bending (20), and DNA unwinding is very small (ϳ6°) (34). In addition, the conformational alterations induced by the interstrand CL of transplatin are much less severe than those induced by the CLs of cisplatin (33,35). The duplex is slightly distorted on both sides of the CL, but all bases are still paired and hydrogen-bonded. The CL of transplatin only unwinds the double helix by ϳ12°and induces a slight, flexible bending of ϳ20°of its axis toward the minor groove. We conclude that these relatively subtle structural perturbations induced by transplatin in the CDRE have no substantial effect on the formation of the tetrameric complex of p53 with the CDRE.
1,2-Intrastrand Cross-links of Cisplatin Formed in DNA Lacking the Consensus Nucleotide Sequence Increase DNA Binding Affinity to Active wt p53 Protein-The results of the present work also demonstrate enhancement of the binding affinity of p53 to DNA lacking CDRE due to its modification by cisplatin. However, this binding affinity of p53 is considerably lower (roughly by one order of magnitude) than that to the unplatinated CDRE. Importantly, no change of the binding affinity to p53 of DNA lacking CDRE is observed due to its modification by clinically ineffective transplatin; thus this enhancement is specific for cisplatin-modified DNA. Also importantly, among DNA adducts of cisplatin only the 1,2-intrastrand CLs are responsible for this increase in DNA binding to p53. Hence, 1,2-intrastrand CLs distort DNA lacking the CDRE in a specific way, producing a structural motif recognized by p53. We propose that directional and stable bending of DNA lacking the CDRE toward its major groove due to formation of the 1,2-intrastrand CL of cisplatin affords a structural element exhibiting this specific affinity to p53. In this way a stable flexure of DNA by 1,2-intrastrand CL provides an opportunity for more stable contacts between p53 and DNA lacking CDRE. In other words, the lesions such as 1,2-intrastrand CLs that efficiently induce the directional and fixed bend in DNA toward the major groove, thus providing a stable pre-bent site on DNA to p53, serve as a structural motif for recognition of DNA lacking CDRE by p53.
The observation that other minor bifunctional DNA adducts formed by cisplatin in usual B-DNA lacking CDRE, which also bend and unwind DNA, are not recognized by p53 deserves further discussion. The intrastrand CL formed by cisplatin between two G residues separated by a third base also bends the helix axis toward the major groove by ϳ30°(36) and locally unwinds DNA by ϳ19° (26). However, another important feature of the conformational alteration induced by this lesion is that DNA is locally denatured and flexible at the site of the adduct (36), the structural feature much less pronounced in the 1,2-intrastrand adduct. Furthermore, interstrand CLs of cisplatin also distort DNA distinctly differently and more severely than intrastrand adducts (see above). These significant structural differences between the latter minor adducts and the major 1,2-intrastrand CLs suggest that their recognition and binding by active p53 will be different and restricted to the structural features identical or similar to those induced in DNA by the 1,2-intrastrand adduct of cisplatin. Consistent with this idea is the observation that there are also significant structural differences between 1,2-intrastrand CLs of cisplatin and the adducts formed on DNA by transplatin (see above), which do not form the lesions recognized by p53 on DNA lacking CDREs. Thus, the data base on the structures of cisplatin and transplatin adducts reveals their propensity to distort DNA in very different ways. The distinctive structural features of 1,2-intrastrand CLs of cisplatin suggest a unique role for this adduct in the enhancement of the binding of p53 to platinated DNA segments lacking CDRE, which is, however, weaker than the binding of p53 to unplatinated CDRE.
Biological Implications-There is substantial evidence suggesting that p53 plays a central role in the cellular response to DNA damage. It is also clear that p53 function may only be one of many factors that modulate cisplatin sensitivity, and the effects may be different for various cell types. In addition, sequence-specific DNA binding is one of the key biochemical activities responsible for much of the biological function of p53. Hence, the observation described in the present work demonstrating that cisplatin adducts formed in the CDRE reduce its binding affinity to active p53 may affect these key biochemical activities. Because the affinity of p53 to the CDRE is not affected by the adducts of clinically ineffective transplatin, we suggest that the reduced affinity of the active p53 protein to the CDRE due to its modification by cisplatin is relevant to the biological activity of this drug.
A possible intriguing scenario for response to cellular exposure to cisplatin is also that associated with another important result of the present work demonstrating a relatively weak but significant increase in the binding affinity of p53 to major adducts of cisplatin-1,2-intrastrand CLs formed in the parts of DNA lacking CDRE. This increase in the binding affinity of p53 is specific for DNA modified by cisplatin and does not occur if DNA is modified by clinically ineffective transplatin. This result is consistent with the view that the mechanism of antitumor activity of cisplatin may also be associated with its effi-ciency to promote binding activity of p53 in the segments of DNA lacking CDRE. We suggest that 1,2-intrastrand CLs formed by cisplatin in DNA segments lacking CDRE may be sufficient to hijack p53 protein, keeping it away from its natural targets. Thus, the resultant complexes may divert p53 from its natural functions and/or may protect the cisplatin damage from its recognition by other cellular components. Alternatively, it has been shown (37) that p53 enhances binding of DNA modified by cisplatin to chromosomal high mobility group (HMG) 1 protein. HMG domain proteins are known to specifically bind 1,2-intrastrand CLs of cisplatin and thus mediate antitumor effects of this drug (3). The details of how the binding of HMG domain proteins to cisplatin-modified DNA sensitizes tumor cells to cisplatin are still not completely resolved, but possibilities such as shielding cisplatin-DNA adducts from excision repair or titrating away these proteins from their transcriptional regulatory function have been suggested as clues to how these proteins are involved in the antitumor activity (9). It is therefore possible that p53 and HMG1 proteins encounter each other at the sites of 1,2-intrastrand adducts of cisplatin (formed outside the CDRE). This interaction could further promote keeping p53 away from its natural target and/or promote the binding affinity of HMG1 protein to cisplatin-modified DNA, with the consequences for the antitumor effect of cisplatin mentioned above. More detailed information about cellular consequences of interactions of platinated DNA and p53 protein is required before any definite conclusions can be drawn about how these interactions contribute to the mechanism of the biological activity of cisplatin.