Kinetic studies of the TATA-binding protein interaction with cisplatin-modified DNA.

The TATA-binding protein (TBP) recognizes the TATA box element of transcriptional promoters and recruits other initiation factors. This essential protein binds selectively to cisplatin-damaged DNA. Electrophoretic mobility shift assays were performed to study the kinetics of TBP binding both to the TATA box and to cisplatin-damaged DNA in different sequence contexts. TBP binds with high affinity (K(d) = 0.3 nm) to DNA containing site-specific cisplatin 1,2-intrastrand d(GpG) cross-links. The k(on) and k(off) values for the formation of these TBP complexes are 1-3 x 10(5) m(-1) s(-1) and approximately 1-5 x 10(-4) s(-1), respectively, similar to the corresponding values for the formation of a TBP-TATA box complex. In electrophoretic mobility shift assay competition assays, cisplatin-damaged DNA extensively sequesters TBP from its natural binding site, the TATA box. Nine DNA probes were prepared to determine the flanking sequence dependence of TBP binding to cisplatin-modified DNA. TBP clearly displays sequence context selectivity for platinated DNA, very similar to but not as dramatic as that of the high mobility group protein HMGB1. When TBP was added to an in vitro nucleotide excision repair assay, it specifically shielded cisplatin-modified 1,2-(GpG) intrastrand cross-links from repair. These results indicate that TBP is likely to be a key protein in mediating the cytotoxicity of cisplatin.

After the discovery of the anticancer activity of cisplatin, many studies have focused on elucidating its mechanism of action (1,2). The formation of covalent cisplatin-DNA adducts, especially the 1,2-intrastrand d(GpG) cross-link, correlates with the cytotoxicity of the drug (1)(2)(3). Attention now focuses on understanding how cells react to the presence of the cisplatin-DNA lesions and, using this information, designing more effective anticancer treatments. Upon cisplatin binding, the DNA duplex is bent and unwound, and the minor groove becomes wide and shallow. These structural changes inhibit essential DNA metabolic processes such as replication and transcription (4 -7). The distorted DNA duplex can also interact with a number of cellular proteins (1,2,8,9), an activity postulated to mediate the processing of cisplatin-DNA lesions (1-3, 10 -14).
The identification and characterization of proteins that bind selectively to cisplatin-damaged DNA has therefore become one of the main thrusts of research in this field.
The TATA-binding protein (TBP) 1 is a key component of transcription factor IID, which is required for transcription initiation of all three eukaryotic RNA polymerases (15). As the first step in the process, TBP recognizes a TATA box element located ϳ30 base pairs (bp) upstream from the transcription start site and eventually recruits other transcription factors. Structural analyses of several TBP-TATA box complexes reveal that TBP binds at a widened minor groove and bends the duplex DNA toward the major groove by intercalation of two pairs of phenylalanine residues (16 -18). Because sequence-specific DNA-binding proteins usually reside in the major groove, the minor groove binding by TBP was initially puzzling. Subsequent studies demonstrated that the flexibility of the TATA box element primarily determines its binding affinity to TBP (19,20). In addition to the specificity of its DNA binding, TBP has distinctive DNA binding kinetics. The formation of the TBP-DNA complex is characterized by relatively slow on and off rates (21).
TBP binds selectively to cisplatin-damaged DNA (22). The sequestration of TBP by cisplatin DNA adducts inhibits transcription in vitro, which could be restored in a reconstituted system by addition of extra TBP (22). Transcription inhibition by exogenously added cisplatin-damaged DNA has also been reported (23). More recently, enhanced binding of TBP to the TATA element containing flanking cisplatin 1,2-intrastrand cross-links was demonstrated (24). This series of experiments suggests a possible role for TBP-cisplatin-DNA ternary complexes in mediating the anticancer activity of the drug.
Because a variety of proteins interact with cisplatin-DNA adducts (9), many studies have focused on determining their binding affinities (25)(26)(27)(28)(29)(30). To evaluate the relative importance of the different protein-platinated DNA interactions in the cell, however, detailed kinetic parameters are also required. Despite the importance of such information, few attempts have been made to examine the kinetics of protein binding to cisplatin-damaged DNA (31)(32)(33). The interactions between the two domains (A and B) of HMGB1 with cisplatin-modified DNA were investigated by using stopped-flow fluorescence and fluorescence resonance energy transfer methods (31,32). In addition, a stopped-flow kinetic analysis was performed for replication protein A (RPA) binding to cisplatin-damaged DNA (33).
In the present study we employed electrophoretic mobility * This work was supported by a grant from NCI, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc. org) contains a table of electrospray ionization-mass spectrometry data for platinated oligodeoxyribonucleotides (Table S1) and the hydroxyl radical footprinting assays (Figs. S1-S3).
‡ Recipient of Japanese Society for the Promotion of Science postdoctoral fellow for research abroad (1999 shift assays (EMSAs) to examine the kinetics of TBP binding to cisplatin-damaged DNA and, for comparison purposes, the TATA box. TBP binding to DNA containing a site-specific 1,2intrastrand d(GpG) cross-link was evaluated by using several platinum-modified DNA probes with various flanking sequences. We also performed an in vitro repair assay to examine the effect of TBP in modulating nucleotide excision repair (NER) of platinated DNA. The results provide strong evidence for a biological role of TBP in mediating the cytotoxicity of cisplatin.

EXPERIMENTAL PROCEDURES
Preparation of Yeast TBP-The C-terminal DNA binding domain of recombinant yeast TBP was provided by Dr. S. M. Cohen in our laboratory and prepared as described previously (24). The concentration of active TBP was determined as reported in the literature (34).
Preparation of Oligonucleotides Probes- Table I lists the 25-bp oligonucleotides used in this study together with their abbreviations. The synthesis, platination, and purification of site-specifically platinated single-stranded oligonucleotides were carried out as described previously (26). Platinated top strands, (5Ј-CCTCTCCTCTCN 1 G*G*N 2 TCT-TCTCTCC-3Ј, N 1 and N 2 ϭ A, T, or C), where the asterisks indicate the formation of Pt-N(7) bonds, were annealed with their complementary bottom strands in 10 mM Tris (pH 7.0), 50 mM NaCl, and 10 mM MgCl 2 , heated to 90°C, and slowly cooled to 4°C over several hours. The resulting cisplatin-modified duplex probes were purified by ion-exchange HPLC using a Dionex DNApac PA-100 column. These oligonucleotides were desalted by dialysis and concentrated to 5-10 M. The TATAMLP probe (Table I) was prepared by the same method except for the platination step. The TATA element of adenovirus major late promoter, one of the strongest promoters, was used for this study. The purity of the platinated single-stranded DNA was confirmed by analytical HPLC. Atomic absorption spectroscopy combined with UV-visible spectroscopy and electrospray mass spectrometry verified the existence of singly platinated probes (see the supplemental material on-line).
EMSA-All platinated and TATAMLP duplex probes (ϳ10 pmol) were radioactively labeled at their 5Ј-ends by using 50 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences) and 20 units of polynucleotide kinase (New England Biolabs). Labeled probes were separated from small nucleotides by passage through G-25 Sephadex Quickspin columns (Roche Molecular Biochemicals). For each binding reaction, DNA probes (0.5-2 nM, ϳ10,000 cpm) and the indicated concentrations of TBP were mixed and incubated in a buffer solution containing 60 mM KCl, 20 mM Tris (pH 7.9), 5.0 mM MgCl 2 , 10 mM dithiothreitol, 0.2 mg/ml bovine serum albumin, and 10% glycerol (24). After incubation at 30°C for 30 min, binding mixtures were directly loaded onto 6% native polyacrylamide gels and electrophoresed for 1.5 h at 150 V in 25 mM Tris (pH 7.9), 190 mM glycine, 1.0 mM EDTA, and 4.0 mM MgCl 2 running buffer. The gels were dried at 80°C and then exposed to phosphorimaging plates for 15-20 h. Quantitative analysis was performed with a Bio-Rad GS-525 Molecular Imager employing Multi-analyst software (Bio-Rad).
Kinetic EMSA Analysis-For the measurement of the protein-DNA association constants by kinetic methods, the increase in the amount of complex was monitored at various times after the addition of 5-10 nM TBP to ϳ1 nM probe. As described previously (35), the raw data were fit to Equation 1, which is derived from the bimolecular binding equation under conditions of excess protein (36). The left side of the equation was plotted against time in s, and the k on value for each probe was obtained from the slope of the plot. In the equation, [TBP] 0 and [probe] 0 indicate the initial concentrations of TBP and DNA probe, respectively, and [complex] is the concentration of complex at each time point.
To determine the dissociation rate constants (k off ), ϳ10 nM TBP and ϳ1 nM probe were mixed at 30°C for 30 min to reach the equilibrium state. Dissociation of the protein-DNA complexes was initiated by the addition of 200 ng of poly[dI⅐dC] at different time points to assure that all dissociation reactions were complete at the same time. The decrease in the amount of complex was followed over a 1-or 2-h time period. The data were fit to Equation 2 to obtain the first-order rate constant for the dissociation reaction, where [complex] indicates the concentration of complex at time t, and [complex] 0 is the complex concentration under the initial binding conditions. The natural logarithm of [complex] divided by [complex] 0 was plotted versus time, and the negative slope of the fit provided the k off value. The dissociation constant, K d , values were calculated from these results (K d ϭk on /k off ). Competition Assay-Competitor DNA of varying concentrations was mixed with a fixed amount of TBP and radioactively labeled TATAMLP, and the disappearance of the TBP-TATAMLP complex was analyzed by EMSA.
Competition assays were also used to obtain the K d values for weak binding probes. In such a competition experiment, two binding reactions proceed simultaneously and follow the relationship in Equation 3 (37), where P t , C t , and T t represent the total concentrations of TBP, competitor, and TATAMLP, respectively. K t and K c are the respective dissociation constants of TATAMLP and competitor. At different C t values, the binding fraction () was determined. The K d value can be calculated by using Equation 4, where C 1/2 represents the concentration of competitor at ϭ 1/2 (37).
Excision Repair Assay-Whole cell-free extracts (CFE) from HeLa cells were prepared by a reported method (38) and stored at Ϫ80°C. DNA duplexes 161 bp in length containing a site-specific cisplatin 1,2-d(GpG) or 1,3-d(GpTpG) cross-link with a radiolabeled phosphate located six or seven bases to the 5Ј side of the cisplatin binding site were prepared as previously described (39,40). The probes contained three phosphorothioate residues at their 3Ј ends to minimize nonspecific nuclease degradation. Whole cell extracts were incubated with damaged DNA in excision repair buffer (39), and excision fragments were resolved by 10% denaturing polyacrylamide gel electrophoresis. For the repair experiment in the presence of TBP, the protein was pre-incubated with the damaged DNA probe in excision repair buffer for 30 min on ice unless otherwise indicated. The extent of NER was measured by comparing the signal intensity corresponding to the 25-30-bp excised fragment with that of the entire lane. The small amount of DNA degradation due to nonspecific nuclease activity was subtracted from the repair signal by using the area corresponding to the 32-37-bp region as background (41).

Kinetics of TBP Binding to the TATA Box and
Cisplatindamaged DNA-To optimize conditions for the EMSA experiments, probes of three different lengths (15,25, and 35 bp) were investigated. The 15-bp probe showed weak binding affinity to TBP compared with longer probes. In addition, this short oligonucleotide was partially melted under the experimental conditions (data not shown). The 35-bp probe displayed a considerable level of nonspecific binding (data not shown). The 25-bp probe was stable under the experimental conditions and did not exhibit any nonspecific binding. Therefore, 25-bp probes were used for further studies.
EMSA analyses were performed to investigate the kinetics of TBP binding to the TATA box (TATAMLP) and cisplatin-damaged DNA (AGGA, Table I). Fig. 1A shows EMSA data used to determine the association rate constants for two probes, and the data analysis is presented in Fig. 1B. The bimolecular binding reactions reach the equilibrium state within 10 min for both probes. As shown in Fig. 1A, however, the TBP-AGGA complex clearly forms more rapidly than the TBP-TATAMLP complex. Data collected within 2-3 min were fit to Equation 1, yielding k on values of 1.3 ϫ 10 5 M Ϫ1 s Ϫ1 for TBP binding to TATAMLP and 3.0 ϫ 10 5 M Ϫ1 s Ϫ1 for the AGGA probe (Fig. 1B). The k on value for TATAMLP is in good agreement with published values, which fall in the range 1.0 -3.0 ϫ 10 5 M Ϫ1 s Ϫ1 (21,35,(42)(43)(44).
After the addition of excess of poly[dI⅐dC] to the TBP-DNA complex, decreased amounts of the complex were detected (Fig.  2). The dissociation reaction data were fit to Equation 2. The AGGA and TATAMLP probes dissociate from TBP with halflives, t1 ⁄2 , of 120 min and 190 min, respectively. Although TBP associates with AGGA more rapidly than with TATAMLP, the latter probe has a slightly lower off-rate (Table II). From these k on and k off results, dissociation constants K d (k off /k on ) for each probe were calculated (Table II). Compar-ison of K d values for AGGA and TATAMLP reveals that cisplatin-damaged DNA has about a 1.5-fold higher binding affinity to TBP compared with that of its natural binding site, the TATA box, under the same binding conditions. The K d value (0.44 nM) for the TATA box binding to adenovirus major late promoter is consistent with previous reports (21,34,35,(42)(43)(44)(45).
Flanking Sequence Preference of TBP Binding to Cisplatindamaged DNA-Previously we demonstrated that HMG domains display a distinct selectivity for platinated DNA containing different flanking sequences (26,46). Moreover, different flanking sequences can affect the conformational and thermodynamic properties of cisplatin-damaged oligonucleotides (47). Here, we studied the kinetics of TBP binding to cisplatindamaged probes in diverse sequence contexts (Table I) a Boldface font denotes the TATA box, cisplatin d(GpG) lesion (N 1 GGN 2 ), and flanking sequences. shows EMSAs of TBP binding to TATAMLP and nine different cisplatin-damaged probes. Sequence selectivity is clearly manifest for TBP binding to platinated DNA.
Kinetic EMSA experiments were performed to determine k on and k off values for each probe. Fig. 4A indicates fits to EMSA data obtained from the association kinetics for six different cisplatin-damaged probes. All k on values lie between 2.1 ϫ 10 5 M Ϫ1 s Ϫ1 and 1.2 ϫ 10 5 M Ϫ1 s Ϫ1 , indicating that most of the cisplatin-damaged DNA complexes have faster association rates than that of the TATA box (Table II). The k off values were also calculated by fitting the dissociation EMSA data to Equation 2 (Fig. 4B). The two more weakly binding probes, TGGC and CGGC, did not allow reliable kinetic data to be obtained. We therefore carried out competition assays to obtain relative K d values for these two probes. At various concentrations of TGGC or CGGC competitor, the binding fraction of the TBP to TATAMLP was examined by EMSA and quantitated (Fig. 5). From the curve fitting shown in Fig. 5B, the concentration of competitor at ϭ 1/2, C 1/2 , was calculated. Equation 4 was employed to compute K d values of 12 and 10 nM, respectively, for TGGC and CGGC as competitors.
Our results reveal a clear flanking sequence dependence of TBP binding to cisplatin-damaged DNA (Table II). At the N 1 position, TBP prefers dA rather than T or dC, and TBP forms more stable complexes when N 2 is either dA or T.
Cisplatin-damaged DNA Sequesters TBP from the TATA Box-A DNase I footprinting assay previously demonstrated reduced binding of TBP to the TATA box after the addition of excess cisplatin-damaged DNA (48). In the present study, competition EMSA experiments were performed to study the rela-tive binding affinities of TBP to the TATA box and cisplatindamaged DNA. TBP and the two DNA probes were mixed, and the amounts of free and bound TATA box were analyzed by using radioactively labeled TATA box DNA. Fig. 6 provides clear evidence that TBP dissociates from the TATA box upon addition of cisplatin-damaged DNA. At 1 nM TBP, more than half of the TBP-TATAMLP complex has dissociated when equivalent amounts (4 nM) of AGGA and TATAMLP are present. Upon the addition of a 3-fold excess of AGGA, most of the TBP is sequestered from the natural TATAMLP binding site.
TBP Blocks NER of the Cisplatin 1,2-d(GpG) Cross-link-Cisplatin-modified DNA is repaired by the NER machinery. When HMG-domain proteins are added to HeLa cell-free extracts or reconstituted NER components, a significant amount of the repair of cisplatin 1,2-d(GpG) intrastrand cross-links is inhibited (39). Repair of cisplatin 1,3-d(GpTpG) intrastrand cross-links, which do not bind HMG-domain proteins, was not significantly affected under the same conditions. Because TBP has a high binding affinity for cisplatin 1,2-d(GpG) intrastrand-cross-linked DNA, we next examined repair of this ad- duct by NER using HeLa whole cell-free extracts in the presence of TBP.
For this experiment we employed the CGGC probe because its repair signal is larger by a factor of two than that of AGGA (data not shown). Fig. 7 reveals that repair of the cisplatin 1,2-d(GpG) intrastrand cross-link in CGGC is indeed inhibited by preincubation with TBP, whereas a probe containing the cisplatin 1,3-d(GpTpG) intrastrand cross-link is not. In the presence of a great excess of TBP (Ͼ200 M), repair of the cisplatin 1,3d(GpTpG) cross-link is also shielded to a similar extent as the 1,2-d(GpG) cross-link, probably because of nonspecific binding (data not shown). DISCUSSION TBP Binding to Cisplatin-damaged DNA-The present study reveals that TBP very slowly associates with and dissociates from cisplatin 1,2-d(GpG) intrastrand cross-links, behavior that is very similar to its binding to the TATA box. Including these results, kinetic parameters for the binding of three proteins, HMGB1-domain proteins, TBP, and RPA, to cisplatin-modified DNA are now available (31)(32)(33). Although it is hard to compare the kinetic parameters directly, because of the different binding conditions and temperatures employed, TBP clearly exhibits very different properties for the binding to cisplatin-damaged DNA. The k on values reveal that the HMGB1 domains bind to cisplatin-modified DNA near the diffusion limit (k on ϭ ϳ5 ϫ 10 8 M Ϫ1 s Ϫ1 ) more than 1000-fold more rapidly than the association of TBP (k on ϭ 3.0 ϫ 10 5 M Ϫ1 s Ϫ1 ). RPA also has about a 100-fold higher association rate constant (k on ϭ 2.2 ϫ 10 7 M Ϫ1 s Ϫ1 ) than TBP. If the association rate were the determining factor in protein binding to target lesions in the cell, TBP would be an unlikely participant in the competition for cisplatin-modified DNA. However, HMGB1 domains and RPA also dissociate much more rapidly than the extremely stable TBP-platinated DNA complex. Under relatively similar binding conditions, TBP has more than a 100-fold higher binding affinity for cisplatin-damaged DNA than RPA (33). In addition, TBP has the highest specificity for cisplatin-damaged DNA among all the identified proteins (1). Because the K d for TBP binding to genomic DNA is 6 ϫ 10 Ϫ6 M, this protein prefers to bind platinum-damaged DNA by a factor of Ͼ3000 compared with undamaged DNA (34). RPA and HMGB1 have ϳ15and 100-fold preferences for damaged over undamaged DNA, respectively (25,28,49).
Several protein-protein interactions involving TBP, RPA, and HMGB1 occur that can affect the DNA binding kinetics in the cell. XPA, another damage recognition protein, stabilizes the RPA-damaged DNA complex (28,50). Interactions between human TBP and HMGB1 occur by which the latter inhibits transcription (51). Subsequent studies revealed that HMGB1 enhances the binding affinity of TBP to the TATA box by about 20-fold, with the C-terminal acidic domain of HMGB1 and the non-conserved N terminus (residues 55-95) of human TBP essential to complex formation (52)(53)(54). Because these two highly abundant proteins can bind to cisplatin-damaged DNA and the interaction between them occurs through their non-DNA binding domains, the binding kinetics of the HMGB1-TBP complex to cisplatin-DNA lesion would be interesting to evaluate. Ultimately, information about the DNA binding kinetics of RPA-XPA and HMGB1-TBP complexes will be useful for assessing likely protein binding priorities in the cell.
Flanking Sequence Dependence of TBP Binding to Cisplatindamaged DNA-Kinetic analyses of TBP binding to 10 different probes were performed, and the results are summarized in Table II. The trends in the kinetically determined K d values of the individual probes agree well with the EMSA data in Fig. 3. Both k on and k off values generally affect the binding affinities of the probes studied. This binding is unlike that encountered in classical TATA box kinetic studies in which different TATA boxes have different k off values (24,35,44). Specifically, it is the differences in k on values that mainly determine the relative TBP binding affinities between the three probes AGGA, AGGT, and AGGC.
The flanking sequence dependence of TBP binding to cispla-  tin-damaged DNA reveals a preference for dA rather than T or dC at the N 1 position. For the N 2 position, TBP shows higher binding affinity for dA or T compared with dC. These flanking sequence preferences for TBP binding are surprisingly similar to those for HMGB1 domain A binding (26,46). The two proteins, HMGB1 domain A and TBP, share no sequence or struc-tural homology, but they display similar selectivities for DNA sequence context of cisplatin-DNA 1,2-intrastrand cross-links. To address the reasons for this selectivity, we now consider the probable binding mode of TBP to cisplatin-damaged DNA.
TBP binds to the widened minor groove of the TATA box mainly by hydrophobic interactions. Two key features of such binding are intercalation by two pairs of phenylalanine residues and several hydrogen-bonding interactions involving two asparagine residues at the middle of the complex (19). The TATA box used in the present study and several of its specific interactions with TBP are shown in Fig. 8A. In previous work, a DNase I footprinting assay was performed for TBP bound to the TATA element and to cisplatin-damaged DNA (48). TBP protected 12 bp of DNA in a region including the TATA box, as indicated in Fig. 8A. Similarly, a 4-bp DNA fragment at the 3Ј side and at least a 6-bp DNA fragment at the 5Ј side of the cisplatin d(GpG) 1,2-intrastrand cross-link were protected by TBP (Fig. 8B). We carried out hydroxyl radical footprinting assays, which provide higher resolution footprints of DNAprotein complexes (55,56). Consistent with previous results, the same 12-bp DNA fragments around the cisplatin d(GpG) lesion and the TATA box were protected by TBP ( Fig. 8 and Figs. S1 and S2 in the on-line supplemental material). Evidently, these footprinting results cannot reveal the exact intercalating positions of TBP bound to cisplatin-damaged DNA, which determine its precise binding mode. Nevertheless, the approximate positioning of the N 1 and N 2 nucleotides is possible to delineate. The N 2 nucleotide is near the intercalating position, and the N 1 nucleotide is at the hydrogen-bond contact region at the center of the complex (Fig. 8B). The preference of dA or T at the N 2 position for TBP binding suggests that it is the flexibility of base pairing at N 2 that determines the binding affinity (19,20). For the N 1 position, TBP prefers to bind probes containing dA at N 1 rather than T or dC. Specific interaction between dA and residues in TBP possibly contribute to this selectivity since N 1 is near the two-asparagine contact region. More details about the binding mode of TBP-platinated DNA complexes must await an x-ray structure determination.
Despite the similarity in flanking sequence preferences, TBP is much less selective than HMGB1 domain A. The highest affinity probe, AGGA, has only 40-fold lower K d values compared with the weakest (TGGC) probe for the interaction with TBP. In contrast, the binding affinities of HMGB1 domain A vary by several orders of magnitude depending upon the sequence context of the cisplatin-DNA lesion, and the lowest affinity probe has only severalfold higher binding affinity than that of undamaged DNA (26). Because different flanking sequences have the same likelihood of occurring in platinated DNA (57), some cisplatin DNA lesions may not participate at all in interactions with HMGB1 in the cell. A crystal structure of HMGB1 domain A bound to a cisplatin-modified DNA duplex revealed the intercalation of a single phenylalanine residue into the hydrophobic wedge created in the minor groove across from the platinum-induced d(GpG) cross-link, with its binding extending mainly to 3Ј side of the adduct (58). Because this intercalation interaction is critical for tight binding of HMG1domain A, the flanking sequence significantly affects the binding affinity of the protein to cisplatin-modified DNA (26). In contrast, in the TBP complex, the cisplatin d(GpG) adduct is located between the two intercalation sites, resulting in a much less significant flanking sequence influence on binding affinity. Although the two proteins have different degrees of selectivity for flanking sequence, their preference trends are similar. In both cases, flexibility at the flanking base pairs is a primary contributor to binding selectivity.
Biological Implications of TBP Binding to Cisplatin-damaged DNA-In the cell the TBP-TATA interaction is required for the recognition and utilization of eukaryotic promoters in basal transcription. The formation rate and stability of TBP-TATA complexes are major determinants that regulate transcription (43,59,60). Disruption of this important interaction is an attractive candidate for explaining the cytotoxicity of cisplatin.
In the present investigation, one of the site-specifically cisplatin-modified probes, AGGA, showed a K d value of 0.3 nM, which indicates 1.5-fold higher binding affinity to TBP than a strong TATA element. Such high binding affinity of cisplatindamaged DNA for TBP suggests that the drug can interfere with cellular TBP-TATA interactions. Furthermore, cisplatin lesions with different flanking sequences show similarly high binding affinity to TBP (Table II), as discussed above. Even the most weakly binding probe, TGGC, has sufficient binding affinity (K d ϭ 12 nM) to disturb the TBP-TATA interaction, because many weak promoters have TATA elements with K d values higher than 12 nM (19,43). In the mammalian genome, ϳ10,000 transcribed genes contain the TATA box in their promoters (2,61). Tumor cells isolated from cisplatin-treated cancer patients have 10,000 -100,000 platinum lesions per genome (2,62,63). These data and our kinetic results suggest that the hijacking of TBP from its natural binding site, the TATA element, might contribute to the cytotoxicity of cisplatin.
NER is a major pathway for removing DNA lesions including those of cisplatin (64). Specific interactions between cisplatindamaged DNA and damage-detection components of NER such as XPA and RPA have been extensively investigated (65)(66)(67).
In addition, a study demonstrated that diminished repair in tumor cells sensitizes them to cisplatin (68). Apart from the NER components, there has been much interest in the role of other cellular proteins binding to cisplatin-damaged DNA. Previous work revealed that HMGB1 proteins block cisplatin lesions from NER, as monitored by a repair assay conducted with whole cell extracts (12,39). In addition, a yeast mutant lacking the HMG-domain protein Ixr1 was less sensitive to cisplatin than wild type cells (11). HMGB1 has been implicated as an important protein in the modulation of such repair shielding because of its abundance in the cell and its relative high binding affinity for cisplatin lesions (1,11,12,25,39). From this work has emerged a strategy for enhancing cisplatin activity in the clinic (13).
In the present study, TBP is demonstrated to shield cisplatin-damaged DNA from NER. The repair of the intrastrand 1,2-d(GpG) cross-link is specifically reduced after preincubation with TBP, whereas repair of the intrastrand 1,3-d(GpTpG) cross-link remains relatively unaffected. Based on the K d value discussed above, TBP binds strongly enough to cisplatin-modified DNA to protect the platinated site from NER, but k on is also an important parameter to consider. When TBP and CFE were added simultaneously to probe DNA in the NER buffer, repair shielding was diminished. Moreover, when TBP was incubated with the DNA probe for a short period of time, its slow binding rate resulted in loss of repair shielding (supplemental Fig. S3). These results suggest that there is a competition between TBP and repair proteins for the cisplatin-modified site. It has been estimated that there are 30,000 -100,000 TBP molecules in a mammalian cell, which is similar to the level of HMGB1 (69,70). RPA is also an abundant protein in human cells, expressed in the range of 30,000 -200,000 copies (71). Interestingly, a recent study reported that TBP is 5-10-fold more highly expressed at the protein level in testicular compared with other tissues (72). Based on this information, it is likely that TBP can protect cisplatin-DNA lesions from the NER in cancer cells that express high levels of TBP.