Synthesis and Structural Characterization of the N2G− Mitomycin C

Chem. Res. Toxicol. 1996, 9, 1063-1071

1063

Synthesis and Structural Characterization of the N2G-Mitomycin C-N2G Interstrand Cross-Link in a Model Synthetic 23 Base Pair Oligonucleotide DNA Duplex Amy J. Warren†,‡ and Joshua W. Hamilton*,†,‡,§ Department of Chemistry, Dartmouth College, Hanover, New Hampshire 03755-3564, Department of Pharmacology and Toxicology, Dartmouth Medical School, Hanover, New Hampshire 03755-3835, and Norris Cotton Cancer Center, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire 03756-0001 Received April 22, 1996X

Mitomycin C (MMC) is a genotoxic cancer chemotherapeutic agent that reacts principally at the N2 position of guanine to form one of two predominant monoadducts, or a G-G interstrand cross-link at CpG sites, or a G-G intrastrand cross-link at GpG sites. Previous studies of MMC adduction have principally used very short duplex oligonucleotides (5-15 bp) or very long native duplex DNAs. We examined the formation and structural features of the MMC CpG interstrand cross-link on a model 23 bp synthetic oligonucleotide duplex having the (upper strand) sequence 5′-ATAAATACGTATTTATTTATAAA-3′. MMC was reacted with the duplex oligonucleotide in the presence of sodium dithionite at ratios of 6 mM dithionite:1.5 mM MMC:0.03 mM duplex. The yield of cross-link in the reaction was determined to be approximately 4.8% by denaturing gel electrophoresis, which represented approximately 75% of the total bound MMC. The crosslinked DNA was isolated to greater than 97% purity in a single step by high temperature size exclusion column chromatography. Characterization of the purified product confirmed that the complex contained exclusively the N2G-MMC-N2G cross-link at the single central CpG site. CD spectroscopy demonstrated a negative band at approximately 290-320 nm which has previously been shown to be characteristic of the MMC cross-link. The relative intensity of this band compared to those reported for shorter duplexes suggested that the majority of the duplex is in a normal B-DNA helical configuration. Base-specific chemical footprinting techniques also indicated that there were subtle but distinct structural perturbations principally within the central four to six base pairs containing and adjacent to the cross-link.

Introduction (MMC)1

Mitomycin C is a genotoxic cancer chemotherapeutic agent which has been used as a primary agent for anal, lung, and superficial bladder cancers and as a secondary agent in breast, colon, gastric, and pancreatic cancers (1). As with cisplatin, the preferential killing of cancer cells by MMC is thought to be a result of its ability to damage the DNA of a cell, principally due to formation of DNA cross-links (2-5). Unrepaired DNA interstrand cross-links are known to specifically, potently, and irreversibly inhibit DNA synthesis in dividing cells, which is a lethal event (2, 3). In addition, MMC and other cross-linking agents have potent biological effects at noncytotoxic doses (6-8). MMC is a bifunctional cross-linking agent which requires sequential chemical or enzymatic reductions to form covalent adducts with DNA (4). In living cells, MMC activation occurs principally by flavoenzymes (9, 10); this activation * To whom correspondence and reprint requests should be sent: Joshua W. Hamilton, Ph.D., Department of Pharmacology and Toxicology, Dartmouth Medical School, 7650 Remsen, Hanover, NH 037553835. Telephone 603-650-1316; fax 603-650-1129; e-mail josh.hamilton@ dartmouth.edu. † Dartmouth College. ‡ Dartmouth Medical School. § Norris Cotton Cancer Center. X Abstract published in Advance ACS Abstracts, August 15, 1996. 1 Abbreviations: mitomycin C, MMC; polyacrylamide gel electrophoresis, PAGE; sodium dithionite, SDT; dimethyl sulfate, DMS; formic acid, FA; hydrazine, HZ; diethyl pyrocarbonate, DEPC; hydroxylamine, HA.

S0893-228x(96)00070-7 CCC: $12.00

can be mimicked in vitro by chemical reducing agents such as by H2 and PtO2, which favor the formation of a monofunctional monoadduct with DNA, by Na2S2O4, which favors the formation of bifunctional adducts, or by various NAD(P)H-dependent flavoreductases (11). Reduced MMC reacts almost exclusively with the N2 position of guanine in duplex DNA (12). Reaction of MMC with DNA in vivo principally produces four major types of DNA adducts: G-MMC-G interstrand (4) or intrastrand (13) cross-links at CpG or GpG sites, respectively, and the bifunctional and monofunctional MMC-G monoadducts (4). The relative proportion of adducts in vivo depends on the sequence of the DNA and on the metabolic and redox status of the cells (5, 14, 15). However, despite our current detailed knowledge about the chemistry of MMC DNA adduction, the relationship between specific DNA adducts and observed biological effects in vivo is still not clear. Our laboratory has recently demonstrated that treatment of cells in culture and whole animals in vivo with MMC and other genotoxic carcinogens significantly and preferentially alters expression of certain inducible genes, whereas expression of constitutively expressed genes is refractory to these same treatments (6-8, 16-18). These effects are transient, occur principally at the level of gene transcription, and are closely associated temporally with DNA damage and repair of chemically-induced DNA lesions (6-8, 16-18). Recent results suggest that alterations in gene expression by MMC in vivo are most © 1996 American Chemical Society

1064 Chem. Res. Toxicol., Vol. 9, No. 7, 1996

Figure 1. Sequence of the model 23 base pair oligonucleotide duplex 7/8. Synthetic oligonucleotides 7 (top) and 8 (bottom) are nonidentical and are non-self-complementary. The duplex 7/8 was annealed and the Tm was determined experimentally to be approximately 48 °C (in 0.1 M Tris, pH 7.4; see also Figure 5). Duplex 7/8 contains a single off-center CpG site for MMC interstrand cross-link formation at opposing guanines (box).

closely associated temporally with formation and removal of the MMC interstrand cross-link, and that chromatin structure also seems to be important for these effects.2 We were therefore interested in whether structural changes in chromatin following formation of the MMC cross-link could be involved in MMC-induced alterations in expression of susceptible genes. Structural characterizations of the MMC interstrand cross-link to date have used very short duplex oligonucleotides (5-15 bp). For example, two- and three-dimensional NMR characterization of the CpG MMC cross-link by Patel and co-workers suggested that there was little or no perturbation of the DNA helix by the adduct (19); however, these studies used a 6 bp oligonucleotide which represents only half of a full turn of duplex DNA. Similarly, circular dichroism studies of the MMC crosslink have used DNA duplexes in the size range of 6-10 bp, and the characteristic signal of the mitosene chromophore cross-linked to DNA appeared to be dependent on DNA length (20). To determine the effects of the MMC cross-link on DNA structure within a longer DNA, we designed a non-self-complementary, model 23 bp duplex oligonucleotide containing a single off-center CpG site. We also optimized the reaction conditions of MMC with this DNA to increase the yield of cross-link in the reaction and developed a single step purification scheme for isolation of the cross-linked DNA from unreacted and monoadducted DNA after the reaction. The purified MMC-cross-linked duplex was then characterized by HPLC, circular dichroism (CD), and chemical footprinting to examine the structure of the cross-link in this model sequence.

Materials and Methods Synthesis and Purification of the MMC-DNA Interstrand Cross-Link. Oligonucleotides 7 and 8 (see Figure 1) were commercially synthesized on a 1-µmol scale (DNA Express, Macromolecular Resources) and were purified by 20% denaturing polyacrylamide gel electrophoresis (PAGE). Complementary oligos 7 and 8 were annealed in 0.1 M Tris (pH 7.4), in a Cetus Perkin-Elmer thermal cycler using a 4 h multistep thermal program. Proper annealing was confirmed by UV-vis spectroscopic analysis of thermal denaturation/renaturation and nondenaturing PAGE. For studies requiring autoradiography, one strand of the duplex was 5′-end labeled with [γ-32P]ATP (New England Nuclear, 3000 Ci/mmol) and T4 kinase (Gibco-BRL) and purified by NEN20 chromatography (New England Nuclear) essentially as previously described (21) prior to annealing of both strands. Duplex oligo was reacted with MMC essentially as previously described (15, 20, 22) with some modifications. Caution: MMC is a genotoxic mutagen and suspect carcinogen and should be handled in an appropriate manner. The reagents used for chemical footprinting (DMS, FA, HZ, HA, KMnO4, DEPC, and piperidine) are hazardous chemicals; therefore, protective cloth2 R. M. Caron, A. J. Warren, M.A. Ihnat, and J. W. Hamilton, unpublished results.

Warren and Hamilton ing should be worn and appropriate safety measures should be followed when working with these compounds. Briefly, MMC (Sigma) was chemically reduced by sodium dithionite (SDT, Aldrich) in the presence of the duplex oligo at ratios (final) of 6 mM SDT:1.5 mM MMC:0.03 mM duplex (1.5 mM in nucleotides, but with only 1 CpG site per duplex) in 0.1 M Tris (pH 7.4 at 4 °C), on ice under argon. SDT was added in four aliquots over a 1 h period with constant stirring. The argon was passed through a CuO oxygen scrubber, and was saturated with water to prevent dehydration of samples. Solid SDT was weighed, placed in a sealed vessel, and immediately purged with argon. Argon-purged buffer was added via syringe to the solid SDT under an argon atmosphere. Dissolved SDT was used within 15 min of preparation. The reaction was stopped by the introduction of air, and the crude product was separated from the unreacted duplex by size exclusion chromatography. Sephacryl S-100-HR (Sigma) was packed under gravity flow in a water-jacketed column (2.5 × 93 cm, Pharmacia) which was connected to a circulating water bath set at 60 °C. The sample was loaded without dilution and eluted in 0.02 M NH4HCO3 (pH 7.4), at a flow rate of 0.8 mL/min using a peristaltic pump (Gilson). Eluant was collected in 3 mL fractions, and the 32Pradiolabeled DNA was monitored by liquid scintillation and 20% denaturing PAGE followed by phosphorimaging (Molecular Dynamics). The cross-link-containing fractions were combined and reduced in volume by lyophilization (Speed Vac), followed by dialysis and ethanol precipitation (6 volumes of ethanol, 10 mM MgCl2). Unmodified duplex and purified MMC-reacted products were characterized by UV-vis spectroscopy, thermal denaturation, and circular dichroism (CD) in 10 mM Tris/1 mM EDTA (pH 7.4). Enzymatic Digestion and HPLC Analysis. For HPLC analysis, the purified cross-linked duplex was digested with nuclease P1 (Worthington), phosphodiesterase I (Sigma), and alkaline phosphatase (Sigma) to the nucleoside level essentially as previously described (13, 15). Briefly, three A260 OD units/ mL of duplex or cross-linked duplex were digested with nuclease P1 (0.5 U/A260 unit) in sodium acetate (pH 5.3), for 2 h at 55 °C. The pH was adjusted to 8.5 with 1 M Tris, and MgCl2 was added to 0.9 mM. The samples were then further digested with phosphodiesterase I (4.5 U/A260 unit) for 1 h at 37 °C followed by alkaline phosphatase (1.6 U/A260 unit) for 2 h at 37 °C. Samples were then analyzed by HPLC (Hewlett Packard 1090 or 1050) using a C18 reversed phase column (Rainin MicrosorbMV, 5 µm, 4.6 mm × 25 cm) and an isocratic solvent system of 5% CH3CN/95% 0.1 M TEAA (pH 7.0), at a flow rate of 1 mL/ min. Elution was monitored at 254 nm for DNA and at 313 nm for the mitosene chromophore (20). Retention times were determined using internal standards, and molar ratios were calculated by dividing the area of each peak by the extinction coefficient for each nucleoside. Cytosine was arbitrarily chosen as the internal standard for computation of the molar ratios. Chemical Footprinting. Chemical footprinting of the crosslink was performed using various base-specific chemical reactions essentially as previously described (23-26). MaxamGilbert sequencing, consisting of five separate chemical reactions, was performed as previously described (23) with modifications described by the manufacturers of the chemical sequencing kit containing all reagents used in this procedure (New England Nuclear). Briefly, 50 000 cpm (∼0.5-1 µg in 10-20 µL of H2O) of DNA or DNA-MMC was incubated with 1 µL of dimethyl sulfate (DMS; NEN) in 200 µL of DMS reaction buffer (50 mM cacodylate, pH 7.0, 1 mM EDTA, pH 8.0; NEN), 50 µL of 88% formic acid (FA; NEN), or 30 µL of hydrazine (HZ; NEN) (with or without 15 µL of 5 M NaCl), for 15 min at 20 °C (ref 23 and NEN protocol). In a similar manner, 50 000 cpm (∼0.5-1 µg in 10-20 µL of H2O) of DNA or DNA-MMC were also treated with diethyl pyrocarbonate (DEPC; Sigma) as described (25). Briefly, samples (10 µL) were mixed with 5 µL of DEPC and incubated at 37 °C for 15 min with occasional mixing. For the hydroxylamine reactions (HA; Sigma), the hydroxylamine reagent was freshly prepared by titrating 3 M hydroxylamine hydrochloride to pH 6 with diethylamine (24). Samples (7 µL)

Structure of Mitomycin C DNA Interstrand Cross-Link

Figure 2. Analysis of the reaction of MMC with duplex 7/8 and purification of the interstrand cross-link. Oligo 7 was 5′end-labeled with 32P prior to annealing, and duplex 7/8 was reacted with MMC under chemical reducing conditions that favor formation of the interstrand cross-link (see Materials and Methods). Following the reaction, an aliquot of the crude reaction mixture was removed and electrophoresed on a 20% denaturing polyacrylamide gel (lane labeled “crude”) followed by phosphorimaging and digital densitometry as described in Materials and Methods. The denatured single-stranded unmodified and monoadducted DNA (labeled “SS”) was well-resolved from the putative interstrand cross-linked DNA (labeled “XL”). In this representative experiment, the yield of interstrand crosslink was 5.0%. The interstrand cross-linked DNA was purified from the crude reaction mix by high temperature size exclusion column chromatography (see Figure 3). Aliquots were removed from collected column fractions eluting in three different regions (lanes A, B, and C, corresponding to regions A, B, and C of Figure 3) and were analyzed. Lane A represents several combined fractions in the peak containing the interstrand crosslink; lane B represents the fraction containing principally crosslink but also containing some single-stranded and MMCmonoadducted DNA (monoadducted DNA cannot be resolved from unmodified DNA on this gel); lane C represents the large peak in Figure 3 principally containing single-stranded and MMC-monoadducted DNA. were incubated with 30 µL of HA for 30 min at 37 °C. After chemical modification, reactions were stopped with 0.5 M ammonium acetate and reagents were removed by ethanol precipitation twice using tRNA as a carrier to aid in precipitation of small fragments. Samples were heated for 30 min in 1 M piperidine (NEN) to displace chemically modified bases and cause cleavage of the sugar-phosphate backbone. KMnO4 reactions were performed on MMC-DNA samples or DNA alone as described (25). Briefly, samples (10 µL) were incubated with 5 µL of 100 mM KMnO4 (Sigma) for 2 min at room temperature. Reactions were stopped by the addition of 2 µL of β-mercaptoethanol and precipitated and cleaved with hot piperidine as described above. Reaction products were separated on a 16% denaturing polyacrylamide gel (0.4 mm × 17 cm × 32 cm; GibcoBRL, Model SA Sequencing Gel Electrophoresis System). Gels were run for 45 min at 40 W, in 0.89 M Tris/0.89 M borate/2.0 mM EDTA (pH 8.3). The bands were quantified by autoradiography followed by scanning densitometry. To compensate for interlane variability, each band was normalized to the total density in that lane. The fold change in reactivity was then calculated by dividing the percent total DNA cleavage at each base position in the cross-linked duplex by the percent total DNA cleavage at the same position in the unmodified duplex.

Results The MMC-DNA interstrand CpG cross-link was synthesized in a 23 bp oligonucleotide (Figure 1) under chemical reducing conditions with a yield of 4.8%, as determined by denaturing PAGE of radiolabeled DNA (Figure 2). Lane A of Figure 2 represents an aliquot of the reaction mixture prior to purification. Under these denaturing gel conditions, the interstrand cross-link migrates with about half the mobility of unmodified DNA,

Chem. Res. Toxicol., Vol. 9, No. 7, 1996 1065

Figure 3. Purification of the 23 base pair duplex DNA containing the MMC interstrand cross-link by high temperature size exclusion column chromatography. Following reaction of the radiolabeled DNA duplex 7/8 with MMC, the reaction mixture was subjected to size exclusion chromatography through Sephacryl S-100-HR at 60 °C, in 0.02 M NH4HCO3 (pH 7.4), as described in Materials and Methods. Radiolabeled material in each fraction was monitored by liquid scintillation, and the elution volume was measured. The region labeled “A” represents the elution peak of the interstrand cross-linked DNA at approximately 220 mL, as confirmed by denaturing 20% polyacrylamide gel electrophoretic analysis (see Figure 2). The region labeled “B” indicates an area between the two major peaks that contained a mixture of cross-linked DNA, single-stranded DNA, and monoadducted DNA, and the fraction labeled “C” from the large second peak contained principally unmodified and monoadducted DNA (see Figure 2).

and the monoadducted DNA comigrates with unmodified DNA as single-stranded oligonucleotides. Reaction products were purified by high temperature, high resolution Sephacryl S-100 size exclusion column chromatography. Figure 3 demonstrates the elution profile of the reacted duplex. Duplex containing the MMC interstrand crosslink (region A) eluted in the void volume at approximately 220 mL. This well-resolved single peak contained approximately 97% pure interstrand cross-link (lane A, Figure 2). An aliquot taken between the two peaks, corresponding to region B, contained both cross-link and a mixture of monoadducted products and unmodified DNA (lane B, Figure 2). This region was excluded from the pooled samples taken for the cross-link. The large peak centered at approximately 240 mL in Figure 3 (region C) contained a mixture of unmodified and monoadducted DNA running as single-stranded oligonucleotides in both the chromatographic column and the denaturing gel (lane C, Figure 2). To determine the relative amount of cross-link to monoadducted products, a parallel reaction was performed in which a small amount of [3H]porfiromycin was added (2.2 × 106 dpm, 350 mCi/mmol, Moravek Biochemicals, Brea, CA). Porfiromycin is an N-methylated MMC analog and was substituted because radiolabeled MMC is not readily available. Porfiromycin has been previously demonstrated to have similar reactivity with DNA as MMC (27). The reaction products were precipitated with 10 mM MgCl2 and 6 volumes of ethanol prior to analysis by denaturing 20% PAGE. The reaction sample was electrophoresed in a lane next to a 32P-labeled MMC-DNA reaction sample. Using the bands on the autoradiogram in the lane containing the [32P]-labeled DNA as a guide, the regions in the gel corresponding to cross-links and monoadducts were excised, mashed, and quantified by liquid scintillation. Approximately 75% of the bound drug was present as the interstrand cross-link, and the remaining 25% was in the form of monoadducts (data not shown).

1066 Chem. Res. Toxicol., Vol. 9, No. 7, 1996

Figure 4. UV-visible absorbance spectrum of the purified MMC-cross-linked duplex 7/8 DNA. The absorbance spectrum was determined at room temperature in 0.1 M Tris (pH 7.4), as described in Materials and Methods. In addition to the characteristic maximum absorbance peak of DNA at 260 nm, a shoulder centered at 313 nm and extending out to 500 nm was observed (labeled as “M”) which is characteristic of the mitosene chromophore (20). The extinction coefficient of the mitosene moiety in this complex was calculated to be 30 900 M-1 cm-1.

UV-visible spectroscopy of the purified cross-linked duplex indicated a DNA peak at 260 nm and a second absorbance shoulder at 313 nm and continuing to approximately 500 nm which is characteristic of the mitosene chromophore (20) (Figure 4). The A260 was used to quantify the amount of DNA in the purified cross-linked duplex, assuming negligible absorbance of the mitosene chromophore at this wavelength in this 23-mer. Based on the magnitude of the 313 nm peak and an assumption of a 1:1 ratio of duplex to MMC in the purified duplex, the extinction coefficient for the mitosene ring at 313 nm was calculated to be 30 900 M-1 cm-1. This value is much larger than the previously calculated value of 11 000 M-1 cm-1 reported by Tomasz and co-workers (28), although it is in good agreement with our independent analyses of the amount of cross-link as determined by digital densitometry of radiolabeled duplex following PAGE (see Figure 2). However, it should be noted that since the 313 nm peak is relatively broad and much smaller in magnitude than the 260 nm DNA peak, the mitosene extinction coefficient is difficult to accurately determine and is not particularly useful experimentally as a means of quantifying MMC cross-link formation. The PAGE technique was found to be a much more sensitive, accurate, and reproducible measure of cross-linking, and was used to determine cross-link yield in all of these studies. Thermal denaturation/renaturation studies of the unmodified duplex and the purified cross-linked duplex demonstrated that the presence of the cross-link stabilizes the duplex to denaturation, and also allows quick renaturation due to the covalently bound complementary strands (Figure 5). The calculated Tm values for the unmodified and cross-linked duplexes were 48 and 59 °C, respectively. Circular dichroism (CD) spectroscopy of the purified cross-linked duplex was very similar to that of the unmodified duplex with characteristic DNA peaks at 248 and 272 nm (Figure 6). The cross-linked DNA contained an additional negative band at approximately 290-320 nm (labeled as region A in Figure 6) which has previously been shown to be characteristic of the MMC cross-link (20). The relative magnitude of this band as compared to the normal DNA bands was significantly smaller for this 23-mer than those previously reported for shorter duplexes containing a MMC cross-link (20). Comparison of CD data for 6- to 10-mers (20) with our 23-mer

Warren and Hamilton

Figure 5. Thermal denaturation analysis of unmodified and purified MMC-cross-linked duplex 7/8 DNA. Spectra of unmodified and MMC-cross-linked 7/8 duplex DNAs were determined in 100 mM NaCl/10 mM NaH2PO4/0.1 mM EDTA (pH 7.4), at increasing temperature as described in Materials and Methods. Absorbance values at each temperature were divided by the absorbance of the sample at room temperature (21 °C), and the data were normalized to the maximum absorbance of each duplex for graphing purposes. The calculated Tm values for the unmodified duplex DNA and the MMC-cross-linked duplex DNA were 48 and 59 °C, respectively. Similar results were obtained from renaturation studies of denatured duplex DNAs (data not shown).

Figure 6. Circular dichroism spectral analysis of unmodified and purified MMC-cross-linked duplex 7/8 DNA. Spectra of unmodified and MMC-cross-linked 7/8 duplex DNAs were determined in 10 mM Tris/1 mM EDTA, pH 7.4, at room temperature on a Jobain-Yvon Auto-Dichrograph Mark V Instrument. The characteristic spectrum of normal B-DNA was observed for the unmodified 7/8 duplex. The spectrum of the cross-linked duplex was slightly shifted toward the blue region and contained a negative band (labeled as region “A”) at approximately 290-320 nm which is characteristic of MMC cross-linked DNA (20).

indicated that the relative magnitude of this band was inversely proportional to the length of the duplex (R2 ) 0.93; data not shown), suggesting that this band represents a local structural feature (the mitosene-guanine region (20)) that is independent of DNA duplex length. We concluded that the majority of the cross-linked 23mer duplex in this study was present in a normal B-DNA conformation, further supporting the model that the guanine-MMC-guanine cross-link represents a small local feature in the otherwise normal duplex. The purified cross-linked duplex was digested to the nucleoside level by nuclease P1, phosphodiesterase I, and alkaline phosphatase, followed by HPLC analysis (Figure 7). A comparison of the molar ratios of nucleosides from digested unmodified duplex (Figure 7A) to those of the digested cross-link (Figure 7B) is shown in Figure 7C, demonstrating a 75% reduction of the guanine peak with

Structure of Mitomycin C DNA Interstrand Cross-Link

Chem. Res. Toxicol., Vol. 9, No. 7, 1996 1067

Figure 7. HPLC analysis of enzymatically digested unmodified and purified MMC-cross-linked duplex 7/8 DNA. Duplex DNAs were digested with nuclease P1, phosphodiesterase I, and alkaline phosphatase followed by separation by HPLC as described in Materials and Methods. Shown are typical chromatograms of unmodified duplex (panel A) and purified MMC-cross-linked duplex (panel B) digested to individual nucleosides (labeled). Molar ratios of nucleosides are shown in panel C and were calculated by dividing the area of each peak of a chromatogram by the extinction coefficient for each nucleoside (horizontal hatches, panel A, unmodified; diagonal hatches, panel B, cross-linked). Bars represent the mean ( range of values from two separate experiments, and duplicate HPLC analyses were performed for each experiment. Nucleoside molar ratios were dC:dG:dT:dA ) 1.00:1.01:9.83:9.13 for the unmodified duplex and 1.00:0.25:9.88:8.93 for the MMC-cross-linked duplex, demonstrating a 75% reduction of the guanine peak. Cytosine was arbitrarily chosen as the internal standard for computation of the molar ratios.

no significant decrease in the ratios of the other nucleoside peaks. We concluded that the principal sites of MMC cross-linking involve the lone guanine in each strand of the duplex. The lack of complete loss of the guanine peak is likely due to a partial reversal of one of the C10 mitosene-guanine linkages in the cross-link which has been shown to occur under certain chemical conditions (20). Chemical footprinting was used to further characterize the purified cross-linked duplex. The Maxam-Gilbert reagents, dimethyl sulfate (DMS), formic acid (FA), and hydrazine (HZ), and the chemical probes diethyl pyrocarbonate (DEPC), hydroxylamine (HA), and KMnO4 were used both to determine the sequence position of the MMC cross-link and to structurally characterize the cross-linked duplex DNA. Figure 8 represents an example of a typical autoradiogram from these footprinting experiments. Band intensities were compared by digital densitometry between modified and unmodified duplexes electrophoresed on the same gel and run in parallel reactions using the same reagents, normalizing for percent of total DNA modified at each site. Results from several experiments are summarized in Figure 9 and Table 1. Labeling of each strand in separate reactions and comparison of loss of signal along the duplex between the cross-linked and unmodified DNA indicated that the only detectable adduct was a G-MMC-G cross-link at the single central CpG site. For example, in Figure 8, bands in lane 7 (contrasted with lane 2) indicate fragments up to the guanine resulting from cleavage 5′ to the guanine. However, in the cross-linked duplex all other cleavage events 3′ to the cross-link are shifted to the top of the gel due to the higher molecular weight of the cross-linked complex. The data summarized in Figure 9 and Table 1 indicate that only the base pairs in and immediately adjacent to the central CpG site of the cross-link were altered in their chemical sensitivity. Other than a significant increase in DMS-mediated cleavage at the guanine of each strand, these alterations were principally minor reductions in reactivity which represent subtle but distinct structural changes in the major groove opposite the MMC cross-link which is bound to the guanines in

Figure 8. Chemical footprinting of unmodified and purified MMC-cross-linked duplex 7/8 DNA. Chemical modification reactions were performed as described in Materials and Methods. This figure shows a representative autoradiogram in which the upper strand (oligo 7) was radiolabeled. Reactions were performed on the native duplex DNA (lanes 1-5) and on the purified MMC-DNA interstrand cross-link (lanes 6-10). Reactions were as follows: lanes 1 and 6, dimethyl sulfate (G); lanes 2 and 7, formic acid (G+A); lanes 3 and 8, hydrazine (C+T); lanes 4 and 9, diethyl pyrocarbonate (G+A); and lanes 5 and 10, hydroxylamine (C). DMS, dimethyl sulfate; FA, formic acid; HZ, hydrazine; DEPC, diethyl pyrocarbonate; HA, hydroxylamine. Reactions were also performed with KMnO4 and HZ + NaCl (data not shown). Each band was quantified by scanning digital densitometry and normalized to the total density in that lane. A summary of these footprinting experiments is shown in Figure 9 and Table 1.

1068 Chem. Res. Toxicol., Vol. 9, No. 7, 1996

Warren and Hamilton

Figure 9. Summary of chemical footprinting results for the purified MMC cross-linked duplex 7/8 DNA. Chemical footprinting of the cross-linked duplex followed by electrophoresis and autoradiographic analysis was performed as shown in Figure 8. Each reaction was performed on unmodified and cross-linked duplex a minimum of 4 times and the data were averaged; results are summarized above and in Table 1. Data are expressed as the fold positive or negative change in reactivity of each nucleotide position by each agent in the cross-linked duplex relative to the unmodified duplex. DMS, dimethyl sulfate; FA, formic acid; HZ, hydrazine; DEPC, diethyl pyrocarbonate; HA, hydroxylamine; KMnO4, potassium permanganate. Table 1. Summary of Fold Changes in Base-Specific Chemical Reactivity of the MMC Cross-Linked Duplex DNAa fold change in base-specific chemical reactivity T

A

C

G

DMS +2.30 ((1.13) FA -1.38 ((0.74) -2.36 ((1.10) HZ -2.31 ((0.84) -1.64 ((0.78) DEPC -2.56 ((1.89) -3.25 ((1.85) HA -2.80 ((1.76) +1.10 ((0.19) KMnO4 -1.20 ((.39) a The data from Figure 9 are shown in tabular form. Results are expressed as the positive or negative fold change in reactivity of the cross-linked duplex relative to the unmodified duplex at each nucleotide position (average values for n ) 4-10 with SEM shown in parentheses). There was no significant change in reactivity of nucleotides 5′ to this four base pair region. Data for the upper strand only are shown for simplicity; similar values were obtained for the comparable positions of the lower strand. A positive value indicates an increased sensitivity compared to the unmodified duplex, whereas a negative value indicates a decreased sensitivity to chemical attack.

the minor groove. These results are consistent with the circular dichroism results and indicate that the majority of the cross-linked duplex is in a normal B-DNA form. However, the MMC interstrand cross-link also causes minor perturbations in DNA helical structure principally within a central four to six base pair region centered around the site of the cross-link.

Discussion Previous studies of MMC-DNA interactions have principally used very short (6- to 15-mer) oligonucleotide duplexes (19, 20). In order to determine the effects of the MMC cross-link on DNA structure within a longer oligonucleotide duplex DNA, we designed and synthesized a non-self-complementary 23 bp duplex DNA. This sequence was designed to provide two full turns of the

DNA helix, to contain a central but off-center single CpG site for MMC interstrand cross-link formation, to contain optimal flanking bases around the site of adduction to optimize MMC binding efficiency based on previous work (14, 15, 20), and to be thermally stabile under physiological conditions such that the duplex would remain fully double-stranded in vivo. This last parameter was included so that the purified cross-linked duplex could be used in a separate study as an antigen for raising antibodies to the native MMC cross-link in mice and rabbits. The reaction of MMC with DNA has been well described (14, 20, 22, 29). In confirmation of other reports (20, 22), we had observed that the yield of cross-link progressively decreased as the length of duplex oligonucleotide containing a single CpG site was increased. To optimize the reaction conditions for our model 23-mer, the ratios of reactants were examined based initially on the procedures of Teng and Crothers (22) and Tomasz and co-workers (20). The most significant difference between the two previous methods is that the latter group used a larger excess of MMC to nucleotides. Although a large yield was reported with this method for small duplexes, for our larger 23-mer the amount of excess dithionite required to reduce the larger amount of MMC caused an acidic shift in the pH and extremely low amounts of cross-link (approximately 1%), confirming the previous observations of Teng and Crothers (22). In addition, Teng and Crothers reported that a 2- to 3-fold excess of SDT to MMC was required for optimal crosslinking while the Tomasz approach employs a lower ratio of 1.5-fold excess of SDT to MMC. We determined empirically that optimal yields with our 23-mer were obtained with ratios of 6.0 mM SDT:1.5 mM MMC:1.5 mM nucleotides (0.03 mM duplex). As in the Tomasz method, we found that performing the reaction at 4 °C (on ice) and addition of the reducing agent in 4-5 aliquots over a 1 h period produced optimal cross-linking. Preparation of the reducing agent under completely anaerobic conditions was also determined to be critical to obtaining high yields, by use of argon and a CuO scrubber to remove all trace amounts of air. We also found that saturation of the argon with water prior to introduction of argon into the sealed reaction vessel significantly reduced the problem of dehydration of the small reaction volumes over the course of the reaction. Final reaction conditions consistently resulted in 75% of the total adducted MMC being in the form of the interstrand crosslink and a 4.8% ((1.0) overall yield of cross-link. This is among the highest yields reported to date for this reaction with an oligonucleotide containing a single CpG site (20, 22, 30, 31). Previous methods for purifying the cross-link have primarily involved column chromatography followed by HPLC, or recovery of the cross-link from a denaturing polyacrylamide gel (14, 20, 31, 32), both of which are labor intensive and provide low sample recovery. We developed a single step purification method involving high resolution Sephacryl S-100 resin size exclusion chromatography in a water-jacketed column heated to 60 °C. Under these conditions, unmodified and monoadducted duplex DNAs denature to single strands but the cross-link is incapable of full separation. The column can then resolve the cross-linked duplex, which elutes in the void volume, from the unmodified and monoadducted oligos which coelute after the void volume. The entire yield of cross-linked DNA can be collected in a single well-

Structure of Mitomycin C DNA Interstrand Cross-Link

resolved peak. Analysis by 20% denaturing PAGE demonstrated that the cross-link was purified to greater than 97% in this single step method. UV-vis spectroscopy of the purified cross-linked duplex confirmed the presence of the characteristic mitosene chromophore as previously reported (20). By thermal denaturation curve analysis, the cross-linked 23-mer was shown to have a Tm approximately 10 °C higher than that of the unmodified duplex. This increase in Tm is likely to be a result of increased stacking interactions, hydrogen bonds, and/or other interactions around the site of the cross-link, which are reinforced by covalent attachment of both strands. The psoralen interstrand cross-link similarly causes an increase in the Tm of oligonucleotides (33), whereas the cisplatin intrastrand cross-link, which significantly disrupts the helix, dramatically decreases the Tm of oligonucleotides (34). Circular dichroism (CD) spectroscopy of the cross-link demonstrated a characteristic negative band at approximately 290-320 nm as previously described (20). The relative magnitude of this band was smaller than that previously reported for the shorter 6- to 10-mer duplexes containing a MMC crosslink (20). Comparison of the spectrum of the cross-linked 23-mer to those reported by Tomasz et al., indicated that the relative magnitude of this negative band was inversely proportional to oligonucleotide length. This is consistent with a model in which the majority of the DNA duplex is in a normal B-DNA form and the mitoseneguanine linkage causes an induced CD feature which is independent of DNA length. Enzymatic digestion of the MMC-DNA cross-link followed by HPLC analysis confirmed that MMC was bound exclusively to the guanine, as had been shown previously (4, 14, 15, 19, 20). Chemical footprinting was used to more precisely demonstrate the location of the cross-link and to examine structural features within the DNA duplex surrounding the cross-link. Millard and Hopkins recently described a method for determining the base sequence preferences of DNA interstrand cross-linking drugs (31, 32, 35). They demonstrated, initially using oligonucleotides crosslinked at a single site, that 5′-end labeled cross-linked DNA which had been randomly cleaved by a hydroxyl radical 5′ to the cross-link could be separated as a ladder of fragments by gel electrophoresis. Cleavage of the DNA 3′ to the cross-link resulted in a much larger radiolabeled fragment (containing both strands), which had a significantly retarded mobility in the gel. Thus, the resulting discontinuity in the fragment pattern was diagnostic of the nucleotide position of the cross-link on the labeled strand. We extended this method by using base-specific chemical probes frequently used for studies of unusual DNA structures (i.e., DEPC, HA, and KMnO4) and Maxam and Gilbert chemical sequencing reagents (i.e., DMS, FA, HZ) (23-26, 36, 37). It should be noted that all of the reagents used in this study principally attack sites located in the major groove, and therefore any observed alterations in reactivity were probably not due to simple steric inhibition of the MMC bound in the minor groove. Using this approach, our results clearly demonstrated that the only detectable adduct was a G-MMC-G crosslink at the single central CpG site and that the MMC was bound to the N2 position of guanine rather than the N7. N7-purine adducts are easily cleaved by hot piperidine, resulting in bands comigrating with the guanine specific reactions. In contrast, heating N2-purine adducts in piperidine results in a slow dissociation of the drug

Chem. Res. Toxicol., Vol. 9, No. 7, 1996 1069

from the guanine as previously observed by Tomasz (20). After piperidine treatment, approximately 12.8% of the cross-linked DNA comigrated with single-stranded DNA. Storage of the cross-linked duplex also resulted in some degradation of cross-link to monoadducts, and typically 7% of the untreated cross-link comigrated with singlestranded DNA. The footprinting results indicated that the guanines may be structurally and/or electronically perturbed, as demonstrated by the significant increase in reactivity of the cross-linked duplex to DMS. The decreased reactivity of the cross-linked DNA to FA may be due to an increased rigidity of the guanine glycosidic bond, leading to a decrease in piperidine-catalyzed depurination. A similar suppression in FA-induced depurination has also been observed with cisplatin-DNA adducts (38). Both the cytosine and adenine immediately next to the site of the cross-link demonstrated decreased sensitivity to a variety of chemical probes, which may indicate a decrease in the accessibility of these reactive sites to chemical attack. Similar trends in differential reactivity of the cross-linked duplex relative to the unmodified duplex were observed on both strands. It has been reported that the MMC cross-link can exist as two isomers, differing in whether the mitosene chromophore points toward the 5′ or 3′ end (39). It is possible that the similar reactivity observed in the complementary strands represents the averaged effects of a mixture of the two isomers. These results, taken together with previous studies, suggest that the effects of MMC on DNA structure are localized to the four to six base pairs including and centered around the CpG cross-link site, and that structural perturbations drop off significantly with increasing distance from the cross-link. The cross-linked structure appears to be relatively fixed and independent of DNA length, at least for linear oligonucleotides. The DNA structural perturbations are relatively small, indicating that, unlike the psoralen or cisplatin cross-links, the MMC cross-link does not significantly distort DNA duplex structure. In support of this, as part of another project we have been injecting a complex consisting of the purified MMC-cross-linked DNA into mice and rabbits for the purpose of obtaining monoclonal and polyclonal antibodies to the native MMC lesion in duplex DNA. Following a large number of such experiments, we have been unable to obtain any antibodies that specifically recognize the cross-link, although a number of very high affinity anti-duplex DNA antibodies were raised against the 23-mer itself in each experiment. This difficulty in raising specific antibodies may be due, at least in large part, to the relatively nondistorting nature of the MMC cross-link in duplex DNA, such that it does not present a significant antigenic site. Based on these observed subtle modifications of DNA structure, it is intriguing to speculate as to the mechanisms of MMC-DNA adduct recognition and repair in vivo, which are currently poorly understood. Proteins that recognize more distorting lesions such as the cisplatin-DNA intrastrand cross-link, UV-induced pyrimidine cross-links, and psoralen-DNA interstrand cross-links have been described (40-47). Several excision repair-deficient cell lines have been shown to have increased sensitivity to MMC which may involve damage recognition and/or cross-link repair (48). It is likely that there are specific proteins which are able to recognize and signal repair of the MMC-DNA lesions, including the cross-link, since repair of these lesions is

1070 Chem. Res. Toxicol., Vol. 9, No. 7, 1996

relatively efficient in non-repair-deficient cell lines and in vivo2 (8). Our laboratory has previously demonstrated that noncytotoxic doses of genotoxic carcinogens cause a significant and preferential alteration in the expression of inducible genes, while having little or no effect on the expression of constitutively expressed genes in both in vivo and cell culture systems (6-8, 16-18). The strongest preferential effects were observed with DNA crosslinking agents, including MMC (8), cisplatin, carboplatin, the new mitomycin analog, BMS181174,2 and the human lung carcinogen, chromium(VI) (6, 7). Most recently, we have demonstrated that expression of several proteins associated with natural and acquired multi-drug resistance in cancer cells, including the drug pumps, pglycoprotein and MRP, the vault protein, LRP, and several glutathione S-transferase isozymes, are downregulated by treatment with MMC and other DNA crosslinking agents.2 Thus, this preferential effect could be a useful tool for improving cancer chemotherapy of refractory tumors. The molecular mechanisms underlying this phenomenon are currently under active investigation. We have hypothesized that subtle alterations in DNA structure due to formation of MMC interstrand cross-links cause significant changes in critical DNA-protein interactions within inducible promoters resulting in alterations in gene regulation. How the relatively nondistorting MMC interstrand cross-link causes such alterations in DNA-protein interactions remains to be determined.

Acknowledgment. The authors gratefully acknowledge the technical assistance of Michael A. Ihnat, Robert Felty, Thomas L. Ciardelli, and Jan Fisher, and valuable discussions with Karen E. Wetterhahn, Frank V. Valone, and Maria Tomasz. This work was supported by grants to J.W.H. from NIH-NCI (CA49002), the Norris Cotton Cancer Center, and the Hitchcock Foundation. J.W.H. was also partially supported by the Norris Cotton Cancer Center, and A.J.W. was partially supported by the Dartmouth College Chemistry Department. Support for the Dartmouth College Molecular Biology Core Facility was provided by the Norris Cotton Cancer Center Core Grant CA23108.

References (1) Doll, C. D., Weiss, R. B., and Issell, B. F. (1984) Mitomycin: ten years after approval for marketing. J. Clin. Oncol. 3, 276-286. (2) Iyer, V. N., and Szybalski, W. (1964) Mitomycins and porfiromycin: Chemical mechanism of activation and cross-linking of DNA. Science 145, 55-58. (3) Roberts, J. J., and Thomson, A. J. (1979) The mechanism of action of antitumor platinum compounds. Prog. Nucleic Acid Res. Mol. Biol. 22, 71-133. (4) Tomasz, M., Lipman, R., Chowdary, D., Pawlak, J., Verdine, G. L., and Nakanishi, K. (1987) Isolation and structure of a covalent cross-link adduct between mitomycin C and DNA. Science 235, 1204-1208. (5) Bizanek, R., Chowdary, D., Arai, H., Kasai, M., Hughes, C. S., Sartorelli, A. C., Rockwell, S., and Tomasz, M. (1993) Adducts of mitomycin C and DNA in EMT6 mouse mammary tumor cells: Effects of hypoxia and dicumarol on adduct patterns. Cancer Res. 53, 5127-5134. (6) Hamilton, J. W., and Wetterhahn, K. E. (1989) Differential effects of chromium(VI) on constitutive and inducible gene expression in chick embryo liver in vivo and correlation with chromium(VI)induced DNA damage. Mol. Carcinog. 2, 274-286. (7) McCaffrey, J., Wolf, C. M., and Hamilton, J. W. (1994) Effects of the genotoxic carcinogen chromium(VI) on basal and hormoneinducible phosphoenolpyruvate carboxykinase gene expression in vivo: correlation with glucocorticoid- and developmentallyregulated expression. Mol. Carcinog. 10, 189-198.

Warren and Hamilton (8) Caron, R. M., and Hamilton, J. W. (1995) Preferential effects of the chemotherapeutic DNA crosslinking agent mitomycin C on inducible gene expression in vivo. Environ. Mol. Mutagen. 25, 4-11. (9) Tomasz, M., and Lipman, R. (1981) Reductive metabolism and alkylating activity of mitomycin C induced by rat liver microsomes. Biochemistry 20, 5056-5061. (10) Pan, S., Andrews, P. A., Glover, C. J., and Bachur, N. R. (1984) Reductive activation of mitomycin C and mitomycin C metabolites catalyzed by NADPH-cytochrome P-450 reductase and xanthine oxidase. J. Biol.Chem. 259, 959-966. (11) Tomasz, M., Chawla, A., and Lipman, R. (1988) Mechanism of monofunctional and bifunctional alkylation of DNA by mitomycin C. Biochemistry 27, 3182-3187. (12) Tomasz, M., and Lipman, R. (1986) Reassignment of the guanine binding mode of reduced mitomycin C. Biochemistry 25, 43374344. (13) Bizanek, R., McGuinness, B. F., Nakanishi, K., and Tomasz, M. (1992) Isolation and structure of an intrastrand cross-link adduct of mitomycin C and DNA. Biochemistry 31, 3084-3091. (14) Borowy-Borowski, H., Lipman, R., and Tomasz, M. (1990) Recognition between mitomycin C and specific DNA sequences for cross-link formation. Biochemistry 29, 2999-3006. (15) Kumar, S., Lipman, R., and Tomasz, M. (1992) Recognition of specific DNA sequences by mitomycin C for alkylation. Biochemistry 31, 1399-1407. (16) Hamilton, J. W., McCaffrey, J., Caron, R. M., Louis, C. A., Treadwell, M. D., Hunt, S. R., Reed, M. J., and Doherty, K. A. (1994) Genotoxic chemical carcinogens target inducible genes in vivo. Ann. N.Y. Acad. Sci. 726, 343-345. (17) Hamilton, J. W., Louis, C. A., Doherty, K. A., Hunt, S. R., Reed, M. J., and Treadwell, M. D. (1993) Preferential alteration of inducible gene expression in vivo by carcinogens that induce bulky DNA lesions. Mol. Carcinog. 8, 34-43. (18) McCaffrey, J., and Hamilton, J. W. (1994) Comparison of effects of direct-acting DNA methylating and ethylating agents on inducible gene expression in vivo. Environ. Mol. Mutagen. 23, 164-170. (19) Norman, D., Live, D., Sastry, M., Lipman, R., Hingerty, B. E., Tomasz, M., Broyde, S., and Patel, D. J. (1990) NMR and computational characterization of mitomycin cross-linked to adjacent deoxyguanosines in the minor groove of the d(T-A-C-GT-A)‚d(T-A-C-G-T-A) duplex. Biochemistry 29, 2861-2875. (20) Borowy-Borowski, H., Lipman, R., Chowdary, D., and Tomasz, M. (1990) Duplex oligodeoxyribonucleotides cross-linked to mitomycin C at a single site: Synthesis, properties, and cross-link reversibility. Biochemistry 29, 2992-2999. (21) Hamilton, J. W., Bement, W. J., Sinclair, P. R., Sinclair, J. F., and Wetterhahn, K. E. (1988) Expression of 5-aminolaevulinate synthase and cytochrome P-450 in chicken embryo hepatocytes in vivo and in cell culture: Effect of porphyrinogenic drugs and haem. Biochem. J. 255, 267-275. (22) Teng, S. P., Woodson, S. A., and Crothers, D. M. (1989) DNA sequence specificity of mitomycin cross-linking. Biochemistry 28, 3901-3907. (23) Maxam, A. M., and Gilbert, W. (1980) Sequencing end-labeled DNA with base-specific chemical cleavages. Methods Enzymol. 65, 499-560. (24) Johnson, B., and Rich, A. (1985) Chemical probes of DNA conformation: Detection of Z-DNA at nucleotide resolution. Cell 42, 713-724. (25) Fox, K. R., and Grigg, G. W. (1988) Diethylpyrocarbonate and permanganate provide evidence for an unusual DNA conformation induced by binding of the antitumour antibiotics bleomycin and phleomycin. Nucleic Acids Res. 16, 2063-2075. (26) Palecek, E., Boublikova, P., Jelen, F., Krejcova, A., Makaturova, E., Nejedly, K., Pecinka, P., and Vojtiskova, M. (1990) Chemical probing of the DNA polymorphic structure in vitro and in situ. In Structure & Methods, Volume 3: DNA & RNA (Sarma, R. H., and Sarma, M. H., Eds.) pp 237-253, Adenine Press, Albany. (27) Tomasz, M., Hughes, C. S., Chowdary, D., Keyes, S. R., Lipman, R., Sartorelli, A. C., and Rockwell, S. (1991) Isolation, identification, and assay of [3H]porfiromycin adducts of EMT6 mouse mammary tumor cell DNA: effects of hypoxia and dicumerol on adduct patterns. Cancer Commun. 3, 213-223. (28) Tomasz, M., Mercado, C. M., Olson, J., and Chatterje, N. (1974) The mode of interaction of mitomycin with DNA and polynucleotides in vitro. Biochemistry 13, 4878-4887. (29) Tomasz, M., Chowdary, D., Lipman, R., Shimtokahara, S., Viero, D., Walker, V., and Verdine, G. L. (1986) Reaction of DNA with chemically or enzymatically activated mitomycin C: Isolation and structure of the major covalent adducts. Proc. Natl. Acad. Sci. U.S.A. 83, 6702-6706.

Structure of Mitomycin C DNA Interstrand Cross-Link (30) Millard, J. T., and Beachy, T. M. (1993) Cytosine methylation enhances mitomycin C cross-linking. Biochemistry 32, 1285012856. (31) Millard, J. T., Weidner, M. F., Kirchner, J. J., Ribeiro, S., and Hopkins, P. B. (1991) Sequence preferences of DNA interstrand crosslinking agents: Quantitation of interstrand crosslink locations in DNA duplex fragments containing multiple crosslinkable sites. Nucleic Acids Res. 19, 1885-1891. (32) Millard, J. T., Weidner, M. F., Raucher, S., and Hopkins, P. B. (1990) Determination of DNA cross-linking sequence specificity of reductively activated mitomycin C at single-nucleotide resolution: Deoxyguanosine residues at CpG are cross-linked preferentially. J. Am. Chem. Soc. 112, 3637-3641. (33) Shi, Y., and Hearst, J. E. (1986) Thermostability of doublestranded deoxyribonucleic acids: Effects of covalent additions of a psoralen. Biochemistry 25, 5895-5902. (34) Kline, T. P., Marzilli, L. G., Live, D., and Zon, G. (1989) Investigations of platinum amine induced distortions in singleand double-stranded oligodeoxyribonucleotides. J. Am. Chem. Soc. 11, 7057-7068. (35) Weidner, M. F., Millard, J. T., and Hopkins, P. B. (1989) Determination at single-nucleotide resolution of the sequence specificity of DNA interstrand cross-linking agents in DNA fragments. J. Am. Chem. Soc. 111, 9270-9272. (36) Guo, Q., Lu, M., Shahrestanifar, M., Sheardy, R. D., and Kallenbach, N. R. (1991) Drug binding to a DNA BZ molecule: Analysis by chemical footprinting. Biochemistry 30, 11735-11741. (37) Mendel, D., and Dervan, P. B. (1987) Hoogsteen base pairs proximal and distal to echinomycin binding sites on DNA. Proc. Natl. Acad. Sci. U.S.A. 84, 910-914. (38) Johnson, N. P., Mayard, A. M., Escalier, J., and Maquet, J. P. (1985) Mechanism of the reaction between cis-[PtCl2(NH3)2] and DNA in vitro. J. Am. Chem. Soc. 107, 6376-6380. (39) Kumar, S., Johnson, W. S., and Tomasz, M. (1993) Orientation isomers of the mitomycin C interstrand cross-link in non-selfcomplementary DNA. Differential effect of the two isomers on restriction endonuclease cleavage at a nearby site. Biochemistry 32, 1364-1372.

Chem. Res. Toxicol., Vol. 9, No. 7, 1996 1071 (40) Jones, C. J., and Wood, R. D. (1993) Preferential binding of the Xeroderma Pigmentosum Group A complementing protein to damaged DNA. Biochemistry 32, 12096-12104. (41) Pil, P. M., and Lippard, S. J. (1992) Specific binding of chromosomal protein HMG1 to DNA damaged by the anticancer drug cisplatin. Science 256, 234-237. (42) Toney, J. H., Donahue, B. A., Kellett, P. J., Bruhn, S. L., Essigmann, J. M., and Lippard, S. J. (1989) Isolation of cDNAs encoding a human protein that binds selectively to DNA modified by the anticancer drug cis-diamminedichloroplatinum(II). Proc. Natl. Acad. Sci. U.S.A. 86, 8328-8332. (43) Bianchi, M. E., Falcioloa, L., Ferrari, S., and Lilley, D. M. J. (1992) The DNA Binding Site of HMG1 Protein is Composed of Two Similar Segments (HMG Boxes), both of which have Counterparts in Other Eukaryote Regulatory Proteins. EMBO J. 11, 10551063. (44) Clugston, C. K., McLaughlin, K., Kenny, M. K., and Brown, R. (1992) Binding of human single-stranded DNA binding protein to DNA damaged by the anticancer drug cis-diamminedichloroplatinum(II). Cancer Res. 52, 6375-6379. (45) Hirschfeld, S., Levine, A. S., Ozato, K., and Protic, M. (1990) A constitutive damage-specific DNA-binding protein is synthesized at higher levels in UV-irradiated primate cells. Mol. Cell. Biol. 10, 2041-2048. (46) Chao, C. C.-K. (1992) Damage-recognition proteins as a potential indicator of DNA-damage-mediated sensitivity of human cells to ultraviolet radiation. Biochem. J. 282, 203-207. (47) Hang, B., Yeung, A. T., and Lambert, M. W. (1993) A damagerecognition protein which binds to DNA containing interstrand cross-links is absent or defective in Fanconi Anemia, Complementation Group A, cells. Nucleic Acids Res. 21, 4187-4192. (48) Hoy, C. A., Thompson, L. H., Mooney, C. L., and Salazar, E. P. (1985) Defective DNA cross-link removal in Chinese hamster cell mutants hypersensitive to bifunctional alkylating agents. Cancer Res. 45, 1737-1743.

TX960070C