Supercoiled DNA Promotes Formation of Intercalated cis-N2

330

Chem. Res. Toxicol. 2004, 17, 330-339

Supercoiled DNA Promotes Formation of Intercalated cis-N2-Deoxyguanine Adducts and Base-Stacked trans-N2-Deoxyguanine Adducts by (+)-7R,8S-Dihydrodiol-9S,10R-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene Guohui Jiang,† Ryszard Jankowiak,‡ Nenad Grubor,‡ Marzena Banasiewicz,‡ Gerald J. Small,‡ Milan Skorvaga,§ Bennett Van Houten,§ and J. Christopher States*,† Department of Pharmacology and Toxicology, University of Louisville, Louisville, Kentucky 40292, Ames Laboratory, USDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011, and NIEHS, Research Triangle Park, North Carolina Received September 11, 2003

The highly reactive and mutagenic benzo[a]pyrene metabolite, (+)-7R,8S-dihydroxy-9S,10Repoxy-7,8,9,10-tetrahydrobenzo[a]pyrene (BPDE), forms predominantly N2-deoxyguanine DNA adducts in two stereoisomeric configurations (cis and trans). In previous in vitro assays using oligonucleotide substrates site specifically modified with cis- and trans-BPDE adducts, the nucleotide excision repair (NER) systems of eukaryotes and prokaryotes incise cis-BPDE adducts more efficiently than trans-BPDE adducts [Hess, et al. (1997) Mol. Cell Biol 17, 7069; Zou, et al. (2001) Biochemistry 40, 2923). We investigated the influence of DNA secondary structure on stereospecificity of BPDE adduct formation, and incision of BPDE adducts by the prokaryotic UvrABC NER endonuclease was examined. BPDE adducts formed at low density on supercoiled plasmids were incised 6-7-fold better by the thermoresistant Bacillus caldotenax UvrABC than were BPDE adducts formed on linear DNA. Linearizing supercoiled plasmid DNAs after BPDE adduct formation did not diminish incision efficiency. These results suggested that configuration and/or conformation of adducts formed on linear and supercoiled DNAs differed. This hypothesis was confirmed by low temperature fluorescence spectroscopy of adducted supercoiled and linear DNAs. Spectroscopic results indicated that intercalated cisBPDE adducts as well as base-stacked trans-BPDE adducts formed more abundantly in supercoiled DNA than in linear DNA. A higher cis to trans adduct ratio in supercoiled DNA was confirmed by high resolution [32P]postlabeling analyses. These results demonstrate that DNA secondary structure influences both configuration and conformation of BPDE adducts formed at low density (∼1 adduct/kbp) and suggests that the ratio of cis- to trans-BPDE adducts and amount of base-stacked trans adducts formed under physiological exposure conditions may be higher than inferred from high dose experiments.

1. Introduction Polycyclic aromatic hydrocarbons such as benzo[a]pyrene (BP; reviewed in ref 1) are a major class of environmental pollutants produced by incomplete combustion. BP is found in urban air, cigarette smoke, and charred foods. Enzymatic activation of BP to electrophilic metabolites, including four stereoisomers of benzo[a]pyrene-7,8-dihyrodiol-9,10-epoxides, is stereoselective. The predominant metabolically produced diol epoxide, BPDE,1 is also the most mutagenic in mammalian cells and the most tumorigenic (reviewed in ref 2). BPDE forms predominantly N2-deoxyguanine DNA adducts that * To whom correspondence should be addressed. Tel: 502-852-5347. Fax: 502-852-2492. E-mail: [email protected]. † University of Louisville. ‡ Iowa State University. § NIEHS. 1 Abbreviations: BPDE, (+)-7R,8S-dihydroxy-9S,10R-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene; NLN, nonline narrowed; FLN, fluorescence line narrowed.

can be removed by nucleotide excision repair (NER), a general repair system that eliminates many structurally dissimilar lesions from DNA (reviewed in ref 3). BPDE forms adducts in two stereoisomeric configurations in which electrophilic attack at the exocyclic amine of guanine bases occurs on the opposite side (trans) or on the same side (cis) as the pyrene ring system (4). The major BPDE adduct formed is commonly believed to be the trans adduct (5, 6). However, this belief was inferred from results of very high dose experiments in supercoiled DNA (7) [at about 1 BPDE adduct/16 bp, which is >10 000-fold greater than adduct densities induced by environmentally relevant exposures to BP (8-11)] and from results of linear DNA treated in vitro (12, 13). The (+)-trans adduct is more mutagenic than the (+)-cis isomer in mammalian cells (14). In addition to differences in BPDE adduct stereochemistry, BPDE adducts can adopt multiple conformations. The preferred conformation is believed to be dependent on configuration (15). Earlier in vitro studies with

10.1021/tx034184h CCC: $27.50 © 2004 American Chemical Society Published on Web 01/28/2004

DNA Conformation Influences Adduct Configuration

BPDE-adducted calf thymus (CT) DNA (12, 13) and BPDE-N2-dG adducts formed in double-stranded oligonucleotides (13) revealed that BPDE-DNA adducts exist in several conformations (12, 16). Low temperature fluorescence studies of BPDE-N2-dG adducts in duplex oligonucleotides identified three BPDE adduct conformations (16) designated by (+)-1, (+)-2, or (+)-3, with the S1 state (fluorescence origin band) energy decreasing with increasing adduct label number. The trans-BPDE-DNA adduct can adopt an external (solvent exposed/groove bound) conformation [(+)-1] with a fluorescence origin band at 378 nm and a base-stacked (in part solvent accessible) conformation [(+)-2] with a fluorescence origin band at 380 nm. In contrast, cis-BPDE-DNA adducts adopt mainly an intercalated (hardly solvent exposed) conformation [(+)-3] with a fluorescence origin band near 381 nm. Red shifting of the fluorescence origin band with increasing adduct label number is a consequence of increasing charge transfer character of the fluorescent state due to interaction of the pyrene moiety with DNA bases (16). NMR studies of double-stranded DNA oligonucleotides site specifically modified with stereospecific BPDE adducts indicate that the predominant conformation of the (+)-trans-N2dG adduct is with the pyrenyl ring system in the minor groove while the predominant conformation of the (+)-cis-N2dG adduct is with the pyrenyl ring system intercalated (reviewed in ref 4). Molecular modeling studies support the NMRderived structures and predict that these conformations have the lowest free energy states for each conformer (15). Many aspects of BPDE-induced carcinogenesis are not well-understood, especially the mechanisms determining adduct configuration and conformation. It is important to study low level exposure to understand the mechanistic basis of DNA adduct formation under physiologically relevant environmental exposures. In this study, we examined the role of DNA secondary structure in the stereospecific formation of BPDE adducts in plasmid DNAs with relatively low lesion density (0.5-1 adduct/1000 bp). Results obtained using enzymatic recognition of the adducts by a thermophilic UvrABC endonuclease, low temperature fluorescence spectroscopy, and [32P]postlabeling analyses are presented. The results indicate that formation of cis adducts [(+)-3 type] and base-stacked trans adducts [(+)-2 type] is enhanced on supercoiled DNA relative to linear DNA.

2. Materials and Methods 2.1. Plasmid DNA Isolation. Plasmids pTHQB04 (5.45 kbp, which contains a 1.5 kb fragment spanning the 5′-flanking region through exon 2 of the human β-globin gene) and pTHQ008 (5.75 kbp, which contains a 1.8 kb fragment spanning exons 5-9 of the human p53 gene) were propagated in XL1Blue. The plasmids were derived by cloning PCR-amplified DNA fragments into the TA cloning site of pCRII (Invitrogen, La Jolla, CA). The plasmids contain a pair of EcoR I sites flanking the insertion site and a unique Bgl II site in the vector. Form I plasmid DNA was purified by CsCl-ethidium bromide sedimentation equilibrium ultracentrifugation. DNA concentrations were determined by measuring the A260. DNA quality was checked by agarose gel electrophoresis as well as measurement of the A260/A280 ratios (which were 1.8-2.0). 2.2. Preparation of BPDE Damaged DNA Substrates. Polycyclic aromatic hydrocarbons are hazardous chemicals and

Chem. Res. Toxicol., Vol. 17, No. 3, 2004 331

Figure 1. Flow diagram of UvrABC incision of plasmid DNAs treated with [3H]BPDE. The plasmids are linearized by Bgl II digestion, and the insert gene fragment is released by Eco RI digestion. S/S, S/L, and L/L refer to the DNA conformations when treated/incised, respectively. Incision is measured by comparing the amounts of intact restriction fragments by southern blot hybidization analysis and applying the Poisson distribution to the amounts of intact fragment remaining to calculate the average number of incisions per fragment (see Materials and Methods for calculation).

should be handled carefully in accordance with NIH guidelines. To compare the extent of adduct formation on supercoiled and linear plasmid DNA, supercoiled (form I) and linearized (form III, prepared by digestion with Bgl II, single restriction site in each plasmid) pTHQB04 and pTHQ008 DNAs were treated with 0-500 nM [3H]BPDE (specific activity, 1320 mCi/mmol; NCI Chemical Carcinogen Repository, Chemsyn, Kansas City, MO) and THF (solvent for [3H]BPDE) in 10 mM Tris-HCl, pH 7.5s1 mM Na2sEDTA at 37 °C. After overnight incubation, noncovalently bound [3H]BPDE (which had hydrolyzed to tetrols) was removed by repeated extraction with ethyl acetate. The DNA was precipitated by addition of 1/10 volume 3 M Na-Acetate (pH 5.7) and 2.2 volumes of ethanol and collected by centrifugation. The DNA was redissolved in 10 mM Tris-HCl/1 mM NaxEDTA, pH 7.5. Covalently bound [3H]BPDE was quantified by analyzing a sample by scintillation spectrometry. DNA samples (0.5 µg as indicated in the figure legends) were mixed with 10 mL of liquid scintillation cocktail (Beckman Counter Inc., Fullerton, CA), and the [3H]cpm were counted with a liquid scintillation analyzer (Packard Instruments Company, Downers Grove, IL). The average number of adducts per plasmid was calculated using the molecular weight of plasmid DNA and [3H]BPDE specific activity with correction for [3H] counting efficiency. 2.3. Specific Incision by UvrABC of BPDE Adducts Formed on Supercoiled and Linearized DNAs. Adducted DNAs were prepared by incubating linear and supercoiled plasmid DNAs at 10 µg/mL in TE with 250 nM [3H]BPDE overnight. Unbound tetrols were removed by repeated ethyl acetate extraction. The procedure, outlined in Figure 1, was as follows. Adducted supercoiled DNAs were incubated with and without UvrABC before (S/S) and after (S/L) linearization with Bgl II. Adducted linearized plasmid DNAs (L/L) were also incubated with and without UvrABC. Purification of recombi-

332

Chem. Res. Toxicol., Vol. 17, No. 3, 2004

Jiang et al.

Table 1. Oligonucleotide Probes gene

upper strand probe

lower strand probe

p53 β-globin

AGTGTTAGACTGGAAACTTT CCACACTGATGCAATCATTCG

TGTTCACTTGTGCCCTGACT ATAATCTGAGCCAAGTAGAAGACC

nant Bacillus caldotenax UvrABC components and characterization of the incision of BPDE adducts has been described (17). Twenty femtomoles of [3H]BPDE and THF-treated substrates (equivalent to 71 ng of pTHQB04 or 75 ng of pTHQ008) were preincubated with UvrA (5 nM) and UvrB (125 nM) subunits for 30 min at 60 °C. Then, UvrC (50 nM) was added for additional incubation of 30 min at 60 °C in 20 µL of UvrABC buffer (50 mM Tris-Cl, pH 7.5, 50 mM KCl, 10 mM MgCl2, 5 mM DTT, and 1 mM ATP). The reaction was terminated by placing samples in ice water. Supercoiled DNAs were linearized with Bgl II. Half of the linear DNA samples were restricted with EcoR I. Reactions were terminated by adding 2 µL of 10× stop buffer (1% SDS, 200 mM EDTA-Na2). Restriction with Bgl II alone results in either a 5.4 (pTHQB04) or 5.7 (pTHQ008) kb fragment, and restriction with EcoR I results in a 1.5 (pTHQB04) or 1.8 (pTHQ008) kb fragment that hybridizes to the strand specific oligonucleotide probes. Using both large and small target DNA fragments allows us to measure incisions at low (2-6 adducts per plasmid) and high (10-12 adducts per plasmid) adduct densities, respectively. The reaction samples were mixed with one-third volume of formamide loading buffer (96% formamide, 0.05% xylene cyanol, 0.05% bromophenol blue, and 10 mM EDTA). DNAs were denatured at 76 °C for 10 min and resolved by denaturing gel electrophoresis in 1% agarose containing formaldehyde. The DNA was transferred to nylon membrane by capillary blotting. To determine the presence of adducts in the two DNA strands, the blotted DNAs were hybridized sequentially with strand specific [32P]endlabeled oligonucleotide probes as listed in Table 1. Hybridization was quantified by phosphorimage analysis using a Molecular Dynamics Storm 860 (Amersham Pharmacia Biotech Inc., Piscataway, NJ). UvrABC-incised fragments are smaller and run faster in the gels. Thus, the signal from unincised fragment is reduced by incision of adducts (Figure 1). Specific UvrABC incision of BPDE adducts was determined by quantifying the amount of full-length (unincised) restriction fragment present in products of UvrABC incision and control reactions and applying the Poisson distribution. The specific incision, which corrects for nonspecific nuclease acitivity, is defined as:

number of specific incisions per fragment (I) ) [-ln(F1/F2)] - [-ln(F1*/F2*)] (1) where F1 ) amount of intact fragment from UvrABC-incised BPDE-DNA; F2 ) amount of intact fragment from BPDEDNA incubated without UvrABC; F1* ) amount of intact fragment from UvrABC-incised THF-treated DNA; and F2* ) amount of intact fragment from THF-treated DNA incubated without UvrABC. The incision efficiency was calculated by comparing the number of incisions per fragment with the number of adducts per fragment calculated from the [3H] incorporation:

incision efficiency (%) ) 100 × I/A

(2)

where I ) average number of specific incisions per target fragment and A ) average number of adducts per target DNA fragment calculated from the average number of adducts per plasmid assuming random distribution. 2.4. Preparation and Incision of DNA Substrates Containing Site Specific Stereoisomeric BPDE Adducts. Oligonucleotides (11-mers) containing site specifically placed trans- or cis-BPDE adducts (kind gift of Dr. Nicholas Geacintov)

were used to construct BPDE-N2-dG-adducted 50 bp substrates (Figure 2) as described (18). The constructed 50 bp

Figure 2. Construction of stereoisomeric BPDE-adducted 50 bp DNA substrates. The 11-mer containing a single BPDE stereoisomeric adduct (G) was ligated with a 19-mer and a 20mer in the presence of a 50-mer complementary strand. The recognition site of Rsa I is indicated by underlines. substrates were purified by electrophoresis in a 10% native polyacrylamide gel. The double-stranded character and homogeneity of 50 bp substrates were examined by a restriction assay with Rsa I (data not shown). To determine specific incision of Bca UvrABC on stereoisomeric BPDE adducts, 20 fmol of nondamaged (control substrate) and stereoisomeric BPDE-adducted 50 bp substrates were preincubated with UvrA, UvrB subunits for 30 min at 60 °C, and UvrC was added for additional incubation of 60 min at 60 °C in 20 µL of UvrABC buffer (50 mM Tris-Cl, pH 7.5, 50 mM KCl, 10 mM MgCl2, 5 mM DTT, and 1 mM ATP). Reactions were terminated by adding 2 µL of stop buffer (1% SDS, 200 mM EDTA-Na2). DNA samples were precipitated with Na-acetate and ethanol, collected by centrifugation, and redissolved in 96% formamide loading buffer (96% formamide, 0.05% xylene cyanol, 0.05% bromophenol blue, and 10 mm EDTA). DNAs were denatured at 90 °C for 5 min and resolved by electrophoresis in denaturing 12% polyacrylamide gel containing 7.5 M urea. 2.5. Low Temperature Fluorescence Measurements. Adducted DNAs were prepared by incubating linear and supercoiled plasmid DNAs at 200 µg/mL in TE with 1 µM [3H]BPDE overnight at 37 °C. Unbound tetrols were removed by ethyl acetate extraction, and the adduct level was determined by scintillation spectrometry. Treatment was repeated two more times until the adduct levels were approximately 6-7 adducts per plasmid. Final adduct densities were 7.3 adducts/plasmid in supercoiled pTHQ008, 6.8 adducts/plasmid in linear pTHQ008, 7.5 adducts/plasmid in supercoiled pTHQB04, and 6.5 adducts/ plasmid in linear pTHQB04. Adducted plasmid DNA was studied in aqueous buffer (10 mM Tris-HCl, pH 7.4, 1 mM NaEDTA) and buffer/glycerol (v:v::1:1) (ultrapure glycerol from Spectrum Chemical, Gardena, CA). Samples (∼30 µL) were contained in quartz tubes (2 mm i.d.) sealed with rubber septa. A double-nested glass optical cryostat with quartz windows (H. S. Martin Inc., Vineland, NJ) was used to cool samples to 77 K (immersion in liquid N2) and 4.2 K (immersion in liquid He). NLN fluorescence spectra were obtained at 77 K. FLN spectra were obtained at 4.2 K. The laser-based instrumentation is described in detail elsewhere (19, 20). Briefly, excitation was provided by a Lambda Physik Scanmate2 pulsed dye laser pumped by a Lextra 100 XeCl excimer laser (Lambda Physik, Ft. Lauderdale, FL) at a repetition rate of 10-30 Hz. NLN fluorescence spectra were recorded using S0 f S2 excitation of the pyrenyl chromophore at 346 and 355 nm, which are, respectively, selective for external and quasi-basestacked/intercalated conformations of BPDE-DNA adducts (12, 16). S0 f S1 excitation was used to generate FLN spectra. As reviewed in refs 19 and 20, by varying the S0 f S1 excitation wavelength, it is possible to determine the frequencies and relative intensities of all active molecular vibrations. Fluorescence was dispersed by a McPherson (model 2061) 1 m monochromator: for NLN measurements, a 150 G/mm grating and a resolution of 0.8 nm were used; for FLN measurements, a 2400 G/mm grating and a resolution of 0.05 nm were used. Fluores-



cence was detected with an intensified linear photodiode array (model IRY-1024/GRB, Princeton Instruments, Trenton, NJ) operated in the gated mode to reject scattered light. Typical gate delays and widths were 20 and 200 ns, respectively. Fluorescence excitation spectra at 77 K were measured using a Cary Eclipse fluorescence spectrophotometer (Varian Inc., Palo Alto, CA). Room temperature absorption spectra were measured with a UV/vis spectrometer (Perkin-Elmer Lambda 18, PerkinElmer Instruments, Wellesley, MA). 2.6. [32P]Postlabeling Analyses of BPDE-Adducted Plasmid DNAs. BPDE-treated (∼6-7 adducts/plasmid) and THFtreated (solvent control) supercoiled and linearized pTHQ008 DNAs were subjected to enzymatic hydrolysis with nuclease P1 enrichment and 5′-labeled with [32P] essentially as described (21-23). Briefly, 0.6 pmol of THF- or BPDE-treated plasmids were mixed with 2 µg of undamaged genomic DNA. DNA samples in 10 mM sodium succinate/5 mM CaCl2 were hydrolyzed at 37 °C for 6 h with 0.8 µg of micrococcal nuclease and 0.8 µg of spleen phosphodiesterase. Enrichment of DNA adducts was achieved by incubation for 45 min at 37 °C with nuclease P1 (2 µg). The [32P]postlabeling of the DNA digest was for 1.5 h at 37 °C with γ-[32P]ATP (100 µCi, 6000 Ci/mmol, Amersham Biotech) and T4 polynucleotide kinase (15 units). Adducts were resolved by multidimensional thin-layer chromatography on 10 cm × 10 cm polyethyleneimine-cellulose plates using 1.0 M sodium phosphate (pH 6.0) for D1 and 0.6 M NH4OH with 8 M urea for D3 and 2-propanol:4 M NH4OH ) 1:1 for D4 elution. The resulting chromatograms were visualized and quantified on a Molecular Dynamics Storm 860 phosphorimager.

3. Results 3.1. Linear Concentration Dependence of BPDE Adduct Density. Three kinds of BPDE-treated DNA substrates were obtained with both pTHQB04 and pTHQ008, which were (A) BPDE-treated supercoiled DNA (S/S); (B) linearized BPDE-treated supercoiled DNA (S/L); and (C) BPDE-treated linearized DNA (L/L). Adducts formed per plasmid with a linear response to [3H]BPDE concentration (Figure 3). DNA samples were obtained with both DNAs with approximately 1, 2, 5-6, and 9-11 adducts per plasmid. There was no significant difference in the number of adducts formed on the two plasmids, whether supercoiled or linear. 3.2. BPDE-DNA Adduct Incision by B. caldotenax UvrABC on Plasmid DNA Substrates. UvrABC incision efficiency was determined (Figure 1) on DNAs that were (i) supercoiled when adducted and supercoiled when incubated with UvrABC (S/S), (ii) supercoiled when adducted and linear when incubated with UvrABC (S/ L), and (iii) linear when adducted and linear when incubated with UvrABC (L/L). Results obtained with pTHQ008 DNA and pTHQ004 DNA both containing ∼6-7 adducts per plasmid are shown in Figure 4 (similar results were obtained with anti-sense strands, not shown). Incision efficiency (% of adducts recognized; see Materials and Methods for calculation) was approximately 50-60% when supercoiled DNA was treated with BPDE and incised (S/S). Incision efficiency did not change if the DNA was linearized after BPDE treatment but before UvrABC incision (S/L). Thus, efficiency of UvrABC incision adducts formed on supercoiled DNA was independent of DNA secondary structure at the time of UvrABC incubation. In marked contrast, incision efficiency was only approximately 5-10% when linear DNA was treated with BPDE and incised (L/L). The ratios of S/S to L/L incision efficiencies averaged over both strands using both long and short targets were 7.3 ( 1.4 for pTHQ008 and 6.2 ( 3.7 for pTHQB04. These data indicate that UvrABC

Figure 3. BPDE adduct formation. Supercoiled and linear forms of (A) plasmid pTHQ008 or (B) plasmid pTHQB04 DNA were incubated with [3H]BPDE, and the incorporation of covalently bound [3H] into the DNAs was determined by scintillation spectrometry. Means ( SD for triplicate DNA samples are plotted with linear regression.

incision efficency was dependent on the DNA secondary structure when adducts were formed rather than when adducts were incised. UvrABC incision of plasmids with adduct densities of approximately 11 adducts per plasmid also was more efficient on adducts formed on supercoiled DNA than on linear DNA with both pTHQ008 and pTHQB04 (S/S to L/L ratios of 1.9 and 3.7, respectively, data not shown). These results suggest that configuration or conformation of adducts formed on supercoiled and linear DNA is different and that B. caldotenax UvrABC incised adducts in one configuration and/or conformation more efficiently than the other(s). 3.3. B. caldotenax UvrABC Preferentially Incises cis-BPDE Adducts in Site Specifically Constructed 50 bp Oligonucleotide. Escherichia coli UvrABC incises cis-BPDE adducts on oligonucleotide substrates more efficiently than trans-BPDE adducts (18, 24). We tested whether B. caldotenax UvrABC has the same preference. Double-stranded 50-mer oligonucleotides containing either a cis, a trans, or no BPDE adduct in one strand (Figure 2) were site specifically labeled with [32P] to monitor the 5′-incision and used as substrates in UvrABC reactions (Figure 5). The amount of substrate incised was quantified by phosphorimage analysis of the band intensities. The oligonucleotides containing a cis adduct were incised approximately twice as well as those containing a trans adduct. No incision was observed on

334


Figure 4. BPDE DNA adduct incision by B. caldotenax UvrABC. Plasmid DNAs were treated with [3H]BPDE and subjected to UvrABC incision as outlined in Figure 1 (S/S ) treated supercoiled/incised supercoiled; S/L ) treated supercoiled/incised linear; and L/L ) treated linear/incised linear). Incision efficiencies were calculated as described in the Materials and Methods. Average adducts per plasmid: pTHQ008 S/S and S/L, 6.2; pTHQ008 L/L, 5.6; pTHQB04 S/S and S/L, 6.1; and pTHQB04 L/L, 5.8. UvrABC incision of each DNA sample was preformed three times each time in triplicate, and incision efficiencies were calculated. DNAs were probed with sense and anti-sense probes. Results of incisions on sense strands are shown. Means ( SD are plotted.

Figure 5. Incision of stereospecific BPDE adducts on oligonucleotide substrates by B. caldotenax UvrABC. Site specifically modified and strand specifically [32P]-labeled double-stranded oligonucleotides (prepared as described in the Materials and Methods and shown in Figure 2) were preincubated with UvrA and UvrB for 30 min. UvrC was added, and the reaction was incubated an additional 60 min. Reaction products were denatured and resolved on 12% polyacrylamide sequencing gel. Oligonucleotide concentration ) 1 nM. UvrABC concentrations as multiples of 1× ) 2.5 nM UvrA, 62.5 nM UvrB, and 25 nM UvrC. Total reaction volume ) 20 µL. Lengths of oligonucleotide markers are noted.

control substrates that did not contain a BPDE adduct. Thus, B. caldotenax UvrABC and E. coli UvrABC more efficiently incise cis-BPDE adducts. These data taken together with the preferential incision of adducted supercoiled DNA suggest that there likely is a higher ratio of cis to trans adducts formed on supercoiled DNA than on linear DNA.

Jiang et al.

Figure 6. NLN 77 K fluorescence spectra of the BPDE adducts formed in supercoiled and linear pTHQB04 plasmids in aqueous buffer. (A) Spectra a and b were obtained with λex of 346 nm for supercoiled and linear DNA samples, respectively. Spectra c (supercoiled DNA) and d (linear DNA) were obtained with excitation at 355 nm. The inset shows normalized fluorescence excitation spectra of BPDE adducted supercoiled (solid line) and linear (dotted line) DNA samples obtained at 77 K with λobs ) 379 nm (band pass ∼ 5 nm). Arrows indicate excitation wavelengths of 346 and 355 nm. (B) Spectra e and f (λex ) 355 nm) show the corrected (0,0)-bands (at 380.4 nm) of the intercalated and/or quasi-intercalated adducts in supercoiled and linear DNA. The asterisk locates the emission band near 376.5 nm corresponding to BPDE adducts in an unusual conformation. A similar origin band was observed for (-)-cisanti-BPDE-N2-dG adducts in doubled-stranded 5′-d(CCATCGCTACC)‚d(GGTAGCGATGG) oligonucleotide, where G denotes the lesion site (16).

3.4. Stereospecificity of BPDE Adducts Formed on Supercoiled and Linear DNAs. The different incision efficiencies of B. caldotenax UvrABC on DNAs adducted when supercoiled (S/S) and adducted when linear (L/L) suggest that the relative distribution of cis and trans adducts formed on supercoiled and linear DNA is different. To test this hypothesis, BPDE-adducted supercoiled and linear pTHQ008 and pTHQB04 plasmid DNAs were analyzed by NLN and FLN spectroscopy at 77 and 4.2 K, to explore the trans-/cis-BPDE adduct ratio and adduct conformations formed in the two plasmids at an adduct density of ∼6-7 adducts per plasmid. Only results obtained with pTHQB04 plasmid are presented because similar results were obtained with pTHQ008 plasmid. The 77 K fluorescence excitation spectra (normalized at the band maximum) of BPDE-DNA adducts are presented in the inset of Figure 6A as solid (supercoiled DNA) and dashed (linear DNA) curves. The band maxima in both spectra lie near 347 nm, which wavelength marks the onset of the S0 f S2 absorption system, in agreement with the absorption maxima obtained for (+)-transBPDE-N2-dG adducts in CT DNA and various oligonucleotides (25, 26). This result indicates that the majority of adducts in both supercoiled and linear DNA samples are (+)-trans adducts. However, the excitation spectrum for supercoiled DNA is slightly broader with low energy tailing (>350 nm). The tailing is evident in room temperature absorption spectra (data not shown). The difference between the supercoiled and the linear excitation spectra exhibits an absorption band with a maximum near 354-355 nm that is characteristic for (+)cis type adducts (25). This result suggested that (+)-cis adducts are more abundant on supercoiled DNA. NLN fluorescence spectra for BPDE-adducted supercoiled and linear DNA in water/buffer matrix obtained with an excitation wavelength of 346 nm are shown in


Figure 6A as curves a and b, respectively. Spectra c (supercoiled DNA) and d (linear DNA) were obtained with excitation at 355 nm. All spectra were obtained under identical conditions with an excitation intensity of ∼180 mW/cm2. Spectra b and d were multiplied by the ratio of supercoiled to linear DNA mass. The excitation wavelengths of 346 and 355 nm preferentially excite the external and intercalated adduct types, respectively (12, 16, 26). The fluorescence origin maxima in spectra a and b are at 378 nm, which wavelength indicates that the fluorescence is mostly due to external (solvent exposed) (+)-trans-BPDE-N2-dG [(+)-1 type] adducts (26). Although the positions of the origin bands in spectra a and b are very similar, the corresponding R values (defined as the ratio of the intensity of the fluorescence origin band to that of the prominent ∼1400 cm-1 vibronic band) for spectra a and b are somewhat different, i.e., 1.8 and 1.6, respectively. These values increase to 2.2 and 1.8, respectively, when the contribution from (+)-3 type (cis) adducts is subtracted from spectra a and b of Figure 6A (see below). It is well-established that the more hydrophobic the environment of the pyrenyl chromophore, the lower the R value [the R value for (+)-1 BPDE adducts in CT DNA is 1.75] (12). The difference between the R values of 2.2 and 1.8 suggests that the conformation of (+)-1 adducts depends to some degree on DNA secondary structure. The emission maxima in spectra c and d of Figure 6A obtained with excitation at 355 nm are at 380 nm, a red shift of 2 nm relative to the origin maxima of (+)-1 adducts. On the basis of results in prior studies (16, 26), fluorescence with an origin band at ∼380 nm can be assigned to (+)-3 intercalated and/or (+)-2 base-stacked adducts. The broader bandwidth of the ∼380 nm origin band [relative to the 378 nm origin band of the (+)-1 adduct] indicates stronger electron-phonon coupling for the corresponding BPDE-derived adducts (see below). Comparison of the origin band profiles in spectra c and d of Figure 6A reveals that the supercoiled DNA (curve c; see solid arrow) has a significantly larger contribution from base-stacked and intercalated adducts than the linear DNA (curve d), consistent with the excitation spectra shown in the inset. Spectra e and f in Figure 6B were obtained by subtracting spectrum a from spectrum c and spectrum b from spectrum d, respectively. For ease of inspection, spectra e and f were normalized for the same contribution from the (+)-1 adducts. The fluorescence origin bands in spectra e and f are at 380.4 nm. Again, this wavelength is about that expected for intercalated and base-stacked conformations. The ratio of the integrated areas of curves e and f of Figure 6B is about 2.4, which demonstrates that the supercoiled DNA contains about 2.4 times more intercalated and basestacked BPDE-derived DNA adducts than the linear DNA. However, the associated R values are significantly different, 1.9 and 1.5 for supercoiled DNA (curve e) and linear DNA (curve f), respectively. This suggests that the relative contribution from (+)-2 and (+)-3 adducts to spectra e and f must be different. The smaller R value of 1.5 observed for linear DNA is very similar to that observed for cis-BPDE-N2-dG [type (+)-3] adducts in oligonucleotide duplexes [measured from NLN spectra obtained in the experiments that led to ref 16 on the fluorescence properties of sequence-defined duplexes oligonucleotides with specified G BPDE adduct site and with conformations determined by NMR (4)] but about


Figure 7. NLN 77 K fluorescence (0,0)-origin bands of BPDEDNA (supercoiled; pTHQB04 plasmid). (A) Spectra a and b were obtained with an excitation wavelength of 355 nm in water/ buffer and glycerol/water/buffer glasses, respectively. (B) Spectra c and d were obtained with excitation at 355 and 346 nm in glycerol/water/buffer glass in the presence of acrylamide (∼1 M) quencher. Curve d was scaled to reveal the relative contribution from the (+)-1 type adduct to spectrum c. Curve e is the difference spectrum (c - d) representing the contribution of the (+)-3 type adduct.

20-30% higher than that observed for typical intercalated adducts formed by BPDE in CT DNA (12, 26). In contrast, the R value of 1.9 for supercoiled DNA is similar to the R value of 1.95 obtained for the minor [(+)-2] conformation of trans-BPDE-N2-dG adducts in oligonucleotide duplexes (16) but higher (∼35%) than that observed for (+)-2 type adducts in BPDE-adducted CT DNA (12, 26). The presence of (+)-2 type adducts in DNA plasmids can be tested by addition of glycerol to the aqueous buffer, because glycerol can disrupt the weak stacking interaction of the minor (+)-2 base-stacked conformation, thereby increasing the external (+)-1 type adduct contribution (16). This effect is illustrated in Figure 7A, where NLN spectra a and b were obtained for the BPDE-adducted supercoiled DNA with 355 nm excitation in water/buffer and glycerol/water/buffer matrixes, respectively. In the presence of glycerol (curve b), a fraction of adducts (∼2030%) undergoes (+)-2 to (+)-1 conformational transformation, as indicated by the increased fluorescence near 378 nm (see gray arrow). This transformation is believed to be due to an increased solvent compatibility towards the aromatic moiety. These results indicate that (+)-2 type adducts likely contribute to the adducts formed on supercoiled DNA. The (+)-2 to (+)-1 transformation was negligibly small in linear DNA (spectra not shown), suggesting that linear DNA contains relatively little of the (+)-2 adduct type and indicating that the fluorescence with origin band near 380 nm in spectra d and f of Figure 6 is most likely due to a (+)-3 type adduct. On the basis of the fluorescence properties and structures of BPDE adducts of sequence-defined oligonucleotides in (16), we assign the (+)-2 supercoiled adduct as trans-BPDE-N2dG. Because only 20-30% of the internal adducts of supercoiled DNA (excited at 355 nm) undergo a glycerolinduced +2 to +1 transformation, the majority of internal adducts responsible for the fluorescence maximum near 380 nm (Figure 7A) is assigned to (+)-3 intercalated adduct, which, on the basis of the results in ref 16, is further assigned to cis-BPDE-N2-dG. Further support for this (+)-3 intercalated assignment for supercoiled DNA is provided by results of fluorescence quenching (by acrylamide) experiments. Acrylamide efficiently quenches fluorescence of external, but not intercalated, BPDE adducts (12). Spectrum c in Figure 7B is that of supercoiled DNA in a glycerol/water/buffer glass containing 1 M acrylamide with excitation at 355

336


nm. Its fluorescence origin band at 380.2 nm is red shifted by 0.8 nm relative to the maximum of curve b of Figure 7A, also obtained with excitation at 355 nm. A slight red shift is expected when only the external (+)-1 adduct undergoes quenching. Support for this comes from comparison of curve b (glycerol/water/buffer glass) of Figure 7A with the origin bands of spectra d and e of Figure 7B. The latter spectrum (curve e) is the difference between spectra c and d. Spectrum d is that of supercoiled DNA (glycerol/water/buffer, 1 M acrylamide) obtained with excitation at 346 nm and scaled to fit the ∼378 nm high energy shoulder of the origin band in spectrum c. Spectrum d is the contribution of the (+)-1 external adduct to spectrum c, and spectrum e is that of the (+)-3 adduct (cis-BPDE-N2-dG) with its origin band at 380.4 nm. Comparison of the relative fluorescence intensity near 378 and 380 nm in curve b of Figure 7A (glycerol/ water/buffer without acrylamide quencher) and curve c in Figure 7B (glycerol/water/buffer plus 1 M of acrylamide) indicates that mostly fluorescence of the (+)-1 adduct [(+)-trans-N2-dG] is quenched, consistent with spectrum e being due to an intercalated adduct. This was confirmed by spectral deconvolution (data not shown) of the origin band in spectrum b (Figure 7A) into “pure” contributions from (+)-1 and (+)-3 adducts using the origin bands of spectra d and e. Considered next are the R values of spectra d and e of Figure 7B for the (+)-1 trans-BPDE-N2-dG and (+)-3 cisBPDE-N2-dG adduct, respectively. The value for the former adduct is 2.2, identical to that for supercoiled DNA in the water/buffer glass (see above). Thus, glycerol does not appear to affect the conformation of external type adducts. However, the value of 2.2 is considerably larger than the value of 1.8 for the external adduct of linear DNA (see above). The difference indicates that DNA secondary structure affects the conformation of the external adducts. The R value for the intercalated adduct of supercoiled DNA (spectrum e) is 1.5, which is similar to the value of 1.55 for the cis-BPDE-N2-dG intercalated adduct on duplex oligonucleotides (8). We note that the R value of 1.9 for spectrum e of Figure 6B (supercoiled DNA, water/buffer glass) is higher than 1.5 because spectrum e is contributed to by about 20-30% of basestacked trans-BPDE-N2-dG adducts (corrected spectra not shown), whereas spectrum e of Figure 7B is that of the pure, intercalated cis-BPDE-N2-dG adduct. On the basis of spectra e and f of Figure 6B, it was concluded that supercoiled DNA contains about 2.4 times more base-stacked/intercalated adducts than linear DNA. Taking into account that spectrum e of Figure 6B is contributed to by 20-30% base-stacked adducts, we estimate that supercoiled DNA contains 1.7-1.9 times more of the cis-BPDE-N2-dG adduct than linear DNA. This factor is consistent with the more efficient incision observed in supercoiled DNA by UvrABC, which recognizes cis-BPDE adducts more readily than trans-BPDE adducts (Figures 4 and 5). The NLN fluorescence spectra obtained for supercoiled DNA that was linearized after adduction (S/L DNA) are consistent with this more efficient incision. For the sake of brevity, the spectra are not shown. It was found that linearization has essentially the same effect as addition of glycerol to the water/buffer glass forming solvent, i.e., linearization converted 20-30% of the adducts from the (+)-2 base-stacked conformation to the external (+)-1 conformation. Because no significant difference in incision efficiency between supercoiled and S/L

Jiang et al.

Figure 8. FLN spectra of BPDE adducted DNA obtained in water/buffer glass with an excitation at 356.9 nm (T ) 4.2 K). Curve a ) supercoiled DNA; curve b ) linear DNA. The zerophonon lines are labeled with excited state vibrational frequencies.

DNA was observed (Figure 4), this suggests that UvrABC preferentially recognizes the intercalated cis-BPDE adduct over the base-stacked trans-BPDE adduct. Figure 8 shows 4.2 K FLN spectra for supercoiled DNA (spectrum a) and linear DNA (spectrum b) obtained with a water/buffer glass and excitation at 356.9 nm. The sharp zero-phonon lines at 1442, 1521, 1560, and 1615 cm-1 correspond to the excited state (S1) frequencies of the active pyrenyl vibrations. The frequencies and relative intensities of the zero-phonon lines are those expected for the external trans-BPDE-N2-dG adduct (12, 16, 27). The broad fluorescence indicated by the asterisk near 380.4 nm is due to base-stacked and intercalated BPDE-N2-dG adducts. These conformations are characterized by strong electron-phonon coupling, which leads to a loss of line narrowing (12, 16, 20). Consistent with the NLN fluorescence results is that supercoiled DNA contains more internal adducts than linear DNA. Moreover, the intensity of the broad fluorescence only slightly decreased in the glycerol/water/buffer matrix, which indicates that most of the broad fluorescence is due to the (+)-3 intercalated adduct. The NLN spectra shown in Figures 6 and 7 indicate that external trans-BPDE adducts with a fluorescence origin near 378 nm dominate the fluorescence due to internal adducts with origin bands near 380 nm. Thus, the total adduction level of (+)-trans external and basestacked adducts in supercoiled and S/L DNA is greater than that of the (+)-cis intercalated adduct. This was confirmed by [32P]postlabeling. The relative abundance of cis and trans adducts was also determined by [32P]postlabeling analyses of supercoiled and linearized plasmid DNAs containing approximately six adducts per plasmid. Samples were prepared, and stereospecific adducts were resolved and quantified as detailed in the Materials and Methods. Representative chromatograms are shown in Figure 9. No adducts were detected from THF (solvent control)treated plasmids mixed with untreated genomic DNA (Figure 9A,C). We observed two major spots and 3-4 minor spots in the chromatograms from both supercoiled (Figure 9B) and linear (Figure 9D) DNA treated with BPDE. The two major spots (1 and 2) contained 85-90% of the adducts (Table 2) and were identified as cis- and trans-BPDE-N2dG-adducts, respectively, by comparison with the migration of stereospecific standards derived by processing oligonucleotides site specifically modified with stereospecific BPDE adducts (data not shown). The minor adducts (spots 3-6) were not identified. Quantification


Figure 9. [32P]postlabeling analyses of solvent control and BPDE-treated supercoiled and linearized pTHQ008. (A) THFtreated supercoiled pTHQ008, (B) BPDE-treated supercoiled pTHQ008, (C) THF-treated linearized pTHQ008, and (D) BPDEtreated linearized pTHQ008. Table 2. Distribution of BPDE Adducts on [32P]postlabeling Chromatograms

a

adduct

% in supercoileda

% in lineara

1 (cis) 2 (trans) 3 4 5 6

21 70 1 3 6 1

13 72 5 4 5 1

Rounded to nearest integer.

of the adducts (Table 2) showed that the fraction of adducts in the cis conformation (adduct 1) was markedly different in supercoiled and linear DNA (21 vs 13%). In constrast, the percent of adducts in the trans configuration showed little difference in supercoiled and linear DNA (70 vs 72%). Of the minor adducts, only adduct 3 was different in linear vs supercoiled DNA (5 vs 1%).

4. Discussion We have demonstrated using low temperature fluorescence methods, postlabeling analyses, and incision by UvrABC endonuclease that BPDE adduct configuration and conformation are strongly influenced by the secondary structure (supercoiled vs linear) of the DNA double helix. Our results indicate that [3H]BPDE at low adduct density forms ∼20-30% of the trans adducts in the basestacked conformation [i.e., (+)-2] and approximately 2-fold more (+)-3 cis adducts in supercoiled DNAs than in linear DNAs. The significance of these results is that these conditions more closely mimic in vivo exposure conditions because DNA in cells is supercoiled and low BPDE adduct densities are obtained at physiological exposures to BP. Our results show that the B. caldotenax UvrABC incises cis adducts better than trans adducts on short oligonucleotide duplexes similar to the differential incision exhibited by the E. coli UvrABC (18, 24). The differential incision is not as great as seen with human NER extracts using longer oligonucleotide substrates (28). Nevertheless, adducts formed by low concentrations of [3H]BPDE on supercoiled plasmids are incised


6-7-fold more efficiently by UvrABC than adducts formed on linear plasmids. The preferential incision also occurs when the adducted supercoiled DNA is linearized before incision indicating that the UvrABC recognizes adducts equally well on supercoiled and linear DNAs. Similar results were obtained by others using E. coli UvrABC to incise BPDE adducts formed on supercoiled ΦX174 DNA before and after linearization (7). Preferential incision of adducts formed on supercoiled DNAs suggests that formation of cis adducts is favored on supercoiled DNA as compared with linear DNA. This suggestion was strongly supported by low temperature fluorescence spectroscopy and postlabeling analyses of adducted supercoiled and linear plasmid DNAs, which both indicated that supercoiled DNA has about 2-fold greater amount of intercalated cis-BPDE adducts than linear DNA. That is, the ratio of cis to trans adducts is higher in supercoiled than in linear DNAs. It is not readily apparent that a 2-fold increase in cis adducts and a 2-fold greater incision of cis adducts is sufficient to explain the 6-7-fold increase in incision efficiency observed on supercoiled substrates. The discrepancy may reflect the difference in the BPDE treatments between DNAs analyzed for UvrABC incision and those used for fluorescence and postlabeling. The DNAs prepared for fluorescence were adducted at higher concentrations of both DNA and BPDE in the reactions. The UvrABC incision efficiency of these DNA preparations was checked for qualitative differences but was not extensively assayed to confirm quantitative correspondence with incision of DNAs adducted at low concentrations. It is possible that the increase in formation of cis adducts is greater at the low concentration treatment. In addition, some of the discrepancy likely is caused by somewhat lower fluorescence quantum yield of [(+)-2/(+)3 type] adducts whose electronic transitions are known to possess charge transfer character and by uncertainty in estimation of the contribution from (+)-2 type adducts to the supercoiled DNA sample. Furthermore, the incision on oligonucleotide substrates is not as efficient as on plasmid substrates. Even with twice as much UvrABC per adduct and longer incubation in the oligonucleotide incisions, much less than half the adducts are incised. This lack of robust incision on oligonucleotide substrates also is observed by others (18, 24, 28). Another caveat is that incision on oligonucleotide substrates was assayed in a single sequence context. The sequence contexts of the adducts formed on plasmid DNAs is unknown and could be different on supercoiled and linear plasmids. The influence of sequence context on UvrABC incision of cisand trans-BPDE-N2dG adducts also is not well-established. Thus, one cannot directly extrapolate a quantitative relationship between incisions on these oligonucleotide substrates to incisions on plasmids. Rather, at this point, we must simply note the qualitative correspondence between the greater UvrABC incision efficiency of cis adducts and the presence of more cis adducts demonstrated by fluorescence and postlabeling analyses on supercoiled vs linear DNAs and allow that there may be subtle differences in adduct conformation on supercoiled DNAs that UvrABC recognizes that are not detectable by the analytical chemistry techniques. Our observations indicate that the fluorescence spectrum obtained from adducted supercoiled DNA (Figure 6B, spectrum e) may contain more contribution from (+)-2 type adducts than seen in the spectrum of adducted

338


linear DNA (Figure 6B, spectrum f). However, the differences and similarities in the R ratio may be accidental, since the value of the R ratio is also sequencedependent as shown for trans-BPDE adducts in various sequence-defined oligonucleotides (unpublished results). This phenomenon is hardly described in the literature and needs further exploration. We suggest that structurally identical BPDE-DNA adducts can get trapped as energetically inequivalent conformers, and this heterogeneity is responsible for the experimentally observed mutational complexity (29). There are a number of potential factors that could have influenced the formation of specific adduct configurational and conformational isomers. More prominent among these factors are the effects of sequence specificity on specific adduct formation (both configuration and conformation effects), which are not completely understood as yet. Furthermore, sequence context-induced heterogeneity could also be held responsible for the dispersive photodissociation process observed in BPDE-adducted CT DNA (30). The R value for the (+)-1 adducts increases when the contribution from (+)-3 type adducts is subtracted from spectra a and b in Figure 6A; the corrected R values for the adducted supercoiled and linear DNA are 2.2 and 1.8, respectively. This finding is of particular interest, since it reveals that the R ratio for (+)-1 adducts in supercoiled DNA (R ) 2.2) is indeed significantly higher than that observed for the external (+)-1 trans-BPDE adducts in linear DNA (R ) 1.8) and CT DNA (R ) 1.75) (12). Thus, it is feasible that (+)-1 trans-BPDE-DNA adducts in supercoiled and linear DNA might be formed in different regions of DNA, and as a result, the corresponding adducts are characterized by different R values. This difference in adduct formation may also contribute to the greater UvrABC incision efficiency for adducts formed on supercoiled DNA. At this point, three remarks are pertinent as follows: (i) the external (+)-1 adducts in supercoiled DNA appear to be more solvent exposed than those in linear DNA; (ii) supercoiled DNA promotes formation of cis-BPDE [(+)-3] adducts and base-stacked [(+)-2] trans-BPDE adducts; and (iii) DNA conformation contributes to the conformational complexity of adducts. Results presented in this paper indicate that the (+)trans/(+)-cis adduct ratio in supercoiled DNA resulting from adduction with BPDE (∼1 lesion/kbp) is about 4, thus remarkably lower than the (+)-trans-/(+)-cis-BPDE adduct ratio of >9 and 6 observed for poly(dG-dC)‚poly(dG-dC) and poly(dG)‚poly(dC), respectively (31). This is interesting, since even smaller ratios of ∼0.4 and ∼0.6 were obtained in frequently mutated -AAGGAA- and -GAGGAG- sequences (32). In contrast to the commonly held belief gained from experiments with high adduct density (∼100/kbp) (7, 33), it is feasible that the (+)trans-/(+)-cis-BPDE adduct ratio may be significantly lower at very low damage levels (

Supercoiled DNA Promotes Formation of Intercalated cis-N2

Recommend Documents