Sequence dependence of benzo[a]pyrene diol epoxide-DNA adduct conformer distribution: A study by laser-induced fluorescence/polyacrylamide gel electrophoresis

Chem. Res. Toxicol. 1994, 7, 98-109

98

Sequence Dependence of Benzo[ alpyrene Diol Epoxide-DNA Adduct Conformer Distribution: A Study by Laser-Induced Fluorescence/Polyacrylamide Gel Electrophoresis G. A. Marsch,? R. Jankowiak, M. Suh, and G. J. Small* Ames Laboratory-USDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011 Received July 29, 1993” Low-temperature laser-induced fluorescence techniques in combination with polyacrylamide gel electrophoresis (LIF/PAGE) were used to study the binding of (-)-anti-and (+)-anti-benzo[alpyrene 7,8-dihydrodiol 9,lO-epoxide (anti-BPDE) to several sequence-defined duplex oligomers. Two of the oligomers contain central 5’-RAGGAR-3’ sequences (R = purine) which appear to be frequently mutated by racemic (&)-anti-BPDEin endogenous genes of cells cultured in vitro. Two contain a central 5’-CCGG-3’ or 5’-TGGT-3’ sequence which are strongly preferred for covalent binding but appear to be not so frequently mutated. Binding of the two enantiomers to the latter two sequences yielded a distribution of BPDE-N2-dGadduct conformations similar to those from binding to highly polymerized, random sequence DNA in vitro which, for (&)anti-BPDE, means that the helix-external conformation of the N2-dG adduct is dominant. Binding of (-)-anti-BPDE to the 5’-RAGGAR-3’ sequences yielded more partially base-stacked and less base-stacked (quasi-intercalated) conformer than observed for random sequence DNA. Importantly, the (+)-anti-BPDE in binding to the more mutagenically inclined 5’-RAGGAR-3’ sequences yielded little external-type adduct in comparison to the other two sequences and random sequence DNA. Moreover, an unusually high proportion of the (+)-anti-BPDE adducts formed with the 5’-RAGGAR-3’sequences result from cis stereoaddition, which yields a partially base-stacked configuration. Since the (+)-anti-BPDE appears to be the more mutagenic, this result suggests a possible role of internal adduct conformations in mutagenesis.

I. Introduction Benzo[alpyrene (BP)’ is a ubiquitous environmental tumorigen that continues to be studied extensively. Potent mutagens or carcinogensare usually electrophilic reactants that directly bind to DNA, or they require metabolic activation to electrophilic derivatives (the ultimate carcinogen) that may then form covalent DNA lesions (1-5). DNA adducts of many polycyclic aromatic hydrocarbon (PAH) metabolites have been investigated, especiallythose resulting from attack by diol epoxides (1-4)and, more recently, the radical cation (5)of benzo[alpyrene (6,7) and 7,12-dimethylbenz[alanthracene(8,9).There are no direct assays presently existing that test which particular adduct species or configuration initiates cancers, but mutation “hot spots” of PAHs can be directly detected by using carcinogen-modified shuttle vectors to introduce the carcinogen lesions into cultured cells (10-13). Mutations may also be scored by assaying carcinogen-damaged endogenous DNA sequences for sites of high mutagenic potential (14,15).For PAHs, the mutations inEscherichia coli (16,17)or mammalian cell lines were predominantly

* To whom correspondence should be addressed.

t Present address: Biology and Biotechnology Research Program, L-452, Lawrence Livermore National Laboratory, Livermore CA 94551. 0 Abstract published in Advance ACS Abstracts, January 1, 1994. 1 Abbreviations: (A)-anti-BPDE, (+)-anti-benzo[alpyrene 7,&dihydrodiol 9,lO-epoxide; BP, benzo[alpyrene; PAH, polycyclic aromatic hydrocarbon;LIF/PAGE, laser-induced fluorescence/polyacrylamide gel electrophoresis; FLN, fluorescence line-narrowing;NLN, non-line-narrowed; ss, single-stranded oligonucleotides; ds, double-stranded oligonucleotides; BPTs, benzo[alpyrenetetraoIs; ZPL, zero-phonon line; R, purine; Y, pyrimidine.

- -

A T or G T transversions (10-17). Of equal importance, however, is the sequence context in which the mutation occurs, although this aspect of the analysis has not received as much attention. These in vitro mutagenesis assays performed for racemic (i)-anti-benzo[alpyrene 7,&dihydrodiol9,10-epoxide [(&)anti-BPDEI measure not only sites of mutation but also covalent DNA binding, since the formation of carcinogen adducts is thought to be the initiating step in mutagenesis and carcinogenesis. There are multiple sites of strong covalent binding by BPDEs [-CGC-, poly(dG), -AGG-, -TGG-, and -TGY-, where Y = a pyrimidine] (18-23); -AGG- sequences flanked by purines appeared especially susceptible to mutation (10-15).Studies on the mutations formed within endogenous aprt (14)and dhfr (15)gene loci were of the greatest utility in determining most preferred sites of mutagenesis since the DNA sequence scored for mutations was long enough to accurately assess the effect of flanking bases on the mutagenic potential of racemic anti-BPDE. It was found that five of seven 5’AG(G),GA-3’ sequences in the dhfr locus (15)were strongly mutated by racemic anti-BPDE, and it was this finding which led to the synthesis of two oligodeoxynucleotides with similar 5’-RAGGAR-3’sequences used in this study. Mutagenesis studies using shuttle vectors with supF genes contained two -AGG- triplets (10-13).The -AGG- triplet from positions 121-123 was a stronger mutation “hotspot” than the -AGG- at positions 158-160; the latter was 3’flanked by two thymines, while the former was flanked by purines at both ends. -CGG- sites, the strongest triplet site of piperidine-labile covalent lesion formation by BPDE

0893-228~/94/2707-0098$04.50/0 0 1994 American Chemical Society

Sequence Dependence of BPDE-DNA Adducts

Chem. Res. Toxicol., Vol. 7, No. 1, 1994 99

Table 1. Spectral Characteristics of (*)-anti-BPDE Covalent Adducts in DNA fluorescence origin,aXfl (O,O), nm quenching efficiency FWHM,b (O,O),cm-l conformation high 378.0 200 helix-external 379.3 intermediate 200 partially base-stacked low 380.6 -320 base-stacked (quasi-intercalated) a Spectra were obtained in standard glycerol glass at 77 K (see ref 28). Full width at half-maximum.

adduct type

(18), were mutated by the carcinogen (10, 161, but less frequently than -AGG- sites given the facile generation of adducts at CGG sites (14, 15). Racemic anti-3,4-dihydroxy-1,2-epoxybenzo[clphenanthreneand anti-1,2-dihydroxy-3,4-epoxy-5-methylchrysene also appear to induce mutations at -AGG- sites more frequently than average, although not always in a purine tract (24,251. By contrast, shuttle vectors introducing BPDE lesions into repairdeficient E. coli cells usually generated mutations at sites other than -AGG-, especially at -TGG- sequences (26). Since adduct formation is sequence-specific, we wished to ascertain whether BPDE binding to -RAGGAR- sequences results in distributions of adduct conformations that deviate from the distributions observed when binding occurs to 5’-CCGG-3’ or 5’-TTGGTT-3’ sequences. Conformations of anti-BPDE enantiomer-DNA complexes have been studied by circular dichroism, absorbance, and conventional fluorescence spectroscopy and classified as ( 2 , 27-31) either helix-internal (base-stacked and intercalated or quasi-intercalated conformations) or helixexternal. The major stable adducts result from covalent attachment of BPDE to DNA by either a cis or trans addition of the (2-10of BPDE to the exocyclic N2of guanine (32). The ratio of trans/& adducts is sequence-dependent (33-351, and for the limited number of sequences studied with (+)-anti-BPDE, trans addition predominated and resulted in external-type adducts. The translcis binding ratio for (-)-anti-BPDE is much lower than for the (+)isomer, and the cis addition appears to result mainly in quasi-intercalated adducts (33-35). The effects of base sequence on the yield of covalent (+)-anti-BPDE-N2-dG adducts generated when (+)-anti-BPDEwas reacted with single-stranded oligodeoxynucleotides of different sequences have been recently studied (35). The translcis ratio of (+)-anti-BPDE-N2-dGlesions in 5’-d(CTATNGNTATC)-3’oligomers (where N denotes any base) was shown to be in the range of 3-5 (35). Effects of base sequence on the relative yields of cis or trans adducts in duplex oligomers have not received adequate attention. Such effects will be considered in this paper using results obtained with laser-induced fluorescence/polyacrylamide gel electrophoresis (or LIF/PAGE). Laser-induced fluorescence spectroscopies have been extensively applied in the determination and characterization of macromolecular DNA, polynucleotide, and nucleoside adducts derived from BPDE isomers. Our methodology (27,28,36-39) employs both line-narrowing ($31 SO vibronic excitation at 4.2 K) and non-lineSO excitation at 77 K) conditions as narrowing (S2 reviewed in refs 36 and 37. In what follows the former and latter conditions will be referred to as FLN and NLN, respectively. In combination, they provide unprecedented spectral selectivity at a detectability of -1 fmol for moderately strong fluorophores such as pyrene. Of importance to the present paper is that NLN/FLN can distinguish between cis and trans addition products of BPDE (38,42)and between helix-external, partially basestacked, and base-stacked (quasi-intercalated) conforma-

- -

--

tions of the stable N2-dG duplex DNA adducts from (+Iand (-)-anti-BPDE (34-36). These three helix conformations are denoted as (*)-l, (*I-2, and (*)-3, respectively [(*)-j,j = 1-31. Table 1summarizes their relevant spectral and other properties. It is important to note that the shifting to the red of the NLN fluorescence origin of the pyrenyl moiety with increasing base-stacking is paralleled by increasing electron-phonon coupling strength of the optical transition. This coupling presumably reflects the amount of charge-transfer character of the SI state introduced through base-stacking. The coupling is weak, intermediate, and strong for the (*)-l, (*)-2, and (*)-3 adducts, respectively (27, 28, 36). I t was also observed that fluorescence quenching efficiency (by acrylamide) is high, intermediate, and low for the (*)-l, (*)-2, and (f)-3 adducts, respectively ( 2 7 , 2 8 ) . Although not included in Table 1, we emphasize that the vibronically excited FLN spectra have led to a quite complete library of the excitedstate (SI) vibrational mode frequencies and relative Franck-Condon factors of the pyrene moiety for different adducts (27). The mode frequencies and Franck-Condon factors for cis and trans adducts are significantly different ( 2 8 , 3 8 ) . In ref 27 evidence for formation of a minor N6dA adduct from (+)-anti-BPDEwith a fluorescence origin at 382.5 nm was presented. This adduct does not appear to figure importantly in the present work. Very recently, we demonstrated that FLN and NLN spectroscopies in combination with polyacrylamide gel electrophoresis (PAGE) allow for high-resolution analysis of the conformations of the major BPDE-induced lesions of duplex oligomers (39). For the sequences studied, it was shown that the spectral properties of the N2-dG adducts of single-stranded (ss) and double-stranded (ds) oligodeoxynucleotides are similar. In view of this it was deemed appropriate to denote the ss-N2-dGadducts from with j , = 1-3, by analogy with (&)-anti-BPDEas (*)-jB, (*)-j for the duplex adducts. I t is worth noting that the similarity only extends to the preservation of conformations. In general, upon duplex formation a further “red” and “blue” shift of the (0,O) band is observed (in agreement adducts, respectively. with ref 34) for the (*)-3s and (*)-ls Utilization of PAGE allowed for facile quantitation of the adduct conformations. Further discussion of the results of ref 39 is given later.

-

11. Materials and Methods A. Chemicals. anti-BPDE enantiomers were purchased from the National Cancer Institutes Chemical Carcinogen Standard Reference Repository through Chemsyn Science Laboratory (Lenexa, KS) as 1.0 mg/mL solutions in tetrahydrofuran/ triethylamine (19:l v/v). T4 polynucleotide kinase and highly polymerized DNA from calf thymus were obtained from Sigma Chemical, Inc. (St. Louis, MO), and the [-p32P]ATP (3000 Ci/ mmol) was purchased from New England Nuclear (Boston, MA). All oligodeoxynucleotides were synthesized on an Applied Biosystems, Inc. (Foster City, CA), PCR Mate DNA synthesizer a t the Iowa State University Nucleic Acids Facility and purified by a Vydac C4 preparative column (Separation Group, Hesperia, CA). We synthesized a self-complementaryDNA oligonucleotide,

100 Chem. Res. Toxicol., Vol. 7, No. 1, 1994

d(AAACCGGTTT), plus three other G-containing oligonucleotides [d(TTAAGGAATT), d(TTGAGGAGTT), and d(AATTGGTTAA)]. The complements of these oligonucleotideswere also synthesized. The central 5'-NGG-3' sequences are sites of preferred covalent binding by BPDEs (18,43),and C and T are essentially unreactive. In addition, flanking AA are sites of low binding efficiency (41). In what followsthe above oligonucleotides are denoted as d(-CCGG-),d(-AAGGAA-),d(-GAGGAG-),and d(-TGGT-) with the 5'-terminus at the left. B. Oligodeoxynucleotide-BPDE Reactions. The appropriate strand was labeled on the 5'-terminus by [-y-32P]ATPand T4 polynucleotide kinase. The radiolabeled DNA oligonucleotides were repurified by electrophoresis under denaturing conditions (7 M urea, 40 "C) through 20% polyacrylamide gels. To extract purified, radiolabeled oligomer, autoradiography of the gel was done by Kodak XOMAT-AR5X-ray film (Rochester, NY), and the bands correspondingto whole 10-merswere excised. The polyacrylamide fragments imbedded with oligodeoxynucleotides were placed in 2 mL of doubly distilled water and eluted out overnight. Eluted oligonucleotideswere desalted with SepPak C18 cartridges from Waters Associates (Milford, MA), and the resulting methanol/water elutions were dried in a Savant SpeedVac concentrator (Farmingdale, NY). Oligonucleotideslabeled at the 5'-terminus with32P and a 10fold excess of complement unlabeled by 32Pwere mixed together in a buffer solution composed of 20 mM dibasic phosphate (pH 7.0), 100 mM NaCl, and 1 mM Na2EDTA. Oligonucleotide solutions were placed in an 80 "C heating block and cooled to 4 "C over 2 h to induce slow error-free annealing of the strands. In a typical low-dose reaction, anti-BPDE enantiomers were added at a ratio of one carcinogen molecule per 20-30 oligonucleotide bases. High levels of covalent modificationresulted after BPDEs had been five times to the d(32P-TGGT-)solution. Each addition ofBPDErepresenteda0.1 BPDE/baseratio (totalBPDE added = 0.5 BPDE/base); tetraols were removed after each addition of BPDE. BPDEs were reacted with duplex oligomers in the dark (to prevent photolysis of the activated tumorigen) for 2 h at T = 18 "C. This temperature was much lower than the melting temperatures (T,)of the studied duplex strands with the exception of d(-AAGG-)-d(-CCTT-)oligomer, for which T, was -25 "C. This indicates that only the latter oligomer may not have been in complete duplex form during the reaction with (*)-anti-BPDE enantiomers. However, it remains to be established whether the distributions and yields of adducts are different in ss and ds oligomers. The benzo[alpyrene tetraols (BPTs) that result from the hydrolysis of BPDE were removed by at least 5 extractions with water-saturated 1-butanol. After the last extraction, the reaction mixtures were dried under a vacuum with the SpeedVac concentrator and resuspended in 10 pL of gel sample buffer composed of l/lOX TBE buffer, xylene cyanole FF tracking dye, and 50% (v/v) glycerol. The presence of tetraol contamination was routinely checked by FLN and NLN spectroscopies. In ref 39 the results of control experiments with PAGE showed that the adduct distributions from the reactions with d P P TGGT-)-d(-ACCA-)and d(32P-TGGT-)followed by annealing with its complement are very similar. This established that the reaction of (+)- and (-)-anti-BPDE with d(-ACCA-) produced no adduct. In the present work this was checked with (+)-antiBPDE for the other two non-self-complementaryoligonucleotides, d(-AAGGAA-)and d(-GAGGAG-),by NLN and FLN spectroscopy. The results (not shown) established, for the former, that the adduct distributions from the two types of reaction protocols are very similar. For the latter, the situation was less clear. Although the results did not reveal significant reactivity with d(-CTCCTC-),they indicate that thereaction with d(-GAGGAG-) followed by annealing with its complement may produce more (+)-3 of the NZ-dG adduct. C. Separation and Purificationof BPDE-Modified DNA Oligomers. Samples consisting of radiolabeled DNA oligomers were applied to a 38 cm X 80 cm Bio-Rad Sequi-Gen nucleic

Marsch et al. acidssequencingsystem (Richmond,CA) powered by a Pharmacia ECPS 3000/150 power supply (Piscatawaw,NJ). Gels were 20% acrylamide w/v [19:1 acrylamide/bis(acrylamide)ratio] with 1 X TBE buffer and no urea. Pre-electrophoresis was performed for -2 h at 2000 V. The samples were then applied and the electrophoresis was continued for 12 h at 1500 V, with the gel equilibrating to 25 "C. Autoradiography of gels using several exposure times was done at ambient temperature with Kodak XOMAT-AR5X-ray film, in order to locate the BPDE-modified DNA oligomers. Quantitation of BPDE adducts was done by densitometry of autoradiograms with an SL-504-XL scanning densitometer from Biomed Instruments, Inc., or by scintillation counting of eluted radiolabeled BPDE-oligonucleotide covalent complexes with a Beta-Trac liquid scintillation counter from Tm Analytic (Brendon,FL). Bands correspondingto carcinogenmodified oligomers were excised from the gel and placed in Eppendorf tubes with 1mL of glass-distilled water. Oligomers eluted overnight at 4 "C; samples were then centrifuged and the supernatant was removed to new tubes and dried in vacuo. Samples were resuspended in a standard glass-formingmixture (1:1 water/glycerol, v/v) used for low-temperature fluorescence spectroscopy. D. Calf Thymus DNA-BPDE Reactions. Calf thymus DNA was repurified by phenol extraction and dissolved in water. The DNA base pair concentration for reactions with BPDEs was typically 150 pM in 20 mM dibasic phosphate buffer (pH 7.0), with 100 mM NaCl and 1.0 mM NaZEDTA. anti-BPDE isomers were added to DNA solutions such that final carcinogen/DNA bp ratios were 5.0,1.0,0.2,0.04, and0.008. Reactions progressed for 1h in the dark, and the tetraols were then removed from the reactions by >5 extractionswith 1-butanol. The DNA was ethanol precipitated and dried in vacuo, and the samples were finally resuspended in glycerol/waterglass and pipetted in quartz tubes for spectroscopic measurements. E. Fluorescence Instrumentation. A detailed description of the apparatus used for FLN and NLN fluorescence spectroscopy is given elsewhere (36). Briefly, the excitation source was aLambdaPhysikFL-2002 dye laser pumped byaLambdaPhysik EMG 102 MSC excimer laser. For FLN, gated detection was accomplished using a Lambda Physik EMG-97 zero-drift controller to trigger an FG- 100 high-voltage gate pulse generator, allowing the delay time and the width of a Princeton Instruments IRY 1024/G/R/B intensified, blue-enhanced photodiode array's temporal observation window to be adjusted. A 1-m focal length McPherson 2061 monochromator was used to disperse the fluorescence. The monochromator and diode array provided a 10-nm spectral segment with a resolution of -4 cm-l at 400 nm. NLN fluorescence spectra were obtained with the same monochromator, a PMT, and an SR Model 250 boxcar averager (Stanford Research). The boxcar averager was interfaced to a PC for data acquisitionand display by an SR245 interface module. Adducts prepared for fluorescence analysis were dissolved in 30 pL of a 1:l (v/v) glass-formingglyceroljwater mixture which was placed in a 2-mm i.d. quartz tube. For FLN or NLN measurements a double-nestedglass cryostat equipped with quartz optical windows was utilized. Samples were slowly cooled to liquid helium or nitrogen temperature. The delay time for fluorescence measurements was 50 ns, and the width of the observationwindow was 100 ns. F. Safety Considerations. The anti-BPDE carcinogensand 32Pradiolabel used in these experiments were in liquid form. Contamination must be avoided by wearing impermeable gloves, laboratory coats, and splash-proof goggles. Manipulations of hazardous materials should take place in specialareas sequestered from the rest of the laboratory. Radiation exposure is kept ALARA ("as low as reasonably achievable-) by working with plexiglass shielding and the proper dosimeters (Geiger counters and body dosimeters). Wastes must be placed in approved containers and disposed properly according to Environmental Safety and Health standards.

-

Sequence Dependence of BPDE-DNA Adducts

Chem. Res. Toxicol., Vol. 7, No. 1, 1994 101 rl

I

x

5

Wavelength (nm) Figure 1. Fluorescence (0,O) origin bands of double-stranded DNA oligomers (not gel separated)modified by (A)-anti-BPDE. Paneh A and B depict binding by (+)-anti-and (-)-anti-BPDE, respectively. The binding of BPDE isomers is to d(TTAAGGAATT).d(AATTCCTTAA) (short dashed line), d(TTGA-

GGAGTT).d(AACTCCTCAA) (solid line), and d(AAACCGGTM')z (dashedanddotted line) oligomers,respectively. Spectra were obtained with A,, = 345 nm (top) and A,. = 355 nm (bottom); T = I7 K.

376

376

360

362

Wavelength (nm) Figure 2. FLN spectra (T= 4.2 K, &, = 356.9 nm) of three

111. Results and Discussion

duplex oligomers modified by (+)-anti-BPDE(unseparated).The binding of (+)-anti-BPDEis to d(-AAGGAA-).d(-TTCCTT-) (dashed line), d(-GAGGAG-).d(-CTCCTC-)(solid line), and d(-CCGG-)Z(dashed and dotted line) oligomers, respectively.

A. Fluorescence Spectra of Unseparated Duplex Oligomers. Guided by the results of our previous work (39))we explore the extent to which laser-induced fluorescence (LIF) under NLN and FLN conditions can provide insight on the adduct distributions associated with the reaction of (+)- and (-)-anti-BPDE with three of the four sequence-defined duplex oligomers defined in secd(-GAGtion 11. They are d(-AAGGAA-).d(-TTCCTT-), GAG-).d(-CTCCTC-), and d(-CCGG-).d(-CCGG-). As shown earlier (27, 28, 36-39) (cf. Table 1))the NLN fluorescence origins of (*)-j (j = 1-3) adducts of BPDEN2-dG are respectively at -377.5-378.5, -378.5-380.0, and -380.0-382.0 nm. For the NLN fluorescence exSOexcitation waveperiments we continue to use S2 lengths of A,, = 345.0 and 355.0 nm. The former is nonselective, providing a considerable degree of excitation for all conformations, while the latter favors excitation of (&)-3adducts. NLN fluorescence results for the three duplex oligonucleotides are shown in Figure 1, with frames A and B corresponding to (+)- and (-)-anti-BPDE, respectively. In each frame the upper and lower sets of three spectra were obtained with A,, = 345.0 and 355.0 nm, respectively. In the order the duplexes are listed in the preceding paragraph their NLN spectra are the dashed, dash-dot, and solid curves in each frame. It is apparent that the NLN spectra exhibit a dependence of the adduct distribution on sequence. For example, the upper set of spectra in Figure 1A suggest that d(-AAGGAA-)-d(-TTCCTT-) yields relatively more (+)-3type adduct than do the other two duplexes. The spectra in the lower set (Aex = 355 nm) of Figure 1A also show some differences. Although A,, = 355 nm favors excitation of the (+)-3 adduct, excitation of (+)-2 and, to a lesser extent, (+)-l does occur. [The absorption bands of all adducts suffer from significant site-inhomogeneous broadening, 200 cm-l (27, 28).] Again, it appears that d(-AAGGAA-).d(-TTCCTT-)yields relatively more (+)-3 conformation than does d(-GAGGAG).d(-CTCCTC-). The broad shoulder near 382 nm in the A,, = 355 nm spectrum of d(-CCGG-)-d(-CCGG-)raises

the possibility that adenine adducts are produced from this sequence (27,28). The same is true for the reaction with (-)-anti-BPDE (see dash-dot spectrum in the lower set of Figure 1B). Comparison of the spectra in Figure 1A with those of Figure 1B clearly suggests that (-)-anti-BPDE produces more helix-internal (base-stacked, quasi-intercalated) adducts relative to helix-external adduct than does (+)-antiBPDE for all three sequences, as is also the case for d(-TGGT-).d(-ACCA-). Finally, it should be noted that the NLN spectra for d(-CCGG-).d(-CCGG-)reacted with both BPDE enantiomers are very similar to those from reactions with calf thymus DNA (27, 28,40). The above qualitative analyses are supported by the FLN spectra of the unseparated mixtures from reactions of (+I- and (-)-anti-BPDE with the four sequence-defined duplex oligonucleotides. We show here only the results for the binding of (+)-anti-BPDE with d(-AAGGAA-). and d(-TTCCTT-1, d(-GAGGAG-).d(-CTCCTC-), d(-CCGG-).d(-CCGG-) (Figure 2). The vibronically excited FLN spectra were obtained with A, = 356.9 nm, which provides excitation for the higher frequency vibrations (- 1400-1600 cm-') of the pyrenyl moiety associated with the various adduct conformations (derived predominantly from binding to N2-dG of guanine). As discussed in refs 27,28,36-39, and 42, the multiplet FLN origin structure for an adduct must occur in the region of the NLN SI SOfluorescence origin, i.e., -377.5-378.5, -378.5-380.0, and -380.0-382.0 nm for (*)-j, j = 1-3, respectively. Furthermore, because the electron-phonon coupling for (*)-3 adducts is strong, FLN is only weakly observed (27,281. It is readily observed for (f)-1and, to a lesser extent, for ( i ) - 2 type adducts (27,28, 39). The spectra in Figure 2 are normalized with respect to the 1387- and 1445-cm-' zero-phonon lines (ZPL) associated with (35) the (+)-1adduct. The broad, structureless fluorescence observed at -381 nm is due to (+)-3type N2-dG adducts. The spectra show clearly that the (+)-3 adduct has the highest relative yield for d(-AAGGAA-)-

-

-

-

102 Chem. Res. Toxicol., Vol. 7, No. 1, 1994

Marsch et al. 7

f

0

9

e

e e

d

d

Unmodlfied Oligomer

+

+ AAACCGGm mGGCCAAA

- +

- + lTAAGGAATT AA?TCCllAA

lTGAGGAGlT AACTCCTCAA

-

Figure 3. Electrophoresis of 32P-radiolabeled BPDE-modified oligodeoxynucleotides through 20 % polyacrylamide slab gels. Electrophoresis was for 12 h under typically nondenaturing conditions for larger DNAs (T 25 "C, no urea in gel). The anode is toward the bottom of the figure. Lanes 1, 4, and 7: ss oligomers with (-)-anti-BPDE. Lanes 2, 5, and 8: ss oligomers with (+)anti-BPDE. Lanes 3, 6, and 9: DNA oligomers unreacted with ligand. Bands corresponding to separated carcinogen-modified oligodeoxynucleotides are labeled from a (highest mobility) up to i (lowest mobility).

d(-TTCCTT-), in accord with the NLN results, with the other two duplexes forming smaller and similar amounts. Again, we cannot exclude the possibility that adenine adducts are contributing to emission near 382 nm for d(-CCGG-)-d(-CCGG-),since the fluorescence excitation and emissionspectra of BPDE-poly(dA-dT)*poly(dA-dT) adducts are known to be red-shifted (27,41). Concerning the presence of the (+)-2 adduct, we note that if one subtracts the lowest spectrum of Figure 2 from the middle spectrum, one obtains an FLN spectrum similar to that of the (+)-2 conformer (see middle spectrum of Figure 6). This indicates that d(-GAGGAG-)-d(-CTCCTC-)yields relatively more (+)-2adduct than d(-CCGG-)-d(-CCGG-), in basic agreement with the NLN/FLN spectra obtained for PAGE-purified adducts, vide infra. Although fluorescence spectra, obtained under NLN and FLN conditions, provide considerable insight on adduct distributions from reactions of (+)- and (-)-antiBPDE with sequence-defined duplex oligonucleotides, higher selectivity and quantitation are clearly desirable. To this end, we further explore the utility of the marriage between low-temperature LIF and PAGE. B. PAGE of BPDE-Oligonucleotide Adducts. We applied PAGE under classically "native" conditions (cf. section 11)to the product mixtures from the reactions of (+)- and (-)-anti-BPDE with each of the previously defined double-stranded (ds) and 32P-labeled oligonucleotides. Nevertheless, under these conditions the ds oligonucleotides denature on the gel into single-stranded (ss) oligonucleotides as discussed in considerable detail elsewhere (39). In this reference it was also shown that ssd(32P-?TAAGGAATT)covalently bound by BPDEs yields a distribution of BPDE-N2-dG adducts consisting of external, partially base-stacked, and base-stacked con-

formations. The presence of carcinogen base-stacking indicates that the ss oligodeoxynucleotideswere not in a random-coilconformation,but retained a defined structure during electrophoresis. By analogy with the conformations observed for duplex oligonucleotides and DNA these ( f ) - 2 , , and (&)-3,, conformations were denoted as (&)-la, respectively. Qualitatively, it appeared, on the basis of FLN and NLN fluorescenceanalysis, that the distributions of external and internal N2-dG adducts for the ds and ss oligonucleotides are very similar. Furthermore, it was observed that (39) the electrophoretic mobility increases with increasing base-stacked character of the N2-dG conformer. In what follows in this and the following subsection it will be shown that these observations hold for the new oligonucleotidesreported in here. On the basis of the analysis of all fluorescence spectra, it appears (in agreement with data presented in ref 34) that origin fluorescence bands of (f)-1,-trans and (f)-2,-trans type adducts (depending on sequence and cooling rate) shift slightly to higher energy upon duplexation while those for (f)-2,-cis and (*)-3,-cis shift slightly to lower energy. However, the shift of the (0,O) origin band for (&)-2,type adducts (upon duplexation) seems to be sequence and cooling rate dependent, and detailed characterization requires further studies. Figures 3 and 4 show that the electrophoretic patterns (electropherograms) of the four ss oligonucleotides are significantly different, as might have been expected on the basis of the results of the preceding subsection and those of ref 39. For example, (+)-anti-BPDE with ssd(32P-CCGG-)yields two major adducts (d, f), three major (a,b, d), and four major adducts with SS-~(~~P-AAGGAA-) adducts with SS-~(~~P-GAGGAG-) ( b e ) . As another example, we note that (-)-anti-BPDE forms three major

Sequence Dependence of BPDE-DNA Adducts

A

1

2

Chem. Res. Toxicol., Vol. 7, No. 1, 1994 103

B

3

2

1

3

-411

f I

h

e d

f + e

+ d + c

c +

4-tl

b+

a

Unmodified Oligomer

+

-

+

-+ AAlTGGTTAA TTAACCAATT

AAlTGGlTAA lTAACCAAlT

Figure 4. Autoradiogram showing the separation of d(S2P-AATTGGTTAA)modified by BPDE enantiomers a t low (panel A) and high (panel B) levels of covalent binding. The anode is toward the bottom of the figure. Lane 2 of panel A shows the migration of ss oligomer unreacted with BPDE, and lanes 1and 3 show d(32P-AATTGGTTAA)reacted a t low levels with (-)-anti- and (+)-antiBPDE, respectively. Lanes 2 and 3 of panel B show the binding by (-)-anti- and (+)-anti-BPDE a t high levels of covalent modification, while lane 1shows the migration of ss oligomer unreacted with BPDE. Note that the autoradiography exposure time for the separated oligomer with low levels of adduct was 50 times longer than for approximately the same amount of oligomer covalently bound a t high levels of adduct (panel B). Table 2. Quantitation of Purified Adduct Conformations and Stereoaddition Products

(+)-anti-BPDE Xfl,(O,O),d

nm

(10.2nm)

PvenYl conform.e

bandb

76 t o t a l c

a b d e

25 44 22 4

380.9 378.6 377.4 377.3

(+)-3e (+I-% (+)-le (+)-le

a b

380.0 379.0 379.4 378.5 377.4

(+)-3e (+)-2e (+)-2e

d e

3 17 34 18 15

d

22

f

51

377.5 377.4

(+)-la (+)-le

bh ch

27 24

381.0 -380.2

(+)-3e (+)-3e

C

(+I-% (+)-le

(-)-anti-BPDE stereoadditionf band TTAAGGAATP cis a cis C trans d trans e TTGAGGAGTP cis a b cis cis C trans e g trans AAACCGGTTT" trans a trans C AATTGGTTAAg cis c' k

7% total

Xfl,(O.O),

nm

(10.2nm)

pyrenyl conform.

stereoaddition

9 74 4 4

381.6 379.2 377.5 377.3

i

6 8 63 9 5

380.1 380.0 379.6 377.8 377.4

(-1-30 (-1-30 (-)-2t3 (-)-hi (-)-le

cis trans cis trans trans

20

381.5 379.5

(-1-30 (-)-2e

cis

41 76

380.6

(3-38

Cis

(-)-2e (-)-le (-)-le

i Cis

trans trans

k

a The same electropherogram patterns of covalently modified oligomers were obtained independent of modification level. b Bands from which adducts are eluted are shown in Figure 3. Percent of the total binding contributed by each adduct type. (See also captions of Figures 3 and 4.) The wavelength of the fluorescenceorigin band, (0,O)n. e The conformation of the adduct about the single-strandedDNA oligomer. (*)-ledenotes an external solvent-exposed adduct, (1)-2,denotes an external partially base-stacked adduct, and (1)-38denotes a highly base-stacked adduct. f Denotes whether the adduct was formed by a trans or cis addition of the BPDE to the guanine exocyclic amine. For the major RAGGAR adducts (or minor ones that were intercalated), we were able to make a direct cis or trans assignment. 8 Data in this part of the table are valid only for high (- 10% ) levels of BPDE modification. At low levels of modification (519%) a different distribution of adduct was obtained (see Figure 4). Bands from which adducts are eluted are shown in Figure 4B,lane 3. For band c see Figure 4B,lane 2. j Most probably adenine adduct, stereochemistry not established. Probably trans addition of the BPDE.

adducts (a-c) with d(S2P-CCGG-)but only one with either of the two d(32P-RAGGAR-)sequences. Extremely weak bands in the electropherograms (Figures 3 and 4), though not analyzed, were labeled for completeness.

Table 2 gives the percentage contribution from PAGE bands for which FLN and NLN fluorescence analyses yielded a conformational assignment. Quantitation was carried out radiographically.

Marsch et al.

104 Chem. Res. Toxicol., Vol. 7, No. 1, 1994

-

Scintillation counting of the adduct bands for all four sequences showed that 10-15 % of oligonucleotideswere bound by both BPDE enantiomers, indicating, as expected, that the levels of covalent binding are not significantly dependent on enantiomer or sequence since all four sequences are preferred for lesion formation. To end this subsection, we note that Figure 4 establishes that the distributions of adducts from d(32P-TGGT-) obtained under the typical low-dosage (panel A) and atypical high-dosage (panel B) conditions are significantly different. At high dosage (+)-anti-BPDE forms a very heterogeneous distribution of adducts as is the case for dPP-GAGGAG-) at low dosage. However, at low level (+)-anti-BPDE yields a distribution of d(32P-TGGT-) adducts almost identical to that observed for d(32P-CCGG-) with two slowly migrating ss oligonucleotide adducts. Significantly, the other three sequence-defined oligonucleotides did not exhibit a dependence of adduct distribution on dosage level (results not shown). We return to the matter of dosage level in section IIID. C. Laser-Induced Fluorescence from PAGE-Purified Oligonucleotide Adducts. We begin with the NLN fluorescence results which lead to a characterization of the N2-dGadducts separated by PAGE as external [ (&)l,)], partially base-stacked [(f)-2,)3, and base-stacked [(&)-3,)1in conformation. Next, we present FLN spectra which are used to assign the stereochemistry (cis vs trans) of the adducts associated with the four sequence-defined oligonucleotides. The results are summarized in Table 2. In this table only PAGE-purified adducts which have been studied in detail by both FLN and NLN fluorescence spectroscopies are listed. Even so, the number of listed adducts prevents us from presenting relevant fluorescence spectra for all adducts. The spectra which are presented are quite typical and serve, it is hoped, to illustrate our methods of analysis. 1. Conformational Assignments by NLN Fluorescence at 77 K. Fluorescence origin bands of six adducts isolated from gels whose electropherograms are shown in Figures 3 and 4 are shown in Figure 5 (Aex = 345.0 nm). Spectra labeled as 1and 2,3 and 4, and 5 and 6 correspond, respectively, to adducts with low, intermediate, and high electrophoretic mobility. Spectra 1-5 correspond, respectively, to the following gel lanelbands of Figure 3: 4ld; ale; 4/c; 8Ib; and 4/a. Spectrum 6 corresponds to 3Ic of Figure 4B. On the basis of our earlier work which led to Table 1 and our first paper (39) on the LIFIPAGE methodology, we can assign spectra 1,2, 3,4, and 5,6 to external, partially base-stacked, and base-stacked conformations of the (+)- or (-)-anti-BPDE adduct of N2-dG, respectively. For example, the expected position of the NLN origin band for (*)-ls, (f)-2s, and (*)-3, N2-dG conformations is -377.5, -379.0, and -381.0 nm, respectively. We note also that the fluorescence origin bandwidths for the external (*)-l,adducts are considerably narrower than those of (f)-2, and (h1-3, conformations. This is consistent with our earlier results (27, 28) and due mainly to the fact that the contribution to the bandwidth from linear electron-phonon coupling is significantly stronger for interior- than for external-type adducts (cf. the Introduction). The results of Figure 5 and corresponding results for all other PAGE-purified adducts listed in Table 2 reveal a striking correlation between electrophoretic mobility and conformation for the four oligonucleotide sequences

3+5

380

365

Wavelength (nm) Figure 5. Inhomogeneously broaded (0,O) origin fluorescence bands of PAGE-purified oligodeoxynucleotidesmodifiedby antiBPDE isomers. Top: NLN spectra of slowly-migratingcovalent BPDE-oligonucleotide complexes. Spectrum 1: a covalent complex of (-)-anti-BPDE and one of the guanines in d(32PTTAAGGAATT) (band d, lane 4, Figure 3); spectrum 2: a covalent complex of (+)-anti-BPDE and d(32P-TTGAGGAGTT) (band e, lane 8, Figure 3). Middle: NLN spectra of covalent BPDE-oligodeoxynucleotide complexes with intermediate mobilities. Spectrum 3: a covalent complex of (-)-anti-BPDE with d(32P-TTAAGGAATT) (band c, lane 4, Figure 3); spectrum 4: a complex of (+)-anti-BPDE with one of the guanines of d(a2PTTGAGGAGTT) (band b, lane 8, Figure 3). Bottom: NLN spectra of rapidly migrating covalent BPDE-oligonucleotide complexes. Spectrum 5: a complex of (-)-anti-BPDE to d(32PTTAAGGAATT) (band a, lane 4, Figure 3); spectrum 6: a complex of (+)-anti-BPDE to d(32P-AATTGGTTAA) (band c, lane 3, Figure 4B). T = 77 K, .,A = 345 nm.

studied. That is, electrophoretic mobility increases with increasing interior character of the conformation. Although such a correlation may be a good rule of thumb, it cannot be genera1izeds2Taken as a whole, these results nicely confirm that the fluorescencespectra of unseparated duplex oligonucleotide adducts shown in Figure 1correspond to complex mixtures of adduct conformations. A few more comments on the results of Table 2 are appropriate. External adducts of (+)-anti-BPDE are formed with much lower yield for the d(-RAGGAR-) sequences than for d-(CCGG-),where at least 73% of the adduct. Furthermore, total distribution is the (+)-l, d(-GAGGAG-) yields more (+)-2, adduct than does d(-AAGGAA-) and d(-(CCGG-), as deduced earlier in section IIIA. We note also that the results in Table 2 for d(-TGGT-)are for a high BPDE level of modificationsince fluorescence from eluted adducts at low dosage was very weak. Nevertheless, the results for d(-TGGT-) are consistent with the aforementioned correlation between electrophoretic mobility and adduct conformation. Other results (not shown) suggest that the two major d(-TGGT-) adducts from (+)-anti-BPDEat a low level of modification (see lane 3 of Figure 4A) have the external conformation. To conclude this part of the subsection, we report that the fluorescence quantum yields of adducts (eluted from 2

Manuscript in preparation.

Chem. Res. Toxicol., Vol. 7,No. 1, 1994 105

Sequence Dependence of BPDE-DNA Adducts N

x

376

378

380

382

h

Figure 6. FLN spectra of ss oligonucleotides modified by (+)anti-BPDE eluted from 20% polyacrylamide gels. The electropherogram depicting the separation is lane 5 of Figure 3. The ZPLs of the upper spectrum are labeled with excited-state

vibrationalfrequencies. For given ZPLs, there is a slight adductto-adduct frequency shift. Eluate from band d (solid line spectrum), eluate from band b (dashed line), and eluate from band a (dashed and dotted line). T = 4.2 K, &, = 356.9 nm. the polyacrylamide gels) with an external [(&)-la] or solvent-exposed conformation are significantly lower than those for partially base-stacked or base-stacked conformations. Detection of (A)-1 type N2-dG adducts in our previous studies (non-PAGE) had posed no difficulty. Thus, it is possible that the low fluorescence quantum adducts are due to quenching by acrylyields for (&)-18 amide monomer impurity in the gel. However, it was also observed that all adducts from d(-CCGG-)and d(-TGGT-) yielded relatively weak fluorescencesignals. This suggests that future work on the dependence of photophysics of BPDE-N2-dGadducts on at least triplet base composition is warranted. 2. FLN Assignments of Adduct Stereochemistry. Here we take advantage, for example, of the ability of FLN to clearly distinguish between cis and trans isomers of BPDE-N2-dG adducts on the basis of excited-state vibrational frequencies and intensities of the pyrenyl moiety (36,381.First, however, we present, for completeness, some FLN results in Figure 6 which help make a transition with the results of section IIIC1. All three FLN spectra correspond to the major PAGE-isolated, (+)-antiBPDE adducts of d(-AAGGA-). From bottom to top the spectra (Aex = 356.9 nm) correspond to bands a, b, and d of lane 5 of Figure 3 which have respectively high, intermediate, and low electrophoretic mobility. The dominant vibronic ZPL structure between 1445 and 1605 cm-l [with a centroid slightly to higher energy of 378 nm (3911 in the topmost spectrum establishes that the adduct In comparing this spectrum with of gel band d is (+)-la. the bottom two, one sees that the ZPL structures in the latter are significantly diminished in intensity and that this is accompanied by the appearance of intense and broad lower energy fluorescence which, in the bottom spectrum, is centered slightly to higher energy of 381 nm. The broad underlying fluorescencein the middle spectrum is centered at -379 nm. For reasons discussed earlier, the middle and lowest spectra are assigned to (+)-2, and (+)-3, adducts of N2-dG. As the first example of how we assign stereochemistry, we present Figure 7, whose topmost and lowest FLN spectra are from a (+)-cis-BPDE adduct (HPLC-purified)

=356.9nm

c/

Wavelength (nm)

376

378

380

Wavelength (nm) Figure 7. Characterizationby FLNS of ss highly base-stacked (+)-cis-BPDEadducts purified by HPLC or PAGE. The upper and lower spectra are of (+)-cis-BPDE-N2-dG standardspurified by HPLC. The pyrenyl is bound to the central G of d(CCATCG*CTACC) and d(CACATG*TACAC),respectively. Note that both oligomers have a single BPDE adduct at the starred residue. For comparison, the middle spectrum is of the eluate from band a, lane 5 (Figure 3). ZPLs of the top spectrum are labeled with excited-statevibrationalfrequencies. T = 4.2 K, &= = 356.9 nm.

for ss-d(CCATCG*CTACC) and ss-d(CACATG*ACAC) (42). Here G* indicates N2-dG adduct. Both adducts show weak ZPL structure which is dominated by broad fluorescence near 381 nm. (Upon duplexation of oligonucleotides, both adducts show a further red-shifting of the centroids of fluorescence as well as stronger electronphonon coupling.) The middle spectrum of Figure 7 is for eluate a from lane 5 of Figure 3 which was assigned in the preceding paragraph as (+)-3,. This spectrum is very similar to the other two, and thus an assignment of the (+)-3, adduct as cis is made. However, for unequivocal assignment of adducts as cis or trans it was necessary to use ,A, = 369.6 nm which exposed, in the FLN spectrum, the lower frequency pyrenyl excited-state vibrational mode structure that is distinctly different for the cis and trans isomers of N2-dG adducts (38,42). An example is shown in Figure 8. For consideration of the FLN spectra we note that for Aex = 369.6 nm the spectrum of the trans isomer is dominated by a mode at 579 cm-l with weaker modes at 468 and 758 cm-l. The spectrum for the cis isomer is richer and notably different. This can be seen again by comparing the topmost spectra of frames A and B of Figure 8 for HPLC-purified (+)-cis and (+)-trans N2-dG adduct of d(CACATG*TACAC)(42). Aside from differences in the vibronic intensity distribution (e.g., the 614-cm-l mode for cis is much stronger), there are significant trans to cis 465 cm-l). The frequency shifts (758 742 nd 468 bottom spectrum of frame B is our previously published spectrum of the (+)-trans N2-dG adduct of calf thymus DNA. These spectrashould be viewed asstandard spectra against which the lowest two spectra of frame A and middle spectrum of frame B are to be compared. These last three spectra are of PAGE-purified adducts corresponding to lane 8lband b, lane 7lband c, and lane 8lband e, respectively, of Figure 3. Results other than those in Figure 8 led us to assign these three adducts as (+)-2,, (-)-2B, and

-

-

Marsch et al.

106 Chem. Res. Toxicol., Vol. 7, No. 1, 1994

(+>2 s cis-BPDE

n

's

(+)-1 s m - B P D E in T&

... ...

band c lane I

I

WAVELENGTH (nm) Figure 8. Representative FLN spectra of PAGE-purified ss (k)-anti-BPDEoligodeoxynucleotidecomplexes. Displayed in panels A and B are various cis- and trans-BPDE adducts, respectively. The top spectra of panels A and B are of HPLC-purified (+)-cis and (+)-trans standards of d(CACATG*TACAC),respectively. The middle and lower spectra of panel A are of PAGE-purified (+)-cis and (-)-cis adducts of d(S2P-TTGAGGAGTT),correspondingto the eluates from band b of lane 8 (Figure 3) and from band c of lane 7 (Figure 3). The middle spectrum of panel B is of a PAGE-purified(+)-transadduct, which is the eluate from band e of lane 8 (Figure

3). The binding of (+)-anti-BPDEto calf thymus DNA at low (+)-anti-BPDE/bpDNA ratios results in predominantly tram external [(+)-I type] adducts (lower spectrum of panel B). Bands are labeled with the excited-state vibrational frequencies. T = 4.2 K,A,. = 369.6 nm.

(+)-lB,respectively. The results of Figure 8 firmly establish that their stereochemistries are cis, cis, and trans. However, with reference to Table 2 it should be pointed out that the (*)-lB type adducts are always in trans stereoaddition,while the (f)-2,and (*)-3, adducts (which have the strong electron-phonon coupling) are all assigned as cis isomers, and there is one (+)-2, and one (-)-3, adduct from d(-GAGGAG-) that is assigned as trans. D. Dependence of Adduct Distribution on Level of Modification. As discussed in section IIIB, no dependence of the adduct distribution on the BPDE dosage level (level of modification) was observed for d(-AAGGAA-)*d(-TTCCTT-), d(-GAGGAG-).d(-CTCCTC-), and d(-CCGG-)-d(-CCGG-). On the other hand, the d(-TGGT-).d(-ACCA-) duplex exhibited a quite remarkable and reproducible dependence for (+)-anti-BPDEand, to a much lesser extent, for (-)-anti-BPDE (Figure 4). For (+)-anti-BPDE it appears that at the high level of modification much more base-stacked [(+)-3,1 adduct is produced (see Figure 4 and Table 2). Why only this duplex of the four studied exhibits a dependence is an interesting question for which we have no data that provide an answer. In any event, the results raised, in our minds, the question of whether or not random sequence, highly polymerized DNA would show a comparable dependence. Our experiments with (-)-anti-BPDE and calf thymus DNA at the BPDE/bp DNA solution concentrations given in the caption to Figure 9 revealed no dependence. However, this is not the case for (+)-anti-BPDE as the spectra in Figure 9 show. The FLN spectra (kex= 356.9 nm) from top to bottom are for an initial BPDE/bp DNA concentration ratio ( r ) of 5, 0.2, and 0.008, respectively. The lowest spectrum (r = 0.008) is due prominantly to the

4 2,"-hex = 356.9nm

t

376

378

380

382

Wavelength (nm) Figure 9. FLN spectra of (+)-anti-BPDE-modifiedcalf thymus DNA at three doses of carcinogen added to DNA. Top spectrum: (+)-anti-BPDEbinding to DNA at initial solution ratios of 5 BPDE/bp DNA; middle spectrum: 0.2 BPDE/bp DNA and lower spectrum: 0.008 BPDE/bp DNA. T = 4.2 K,A,, = 356.9 nm. (+)-l BPDE-N2-dG adduct and is identical to our previously published (27,28) spectra of (+)-anti-BPDEbound to calf thymus DNA. Comparison of this spectrum with the middle ( r = 0.2) and top ( r = 5) spectrum clearly

Chem. Res. Toxicol., Vol. 7, No. 1, 1994 107

Sequence Dependence of BPDE-DNA Adducts

hex =356.9 nm

hex=369.6 nm 3%

3j8

380

382

Wavelength (nm) Figure 10. FLN spectra of (+)-anti-BPDE covalently bound to at a high unseparated d(AATTGGTTAA).d(TTAACCAATT) level of covalent modification (0.2 BPDE/bp DNA, similar to conditions that generated the middle spectrum of Figure 9). Zerophonon lines are labeled with the excited-state vibrational frequencies. T = 4.2 K, A,, = 356.9 nm (top spectrum) and A,, = 369.6 nm (lower).

indicates that the relative amount of quasi-intercalated [(+)-31 BPDE-N2-dG increases with increasing BPDE dosage. Whether this is directly related to the result above for the -TGGT- containing oligonucleotide is unclear. It is interesting that the -CCGG-containing sequence, which also is highly susceptible to lesion formation, does not exhibit a dependence. As a final illustration of the selectivity of FLN we show, in Figure 10, FLN spectra obtained from the adduct mixture from the reaction of (+)-anti-BPDE with d(-TGGT-).d(-ACCA-),resulting in a high level of modification. The top spectrum, obtained with A,, = 356.9 nm, is consistent with the results of Table 2, which shows, for ss-d(-TGGT-), that a high percentage of (+)-3 type N2-dGadduct is formed. The broad, intense fluorescence in the spectrum near 381 nm, together with the relatively weak ZPLs, is due to (+)-3 (the top spectrum of Figure 10 can be compared with those of Figure 7). For discussion of the lower spectrum of Figure 10 our discussion above of Figure 8 (last paragraph of section IIIC2) is useful. The presence of the 742-cm-1mode and considerable intensity of the 614-cm-1 mode in the lower spectrum of Figure 10 indicate the presence of (+)-cis-isomer,which is consistent with the presence of the (+)-3 conformer of BPDE-N2dG. However, the (+)-cis isomer could also be contributed to by the (+)-2 conformation. One last point is that the presence of the mode at 760 cm-1 indicates the presence also of (+)-trans isomer (see Figure 8) which is most likely due to the (+)-l conformation.

IV. Conclusions Laser-induced fluorescence (LIF) under line-narrowing (FLN, 4.2 K) and non-line-narrowing (77 K) conditions was applied to unseparated and PAGE-separated adducts from the reactions of (+I- and (-)-anti-BPDE with four

sequence-defined duplex oligonucleotides formed from d(AAACCGGTTT),d (TTAAGGAATT),d(TTGAGGAGTT), and d(AATTGGTTAA1 and the respective complements. They were selected because the central 5’RAGGAR-3’ (R purine) sequence appears to be frequently mutated and the 5’-CCGG-3’ and 5’-TGGT-3’ sequences are strongly preferred for covalent binding but less prone to lead to mutation (14,151. The results confirm (39)that our LIF techniques, in combination with PAGE, are a powerful approach for high-resolution analysis of the adduct conformation distribution. However, LIF (under FLN and NLN conditions) by itself was shown to provide considerable insight on the contribution from external, base-stacked, and quasi-intercalated conformations [(&)-j,j = 1-31 of the BPDE-N2-dG adduct to the unseparated product mixture as might have been expected on the basis of our earlier work. But utilization of PAGE allows for quantitation of the adduct conformations. Some of the more important findings are as follows: (i) under normal PAGE conditions (-25 OC, no urea) all four duplex (ds) oligonucleotidesdenature into single-stranded (ss) oligonucleotides; (ii) the complex electropherograms and the BPDE base-stacking revealed by FLN and NLN analysis of the adducts eluted from the gel suggest that the four adducted ss oligonucleotides must retain an ordered, rather than random-coil, configuration in the gel; (iii) similar to the situation for (+)- and (-)-anti-BPDE N2-dG adducts of ds oligonucleotides and highly polymerized random sequence DNA, the N2-dG adducts of the ss oligonucleotides can assume external [(&)-la],partially base-stacked [(*)-2J, and base-stacked [(*)-3,1 conformations (see also ref 34); (iv) there is a striking similarity electronic structure as well between the excited-state (SI) as the FLN spectra of the duplex (&)-j and ss (A)-ja adducts, j = 1-3; (v) for all four ss oligonucleotides the electrophoretic mobility depended strongly on adduct conformation type, increasing in the order external, partially base-stacked, base-sta~ked;~ (vi) for both (+)and (-)-anti-BPDE the distribution of N2-dG adduct conformations (as well as electropherograms) depends strongly on sequence; and (vii) LIF/PAGE allows for stereochemical assignment of the BPDE-N2-dG adduct conformations. We follow up on (vi) and (vii) with additional results. It was observed that, at a low level of binding for both BPDE enantiomers, d(-CCGG-).d(-CCGG-) and d(-TGGT-).d(-ACCA-) produce a distribution of adduct conformations similar to those created by the binding of the enantiomers to highly polymerized DNA in vitro (27,28). The binding of (-)-anti-BPDE produces more (-)-2 and less (-)-3 adduct to oligomers with central -RAGGARsequences, whereas the carcinogenic (+)-isomer forms much less (+)-l-trans adduct to the d(TTRAGGARTT). d(AAYTCCTYAA) molecules than to oligonucleotides with the -CCGG- or -TTGGTT- central sequence motifs. The enhancement of cis base-stacked (+)-anti-BPDE adducts in these sequences, which may be pertinent to mutagenesis (10-15), raises the possibility that these complexes, and not the (+)-l-trans adducts, may be important in chemical carcinogenesis. The translcis product ratios resulting from (+)-anti-BPDE (reacted with duplex oligomers) binding to -AAGGAA- [(+)-trans/(+)0.61 are cis 0.41 and -GAGGAG- [(+)-trans/(+)&

-

3

-

A n exception t o t h i s rule will be published separately.

108 Chem. Res. Toxicol., Vol. 7, No. I , 1994

remarkably low. Clearlythis ratio is sensitiveto the nearest neighbor bases of the adducted G and, quite possibly, the bases beyond that. We note that Geacintov et al. (34) determined (+)-trans/(+)& ratios of >9 and 6 for poly(dG-dC)-poly(dG-dC)and poly(dG).poly(dC),respectively, and very recently Margulis et al. (35) a ratio of 3-5 for single-stranded oligonucleotides. Thus, the relatively high efficiencyof (+)-cis adduct formation to the 5’-RAGGAR3’ sequences suggests that in ds nucleic acids the base sequence effects may be more important in determining the levels of (+)-cis-BPDE adduct formation. Since our oligonucleotides were synthesized to contain central highly-preferred sites of covalent binding by antiBPDEs, it is predicted that the majority of binding will be to the 5’-guanine of the central GG doublet (18-23,43). We do not contend that all separated and purified adducts result from binding to the 5’-G of the doublet. Binding to other guanines or to adenines (minor adducts) may well be responsible for the complexity of some of the electropherograms. It should be noted that previous work has shown that some of the BPDE-induced mutations which occur at 5’-AGG-3’ sites in polypurine tracts are not at the 5’-G. Replication errors that occur at this sequence could be due to BPDE lesions on adenines and guanines flanking the 5’-G. An objection to the notion that interior cis N2-dG adducts are important in BPDE-induced mutagenesis might be based on the fact that (-)-anti-BPDE is significantly less mutagenic than (+)-anti-BPDE but with random sequence DNA (-)-anti-BPDE produces much more helix-interior [(-)-2 and (-1-33 N2-dG adduct than (+)-anti-BPDE. This objection may not be valid. For example, Geactinov and co-workers (33)showed that (-1cis adduct spontaneously decomposed to BPT in standard buffered solution within an hour at ambient temperature, regenerating undamaged DNA. They found that (+)-cisBPDE is quite stable, as are the trans adducts for both enantiomers. Perhaps the cell need not repair the (-)-cis adduct; the repair might occur spontaneously if (-)-cisBPDE adduct is unstable in cellular environments. With the (+)-cis adduct persistent, the cellular repair system may falter in its repair, resulting in base transversion and permanent biological insult. Relevant are our very recently obtained results on DNA adducts from benzo[alpyrene mouse skin experiments, in which both trans- and cisBPDE adducts (at rather low translcis ratio) were founde4

Acknowledgment. We are grateful to Prof. Nicholas Geacintov for the kind gift of the HPLC-purified d(CACATG*TACAC) modified by anti-BPDE isomers, to Mr. John Farhat for his assistance at the early stages of these experiments, and to Dr. F. Ariese for useful discussions. G.A.M. is grateful for an appointment to the Alexander Hollaender Distinguished Postdoctoral Fellowship Program sponsored by the USDOE, Office of Health and Environmental Research, and administered by Oak Ridge Associated Universities. Ames Laboratory is operated for the US. Department of Energy by Iowa State University under Contract W-7405-Eng-82. This work was supported by the Office of Health and Environmental Research, Office of Energy Research. References (1) Conney, A. H. (1982) Induction of microsomal enzymes by foreign

chemicals and carcinogenesis by polycyclic aromatic compounds. Cancer Res. 42, 4875-4917. 4

Unpublished data.

Marsch et al. Griislund, A., and Jernstrom, B. (1989) DNA-carcinogen interactions: covalent DNA adducts of benzo[alpyrene 7,8 dihydrodiol 9,lO epoxides studied by biochemical and biophysical techniques. Q. Rev. Biophys. 22, 1-37. Harvey, R. G. (1981) Activated metabolites of carcinogenic compounds. Acc. Chem. Res. 14, 218-226. Phillips, D. H. (1983) Fifty years of benzo[alpyrene. Nature 303, 468-472. Cavalieri, E., and Rogan, E. (1985) Role of radical cations in aromatic hydrocarbon carcinogenesis. Enuiron. Health Perspect. 64,69-84. Rogan, E. G., Cavalieri, E., Tibbels, S., Cremonesi, P., Warner, C., Nagel, D., Tomer, K., Cerny, R., and Gross, M. (1988) Synthesis and identification of benzo[alpyrene-guanine nucleoside adducta formed by electrochemical oxidation and by horseradish peroxidase-catalyzed reaction of benzo[a]pyrene with DNA. J. Am. Chem. SOC. 110,4023-4029. Zamzow, D., Jankowiak, R., Cooper, R. S., Small, G. J., Tibbels, S. R., Cremonesi, P., Devanesan, P., Rogan, E. G., and Cavalieri, E. L. (1989) Fluorescence line narrowing spectrometric analysis of benzo[alpyrene-DNA adducts formed by one-electron oxidation. Chem. Res. Toxicol. 2, 29-34. RamaKrishna, N. V. S., Devanesan, P. D., Rogan, E. G., Cavalieri, E. L., Jeong, H., Jankowiak, R., and Small, G. J. (1992) Mechanism of metabolic activation of the potent carcinogen 7,12-dimethylbenz[alanthracene. Chem. Res. Toxicol. 5, 220-226. RamaKrishna, N. V. S., Cavalieri, E. L., Rogan, E. G., Dolnikowski, G., Cerny, R. L., Gross, M. L., Jeong, H., Jankowiak, R., and Small, G. J. (1992) Synthesis and structure determination of the adducts of the potent carcinogen 7,12-dimethylbenz[alanthraceneand deoxyribonucleosides formed by electrochemical oxidation: Models for metabolic activation by one-electron oxidation. J. Am. Chem. SOC. 114, 1863-1874. Dixon, K., Roilides, E., Hauser, J., and Levine, A. J. (1989) Studies on direct and indirect effects of DNA damage on mutagenesis in monkey cells using an SV40-based shuttle vector. Mutat. Res. 220, 73-82. Maher, V. M., Yang, J.-L., Mah, M. C.-M., and McCormick, J. J. (1989) Comparing the frequency and spectra of mutations induced when an SV-40 based shuttle vector containing covalently bound residues of structurally-related carcinogens replicates in human cells. Mutat. Res. 220, 83-92. Yang, J.-L., Maher, V. M., and McCormick, J. J. (1987) Kinds of mutations formed when a shuttle vector containing adducts of (*)7~,8a-dihydroxy-9a,lOa-epoxy-7,8,9,lO-tetrahydrobenzo[alpyrene replicates in human cells. Proc. Natl. Acad. Sci. U.S.A. 84, 3787-3791. Maher, V. M., Yang, J.-L., and McCormick, J. J. (1987) Kinds of mutations formed when a shuttle vector containingadductsof benzo[a]pyrene-7,8-dio1-9,10-epoxide replicates in COS7 cells. Mol. Cell Biol. 7, 1267-1270. Mazur, M., and Glickman, B. W. (1988) Sequence specificity of mutations induced by benzo[a]pyrene-7,8-diol-9,lO-epoxide a t endogenous aprt gene in CHO cells. Somat. Cell Mol. Genet. 14,393400. Carothers, A. M., and Grunberger, D. (1990) DNA base changes in benzo[alpyrene diol epoxide-induced dihydrofolate reductase mutants of Chinese hamster ovary cells. Carcinogenesis l l , 189-192. Eisenstadt, E., Warren, A., Porter, J., Atkins, D., and Miller, J. H. (1982) Carcinogenic epoxides of benzo[alpyrene and cyclopenta[cdlpyrene induce base substitutionsviaspecific transversions. Proc. Natl. Acad. Sci. U.S.A. 79, 1945-1949. Mackay, W., Benasutti, M., Drouin, E., and Loechler, E. L. (1993) Mutagenesis by (+)-anti-B[alP-NZ-Gua, the major adduct of activated benzo[alpyrene, when studied in anEscherichia coli plasmid using site-directed methods. Carcinogenesis 14, 1415-1425. Rill,R. L., and Marsch, G. A. (1990)Sequence preferences of covalent DNA binding by anti-(+)-and anti-(-)-benzo[alpyrene diolepoxide. Biochemistry 29, 6050-6058. Lobanenkov, V. V., Plumb, M., Goodwin, G. H., and Grover, P. L. (1986) The effect of neighboring bases on G-specific DNA cleavage mediated by treatment with the anti-diol epoxide of benzo[alpyrene in vivo. Carcinogenesis 7. 1689-1695. Boles, T. C., andHogan, M. E. (1986) High-resolution mapping of carcinogen binding sites on DNA. Biochemistry 26, 3039-3043. (21) Dittrich, K. A., and Krugh, T. R. (1991) Analysis of site-specific binding of (~)-anti-benzo[alpyrenediol epoxide to restriction fragments of pBR322 DNAvia photochemical mapping. Chem. Res. Toxicol. 4, 270-276. (22) Dittrich, K. A., and Krugh, T. R. (1991) Mapping of (*)-anti-benzo[alpyrene diol epoxide adducts to human c-Ha-rasl protooncogene. Chem. Res. Toxicol. 4, 277-281. (23) Reardon, D. B., Bigger, A. H., Strandberg, J., Yagi, H., Jerina, D. M., and Dipple, A. (1989) Sequence selectivity in the reaction of optically active hydrocarbon dihydrodiol epoxides with rat H-ras DNA. Chem. Res. Toxicol. 2, 12-14.

Chem. Res. Toxicol., Vol. 7, No. 1,1994 109

Sequence Dependence of BPDE-DNA Adducts (24)Carothers, A. M., Urlaub, G., Jitka, M., Harvey, R. G., Chasin, L. A., and Grunberger, D. (1990)Splicingmutations in the CHO DHFR gene preferentially induced by (+3a,4@-dihydroxy-la,2a-epoxy1,2,3,4-tetrahydrobeno[c]phenanthrene.Proc. Natl. Acad. Sci. U.S.A. 87,5464-5468. (25) Bigger, C. A. H., Flickinger, D. J., Strandberg, J., Pataki, J., Harvey, R. G., and Dipple, A. (1990)Mutational specificity of the anti 1,2dihydrodiol 3,4-epoxide of 5-methylchrysene. Carcinogenesis 11,

2263-2265. (26)Bernelot-Moens, C., Glickman, B., and Gordon, A. J. E. (1990) Induction of specific frameshift and base substitution events by benzo[a]pyrene diol epoxide in excision-repair-deficientEscherichia coli. Carcinogenesis 11, 781-785. (27)Lu, P., Jeong, H., Jankowiak, R., Small, G. J., Kim, S. K., Cosman, M., and Geacintov, N. E. (1991)Comparative laser spectroscopic study of DNA and polynucleotide adducts from the (+)-anti-diol epoxide of benzo[a]pyrene. Chem. Res. Toxicol. 4, 58-67. (28)Jankowiak, R., Lu, P., Small, G. J., and Geacintov, N. E. (1990) Laser spectroscopic studies of DNA adduct structure types from enantiomeric diol epoxides of benzo[alpyrene. Chem. Res. Toxicol. 3, 39-46. (29)Geacintov, N. E., Ibanez, V., Gagliano, A. G., Jacobs, S. A,, and Harvey, R. G. (1984)Stereoselective covalent binding of anti-benzo[alpyrene diol epoxide to DNA. Conformation of enantiomer adducts. J. Biomol. Struct. Dyn. 1, 1473-1484. (30) Geacintov, N. E.,Yoshida, H., Ibanez, V., Jacobs, S. A., and Harvey, R. G. (1984)Conformations of adducts and kinetics of binding to DNA of the optically pure enantiomers of anti-bepzo[alpyrene diol epoxide. Biochem. Biophys. Res. Commun. 122,33-39. (31)Jernstrom, B., Lycksell, P.-O., GriLslund, A., and NordBn, B. (1984) Spectroscopic studies of DNA complexes formed after reaction with anti-benzo[a]pyrene-7,8-dihydrodio1-9,lO-oxide enantiomers of different carcinogenic potency. Carcinogenesis 5, 1129-1135. (32) Cheng, S. C., Hilton, B. D., Roman, J. M., and Dipple, A. (1989) DNA adducta from carcinogenic and noncarcinogenic enantiomers of benzo[a]pyrene dihydrodiol epoxide. Chem. Res. Toxicol. 2,334-

340. (33) Cosman, M., Ibanez, V., Geacintov, N. E., and Harvey, R. G. (1990) Preparation and isolation of adducts in high yield derived from the binding of two benzo[a]pyrene-7,8-dihydroxy-9,10-oxide stereoisomers to the oligonucleotide d(ATATGTATA). Carcinogenesis 11,

1667-1672.

(34) Geacintov, N. E., Cosman, M., Mao, B., Alfano, A., Ibanez, V., and Harvey, R. G. (1991)Spectroscopic characteristics and site I/site I1 classification of cis and tram benzo[alpyrene diolepoxide enantiomer-guanosine adducts in oligonucleotides and polynucleotides. Carcinogenesis 12,2099-2108. (35)Margulii, L. A., Ibanez, V., and Geacintov,N. E. (1993)Base-sequence dependence of covalent binding of benzo[alpyrene diol epoxide to guanine on oligodeoxynucleotides. Chem. Res. Toxicol. 6,59-63. (36) Jankowiak, R., and Small, G. J. (1991)Fluorescence line narrowing: a high-resolution window on DNA and protein damage from chemical carcinogens. Chem. Res. Tolcicol. 4, 256-269. (37)Jankowiak, R., and Small, G. J. (1989)Fluorescence line-narrowing spectroscopy in the study of chemical carcinogenesis. Anal. Chem.

61,1023A-1034A. (38) Small, G. J., Jankowiak, R., Jeong, H., and Marsch, G. A. (1993) Fluorescence line narrowing: high resolution window on DNA damage from BP and DMBA. In Polycyclic Aromat. Compd. Synthesis, Properties, Analytical Measurements, Occurrence and Biological Effects (Garrigues, P., and Lamotte, M., Eds.) pp 741748,Gordon and Breach Science Publishers. (39) Marsch, G. A., Jankowiak, R., Farhat, J. H., and Small, G. J. (1992) Separation and identification of DNA-carcinogen adduct conformers by polyacrylamidegel electrophoresiswith laser-induced fluorescence detection. Anal. Chem. 64,3038-3044. (40) Kim, S.-K., Brenner, H. C., Soh, B. J., and Geacintov, N. E. (1989) Photochem. Photobiol. 50,327-337. (41)Kim, S.-K., Geacintov, N. E., Zinger, D., and Sutherland, J. (1988) inSynchrotronRadiation in Structural Biology, BasicLife Sciences (Sweet, R. M., and Woodhead, A. D., Eds.) Vol. 51, pp 187-205, Plenum Press, New York. (42)Jeong, H. (1991)High resolution fluorescence studies of metabolic pathways of DNA and oligonucleotide adducts from polycyclic aromatic compounds, Ph.D. Thesis, Iowa State University. (43) Marsch, G. A. (1990)Sequence preference motifs of covalent DNA binding by intercalating drugs and carcinogens, Ph.D. Thesis, The Florida State University. (44) Osborne, M. R. (1990) Sequence specificity in the reaction of benzopyrene diol epoxide with DNA. Chem.-Biol.Interact. 75,131-

140.