Formation of Inter-and Intrastrand Imine Type DNA− DNA Cross-Links

University of Texas Medical Branch, Galveston, Texas 77555, Department of Chemistry and ... 3181 SW Sam Jackson Park Road, Portland, Oregon 97239-3098...
0 downloads 0 Views 4MB Size
Download PDF
Chem. Res. Toxicol. 2005, 18, 1683-1690

1683

Formation of Inter- and Intrastrand Imine Type DNA-DNA Cross-Links through Secondary Reactions of Aldehydic Adducts Ana M. Sanchez,† Ivan D. Kozekov,‡ Thomas M. Harris,‡ and R. Stephen Lloyd*,†,§ Sealy Center for Structural Biology and Department of Human Biological Chemistry and Genetics, University of Texas Medical Branch, Galveston, Texas 77555, Department of Chemistry and Center in Molecular Toxicology, Vanderbilt University, Nashville, Tennessee 37235, and Center for Research on Occupational and Environmental Toxicology, Oregon Health & Science University, 3181 SW Sam Jackson Park Road, Portland, Oregon 97239-3098 Received February 24, 2005

Acrolein-derived DNA adducts of guanine have previously been detected in tissues of several species, including humans, and have been shown to be mutagenic in mammalian cells and potentially carcinogenic in higher organisms. In duplex DNA, the predominant acrolein-derived lesion, γ-hydroxy-1,N2-propanodeoxyguanosine (γ-HOPdG), exists in an equilibrated mixture of ring-opened and ring-closed forms. We have previously shown that the acyclic form can undergo secondary chemical reactions to form both interstrand DNA-DNA cross-links in a CpG sequence context and DNA-protein and DNA-peptide cross-links. Investigations using duplex DNAs containing γ-HOPdG in a CpG sequence context reversibly created at least two cross-linked species: an imine, which is a minor species but could be readily reduced by NaBH4, and a major heat labile, nonreducible species that we formulate as a carbinolamine cross-link. The DNA came to equilibrium over several days with the carbinolamine species being significantly more abundant than the reducible imine. In an effort to find other types of DNADNA cross-links, we have developed a high throughput screen to evaluate the effects of DNA sequence and lesion structure on the formation of reducible interstrand and intrastrand crosslinks. These data reveal that four different lesions, two involving deoxyguanosine and two involving deoxyadenosine, can react with nearby bases to form inter- and intrastrand DNA cross-links.

Introduction Numerous DNA lesions arise as a consequence of reactions of bifunctional electrophilic compounds with DNA bases. In particular, adducts are formed with reactive R,β-unsaturated aldehydic compounds, such as acrolein, crotonaldehyde, and 4-hydroxy-2-nonenal (15). Acrolein and crotonaldehyde are widely distributed in the environment, but they, along with 4-hydroxy-2nonenal, also arise as products of cellular lipid peroxidation (6, 7). Acrolein reacts with deoxyguanosine to form adducts through a pair of regioisomeric Michael additions followed by ring closure with initial bond formation occurring at either the N2 or the N1 position to give the pyrimidopurinone type cyclic adducts: γ-hydroxy-1,N2propanodeoxyguanosine (γ-HOPdG, Figure 1) and the isomeric R-hydroxy-1,N2-propanodeoxyguanosine (RHOPdG) (1). The well-known human carcinogen vinyl chloride is metabolized to 2-chlorooxirane, which reacts with deoxyguanosine to form, among other products, the analogous two-carbon cyclic product, β-hydroxy-1,N2ethanodeoxyguanosine (β-HOEdG, Figure 1) (8-10). The * To whom correspondence should be addressed. Tel: 503-494-9957. E-mail: [email protected]. † University of Texas Medical Branch. ‡ Vanderbilt University. § Oregon Health & Science University.

Figure 1. Structures of the modified DNA bases. (A) γ-HOPdG, (B) β-HOEdG, (C) γ-HOPdA, and (D) β-HOEdA.

cross-linking reactions of γ-HOPdG and β-HOEdG and also the analogous adenine adducts, γ-hydroxy-1,N6propanodeoxyadenosine (γ-HOPdA) and β-hydroxy-1,N6ethanodeoxyadenosine (β-HOEdA, Figure 1), are examined in this paper. The formation of γ-HOPdA by the reaction of acrolein with deoxyadenosine has been reported (11), although the structure of the adduct has been questioned (12). Previous studies have demonstrated that the γ-HOPdG lesion exists primarily in a ring-opened form in duplex DNA (Figure 2) (13, 14). This observation suggested that the aldehydic group might be capable of undergoing further reactions with primary amines in close proximity.

10.1021/tx0500528 CCC: $30.25 © 2005 American Chemical Society Published on Web 10/19/2005

1684

Chem. Res. Toxicol., Vol. 18, No. 11, 2005

Sanchez et al.

light or chemical treatments resulting in cell death or mutagenesis (18, 19). Given the expanding literature on the ability of secondary reactions of enal adducts to form DNA-DNA cross-links, we have designed a new strategy for detection of both inter- and intrastrand DNA cross-links, focusing in particular on the reducible imine type cross-links, which have not been characterized spectroscopically. The versatility of this technique permits any sequence context to be examined in order to detect the full spectrum of possible cross-links between the aldehydic lesion and the exocyclic amino groups of nearby bases. This technique has the flexibility to examine other types of DNA adducts for their possible role in intra- or interstrand DNA crosslinking. In this investigation, we examine the ability of DNAs containing γ-HOPdG, β-HOEdG, γ-HOPdA, and β-HOEdA to form cross-linked DNA structures.

Materials and Methods

Figure 2. Proposed pathway for formation of cross-linked species with γ-HOPdG. In duplex DNA, the lesion exists as an equilibrating mixture of the ring-closed and ring-opened forms. Prior investigations have demonstrated that γ-HOPdG, when in a CpG sequence context, is able to form an interstrand crosslink (14, 16).

Consistent with this prediction, oligodeoxynucleotides containing the γ-HOPdG lesion in a Cp*G sequence context, where *G ) γ-HOPdG, have been shown to form a reversible DNA-DNA cross-link (Figure 2) (15, 16). This cross-link involves the free aldehyde in the ringopened form of the adduct reacting with the exocyclic amino group of the deoxyguanosine in the opposing strand. We have concluded that there is a mixture of cross-linked species in the duplex including imine and carbinolamine structures (16). The interchain carbinolamine intermediate has been characterized by NMR, but the concentration of imine was below the level of detection (14). No evidence for interstrand cross-linking was found when the same assays were performed in a *GpC sequence context (16). Kawanishi et al. have described experiments indicating formation of both interstrand and intrastrand cross-links, presumably involving guanines, when gapped duplexes were treated with acrolein (17). A limitation of that investigation that has been addressed in the current study is that it was uncertain what lesion was the source of the cross-links. Collectively, these data are highly significant since DNA-DNA cross-links constitute one of the most serious DNA lesions. In fact, Kawanishi et al. discovered that the otherwise rare tandem point mutations were a significant portion (12%) of the acrolein-generated mutations (17). There are strong literature precedents for the biological importance of DNA cross-links with both the intra- and the interstrand cross-links formed from exposure to ultraviolet

Materials. Sodium borohydride and sodium cyanoborohydride were purchased from Sigma Chemical Co. (St. Louis, MO). [γ-32P]ATP (3000 Ci/mmol) was purchased from Dupont-NEN Life Science Products Inc. (Boston, MA). Synthetic oligodeoxynucleotides containing the adducted adenine and guanine bases were prepared as described previously (12). Figure 1 shows the structure of the adducts used in this work: A, 3-(2-deoxy-β-Derythro-pentofuranosyl)-5,6,7,8-tetrahydro-8-hydroxypyrimido[1,2-a]purine (γ-HOPdG); B, 3-(2-deoxy-β-D-erythro-pentofuranosyl)-3,5,6,7-tetrahydro-7-hydroxy-9H-imidazo[2,1-a]purin-9-one (β-HOEdG); C, 3-(2-deoxy-β-D-erythro-pentofuranosyl)-3,7,8,9tetrahydro-7-hydroxypyrimido[2,1-i]purine (γ-HOPdA); and D, 3-(2-deoxy-β-D-erythro-pentofuranosyl)-7,8-dihydro-7-hydroxy3H-imidazo[2,1-i]purine (β-HOEtdA). Oligodeoxynucleotides with the following sequences were prepared containing the adducted guanine nucleosides: d(5′-GCTAGC*GAGTCC-3′) and d(5′*GGTGATCGCGCTAGG-3′), where *G is the adducted base. Oligodeoxynucleotides with the following sequences were prepared containing the adducted adenine nucleosides: d(5′*AGTGATCGCGCTAGG-3′). Complementary oligodeoxynucleotides for the modified 12-mers were obtained from the Molecular Biology Core Laboratory, Thomas G. Wood, Director, NIEHS Toxicology Center (University of Texas Medical Branch, Galveston, TX). Analysis of Formation of DNA-DNA Cross-Links Using DNA Hairpin Strategy. The 15-mer oligodeoxynucleotides bearing the adducted nucleosides (*G and *A) were 5′-endlabeled using [γ32P]ATP and T4 polynucleotide DNA kinase and purified through a P-6 Bio-Spin Column (Biorad, Hercules, CA). A total of 10 nM labeled adducted or control 15-mers was annealed to a 5-fold molar excess of the partially duplex hairpin (50 nM) (Figure 3) and incubated at 25 °C for several days in the reaction buffer (100 mM sodium phosphate, pH 7.0, 5 mM MgCl2). Aliquots were withdrawn at the indicated times and treated with NaBH4 (25 mM final concentration) for 15 min at 25 °C. The NaBH4 addition was designed to reduce both the imino intermediate that had been established during the reaction and the unreacted aldehydic substrate. In reactions using uracil-containing 50-mers, residual NaBH4 was removed from the reaction mixtures using P-6 Bio-Spin columns. These DNAs were treated with UDG (New England BioLabs, Beverly, MA) according to manufacturer’s protocol and incubated with 2.9 µM T4-Pdg in 50 mM Tris-HCl (pH 7.0), 50 mM NaCl, and BSA (1 mg/mL) at 37 °C for 1 h. The reaction products were separated through a 15% polyacrylamide gel containing urea (8 M). Bands were visualized by autoradiography of wet gels using Kodak Scientific Imaging Film (Eastman Kodak Co.) and analyzed by a PhosphorImager using ImageQuant 5.0 software. Control experiments using nondenaturing electrophoretic conditions demonstrated that all labeled DNA migrated as duplex DNA (data not shown).

DNA-DNA Cross-Links from Aldehydic Adducts

Chem. Res. Toxicol., Vol. 18, No. 11, 2005 1685

Figure 3. Oligodeoxynucleotide design strategy for the detection of intra- and interstrand cross-links. The rationale for detection of cross-links was based on 50-mer hairpin-forming single-stranded DNAs that also served as scaffolds for annealing varying 15-mer adducted oligodeoxynucleotides (underlined). (A) To assay for γ-HOPdG and β-HOEdG cross-links, a series of 50mer DNAs were synthesized in which relative to the 5′-end: (1) position 15 was a C; (2) position 16 (designated N) was either G, A, T, or C; and (3) position 50 (designated N′) was complementary to the nucleotide at position 16. A uracil base was introduced at position 21 (underlined base) to discriminate between inter- and intrastrand cross-links. To detect crosslinked species that are derived from reaction with γ-HOPdA and β-HOEdA, similar strategies were utilized except that the position 15 was a T. Interstrand DNA-DNA Cross-Linking Reactions. The 12-mer oligodeoxynucleotides containing the adducted nucleotides *G were purified by HPLC (12) and quantified using absorbance values at 260 nm. DNAs were 5′-end-labeled using γ-32P-[ATP] and T4 polynucleotide DNA kinase according to the supplier recommendations. The adducted and control DNAs (10 nM) were annealed with a 5-fold molar excess of the appropriate complementary strands (50 nM). Control experiments using the nondenaturing electrophoresis conditions demonstrated that all labeled DNA migrated as duplex DNA (data not shown). In the experiments designed to detect carbinolamine cross-linked species, the duplex was incubated at 25 °C for several days in a reaction buffer containing 100 mM sodium phosphate (pH 7.0) and 5 mM MgCl2. Aliquots were removed at different time points and frozen at -80 °C until they were analyzed by electrophoresis using 15% denaturing polyacrylamide gels. To detect the presence of the imino cross-linked products, aliquots of the reaction mixtures described were treated with NaBH4 (25 mM final concentration). This NaBH4 concentration is sufficient to reduce not only the imine but also the aldehyde group of remaining substrate. After the addition of the reducing agent, the samples were incubated for 15 min at 25 °C, an equal volume of loading buffer (95% v/v formamide, 20 mM EDTA, 0.02% w/v bromphenol blue, and 0.02% w/v xylene cyanol) was added, and samples were frozen at -80 °C, until they were analyzed by electrophoresis using 15% denaturing polyacrylamide gels. Samples were heated at 90 °C for 10 min prior to loading. Radiolabeled DNA bands were visualized by autoradiography of wet gels using Kodak Scientific Imaging Film (Eastman Kodak Co.) and analyzed by a PhosphorImager using Image Quant 5.0 software.

Results Detection of Carbinolamine Type Interstrand Cross-Links by γ-HOPdG. To provide a measure of the kinetics of formation of the carbinolamine cross-link, 32Plabeled single-stranded 12-mer oligodeoxynucleotides containing a site specific γ-HOPdG or a control G in a CpG sequence context were annealed with a complementary 12-mer. Solution equilibria were established in master mixes, and aliquots were withdrawn at 0, 1, 3, 5, and 7 days. Analyses of DNA samples containing γHOPdG adducts revealed a major cross-linked species when electrophoresis was carried out without treatment with NaBH4 or heat (Figure 4A,B). The data in Figure

Figure 4. Detection of interstrand cross-links. (A) Visualization of the kinetics of formation of interstrand cross-links. 5′-32P endlabeled γ-HOPdG-containing 12-mers were annealed with a 5-fold molar excess of complementary 12-mer and incubated for up to 7 days. Aliquots were withdrawn and stored at -80 °C until DNAs were separated using a 15% polyacrylamide gel containing 8 M urea. The aliquots were neither treated with NaBH4 nor heated prior to electrophoretic separation. Lane 1, control single-stranded 12-mer containing γ-HOPdG; lanes 2-6, aliquots of reaction mixtures removed at 20 min and 1, 3, 5, and 7 days of incubation, respectively. (B) Quantitation of carbinolamine cross-link. Three independent experiments were analyzed for the kinetics of accumulation of the carbinolamine cross-link. (C) Differential mobilities of the DNA-DNA crosslinked products. 32P-labeled γ-HOPdG 12-mer DNAs were annealed with a 5-fold molar excess of complementary strands of lengths 15, 18, and 21 nucleotides (lanes 1-3, respectively), as described in panel A. After 5 days of incubation, DNAs were separated using a 15% polyacrylamide gel containing 8 M urea. Lane 4, molecular length size markers (8-32-mers).

4B show a slow rate of formation of this cross-link; data represent an average of three independent experiments. Heat treatment (90 °C) of these samples resulted in a total loss of the gel-shifted product (data not shown). Furthermore, data from identical reaction conditions and gel analyses using nonadducted G in place of γ-HOPdG yielded no cross-linked species (data not shown). To confirm the identity of the slowly migrating band in Figure 4A as a reversibly cross-linked product, the experimental design was changed to one in which the 32Plabeled γ-HOPdG-containing 12-mer strand was annealed to complementary strands of increasing length (15-, 18-, and 21-mer). Analyses were performed as

1686

Chem. Res. Toxicol., Vol. 18, No. 11, 2005

previously described, and the data in Figure 4C show that the relative mobilities of the cross-linked species were consistent with the increasing length differences of the complementary strand. These data were internally consistent with the interpretation that the carbinolamine cross-linked species migrates anomalously fast relative to unbranched oligonucleotides of the same size (20). Thus, consistent with previously reported NMR analyses and thermal lability studies (14), it is concluded that the cross-linked bands in panels A and C represent a carbinolamine cross-link. Additionally, the kinetics of the appearance of this cross-linked species were in good agreement with a prior report using HPLC analyses (15, 16). We attribute the differences in the yield of the reaction reported in the present work (15%) and in the previous studies (40-50%) to different experimental methodologies. Formation of Imine Type Cross-Links by γHOPdG. Imine type cross-links in equilibrium with carbinolamines in oligonucleotide duplexes in which γ-HOPdG in a CpG sequence context have thus far not been observed by NMR due to limitations of sensitivity (14). Nevertheless, interstrand imine type cross-links must be in equilibrium with carbinolamines, at least in small quantities, because they can be detected chemically by reduction with NaCNBH3 (15, 16). An inherent limitation of those studies was that the experimental design precluded observation of intrastrand cross-links. To probe cross-links that can arise in DNAs bearing aldehydic adducts, a versatile new assay system was developed that facilitated the detection of both inter- and intrastrand imine type cross-links in a variety of sequence contexts. This strategy relies on the use of stable hairpin-forming single-stranded DNAs (Figure 3) that also serve as scaffolds for annealing 15-mer adducted oligodeoxynucleotides. To assay for the formation of γ-HOPdG- and β-HOEdG-induced cross-links, a series of 50-mer oligodeoxynucleotides were synthesized in which relative to the 5′-end: (1) position 15 was a C; (2) position 50 was either G, A, T, C, or I; and (3) position 16 was complementary to the nucleotide at position 16, i.e., C, T, A, G, and C (Figure 3A). An additional set of four 50mer hairpin-forming oligodeoxynucleotides were also designed to detect DNA-DNA cross-links involving γ-HOPdA and β-HOEdA (Figure 3B); the design utilized the same strategy as in Figure 3A except that position 15 was a T. 32P-Labeled 15-mers were annealed at a molar ratio of 1:5, 15-mer:50-mer, to maximize annealing to the single-stranded overhang of the unlabeled 50-mer. The design of these DNAs facilitated detection of intrastrand cross-links between the adducted nucleotide and the bases in the region of nucleotide 50 of the 50-mer, while interstrand cross-links could be formed between the adducted nucleotide and the nucleotide 16. In both cases, following cross-linking, the labeled 15-mer would migrate with a mobility approximating a 65-mer. Thus, this assay has significant versatility because (1) both interstrand and intrastrand cross-links can be detected, (2) sequence context effects can be readily assayed, and (3) a variety of lesions can be investigated for their ability to form cross-linked species. However, a limitation of the assay is that only reducible imine cross-links can be detected since conditions required to destabilize the hairpin structure also reverse any carbinolamine cross-links. Figure 5A shows a representative autoradiogram of kinetic analyses of the rate of establishing the equilib-

Sanchez et al.

rium for NaBH4 reducible cross-linked DNA product in which the 32P-labeled 15-mer containing a 5′-γ-HOPdG was annealed to the 50-mer hairpin DNA. DNA duplexes were incubated at 25 °C for up to 7 days, and at various time points, aliquots were treated with 25 mM NaBH4 to rapidly reduce any imine type cross-linked DNA product. It should be emphasized that these assays only reveal the steady state equilibrium of reducible crosslink formation and that the assay conditions reverse all nonreducible cross-links. All aliquots were stored at -80 °C until immediately prior to electrophoretic separation, at which time, samples were heated to 95 °C in the presence of 95% formamide, reversing any unreduced cross-linked DNA and fully separating the unreacted 15and 50-mers. The kinetics of the appearance of a DNA fragment with a size approximating the 65-mer represents the rate of formation of imine type cross-linked DNA. Analysis of data in Figure 5A reveals that the formation of imine cross-links was slow and reversible; the average of five independent experiments showed that the percentage of reducible product(s) never exceeded ∼3% (Figure 5B). Control experiments using unadducted 15-mer oligodeoxynucleotides showed no formation of cross-linked species, thus ruling out questions of reagent purity. Furthermore, failure to reduce the samples with NaBH4 prior to analyses yielded no 65-mer products (data not shown). Experiments were also conducted in which the 50-mers were labeled and the 15-mers were unlabeled; although the data were similar, accurate quantitation was not possible due to the relatively small mobility differences between the 50-mer and 65-mer (data not shown). NMR analyses of DNAs containing the same adduct had not detected the presence of the imine. However, the minimum threshold of detection by the methodology used in the NMR study was in the 3-5% range (14). Thus, the steady state equilibrium concentration of the imine type interstrand DNA cross-link detected in the present study lies below the NMR detection limit. Data in Figure 5C show that similar conclusions can be drawn if 12-mer duplex DNAs containing γ-HOPdG were assayed for reducible cross-links. DNA duplexes containing a centrally placed γ-HOPdG in a CpG sequence context were employed with the adducted oligodeoxynucleotide bearing the 32P label. Samples were withdrawn periodically over a 5 day period and then treated with NaBH4, heated to 90 °C, and analyzed using a denaturing 15% polyacrylamide gel (Figure 5C). These data showed a gradual appearance of a slower migrating band, attributed to the cross-linked duplex, in which the two chains are tethered together via the reduced imine. To confirm that the slower migrating species is in fact an interstrand cross-link, the protocol was changed such that the complementary 12-mer DNA was 32P-labeled. The experiments were carried out as before, except that the reaction was monitored for 7 days. As shown in Figure 5D, a reduced cross-linked species was detected that displayed appearance kinetics and product mobility identical to those observed for the data shown in Figure 5C. Consistent with data in Figure 5A,B, a steady state concentration of the imine type cross-linked product was established over several days of incubation, and an average of seven independent experiments revealed that the equilibrium concentration reached a plateau near ∼3% (Figure 5E).

DNA-DNA Cross-Links from Aldehydic Adducts

Chem. Res. Toxicol., Vol. 18, No. 11, 2005 1687

Figure 5. Kinetic analyses of the formation of imine type cross-links. (A) A representative autoradiogram of kinetic analysis of the formation of reducible cross-linked 65-mer species, in which the 32P-labeled γ-HOPdG-adducted 15-mer was annealed to the 50-mer hairpin DNA containing a C at position 50. Reactions were incubated for up to 7 days, followed by quenching with 25 mM NaBH4. Prior to electrophoretic separation of the labeled 15-mers from the cross-linked 65-mers, samples were denatured in 95% formamide and heated for 10 min at 90 °C. Lane 1, control unmodified single-stranded 65-mer; lane 2, single-stranded γ-HOPdG-adducted 15-mer; lanes 3-7, 20 min and 1, 3, 5, and 7 days of incubation, respectively. (B) Quantitation of the formation of reducible crosslinked species. A total of five independent experiments were analyzed for formation of reduced cross-linked species. (C) Kinetics of reducible cross-link formation using a 32P-labeled 12-mer containing a centrally located γ-HOPdG, annealed to a 5-fold molar excess of an unlabeled complementary 12-mer DNA. Control 12-mer without complement (lane 1) and aliquots of the cross-linking reaction were taken after 20 min and 1, 3, and 5 days (lanes 2-5), respectively. Prior to electrophoretic separation, samples were denatured with 95% formamide and heated for 10 min at 90 °C. (D) Kinetic analyses of reducible interstrand cross-link formation in which, in contrast to the analyses shown in panel C where the γ-HOPdG-containing strand was labeled, the complementary, unadducted strand was 32P-labeled and annealed to a 5-fold molar excess of unlabeled γ-HOPdG 12-mers; otherwise as described in panel C. (E) Quantitation of the accumulation of reducible cross-linked species from data sets shown in panel C. A total of seven independent experiments is shown.

Sequence Context Effects on Imine Cross-Link Formation by γ-HOPdG. To investigate the sequence effects on formation of reducible cross-links, an experimental design was employed in which positions 16 and 50 of the 50-mers were changed to represent all four of the common base pairs. Because the previous study had not revealed significant alterations in the equilibrium concentration of imine cross-links after 5 days, the labeled 15-mer oligodeoxynucleotides were annealed and incubated for 5 days using the previously described conditions and subsequently treated with NaBH4. The data shown in Figure 6 reveal that in every sequence context, γ-HOPdG was capable of forming small but reproducible amounts of reducible cross-linked species. Interestingly, the reduced product formed in the Cp*G sequence context (Figure 6, lane 4) had a slightly faster mobility than the reduced cross-linked DNAs in either the Tp*G, Ap*G, or Gp*G sequence contexts (Figure 6, lanes 1-3, respectively). These mobility differences could potentially be accounted for by differential mobilities of

Figure 6. Formation of cross-linked species as modulated by sequence context effects. The experimental design was the same as described in Figure 5. γ-HOPdG-adducted or unadducted 5′end-labeled 15-mers were annealed to a 5-fold molar excess of the partially duplex hairpin, incubated at 25 °C for 5 days, and then quenched with NaBH4 for 15 min at 25 °C. The reaction products were separated through a 15% polyacrylamide gel containing urea (8 M). Bands were visualized by autoradiography of wet gels. Lanes 1-4, 15-mer oligodeoxynucleotides containing γ-HOPdG were annealed to four different 50-mers in which position 50 of the hairpin structure was T (lane 1), A (lane 2), G (lane 3), and C (lane 4).

interstrand vs intrastrand cross-links. In control reactions using the unadducted 15-mer, no product bands were observed (data not shown).

1688

Chem. Res. Toxicol., Vol. 18, No. 11, 2005

Figure 7. Experimental resolution of inter- and intrastrand DNA cross-links. The scaffolding 50-mer oligodeoxynucleotides was as shown in Figure 3 with an exception that a C at position 21 was substituted by uracil. Reactions were carried out for 7 days using γ-HOPdG in all four sequence contexts, followed by NaBH4 quenching. DNAs were sequentially treated with UDG and T4-pdg. DNAs were separated using a denaturing polyacrylamide gel, and their relative migration was visualized by PhosphorImager analyses. Lanes 1 and 10, control singlestranded 44-mer oligodeoxynucleotide; lanes 2 and 9, molecular length size markers (8-32-mers); lane 3, γ-HOPdG-containing 15-mer; lane 4, γ-HOPdG-containing 15-mer treated with UDG and T4-pdg; lanes 5-8, 15-mer oligodeoxynucleotides containing γ-HOPdG were annealed to four different 50-mers in which position 50 of the hairpin structure was A (lane 5), G (lane 6), C (lane 7), and T (lane 8).

To test whether the differential electrophoretic mobilities of the four cross-linked species could be accounted for by interstrand vs intrastrand cross-links, the 50-mer hairpin was redesigned to include a uracil at position 21 (Figure 3, the underlined base). The rationale for positioning uracil at this site was to introduce a potential site specific single strand break in the 50-mer, such that upon thermal denaturation, the two cross-linked species could be readily resolved. After sequential treatment of the duplex DNAs with uracil DNA glycosylase (UDG) to create an abasic site and with a T4 pyrimidine dimer glycosylase/abasic site lyase (T4-Pdg) to cleave the DNA strand containing the abasic site, the radiolabeled strand would migrate approximately as a linear 44-mer for DNAs containing an intrastrand cross-link, while DNAs containing an interstrand cross-link would migrate as a branched 35-mer (labeled 15-mer plus a 20-mer). Hairpin 50-mer oligodeoxynucleotides containing each of the four bases (T, A, G, and C) at position 16, complement base at position 50, and uracil at position 21 were annealed to γ-HOPdG-modified 15-mer oligodeoxynucleotides, and the cross-linking reactions were carried out for 7 days as previously described. DNAs were treated with NaBH4, UDG, and T4-pdg as provided in the Experimental Procedures section and separated by electrophoresis, using as sizing references both the 8-32 standard markers and a 44-mer oligodeoxynucleotide with the predicted sequence of the intrastrand DNADNA cross-link. The data in Figure 7 (lane 7) clearly demonstrate that the cross-linked species in the Cp*G sequence context does not migrate as a 35-mer but rather comigrated with a 26-mer length standard. Anomalously rapid migration might be expected for a branched (20 + 15)-mer (20). Our interpretation of these data is that γ-HOPdG forms an interstrand imine cross-link in the Cp*G sequence context. This conclusion is in agreement with the data in Figure 5 and with previous studies from which it was concluded that this lesion forms carbinolamine and imine cross-links in the present Cp*G sequence context (14-16). In contrast to these results, analysis of cross-linking in a Gp*G context revealed formation of a labeled DNA migrating with the 44-mer

Sanchez et al.

Figure 8. Detection of the formation of reducible cross-links by γ-HOPdG and β-HOEdG adducts. The experimental design and implementation were as described in Figures 5 and 6, respectively. Panels A and B show the kinetics for establishing an equilibrium of NaBH4 reducible cross-linked DNAs in which the adducted nucleotide is γ-HOPdG and β-HOEdG, respectively. Nucleotide 50 (N′) of the 50-mer was varied and from top to bottom was G, A, T, C, and I.

Figure 9. Detection of the formation of reducible cross-links by γ-HOPdA and β-HOEdA. Experimental design and implementation were as described in the Figure 8 legend.

standard Figure 7 (lane 6), a result that suggests that an intrastrand cross-link was formed in the Gp*G context, rather than an interstrand one. Similarly, in the Ap*G and Tp*G sequence contexts (Figure 7, lanes 5 and 8, respectively), 44-mer products were observed in which those mobilities were consistent with intrastrand crosslinks. It should be noted that the band migrating as a 32-mer was present in the starting sample of γ-HOPdG used in these studies (Figure 7, lanes 3 and 4) and the relative intensity of that band did not change during the time course of the reaction. Formation of Imine Type Cross-Links by Related Aldehydic Adducts. To exploit the high throughput capacity of the 15-mer + 50-mer cross-linking assay, studies were performed in four sequence contexts using γ-HOPdG, β-HOEdG, γ-HOPdA, and β-HOEdA. The data in Figures 8 and 9 show the formation of imine crosslinks as a function of time, using the hairpin oligonucleotide cross-linking assay described above, in which 32 P-labeled 15-mers with various nucleotides at the 5′end were assayed for formation of a cross-link that could be reduced by NaBH4. In these experiments, nucleotide 50 of the 50-mer was either a G, A, T, C, or I. Figure 8A,B shows representative autoradiograms of the crosslinked 65-mer DNA (Figure 3A) using a 32P-labeled 15mer containing either γ-HOPdG or β-HOEdG at the 5′end. Figure 9A,B shows representative autoradiograms

DNA-DNA Cross-Links from Aldehydic Adducts

of the kinetics of 65-mer formation (Figure 3B), using a 32 P-labeled 15-mer containing either γ-HOPdA or β-HOEdA at the 5′-end. Consistent with data in Figure 5A,B, Figure 8A shows that if nucleotide 50 is a C (nucleotide 16 is a G), the formation of a reducible cross-linked species is established over a period of several days. Somewhat surprisingly, a cross-link could still be observed, although somewhat weaker, when I was substituted for the G at position 50. Figure 8A also shows strong cross-linking when nucleotide 50 is a G. When it is a T, a reducible cross-link still forms. However, the kinetics of formation are different, reaching a steady state within 20 min. The poorest cross-linking occurred when position 50 was an A, although the cross-link could be reproducibly detected. When β-HOEdG was analyzed for its capacity to form reducible cross-linked species, the kinetics in various sequence contexts were similar to those observed with the γ-HOPdG, except that the extent of cross-linking was somewhat less in every case except A where crosslinking was somewhat more efficient (Figure 8B). Figure 9A,B shows representative data for 65-mer cross-linked DNA using γ-HOPdA and β-HOEdA, respectively. Cross-links formed in every sequence context and showed a gradual increase in concentration to a plateau over the 5 day duration of the experiment. Close examination of the kinetics of formation of these cross-links revealed some differences in the rate of formation and final concentrations, but interpretation of these differences, as with the γ-HOPdG or β-HOEdG cross-links, must await elucidation of the structures of the crosslinked species.

Discussion Exposure of cells to either endogenous or exogenous sources of bis-electrophiles, such as acrolein, crotonaldehyde, and 4-hydroxy-2-nonenal, results in the formation of DNA adducts, mainly of guanine. Following treatment of DNA with acrolein, R- and γ-HOPdG are formed with the latter being the dominant species in DNA. NMR investigations of the solution structure of duplex DNA containing γ-HOPdG have demonstrated that the three-carbon exocyclic adduct exists in an equilibrium between a ring-opened aldehydic form and, under acidic conditions, a ring-closed pyrimidopurinone form. The ring-opened form predominates in doublestranded DNA near neutrality. This equilibium will be present with the adducts of higher enals and also is expected to favor the open chain form of the adducts in duplex DNA. It should be noted that at the nucleoside level the equilibrium for γ-HOPdG and other γ type enal adducts of dG lies far toward the ring-closed form such that the ring-opened species cannot be detected by NMR, even though the presence of a dynamic equilibrium can be demonstrated by reduction with NaCNBH3. In DNA, the presence of the reactive aldehyde in the minor groove leads to the possibility that secondary chemical reactions can occur with amines and other nucleophiles. Such reactions could give rise to DNADNA and DNA-protein cross-links (21-23). These secondary lesions could be much more deleterious to the viability of cells than the original base modification. This investigation sought to obtain qualitative data for the ability of γ-hydroxypropano and β-hydroxyethano adducts of dG and dA to form DNA-DNA cross-links. The cross-

Chem. Res. Toxicol., Vol. 18, No. 11, 2005 1689

links will exist as an equilibrium mixture of carbinolamines, imines, and possibly pyrimidopurinones. At least for interchain cross-links, we regard the latter to be least likely on steric grounds and due to disruption of WatsonCrick base pairing. Nevertheless, it should be noted that in a publication appearing after submission of this manuscript, Lao and Hecht presented arguments as to why the pyrimidopurinone might be the dominant crosslink species (24). NMR studies of the acrolein and crotonaldehyde cross-links have led to the conclusion that carbinolamines are the dominant species present; imines, although certainly present, are at concentrations below the level of detection. The pyrimidopurinone species, if present at all, is also below the level of detection.1 Three goals of the present study involved development of methodology for (1) direct detection of the presence of imine, (2) detection of intrastrand as well as interstrand cross-links, and (3) probing for DNA-DNA cross-links that can be formed by other types of aldehydic lesions. In the earlier studies of cross-linking by γ-HOPdG and higher homologues, reductions were carried out with NaCNBH3 for extended reaction periods. This technique permitted efficient trapping of the nonreducible carbinolamine via equilibrium with the reducible imine but shed no light on the equilibrium concentrations of imine. The present study used NaBH4 as the reducing agent. NaBH4 is significantly more reactive than NaCNBH3. This reactivity is exhibited not only in its reactions with imines and carbonyl compounds but also with water where an acid-catalyzed decomposition leads to a short half-life in aqueous solutions near neutrality. In the present study, we take advantage of this property of NaBH4 to create a “snapshot” assay of the level of imine present in the equilibrating mixtures provided by the aldehydic adducts in DNA. For γ-HOPdG in a CpG context in a DNA duplex, we were able to reproduce the earlier observation that high levels of an interchain carbinolamine type cross-link were formed, and we were also able to detect the presence of small amounts (