Formation of Deoxyguanosine Cross-links from Calf Thymus DNA Treated with Acrolein and 4-Hydroxy-2-nonenal

Abstract

Acrolein (AC) and 4-hydroxy-2-nonenal (HNE) are α,β-unsaturated aldehyde (enal) endogenous bis-electrophiles that arise from the oxidation of polyunsaturated fatty acids. AC is also found in high concentrations in cigarette smoke and automobile exhaust. These reactive enals covalently modify nucleic acids, to form exocyclic adducts, where the three-carbon hydroxypropano unit bridges the N1 and N² -positions of deoxyguanosine (dG). The bifunctional nature of these enals enables them to undergo reaction with a second nucleophilic group and form DNA cross-links. These cross-linked enal adducts are likely to contribute to the genotoxic effects of both AC and HNE. We have developed a sensitive mass spectrometric method to detect cross-linked adducts of these enals in calf thymus DNA (CT DNA) treated with AC or HNE. The AC and HNE cross-linked adducts were measured by the stable isotope dilution method, employing a linear quadrupole ion trap mass spectrometer and consecutive reaction monitoring at the MS³ or MS⁴ scan stage. The lower limit of quantification of the cross-linked adducts is ~1 adduct per 10⁸ DNA bases, when 50 µg of DNA is assayed. The cross-linked adducts occur at levels that are ~1–2% of the levels of the monomeric 1,N²-dG adducts in CT DNA treated with either enal.

Introduction

Acrolein (AC), crotonaldehyde, 4-hydroxy-2-nonenal (HNE) and other α,β-unsaturated aldehydes (enal) are endogenous bis-electrophiles that arise from the degradation of polyunsaturated fatty acids initiated by reactive oxygen species (1). Acrolein is also found in high concentrations in cigarette smoke and automobile exhaust (2). Enals are reactive electrophiles that covalently modify proteins, peptides, and nucleic acids. As a result, many enals have been shown to be cytotoxic and genotoxic (3–5).

The bifunctional nature of enals enables them to undergo reaction with two nucleophilic groups. In the case of deoxyguanosine (dG), this reactivity results in a so-called exocyclic adduct in which a three-carbon hydroxypropano unit bridges the N1 and N² -positions (Figure 1). The reaction of AC with dG affords two regioisomeric products, the 8-hydroxy-1,N² -propano-dG (1, 8-HO-PdG) and 6-hydroxy-1,N²-propano-dG (2, 6-HO-PdG) adducts in approximately a 2:1 ratio. Thus far, only the 8-hydroxy isomers have been identified from the higher enals. The reaction of crotonaldehyde with dG results in two stereoisomeric adducts (3 and 4), while HNE produces four stereoisomers (4–8). AC, crotonaldehyde, and HNE adducts of dG have been found in background levels in rodent and human DNA (6–11); high levels of the AC and HNE adducts have been detected in the brains of Alzheimer’s patients (12, 13) and both regioisomeric AC adducts were found in the lung tissue from smokers (14). Acrolein, crotonaldehyde, and HNE adducts have been demonstrated to be mutagenic when replicated in mammalian cells using site-specifically modified vector with G→T transversions being the predominate mutation observed. (5).

Figure 1.

Structures of AC, crotonaldehyde, 4-HNE, and their dG adducts

Kawanishi reported electrophoretic evidence that acrolein can cross-link DNA (15); however, the structure of the cross-link was not determined. Wang et al. identified a pyrimidopurinone crosslink from calf thymus DNA (CT DNA) treated with acetaldehyde (Figure 2, 9b) (16, 17). This product is formally a cross-link derived from crotonaldehyde and two dG’s; although crotonaldehyde is the aldol condensation product of acetaldehyde, it was not believed to be involved in the cross-linking reaction (18). Kozekov et al. demonstrated that oligonucleotides containing dG adducts of acrolein and crotonaldehyde formed interstrand DNA cross-links when site-specifically incorporated in a 5′-CpG-3′ sequence context (19, 20); the HNE adduct was subsequently shown to form interstrand cross-links in the same sequence (21). The cross-linking reaction was slow in each case, but most efficient for the AC adduct and highly dependent on the stereochemistry of the crotonaldehyde and HNE adducts (22). The corresponding pyrimidopurinones (9a–c) (Figure 2) were identified from the enzymatic digestion of the cross-links duplex. Treatment of AC and crotonaldehyde cross-linked duplexes with Na(CN)BH₃ prior to digestion resulted in the reduced cross-links where the N²-positions of two dG’s is tethered by a three-carbon chain (10, dG-(CH₂)₃-dG) (19, 22).

Figure 2.

Structures of dG cross-linked adducts of AC, crotonaldehyde and HNE (9), and their chemically reduced analogs (10).

Many cellular processes involving DNA require transient strand separation. Interstrand cross-links are anticipated to interfere with transcription, replication, and DNA repair. The mechanism of action of many anti-cancer agents, such as nitrogen mustards, mitomycin C and cisplatin, are believed to involve DNA cross-links (23–26). Cell lines from patients with the rare genetic disorder Fanconi anemia are highly sensitive to DNA cross-linking agents, but not other DNA damaging agents; this suggests that cells have developed mechanisms to repair DNA crosslinks (26–29). We have hypothesized that interstrand DNA cross-links may contribute to the genotoxic effects of enals. We report here the development of a mass spectrometric method to detect dG enal cross-links in calf thymus DNA (CT DNA) treated with AC or HNE. The cross-linked adducts were quantitated by the stable isotope dilution method, using liquid chromatography-electrospray ionization/multi-stage tandem mass spectrometry (LC-ESI/MS/MSⁿ) with a linear quadrupole ion trap mass spectrometer (LIT MS) and consecutive reaction monitoring at the MS³ or MS⁴ scan stage. We estimate that the cross-linked adducts are present at levels that are ~1% that of the monomeric 1,N²-dG adducts.

Experimental Procedures

Materials

AC was purchased from Aldrich. HNE was prepared according to a literature procedure (30). [¹⁵N₅]-dG (>98% isotopic purity) was purchased from Cambridge Isotope Laboratory, Inc. [²H₆]-1,3-diaminopropane hydrochloride (>98.8% isotopic purity) was purchased from CDN Isotopes. [²H₁₁]-1-bromopentane (>98% isotopic purity was purchased from Isotec (Sigma-Aldrich). Oligonucleotides containing the 8-HO-PdG and HNE-dG adducts were prepared as previously described (21,31). Calf thymus (CT) DNA, dG, adenosine deaminase (Type X from calf spleen), DNase I (Type IV, bovine pancreas), alkaline phosphatase (from E. coli), and nuclease P1 (from Penicillium citrinum) were purchased from Sigma (St. Louis, MO). Phosphodiesterase I (from Crotalus adamanteus venom) was from GE Healthcare (Piscataway, NJ). HyperSep™ filter SpinTips C18 (20 mg) were from Thermo Scientific (Palm Beach, FL). The isotopic purity of the ¹⁵N and ²H-labeled standards was > 99.5%.

HPLC Separations

The purification of nucleosides and oligonucleotides and the analysis of reaction mixtures and nucleosides obtained from enzymatic digestions were performed on a Beckman HPLC system (32 Karat software version 7.0, pump module 125) with a diode array UV detector (module 168) monitoring at 260 nm. An analytical Waters YMC ODS-AQ column (250 mm × 4.6 mm i.d., 1.5 mL/min) was used for monitoring the reactions and a semi-preparative column (250 mm × 10 mm i.d., 5 mL/min) was used for purification of the desired products. The mobile phases were H₂O and CH₃CN for nucleosides and 0.1 M aqueous ammonium formate and CH₃CN for oligonucleotides and some nucleosides with the following gradients: Gradient A: initially 1% acetonitrile, 1–10% acetonitrile over 15 min, 10–20% acetonitrile over 5 min, hold for 5 min, 20–100% acetonitrile over 3 min, hold for 2 min, and then back to 1% acetonitrile over 3 min. Gradient B: initially 1% acetonitrile, 1–5% acetonitrile over 5 min, 5–6% acetonitrile over 15 min, 6–100% acetonitrile over 3 min, hold for 2 min, and then back to 1% acetonitrile over 3 min. Gradient C: initially 18% acetonitrile, 18–26% acetonitrile over 38 min, and then back to 18% acetonitrile over 2 min.

6-HO-PdG/[¹⁵N₅]-6-HO-PdG and 8-HO-PdG/[¹⁵N₅]-8-HO-PdG

Acrolein 31.3 µL (0.468 mmol, 3 eqv.) was added in three portion over one hour to a solution of dG•H₂O (44.5 mg, 0.156 mmol) and L-arginine (81.6 mg, 0.486 mmol) in degassed phosphate buffer (5 mL, 0.1 M, pH8) at 37° C, and reaction was incubated overnight. The dG was completely consumed in the reaction and three products were observed in nearly equal amounts. The reaction mixture was extracted with ethyl acetate (3 × 5 mL), and then purified by HPLC using Gradient B. The first two peaks were collected together and were the two diastereoisomers of 6-HO-PdG (18 mg, 35.5%). The third peak was collected separately and was 8-HO-PdG (11 mg, 21.8%). The 8- and 6-HO-PdG were identified based on comparison of the retention time with known standards (7, 32). 8-HO-PdG: UV: λ_max 260 nm (ε=11650, H2O). 6-HO-PdG, UV: λ_max 260 nm (ε=10,880, H₂O).

The reaction was perform with (¹⁵N₅)-dG(5mg) and the isotopic purity of [¹⁵N₅]-6-HO-PdG and [¹⁵N₅]-8-HO-PdG was > 99.5% by LC-MS/MS.

N²-(3-Hydroxypropyl)-PdG (8-HO-PdG_red)

8-HO-PdG (2.0 mg) in 1:1 (v/v) H₂O/methanol (200 µL) was treated with sodium borohydride (3.0 mg in 39.6 µL 1N NaOH) (32). The mixture was stirred at room temperature for 8 h and then neutralized with 5% acetic acid. HPLC purification (Gradient B) afforded 8-HO-PdG_red (1.5 mg, 75%). UV: λ_max 260 nm (ε= 13,551, H₂O).

The reaction was performed with (¹⁵N₅)-8-HO-PdG (0.5 mg) and the isotopic purity of (¹⁵N₅)-8-HO-PdGred was > 99.5% by LC-MS/MS.

N1-(3-Hydroxypropyl)-dG (6-HO-PdG_red)

NaBH₄ (6 mg, 50 equiv) was added in three portions to a solution of 6-HO-PdG (1.0 mg) in phosphate buffer (200 µL. 50 mM, pH 7.0) over 1 h. The reaction was followed by HPLC and the starting material was completely consumed after 24 h. The reaction was neutralized with 5% acetic acid, and then was purified by HPLC (Gradient A) to afford 6-HO-PdG_red (0.5 mg, 50%). UV: λ_max 260 nm (ε= 11,180, H₂O).

1,N²-Propano-dG (PdG)

A solution of 6-HO-PdG (1.0 mg) and Na(CN)BH₃ (3 mg, 15 equiv) in phosphate buffer (200 µL, 50 mM, pH 7.0) was stirred for 24 h at 45°C, after which time HPLC analysis showed the reaction to be complete. Purification by HPLC (Gradient A) afforded PdG (0.6 mg, 60%). UV: λ_max 260 nm (ε= 13,190, H₂O).

[¹⁵N₅]-1,N²-Propano-dG ([¹⁵N₅]-PdG)

[¹⁵ N₅]-dG (1 mg, 3.7 µmol) and anhydrous K₂CO₃ (5 mg) were placed in glass tube with dry DMSO (300 µL). After 30 min at 45° C 1,3-dibromopropane (0.82 mg, 0.41 µL, 4 µmol) was added and the reaction was kept at 45° C for 24 h. HPLC showed that the dG completely consuming the starting material and the product was purified by HPLC (Gradient A) to afford [¹⁵N₅]-PdG (0.7 mg, 61%).

[²H₆]-1,3-Bis(2'-deoxyguanosin-N²-yl)propane (dG-(CH₂)₃-dG)

The [²H₆]-1,3-diaminopropane hydrochloride (5 mg, 32.8 µmol) was added to excess of O⁶-(trimethylsilyl)ethyl)-2-fluoro-2’-deoxyinosine (25 mg, 67.5 µmol), in DMSO (150 µL) and diisopropylethylamine (200 µL). The mixtures were stirred at 55°C for 24 h. The reactions were stopped and the solvents were evaporated under vacuum. The residues were dissolved in 5% acetic acid (1 mL) and stirred at room temperature for 1 h, neutralized, followed by HPLC purification (Gradient A) afforded [²H₆]-dG-(CH₂)₃-dG (10.5 mg, 55%). UV λmax 260 nm (ε= 23,700, H₂O).

[²H₁₁]-4-HNE

Mg(0) turnings (90 mg, 3.6 mmol) and dry ethyl ether (2 mL) were placed in a three-necked flask. A few drops of a solution of [²H₁₁]-1-bromopentane (420 mg, 2.8 mmol) in dry ether (4 mL) were added, and the flask was gently heated until the reaction started. After the initiation of Grignard reagent formation, the remaining bromide solution was added dropwise. The reaction mixture was heated at reflux for 30 min, and then cooled to 0°C. A solution of fumaraldehyde monodimethyl acetal (300 mg, 2.3 mmol) in dry ether (2 mL) was added dropwise. The monoacetal was obtained by the partial acid hydrolysis of the fumaraldehyde dimethyl acetal (500 mg, 2.8 mmol), using Amberlyst-15 catalyst (0.12 g) in acetone (11.5 mL) and H₂O (0.1 mL). The reaction was stirred for 6 min, and then filtered through a bed of anhydrous sodium carbonate and solvent removed. After all of the monoacetal had been added, the stirring was continued for an additional 2 h at 0°C. The reaction was quenched with saturated aqueous NH₄Cl and extracted with diethyl ether. The extracts were dried with MgSO₄, filtered and evaporated. Purification on a Biotage medium pressure chromatography apparatus using an M25 column and eluting with 0–5% methanol in dichloromethane afforded the dimethyl acetal of [²H₁₁]-HNE. The dimethyl acetal was hydrolyzed with 2% H₂SO₄ (4.2 mL) for 1h at room temperature. The solution was neutralized with 1M NaOH and extracted with diethyl ether. The extracts was dried with MgSO₄, filtered and evaporated. Purification on a Biotage medium pressure chromatography apparatus using an M25 column and eluting with 0–3% methanol in dichloromethane afforded [²H₁₁]-4-HNE (280 mg, 78%).

[²H₁₁]-HNE-dG

[²H₁₁]-4-HNE (17.6 mg, 0.105 mmol) was added to a solution of dG•H₂O (30 mg, 0.105 mmol) and L-arginine (55 mg, 0.315 mmol) in degassed phosphate buffer (4 mL, 0.1 M, pH 8.0) at 37 °C. Two additional portions of [²H₁₁]-4-HNE were added after 15 and 60 min. The reaction vessel was sealed and heated at 37° C for 24h. After cooling, the unreacted HNE and polymerized products were extracted with ethyl acetate (3 × 5 mL). Purification of the aqueous phase by HPLC (Gradient C) afforded [²H₁₁]-HNE-dG (16 mg, 35%) as a mixture of four stereoisomers. UV: λmax 260 nm (ε= 12,000, H₂O).

[²H₁₁]-8-dG-HNE-dG

[²H₁₁]-HNE-dG (3 mg) and dG(6 mg) were dissolved in DMSO (200 µL). The reaction mixture was heated at 100 °C for 5 days, and then the solvent was evaporated under vacuum (21). The residue was dissolved in sodium phosphate buffer (500 µL, pH 7.0, 0.05 M) and purified by HPLC (Gradient C) to afford [²H₁₁]-8-dG-HNE-dG (0.4 mg, 8%). UV λ_max 260 nm (ε= 24,200, H₂O).

Cross-linking of the 5′-GCTAGC(8-HO-PdG)AGTCC-3′ • 5′-GGACTCGCTAGC-3′ Duplex

The 5′-GCTAGC(8-HO-PdG)AGTCC-3′ oligonucleotide (1.0 OD, 8.1 nmol, MW = 3,700.6) was hybridized with complementary strand (1.1 OD, 8.91 nmol, MW = 3,646.4) and incubated in phosphate buffer (200 µL, 50 mM, pH 7.0 containing 1M KCl) at 45 °C. The cross-linking reaction was monitored by CGE; after 10 day the level of interstrand cross-link was ~50% (19, 20). The reaction was treated with Na(CN)BH₃ (0.1 M final concentration). Additional Na(CN)BH₃ (0.2 M final concentration) was added after 30 min, then again after 1h (0.3 M final concentration). The sample was incubated at room temperature for 24 h, and then quenched with 5% CH₃CO₂H. The reaction was filtered through a spin column (Bio-Spin 6 Tris Columns from BioRad) and lyophilized.

Cross-linking of the 5′-GCTAGC(8-HNE-dG)AGTCC-3′ • 5′-GGACTCGCTAGC-3′ Duplex

The 5′-GCTAGC(8-HNE-dG)AGTCC-3′ oligonucleotide (1.5 OD, 12.15 nmol, MW = 3,802.5) was hybridized with complementary strand (1.6 OD, 12.96 nmol, MW = 3,646.4) and incubated in phosphate buffer (200 µL, 50 mM, pH=7.0 containing 1M KCl) at 45° C. The cross-linking reaction was monitored by CGE; the level of cross-link was estimated to be ~50 %, based upon UV response, after 28 days (21).

Capillary Gel Electrophoresis (CGE)

Electrophoretic analyses were carried out using a Beckman P/ACE MDQ instrument system (using 32 Karat software, version 5.0) monitored at 260 nm on a 31.2 cm × 100 µm eCAP capillary with samples applied at 10 kV and run at 9 kV. The capillary was packed with their 100-R gel (for ss-DNA) using the Tris-borate buffer system containing 7 M urea.

Chemical Modification of DNA with AC

CT DNA (0.33 mg/0.33 mL) in 50 mM sodium phosphate buffer (pH 7.0) was treated with 0.1, 0.5, 1 or 5 mol of AC per mol of DNA base for 0, 1, 2, 3, 6 or 10 days, at 37 °C. The unreacted AC and its by-products were removed from the DNA by 3 solvent extractions with an equal volume of ethyl acetate. The AC-cross-linked adduct 8-dG-AC-dG (9a) was chemically reduced by treatment of the CT DNA with Na(CN)BH₃ (10 µL of 3.3 M stock solution). The mixture was incubated at room temperature for 15 min. Thereafter, another 10 µL of the Na(CN)BH₃ solution was added and the reduction continued for 15 min. Finally, another 10 µL of the Na(CN)BH₃ solution was added after 1 h, and the reduction reaction stirred overnight at room temperature. The DNA solution was then quenched with 5% CH₃CO₂H (~50 µL) to attain a pH value between 7.0 and 7.5. The treatment of AC-modified DNA with Na(CN)BH₃ completely reduced the 8-HO-PdG cross-link 8-dG-AC-dG to its ring-opened, reduced form dG-(CH₂)₃-dG (10) (Figure 2) However, 6-HO-PdG and 8-HO-PdG only underwent partial reduction. Therefore, a second chemical reduction with NaBH₄ was done, following enzymatic digestion of the DNA, as described below, to reduce both HO-PdG adducts (Scheme 2). The AC-DNA was precipitated by addition of 0.1 volumes of 5 M NaCl, followed by 1.5 mL C₂H₅OH. The DNA filament was washed with C₂H₅OH:H₂O (7:3) and air-dried.

Scheme 2.

Chemical reduction of AC and HNE-DNA adducts by Na(CN)BH₃/NaBH₄

Chemical Modification of DNA with HNE

CT DNA (1 mg/mL in 50 mM sodium phosphate buffer pH 7.0) was reacted with 2 mol of HNE per mol DNA base. The time course of reaction proceeded for 0, 1, 7, 14, or 21 days at 37 °C. Unreacted HNE and its by-products were removed from the DNA by extraction with an equal volume of ethyl acetate (3×), and the DNA was isolated as described above, except that the DNA was not treated with reducing agents.

Enzymatic Digestion of AC and HNE-DNA Adducts and Interstrand Cross-linked Adducts

The enzyme digestion conditions used for the hydrolysis of AC-modified DNA (10 – 50 µg, in 5 mM BisTris-HCl buffer (pH 7.1, 50 µL) employed DNAse I (13 U) for 1.5 h at 37 °C, followed by incubation with nuclease P1 (200 mU, 0.1 U/µL in 1 mM ZnCl₂) for 3 h, and then by treatment with alkaline phosphatase (100 mU) and phosphodiesterase (4 mU) for 18 h (33). Thereafter, the DNA digest was treated with adenine deaminase (15 mU), for 90 min at room temperature. A second chemical reduction was then done, by treatment of the digest with NaBH₄ (3 × 6 µL of 2 M solution prepared in 1N NaOH), over a time period of 20 min. The treatment with NaBH₄ completely reduced 8-HO-PdG to its ring-opened form N²-(3-hydroxypropyl)-2′-deoxyguanosine (8-HO-PdG_red). Reduction of the 6-HO-PdG resulted in 1,N²-(1,3-propano)-dG (PdG), presumably through the corresponding N5-C6 imine intermediate (14); N1-(3-hydroxypropyl)-2′-deoxyguanosine (6-HO-PdG_red) from the reduction of the ring-opened aldehyde was not observed (Scheme 1). The excess NaBH₄ was then quenched with 5% CH₃CO₂H (~50 µL), to attain a pH value of 7.0 – 7.5.

Scheme 1.

Synthesis of the isotopically labeled internal standards

The 8-dG-HNE-dG cross-link (9c) could not be reduced with either Na(CN)BH₃ or NaBH₄. Moreover, the cross-link underwent partial hydrolysis to form HNE-dG, during the digestion procedure employed for AC-modified CT DNA (unpublished observations). Therefore, the DNA digestion conditions were modified to conserve the 8-dG-HNE-dG cross-link. The amounts of several of the enzymes used for the digestion of DNA were increased, and the time of digestion was decreased to 2.5 h. CT DNA (50 µg in 5 mM BisTris-HCl buffer (pH 7.1, 50 µL)) was simultaneously digested with DNAse I (10 U), nuclease P1 (1.8 U, 0.1 U/µL in 1 mM ZnCl₂), alkaline phosphatase (120 mU), and phosphodiesterase I (200 mU), for 2.5 h at 37 °C.

Solid Phase Extraction (SPE) of AC- and HNE-DNA Adducts

Thereafter, the DNA digest was diluted with high-purity water (200 µL; Burdick and Jackson), and the adducts were purified by SPE, using HyperSep™ filter SpinTips. The AC-modified DNA digests were applied to a SpinTip, which was placed on a vacuum manifold and prewashed with CH₃OH containing 0.1% HCO₂H (2 × 0.25 mL), followed by H₂O (2 × 0.25 mL). The SpinTips were then washed with H₂O (2 × 0.25 mL), to remove non-modified 2′-deoxynucleosides. The desired adducts were then eluted with CH₃OH containing 0.1% HCO₂H (0.3 mL) into silylated glass insert capillary LC vials (Microliter Analytical Supplies, Suwanee, GA). Samples were evaporated to dryness by vacuum centrifugation and reconstituted in H₂O (20 µL). The same procedure was employed for HNE-modified DNA, except that the SpinTips were washed with 0.1% HCO₂H containing 10% CH₃OH (2 × 0.25 mL), instead of H₂O (2 × 0.25 mL). The HNE-DNA samples were reconstituted in 1:1 H₂O:DMSO (20 µL).

Recovery Experiments on 8-HO-PdG and HNE-dG Cross-links in Duplex DNA

The 5′- GCTAGCXAGTCC-3′ • 5′-GGACTCGCTAGC-3′ duplex (X= 8-HO-PdG), which was incubated for 10 days and contained about 50% of the 8-dG-AC-dG cross-link, was used to assess the stability and recovery of the 8-dG-AC-dG cross-link during the enzyme digestion procedure for CT DNA. The duplex was treated with Na(CN)BH₃ to form the reduced dG-(CH₂)₃-dG cross-linked oligomer duplex (vide supra). The recovery of dG-(CH₂)₃-dG cross-link during the enzyme digestion of CT DNA was then determined by LC-ESI/MS/MSⁿ. The 5′-GCTAGCXAGTCC-3′ • 5′-GGACTCGCTAGC-3′ duplex (2.9 pmol, X= 8-HO-PdG), pre-reduced with Na(CN)BH₃ (vide supra), 29 pmol of [²H₆]-dG-(CH₂)₃-dG, 29 pmol [¹⁵N₅]-8-HO-PdG, and 29 pmol of [¹⁵N₅]-8-HO-PdG_red were added to CT DNA (25 µg, 75 nmol DNA bases, to achieve a final level of 3.9 dG-AC-dG cross-linked adducts per 10⁵ bases, assuming 100% of adduct is cross-linked in the duplex DNA. The samples were digested, treated with NaBH₄, and processed by SPE as described above for the AC-modified CT DNA.

Similarly, the same duplex 5′-GCTAGCXAGTCC-3′ • 5′-GGACTCGCTAGC-3′ duplex containing the (6S,8R,11S)-HNE-dG was incubated for 28 days and level of interstrand crosslink was estimated to be 50%, by CGE with UV detection. The cross-linked duplex (15.1 or 151 fmol) and [²H₁₁]-8-dG-HNE-dG standard (15.1 fmol) were added to CT DNA (50 µg DNA, 151 nmol DNA bases, to achieve a level of 1 or 10 8-dG-HNE-dG cross-linked adducts per 10⁷ bases, assuming 100% of adduct is cross-linked in the duplex DNA. The samples were digested and processed by SPE as previously described for the HNE-modified DNA. The estimates of both the AC and HNE cross-linked adducts determined by LC-ESI/MS/MSⁿ ranged between 40–60% of the amount of duplex added to CT DNA (Table 2), which is consistent with the estimate of crosslink formation (50%) observed by gel electrophoresis. Thus, the stable isotope dilution method adequately compensates for partial reversion of the cross-linked adducts back to the monomers, during the chemical reduction and/or enzyme digestion procedures.

Table 2.

Estimates of Levels of interstrand 8-dG-AC-dG (9a) and 8-dG-HNE-dG (10c) Cross-links Recovered from DNA following Enzyme Digestion.

aLevel of spiking 8-dG-AC- dG cross-link in CT DNA	dG-(CH2)3-dG cross-link measured	8-HO-PdGred level measured
3.9 adducts per 105 bases	2.1 ± 0.04 per 105 bases	2.2 ± 0.06 per 105 bases

aLevel of spiking 8-dG-HNE-dG cross-link in CT DNA	8-dG-HNE-dG cross-link level measured
1.0 adduct per 107 bases	6.2 ± 2.6 per 108 bases
1.0 adduct per 106 bases	3.6 ± 1.0 per 107 bases

^aWe assume that only the crosslinked adducts is present in the oligonucleotide duplex N = 4 independent measurements

S1 nuclease treated CT DNA

CT DNA (25 mg, 2 mg/mL, A₂₆₀/A₂₈₀ 1.85) in sodium acetate buffer (50 mM, pH 4.5) with NaCl (0.28 M) and ZnSO₄ (4.5 mM) was treated with S1 Nuclease (10 000 U, Promega) for 30 min at 37°C. The reaction was quenched with EDTA (500µL, 100 mM) and precipitated with ice-cold ethanol (30 mL). The solution was centrifuged at 7000g for 10 min. The pellet was wash with 70% and 100% ethanol, and dried in a centrifugal evaporator for 5 min. The CT DNA was dissolved in sodium phosphate buffer (5 mL, 50 mM, pH 7.0) and added ProCipitate™ solution (2.5 mL, Biotech Support Group, LLC) to remove the protein. The suspension was gently mixed by inversion for 5 min at room temperature and then the sample was centrifuged at 3000g for 15 min. The supernatant was removed and NaCl solution (300 µL, 5M) added. The CT DNA precipitated with ice-cold ethanol (10 mL), and centrifuged at 7000g for 10 min. The CT DNA was dissolved in sodium phosphate buffer (5 mL, 50 mM, pH 7.0) with NaCl (300 mM), precipitated with ice-cold ethanol (10 mL), and centrifuged at 7000g for 10 min. The pellet was wash with 70% and 100% ethanol, and dried in a centrifugal evaporator for 5 min. The CT DNA was dissolved in water, passed through a 30 cm Sephadex G25 column eluting with water, and lyophilized to dryness (15 mg by A₂₆₀). A sample was dissolved in sodium phosphate buffer (50 mM, pH 7.0) and the A₂₆₀/A₂₈₀ value was 1.90.

LC/MS Parameters

Chromatography was performed with an Agilent 1100 Series capillary LC system (Agilent Technologies, Palo Alto, CA) equipped with an Aquasil C18 column (0.32 × 250 mm) from Thermo Fisher (Bellafonte, PA). Samples (2 or 6 µL) were injected, and adducts were separated with the following gradients. For AC-DNA adducts, the solvent conditions were held at 100% A (solvent composition: 0.01% HCO₂H) for 1 min, followed by a linear gradient to 30% B (solvent composition: 95% CH₃CN containing 0.01% HCO₂H) over 40 min at a flow rate of 6 µL/min. For HNE-modified CT DNA adducts, the solvent conditions commenced at 90% A (solvent composition: 0.01% HCO₂H) and 10% B (95% CH₃CN containing 0.01% HCO₂H) and a linear gradient was conducted and arrived at 100% B at 30 min. The mass spectrometer (MS) was a linear quadrupole ion trap LTQ MS (ThermoElectron, San Jose, CA), and Xcalibur version 2.07 software was used for data manipulations. Analyses were conducted in the positive ionization mode and employed an Advance nanospray source from Michrom Bioresource Inc. (Auburn, CA).

MS fragmentations are shown in Figure 3. The product ions obtained at the MS² scan stage for all of the monomeric adducts [M+H-116]⁺ were the aglycone adducts [BH₂]⁺, which arise through the loss of deoxyribose from the protonated molecules. The top two most abundant ions produced in the MS³ scan stage, obtained by fragmentation of the [BH₂]⁺ adducts, were used for quantitative measurements of the monomeric adducts. For the reduced dG-(CH₂)₃-dG cross-link, the product ions monitored at the MS³ scan stage were obtained by collision-induced dissociation of the aglycone, following the loss of the two deoxyribose moieties at the MS² scan stage [M+H]⁺ → [M+H–232]⁺. The 8-dG-HNE-dG cross-link was measured at the the MS⁴ scan stage, by monitoring the transitions due to the successive loss of the two deoxyribose moieties at the MS² and MS³ scan stages: [M+H]⁺ → [M+H-116]⁺ → [M+H–232]⁺. The proposed structures of the product ions of the adducts, produced at the MS/MSⁿ scan stages, are presented in Figure 3, and the LC-ESI/MS/MSⁿ parameter settings employed for adduct measurements are summarized in Table 1.

Figure 3.

Proposed structures of principal product ions arising at the MS/MSⁿ scan stages. Mass is parentheses are the ¹⁵N or ²H labeled internal standards. The sites of isotopic incorporation are noted on Scheme 1.

Table 1.

LTQ/MS/MSⁿ Parameter Settings for Analysis of Enal Adducts

MSn	Adduct	m/z	Isolation width (m/z)	CE	Q	Activation time (ms)	Ions Monitored at final MS/MSn scan stage
MS/MS	6-HO-PdG 8-HO-PdG	324.1	5	22	0.35	10
MS/MS3		208.1	2	34	0.35	10	152.1, 190.1
MS/MS	[15N5]-6-HO-PdG [15N5]-8-HO-PdG	329.1	5	22	0.35	10
MS/MS3		213.1	2	34	0.35	10	157.1, 195.1
MS/MS	6-HO-PdGred 8-HO-PdGred	326.1	5	25	0.35	10
MS/MS3		210.1	2	35	0.35	10	164.2, 192.2
MS/MS	[15N5]-6-HO-PdGred [15N5]-8-HO-PdGred	331.1	5	25	0.35	10
MS/MS3		215.1	2	35	0.35	10	169.1, 197.2
MS/MS	PdG	308.1	5	25	0.35	10
MS/MS3		192.1	2	35	0.35	10	164.2, 192.2
MS/MS	[15N5]PdG	315.1	5	25	0.35	10
MS/MS3		197.1	2	35	0.35	10	169.1, 197.1
MS/MS	8-PdG-(CH2)3-dG	575.1	5	30	0.35	10
MS/MS3		343.2	2	32	0.35	10	164.2, 192.1
MS/MS	[2H6]-8-PdG-(CH2)3-dG	581.1	5	30	0.35	10
MS/MS3		349.2	2	32	0.35	10	166.1, 198.1
MS/MS	HNE-dG	424.2	4	18	0.35	30
MS/MS3		308.1	2	25	0.35	30	152.1, 290.1
MS/MS	[2H11]-HNE-dG	425.2	4	18	0.35	30
MS/MS3		319.2	2	25	0.35	30	152.1, 301.1
MS/MS	dG-HNE-dG	673.1	5	20	0.25	50
MS/MS3		557.1	4	20	0.25	30
MS/MS4		441.1	3	20	0.25	30	152.1, 290.1
MS/MS	[2H11]-dG-HNE-dG	684.2	5	20	0.25	50
MS/MS3		568.1	4	20	0.25	30
MS/MS4		452.3	3	20	0.25	30	152.1, 301.1

Representative optimized instrument tuning parameters were as follows: capillary temperature 185 °C; source spray voltage 1.5 kV and source current 100 µA (HO-PdG’s and dG-(CH₂)₃-dG adducts); source spray voltage 4.0 kV; source current 180 µA (HNE-dG and 8-dG-HNE-dG adducts); no sheath gas, sweep gas or auxiliary gas was employed; capillary voltage 39 V; tube lens voltage 95 V; and in-source fragmentation 10 V. Helium was used as the collision damping gas in the ion trap and was set at a pressure of 1 mTorr. One µscan was used for data acquisition. The automatic gain control settings were full MS target 30,000 and MSⁿ target 10,000, and the maximum injection time was 10 ms. With these MS parameters, about 12 – 15 scans were acquired for each adduct and its corresponding internal standard.

Calibration Curves

Calibration curves were constructed in triplicate for each calibrant level, using CT DNA (50 µg, 151 nmol DNA bases). For the AC-DNA adduct calibration curve, the internal standards of the monomeric adducts (non-reduced and reduced) were added at a level of 1.1 adducts per 10⁵ bases (for CT DNA treated with AC at 0.1 and 0.5 mol AC per mol DNA bases), or 2.2 adducts per 10⁵ bases (for CT DNA treated with AC at 1 and 5 mol AC per mol DNA bases). The unlabeled monomeric adducts were added at the following levels: 0, 0.3, 1.1, 3.3, 6.6, 9.9, 33 or 77 adducts per 10⁵ bases. The [²H₆]-dG-(CH₂)₃-dG was added to the DNA at a level of 1.2 adducts per 10⁶ bases, and the unlabeled dG-(CH₂)₃-dG was added at the following levels: 0, 0.2, 0.6, 0.9, 3.8, 5.7, 19, or 44 adducts per 10⁶ bases. The DNA adducts and internal standards were added to the DNA prior to enzyme digestion and SPE. For estimates of AC adducts in untreated CT-DNA, the HO-PdG internal standards were added at a level of 2.0 adducts per 10⁶ bases, the reduced internal standards were added at a level of 1.0 adducts per 10⁶ bases, and the amount [²H₆]-dG-(CH₂)₃-dG remained at a level of 1.2 adducts per 10⁶ bases.

For the HNE-dG calibration curves, CT DNA (50 µg, 151 nmol DNA bases) was spiked with [²H₁₁]-HNE-dG (15.1 pmol, or 1 adduct per 10⁷ bases, mixture of four stereoisomers), and unlabeled HNE-dG was added at levels of 0, 0.5, 3, 10, 30, 100 or 300 adducts per 10⁷ bases. The [²H₁₁]-8-dG-HNE-dG was spiked into the same DNA sample at a level of 1 adduct per 10⁷ bases and the unlabeled 8-dG-HNE-dG was added at levels of 0, 0.05, 0.1, 0.3, 0.6, 1 or 3 adducts per 10⁷ bases. Because of the instability of 8-dG-HNE-dG, the DNA adducts and internal standards were added to the DNA, following enzyme digestion and were processed by SPE; this procedure allowed us to account for potential ion suppression matrix effects.

The calibration curves were fitted to a straight line (area of response of the adduct/area of response of the internal standard versus the amount of adduct/amount of internal standard), using ordinary least-squares with equal weightings and forcing the Y intercept through the origin. The calibration curve of HNE-dG was non-linear, and the response of the signal of the adduct approached saturation at the highest levels. Therefore, a quadratic equation was used to fit this calibration curve. The coefficient of determination (r²) values of the slopes of calibration curves exceeded 0.98. Representative calibration curves are shown in Supporting Information, Figure S-1.

Results

Synthesis of the Isotopically-Labeled Internal Standards

Isotopically labeled standards were synthesized according to Scheme 1. [¹⁵N₅]-dG was reacted with AC to afford a mixture of the [¹⁵N₅]-8-HO-PdG and [¹⁵N₅]-6-HO-PdG regioisomers, which were separated by HPLC (Scheme 1A) (32). Reduction of [¹⁵N₅]-8-HO-PdG with NaBH₄ provided the [¹⁵N₅]-N²-(3-hydroxypropyl)-dG (HO-PdG_red). [¹⁵N₅]-PdG was synthesized according to a literature procedure from [¹⁵N₅]-dG and 1,3-dibromopropane (Scheme 1B) (34). The reduced AC cross-linked adduct, [²H₆]-dG-(CH₂)₃-dG (10a) was synthesized by the reaction of [²H₆]-1,3-diaminopropane with excess equivalents of O⁶-(trimethylsilyl)ethyl)-2-fluoro-2′-deoxyinosine (Scheme 1C). [²H₁₁]-HNE was synthesized from [²H₁₁]-1-bromopentane and fumaraldehyde bis(dimethylacetal) as previously described (30, 35, 36). [²H₁₁]-HNE was then reacted with dG using arginine to promote the addition reaction to afford [²H₁₁]-HNE-dG as a mixture of four stereoisomers (37–39). The [²H₁₁]-HNE-dG adduct mixture was reacted with dG to afford the [²H₁₁]-HNE-dG pyrimidopurinone cross-link standard (9c, 8-dG-HNE-dG) as a mixture of stereoisomers (Scheme 1D) (21). The isotopic purity of the labeled standards was > 99.5%.

LC-ESI/MS/MSⁿ Characterization of AC and HNE DNA Adducts with a Linear Quadrupole Ion Trap MS (LIT MS)

DNA adducts of AC and HNE were characterized and quantified by LC-ESI/MS/MSⁿ with a LIT MS. The 8-HO-PdG (1), 6-HO-PdG (2), HNE-dG (5–8), the unreduced 8-dG-AC-dG (9a) and reduced dG-(CH₂)₃-dG (10a) cross-links were analyzed at the MS³ scan stage, whereas the unreduced 8-dG-HNE-dG cross-link (9c) (Figure 1 and Figure 2) was assayed at the MS⁴ scan stage. The structures of the proposed principal product ions of these enal adducts, produced at the different MS/MSⁿ scan stages, are depicted in Figure 3, and the MS parameters employed for the analyses of the adducts are summarized in Table 1.

As was previously noted (14), the use of adenine deaminase was critical for the successful analysis of 6-HO-PdG and 8-HO-PdG. The response of the signals for both HO-PdG adducts was very poor, probably due to co-elution with dA or other components, when this enzyme was omitted from the digestion reaction. The response of the signal for the 8-dG-AC-dG (9a) cross-link was weak under the LC-ESI/MS conditions employed, and the adduct underwent extensive deribosylation in the MS source. Because of the poor response 8-dG-AC-dG cross-link combined with its reversion back to 8-HO-PdG during the enzyme digestion of CT DNA, we reduced the cross-link prior to enzyme digestion. The chemically reduced dG-(CH₂)₃-dG (10a) cross-link was stable towards the enzyme digestion procedure (unpublished observations). Moreover, the response of the signal of the reduced dG-(CH₂)₃-dG (10a) cross-link, by LC-ESI/MS, was at least 5-fold greater than the response of the signal of 8-dG-AC-dG (9a).

The 8-dG-HNE-dG cross-link (9c) was resistant to chemical reduction. The isolation widths at all MSⁿ scan stages were increased to maximize the level of sensitivity for the unreduced 8-dG-HNE-dG cross-link (9c), cross-link; however, the wide isolation widths employed resulted in the spillover of some isobaric interference at the MS³ scan stage. Therefore, a MS⁴ scan stage was employed to filter out this interference. The high dissociation efficiency of the LTQ MS permitted scanning of the 8-dG-HNE-dG cross-link at the MS⁴ scan stage with little loss of total ion counts across the MSⁿ scan stages (unpublished observations).

Estimates of Interstrand dG-(CH₂)₃-dG (10a) and 8-dG-HNE-dG (9c) Cross-links in Oligonucleotide Duplexes Containing 5′-CpG-3′ Sequences

The 8-HO-PdG and (6S,8R,11S)-HNE-dG adducts were site-specifically incorporated into 12-mer oligonucleotides as previously described (19,21). The modified oligonucleotides, in which the adduct was situated in a 5′-CpG-3′ local sequence context, were hybridized to its complementary strand and incubated to allow interstrand cross-link formation. The 8-dG-AC-dG cross-link reached ~50% after 10 days, whereas the 8-dG-HNE-dG cross-link reached ~50% after 28 days as judged by CGE analysis with UV detection. These cross-linked oligonucleotide duplexes were utilized to optimize enzymatic digestion conditions for recovery of the intact cross-links. The oligonucleotide duplexes were diluted with unmodified CT DNA (3.8 adducts per 10⁵ bases for 8-dG-AC-dG cross-link and 1 adduct per 10⁶ or 1 adduct per 10⁷ bases for the 8-dG-HNE-dG cross-link), and then digested with phosphodiesterase I, phosphodiesterase II, nuclease P1, and adenosine deaminase (for AC-modified DNA) (vide supra). The AC-dG containing duplex was treated with Na(CN)BH₃ prior to enzyme digestion to reduce the dG-AC-dG cross-link (9a) to the reduced dG-(CH₂)₃-dG cross-link (10a). The Na(CN)BH₃ treatment also partially reduced the 8-HO-PdG to the N²-(3-hydroxypropyl)-dG (8-HO-PdG_red). After DNA digestion, NaBH₄ was added to convert any remaining 8-HO-PdG to 8-HO-PdG_red to simplify the analysis. Reduction of the 8-dG-HNE-dG cross-link with NaB(CN)H₃ or NaBH₄ was slow and incomplete; therefore analysis of the 8-dG-HNE-dG was performed on the unreduced pyrimidopurinone (9c). The reduced dG-(CH₂)₃-dG and unreduced 8-dG-HNE-dG cross-links in the spiked CT DNA were found to be ~50% of the total adduct level (Table 2) as determined by the isotope dilution method and LC-ESI/MS/MSⁿ with the LIT MS. Thus, the estimates of AC and HNE cross-linked adducts obtained by quantitative LC/MS methods are similar to the levels of cross-linked adducts characterized by HPLC and CGE with UV detection.

Time Course for the Formation of 6-HO-PdG, 8-HO-PdG, the Reduced dG-(CH₂)₃-dG (10a) Cross-link, HNE-dG and the Unreduced 8-dG-HNE-dG (9c) Cross-link in CT DNA Treated with AC or HNE

CT DNA was treated with 0, 0.1, 0.5, 1, and 5 mol AC per mol base in CT DNA, and the formation of the dG adducts, including the cross-links, was monitored over 10 days. The formation of the HNE-dG adduct and its cross-link was examined by treating CT DNA with 2 mol HNE per mol DNA base for a period of time up to 21 days. The data are summarized in Figure 4. In untreated CT DNA, the amount of 6-HO-PdG and 8-HO-PdG was estimated at a level of 3.0 ± 0.5 and 4.9 ± 1.0 adducts per 10⁷ DNA bases (mean ± SD, N = 3), respectively. After the Na(CN)BH₃ reduction, digestion, and the NaBH₄ reduction protocol, which also reduces the 6-HO-PdG adduct to PdG, the amounts of PdG and 8-HO-PdG_red due to background levels of 6-HO-PdG and 8-HO-PdG adducts in CT DNA were estimated to be 3.1 ± 0.8 and 5.3 ± 0.5 adducts per 10⁷ DNA bases (mean ± SD, N = 3), respectively. The background level of the HNE-dG adduct in untreated CT DNA was below the limit of quantification (LOQ) (~3 adducts per 10⁹ bases).

Figure 4.

Time course of 6-HO-PdG, 8-HO-PdG, 8-dG-AC-dG (measured as the reduced, 8-dG-(CH₂)₃-dG cross-link), HNE-dG and 8-dG-HNE-dG cross-link formation in CT DNA. Panel A: 1 mol AC reacted per mol base in CT DNA over a time period of 0, 1, 2, 3, 6 and 10 days; Panel B: Levels of 6-HO-PdG, 8-HO-PdG and 8-dG-AC-dG (measured as the reduced, 8-dG-(CH₂)₃-dG cross-link) formed in CT DNA treated with 0.1, 0.5, 1 or 5 mol AC per mol base in CT DNA for 10 days; and Panel C: Time course of HNE-dG and 8-dG-HNE-dG cross-link formation in CT DNA treated with 2 mol HNE per mol base in CT DNA over a time period up to 21 days.The upper panels depict the monomeric DNA adducts and lower panels depict the cross-linked DNA adducts.

The level of 8-HO-PdG formed was about 5 to 10-fold greater than the amount of 6-HO-PdG formed in AC-treated CT DNA at all doses (Figure 4A and 4B). As previously noted (7), the reactivity of AC with CT DNA was considerably greater than HNE (Figure 4C). On a per mol basis (mol enal per mol base in CT DNA), the extent of 8-HO-PdG adduct formation was ~20-fold or greater than the extent of HNE-dG adduct formation. The amounts of the 8-dG-AC-dG (9a) adduct, measured as the reduced dG-(CH₂)₃-dG (10a) cross-link, and the unreduced 8-dG-HNE-dG (9c) cross-link were low and accounted for ~1% of the total adducts of each enal. The 8-dG-AC-dG (9a) cross-link formation reached its maximum level between days 7 and 10, whereas the level of the 8-dG-HNE-dG (9c) cross-link continued to increase during the 3-week incubation period (Figure 4A and 4C). The formation of both the 8-dG-AC-dG (9a) and the 8-dG-HNE-dG (9c) cross-links continued to increase during the incubation period (Figure 4A and 4C). The percentage of the reduced 8-dG-AC-dG (9a) cross-link to total AC-DNA adducts (8-HO-PdG_red and PdG) formed was relatively constant over all dose treatments of AC, at the 10-day time point (Figure 4B).

Representative reconstructed ion chromatograms of the LC-ESI/MS/MS³ traces for the adducts assayed from untreated CT DNA, with and without reduction with Na(CN)BH₃ and NaBH₄, and AC-treated CT DNA (1 mol AC per mol base in CT DNA), followed by reduction, are presented in Figure 5. The reconstructed ion chromatograms of the LC-ESI/MS/MS³ trace for HNE-dG adducts and the reconstructed ion chromatograms for the LC-ESI/MS/MS⁴ trace for the 8-dG-HNE-dG cross-link are shown in Figure 6. The cross-linked adducts are readily observed in the AC- and HNE-treated CT DNA samples.

Figure 5.

A reconstructed ion chromatogram of the LC-ESI/MS/MS³ traces of CT-DNA treated with AC. Panel A depicts untreated CT DNA; Panel B depicts untreated CT DNA chemically reduced with Na(CN)BH₃ and NaBH₄; Panel C depicts CT DNA treated with 1 mol AC per mol base in CT DNA for 10 days, followed by chemical reduction. For untreated DNA, the level of spiking with non-reduced monomer internal standards was 2.0 adducts per 10⁶ bases, reduced monomer internal standards were added at a level of 1.0 adducts per 10⁶ bases, and the spiking level with [²H₆]-dG-(CH₂)₃-dG was 1.2 adducts per 10⁶ bases. The level of spiking with monomeric internal standards was 2.2 adducts per 10⁵ bases and the spiking level with [²H₆]-dG-(CH₂)₃-dG was 1.2 adducts per 10⁶ bases for the AC-treated CT DNA.

Figure 6.

Reconstructed ion chromatograms of the LC-ESI/MS/MSⁿ traces of CT-DNA treated with HNE. A reconstructed ion chromatogram of the LC-ESI/MS/MS³ trace for HNE-dG and the reconstructed ion chromatogram of the LC-ESI/MS/MS⁴ trace for the 8-dG-HNE-dG cross-link. Both [²H₁₁]-HNE-dG and [²H₁₁]-8-dG-HNE-dG cross-link were added to CT DNA at a level of 1 adduct per 10⁷ bases. Panel A depicts untreated CT DNA and Panel B depicts CT DNA treated with 2 mol of HNE per mol base in CT DNA.

There are conflicting reports regarding the usage of the ratio of the UV absorbance at 260:280 nm (A₂₆₀/A₂₈₀) to assess CT DNA purity. The A₂₆₀/A₂₈₀ of the CT DNA used in this study was 1.87, which according to one estimate indicates ~50% protein content (40). However, other reports state that an A₂₆₀/A₂₈₀ ratio > 1.80 is regarded as highly pure DNA, i.e. protein-free (41–43). Enals react readily with nucleophilic sidechains of lysine, cysteine, and histidine residues and the formation of protein adducts and therefore could influence the adduct level observed in our study. Moreover, the 8-HO-PdG and HNE-dG adducts readily form cross-links with peptides (44). It is possible that the formation of DNA-protein cross-links is more favorable than DNA cross-links, resulting in lower than expected levels for the 8-dG-AC-dG and 8-dG-HNE-dG cross-links. Of course, the formation of protein adducts and DNA-protein cross-links of AC and HNE are expected to occur in cells. We were also concerned that single-stranded regions of CT DNA could influence the results since they might be expected to be more reactive toward AC and HNE, and adducts in single-stranded regions would be unable to form interstrand crosslinks.

To address these concerns, the CT DNA was treated with S1 nuclease to remove single stranded regions then precipitated. The S1 nuclease and any other residual proteins were removed by an ion-pairing interaction to an insoluble, cross-linked poly-carboxylic acid polymer (ProCipitate™), followed by centrifugation. The CT DNA was precipitated from the supernatant solution, re-dissolved and precipitated again before passing through a Sephadex G-25 ion-exchange column. The A₂₆₀/A₂₈₀ ratio of this purified CT DNA was 1.90. The modification of the purified CT DNA with AC was repeated. After 2 and 10 days, the CT DNA was treated with Na(CN)BH₃/NaBH₄ and digested as described earlier (Table 3). A 2-fold increase in the level of the reduced dG-(CH₂)₃-dG cross-link (2% versus 1%) relative to the 8-HO-PdG_red adduct formed was observed from the CT DNA that was purified by S1 nuclease and ProCipitate™ treatment versus untreated CT DNA. There was a modest decrease (~10%) in the amount of 8-HO-PdG_red formed from the purified CT DNA. However, the ratio of the 8-HO-PdG_red and PdG increased by ~5-fold from the purified CT DNA and was attributed to a 5-fold reduction in the amount of 6-HO-PdG formed.

Table 3.

AC-DNA Adduct Formation in Untreated CT DNA versus CT DNA that was “Purified” by Treatment with S1 Nuclease, followed by ProCipitate™.

Sample	Days	8-HO-PdGred× 10−5	dG-(CH2)3-dG × 10−7	PdG × 10−6
CT DNA	2	3.9 ± 0.2*	3.6 ± 0.6**	14.8 ± 0.9**
Purified- CT DNAa	2	3.5 ± 0.1	7.6 ± 0.9	2.6 ±0.2
CT DNA	10	11.3 ± 1.3*	11.4 ± 1.3**	33.7 ± 4.5**
Purified-CT DNAa	10	9.7 ± 1.6	21.9 ± 3.9	7.3 ± 0.9

N = 6 independent DNA binding experiments with AC

Unpaired t test (2-tailed)

^*P < 0.014: AC adduct formation from CT DNA versus Purified CT DNA

^**P < 0.001: AC adduct formation from CT DNA versus Purified DNA

Discussion

A variety of bifunctional chemicals have been demonstrated to cause interstrand DNA crosslinks; the interstrand cross-links make up a small fraction of the total DNA adducts formed (typically 1–5%), with the majority of the lesions being the corresponding uncross-linked, monomeric adducts, followed by intrastrand cross-links (23–25). Additionally, these lesions can also form DNA-protein cross-links (45–47). The relatively small contribution of interstrand cross-linked adducts to the total adduct level is likely due to spatial reasons, as many adducts cannot conform to the geometrical requirements for interstrand cross-link formation. Nonetheless, the cellular response to such bifunctional agents is usually ascribed to the minor interstrand DNA cross-link. It has been estimated that as few as 20–40 interstrand cross-links can be lethal to mammalian cells, while a single interstrand cross-link may be sufficient to kill bacteria and yeast cells (23, 24, 26). Thus, biologically relevant levels of interstrand cross-links may be below the limit of detection by current analytical methods.

NMR studies of the 8-HO-PdG adducts, the major DNA adduct of AC, demonstrated that this adduct can undergo a ring opening reaction in duplex DNA when paired opposite dC, which is presumably driven by the potential for Watson-Crick pairing (48). The ring-opened N²-(3-oxopropyl)-dG adduct possesses a reactive aldehyde group in the DNA minor groove, which can react with other nucleophilic groups to form DNA cross-links (Scheme 3). We demonstrated that oligonucleotides containing dG adducts of AC, crotonaldehyde, and HNE form interstrand DNA cross-links when site-specifically incorporated in a 5′-CpG-3′ sequence context (5, 22). The cross-linking chemistry of the AC, crotonaldehyde, and HNE was characterize by NMR spectroscopy and revealed that the chemical nature of the cross-link is a carbinolamine (22). However, the AC and crotonaldehyde cross-links can be reduced with Na(CN)BH₃ (19, 20), suggesting that the carbinolamines are in equilibrium with low concentrations of the imines, which are below the detection limit of NMR spectroscopy. We observed that interstrand crosslink formation was highly dependent on the stereochemistry of the crotonaldehyde-dG and HNE-dG adducts; the (6R,8S)-crotonaldehyde-dG and (6S,8R,11S)-HNE-dG adducts formed interstrand cross-links when incorporated in a 5′-CpG-3′ sequence context, while other stereoisomers formed cross-links at significantly lower levels or not at all (19, 21). NMR analysis suggested that the adduct stereochemistry was critical for orienting the reactive aldehyde toward the N²-amino group of the dG that participated in the cross-linking reaction (49–52).

Scheme 3.

Cross-linking chemistry of enal adducts of dG

Our findings show that the amounts of 8-dG-AC-dG and 8-dG-HNE-dG cross-links formed in CT DNA occur at levels on the order of 1–2% that of the monomeric adducts. While analyses of AC or HNE treated CT DNA demonstrated that these bis-electrophiles induce DNA cross-links, the structural nature of the cross-link – interstrand versus intrastrand – is lost once the DNA has undergone enzymatic digestion. Our previous studies showed that the (6S,8R,11S)-HNE-dG (8) stereoisomers forms interstrand DNA cross-links in a 5′-CpG-3′ sequence contexts, while no evidence of cross-links was observed for the other three stereoisomers (5–7). Only one 8-dG-HNE-dG cross-link was observed when the enzyme digested oligonucleotide duplex containing the (6S,8R,11S)-HNE-dG adduct was analyzed by LC-ESI/MS/MSⁿ (Supplementary Information, Figure S-2). Interestingly, at least two isomers of the 8-dG-HNE-dG cross-link were detected in HNE-treated CT DNA (Figure 6). This observation suggests that both interstrand and intrastrand cross-linked adducts are formed in CT DNA, and both types of crosslinks also likely to occur in AC-treated CT DNA as well (53).

Further purification of the CT DNA prior to AC resulted in a modest increase in cross-links formation (2% versus 1%). The purification involved treatment of the CT DNA with S1nuclease to remove single-stranded regions, removal of any residual proteins with ProCipitate™, and ion-exchange chromatography. Interestingly, the ratio of 8-HO-PdG and 6-HO-PdG adducts (observed as 8-HO-PdG_red and PdG, respectively) increased by ~5-fold for the modification of the purified CT DNA. The formation of 8-HO-PdG and 6-HO-PdG involves initial Michael adduct of either the N²- or N1-position of dG to the β-carbon of AC. Therefore, the reactivity of AC with double-stranded DNA is expected to favor the 8-HO-PdG adduct since N²-position is accessible from the minor groove. These observations are consistent with the S1 nuclease treatment removing single-stranded regions of the CT DNA. The removal of any residual proteins from CT DNA with ProCipitate™ appeared to have marginal effects on the levels of AC adduct formation. This study was not repeated with HNE since the increase in AC cross-link formation was relatively small (2-fold) and HNE does not form the 6-HO regioisomer.

A goal of this study was to establish a lower limit of detection of the AC and HNE cross-links relative to the level of uncross-linked 8-HO-PdG and HNE-dG adducts, which are formed in mammalian cells. For example, high levels of both the 6-HO-PdG and 8-HO-PdG have been detected in DNA from brain tissues of Alzheimer’s disease patients (54, 55). The levels of 6-HO-PdG and 8-HO-PdG ranged from 2,800 to 5,100 adducts per 10⁹ bases. Both regioisomeric AC adducts were also found in the lung tissue from smokers at levels ranging from 10 to 160 HO-PdG adducts per 10⁹ bases (14). The average level of the 6-HO-PdG isomers was slightly higher than the 8-HO-PdG adduct in the 30 human lung samples examined. This is an interesting observation since formation of the 8-HO-PdG adduct is predicted to be in duplex DNA and may reflect differential repair of the between the two AC adducts. The 6- and 8-HO-PdG adducts were also found in human placental and leukocyte DNA at levels of 108 and 78 adducts per 10⁸ bases (56); the regioisomers were integrated together in these analyses. The HNE-dG adduct was also identified in DNA from brain tissue of Alzheimer’s disease patients at levels ranging from 190 to 1260 adducts per 10⁹ bases (57). The levels of these enal adducts in brain tissue are high in comparison to the adduct levels measured in lung tissue. If appropriate measures are not employed during the isolation of DNA from tissues, artifactual enal DNA adduct formation can occur and result in artificially high levels of DNA adducts (58). However, both AC and HNE have been reported to be elevated in brain regions of subjects with mild cognitive impairment and late-stage Alzheimer disease, suggesting that enal-DNA adduct formation in the brain may well be greater than in other organs (12). If the proportion of AC and HNE cross-linked dG-adducts formed relative to the monomeric adducts in enal-modified CT DNA are similar to the proportion that occurs in DNA in cells, and depending on the rate that the cross-linked adducts are removed by repair pathway (26–28), we could expect to detect the 8-dG-AC-dG cross-link adduct at levels ranging from 2–200 adducts per 10⁹ bases in brain tissue of Alzheimer’s patients noted above. The levels of the 8-dG-HNE-dG cross-link may be present at levels of several adducts per 10⁹ bases, which is close to the limit of current detection of MS. Indeed, background levels of a cross-linked adduct derived from the reaction of endogenously produced formaldehyde with the N⁶ atoms of two dA molecules, have been measured at levels of several adducts per 10⁹ DNA bases in rat liver (59). Thus, it may be possible to detect these enal cross-links from brain tissues of Alzheimer patients or tissue from laboratory animals that are deficient in essential cross-links repair factors (60).

In conclusion, we have developed a mass spectrometric method for the detection and quantitation of dG cross-links of AC and HNE in CT DNA. Our current objective is to determine if either one of these DNA cross-links is formed in vivo.

Abbreviations

AC

acrolein

4-HNE

trans-4-hydroxy-2-nonenal

CT DNA

calf thymus DNA

dG

2'-deoxyguanosine

dA

2'-deoxyadenosine

dN

2'-deoxynucleoside

6-HO-PdG

3-(2'-deoxy-β-D-erythro-pentofuranosyl)-5,6,7,8-tetrahydro-6-hydroxypyrimido[1,2-a]purin-10(3H)-one

8-HO-PdG

3-(2'-deoxy-β-D-erythro-pentofuranosyl)-5,6,7,8-tetrahydro-8-hydroxypyrimido[1,2-a]purin-10(3H)-one

dG-(CH₂)₃-dG

N',N''-1,3-propanediylbis-(2′-deoxyguanosine)

8-dG-AC-dG

N-[3-(2-deoxy-β-D-erythro-pentofuranosyl)-3,5,6,7,8,10-hexahydro-10-oxopyrimido[1,2-a]purin-8-yl]-2′-deoxyguanosine

6-HO-PdG_red

N1-(3-hydroxypropyl)-2′-deoxyguanosine

PdG

3-(2-deoxy-β-D-erythro-pentofuranosyl)-5,6,7,8-tetrahydro-pyrimido1,2-a]purin-10(3H)-one

8-HO-PdG_red

N ²-(3-hydroxypropyl)-2′-deoxyguanosine

HNE-dG

3-(2-deoxy-β-D-erythro-pentofuranosyl)-5,6,7,8-tetrahydro-8-hydroxy-6-(1-hydroxyhexyl)-pyrimido[1,2-a]purin-10(3H)-one

8-dG-HNE-dG

N[3-(2-deoxy-β-D-erythro-pentofuranosyl)-3,5,6,7,8,10-hexahydro-6-(1-hydroxyhexyl)-10-oxopyrimido[1,2-a]purin-8-yl]-2′-deoxyguanosine

CGE

capillary gel electrophoresis

LOQ

limit of quantification

LC-ESI/MS/MSⁿ

liquid chromatography-electrospray ionization/multi-stagetandem mass spectrometry

LIT MS

linear quadrupole ion trap mass spectrometry;

SPE

solid phase extraction