1. Introduction
Globally, annual alcohol consumption has been estimated to be 6.4 L per capita in 2016, and it is projected to increase to 7 L by 2025 [
1]. Alcohol is classified by the International Agency for Research on Cancer (IARC) as a Group 1 human carcinogen and is estimated to be responsible for 12.6% of overall cancers [
2,
3], a percentage expected to increase following the growth in consumption. Alcohol is mainly metabolized in the body by alcohol dehydrogenases (ADHs), which oxidize ethanol to acetaldehyde, followed by detoxification to acetate by aldehyde dehydrogenases (ALDHs) [
4]. The variant allele ALDH2*2 encodes for an inactive subunit of the enzyme ALDH2 [
4]. Individuals who are heterozygous carriers of this variant, ALDH2*1/*2, have about 10% residual ALDH2 activity and experience side effects like flushing and nausea when ingesting alcohol [
4]. These individuals inefficiently detoxify acetaldehyde and are at higher risk for developing alcohol-related esophageal and head and neck cancers [
4,
5]. These observations contributed to the reclassification of acetaldehyde associated with alcohol consumption as a Group 1 human carcinogen by IARC [
3]. Acetaldehyde reacts with DNA bases to produce adducts, which are critical in the carcinogenic process because they can cause miscoding resulting in mutated genes and loss of normal cellular growth-control mechanisms [
4]. Although ethanol is mainly metabolized in the liver, the concentration of acetaldehyde in saliva after ingesting ethanol is much higher than in the blood, due to the local metabolism in the oral mucosa and the microflora. Therefore, acetaldehyde genotoxicity may play a specific key role in ethanol-induced carcinogenesis in the oral cavity [
4].
The major adduct formed upon reaction of acetaldehyde with DNA is
N2-ethylidenedeoxyguanosine (
N2-ethylidene-dG), which can be analyzed as its more stable reduced version
N2-ethyldeoxyguanosine (
N2-ethyl-dG) after DNA treatment with NaBH
3CN [
4]. Levels of this adduct showed a positive dose-response relationship in oral-cell DNA collected from volunteers who consumed increasing amounts of alcohol [
6]. However, several other adducts as well as DNA–DNA crosslinks have been identified in the reactions of acetaldehyde with DNA and, together with modifications at other nucleobases, may also play a role in acetaldehyde’s chemically induced carcinogenesis and epigenetic signaling [
7,
8,
9]. To better understand the role of these other lesions, this study optimized and implemented our ultrasensitive data-dependent acquisition constant-neutral-loss triggered-MS
3 (DDA-CNL/MS
3) DNA adductomic method to screen for all acetaldehyde-derived DNA adducts and comprehensively characterize acetaldehyde-derived DNA damage. This approach allowed us to profile acetaldehyde-derived DNA modifications and successfully identify and screen for new markers of alcohol exposure and genotoxicity in humans.
2. Materials and Methods
Caution: acetaldehyde may cause cancer. It should be handled in a well-ventilated hood with extreme care and with personal protective equipment.
2.1. Materials and Chemicals
Acetaldehyde and [ethyl-D5]EtNH3Cl were purchased from Millipore Sigma (St. Louis, MO, USA). In addition, 6-Chloropurine-2′-deoxyriboside was obtained from Carbosynth (Compton, UK). Water (LC-MS grade), methanol (MeOH, LC-MS grade), acetonitrile (ACN, LC-MS grade), 2-propanol (IPA, LC-MS grade), and formic acid (FA, 98% v/v) were purchased from Fisher Scientific (Hanover Park, IL, USA). Distilled water was purified by a Milli-Q system (Milford, MA, USA). Deoxyribonuclease I recombinant expressed by Pichia pastoris (R-DNase, 10,000 U/mg, phosphodiesterase-1 extracted from Crotalus adamanteus (PDE-1, 0.4 U/mg, recombinant alkaline phosphatase expressed by Pichia pastoris (R-ALP, 7000 U/mg, calf thymus DNA (CT-DNA, 5 mg), NaBH3CN, acetaldehyde, Tris base, double-filtration membrane Amicon Ultra (30 kDa cutoff, 0.5 mL), and single-filtration membrane Microcone (10 kDa cutoff, 0.5 mL) were purchased from Millipore Sigma (St. Louis, MO, USA). Silanized vials (0.3 mL, 1.2 mL, 4 mL, 20 mL) were purchased from ChromTech (Apple Valley, MN, USA). Cell lysis solution, protein precipitation solution, RNase A, and proteinase K were obtained from Qiagen (Hilden, Germany).
2.2. General Synthetic Procedures
NMR spectra were recorded on a Bruker 500 MHz spectrometer. Chemical shifts are reported as parts per million (ppm). Residual solvent peaks were used as an internal reference for 1H-NMR (7.26 ppm CDCl3; 2.50 ppm D6-DMSO) and 13C-NMR (77.2 ppm CDCl3; 39.5 ppm D6-DMSO). Peak splitting used the following abbreviations: s = singlet, d = doublet, t = triplet, q = quartet, dd = doublet of doublets, dt = doublet of triplets, ddd = doublet of doublet of doublets, bs = broad singlet, and m = multiplet. All compound structures were evaluated and confirmed with 1H, 13C, COSY, HSQC, and HMBC experiments. High-resolution mass spectrometry (HRMS) for selected compounds was performed on an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, Waltham, MA, USA) and reported as m/z. Thin-layer chromatography (TLC) utilized Polygram precoated silica gel TLC plates (40 × 80 mm, 0.2 mm thick) with 254 nm fluorescent indicator. TLC plates were visualized by UV lamp irradiation. Flash chromatography was performed on SiliCycle 60 (70–150) mesh silica gel. Reactions were performed with oven-dried glassware and under an atmosphere of N2, unless specified otherwise.
2.3. Synthesis of 3′, 5′-bis-O-acetyl-2′-deoxyuridine
To a 25 mL, round-bottom flask equipped with a magnetic stir bar, were added 2′-deoxyuridine (154.9 mg, 0.679 mmol), 4-dimethylaminopyridine (5.6 mg, 0.0458 mmol), triethylamine (275.9 mg, 2.73 mmol, 380 µL), and ACN (3.5 mL). The resulting suspension was treated with acetic anhydride (280.8 mg, 2.75 mmol, 260 µL) and stirred at room temperature for 30 min. The reaction was quenched with MeOH (1 mL) and evaporated in vacuo. The resulting oil was reconstituted in CH2Cl2 and washed once with brine. The organic layer was dried over MgSO4, filtered, and evaporated to a crude foam. Purification by flash column chromatography (1:3:1 hexanes/EtOAc/CH2Cl2) provided pure product as a white foam (183.3 mg, 86.5%).
2.4. Synthesis of 4-chloro-1-N-(3′, 5′-bis-O-acetyl-2′-deoxyribosyl)-2-pyrimidinone
To an oven-dried, two-neck, 25 mL flask equipped with a magnetic stir bar were added 3′,5′-bis-O-acetyl-2′-deoxyuridine (293.3 mg, 0.94 mmol) and anhydrous CHCl3 (10 mL). The flask was purged with argon thrice and then SOCl2 (85.3 mg, 7.17 mmol, 520 µL) and DMF (50 µL) were added. The resulting yellow solution was brought to reflux (~75 °C) for 2 h. After this time, the reaction mixture was cooled to room temperature and quenched with NaHCO3 (~5 mL). Once bubbling ceased, the organic layer was collected, and the remaining aqueous layer was extracted once with CH2Cl2 (10 mL). The pooled organics were dried over MgSO4, filtered, and evaporated in vacuo to a yellow oil. Purification by column chromatography (1% → 2% MeOH in CHCl3) delivered pure product as an off-white solid (186.5 mg, 60%).
2.5. Synthesis of N4-ethyldeoxycytidine
4-Chloro-1-N-(3′, 5′-bis-O-acetyl-2′-deoxyribosyl)-2-pyrimidinone (7.4 mg, 0.0224 mmol), K2CO3 (30.8 mg, 0.223 mmol), and ACN (1 mL) were combined in a two-dram vial equipped with a magnetic stir bar. The cloudy suspension was treated with EtNH3Cl (10.1 mg, 0.124 mmol) and heated to 50 °C for 2 h. After cooling to room temperature, the solvent was removed in vacuo. The residue was then reconstituted in MeOH (1 mL) and stirred at 50 °C for an additional 2 h. The mixture was similarly cooled to room temperature and concentrated to dryness in vacuo. The resulting solid was reconstituted in H2O (3 mL) and purified by HPLC (Agilent 1100 Analytical Flow, Agilent Technologies, Palo Alto, CA, USA). Separation was performed using a Luna C18 column (250 × 4.6 mm, 100 A, 40 °C) with a multistep gradient at a flow rate of 1 mL/min using H2O and MeOH as solvents A and B, respectively. Beginning at 2% B, the eluent was brought to 27% B over 15 min. This was followed by a wash at 95% B for 2 min and re-equilibration. Detection was accomplished using UV-Vis (254 nm). The product eluted at 16.2 min and was collected in a glass vial. After evaporation in vacuo, pure product was isolated as a white solid (3.72 mg, 65.1%).
2.6. Synthesis of [D5]N4-ethyldeoxycytidine
This compound was produced analogously to N4-ethyldeoxycytidine (N4-ethyl-dC), except that [ethyl-D5]EtNH3Cl was used. The product was isolated as a white solid (5.61 mg, 64.2%).
2.7. Synthesis of N6-ethyldeoxyadenosine
6-Chloropurine-2′-deoxyribose (11.3 mg, 0.0417 mmol), EtNH3Cl (5.1 mg, 0.0626 mmol), iPr2EtN (13.3 mg, 0.103 mmol, 18 µL) and DMSO (1 mL) were combined in a two-dram vial equipped with a magnetic stir bar. The solution was heated to 50 °C and stirred for 16 h. The solution was then cooled to room temperature and diluted with H2O (1 mL). The product was purified by HPLC (Agilent 1100 Analytical Flow, Agilent Technologies, Palo Alto, CA, USA). Separation was performed using a Luna C18 column (250 × 4.6 mm, 5 A, 25 °C) with a multistep gradient at a flow rate of 1.5 mL/min using H2O and MeOH as solvents A and B, respectively. The eluent was held at 2% B for 10 min and then brought to 50% B over 25 min. This was followed by a wash at 95% B for 4 min and re-equilibration. Detection was accomplished using UV-Vis at 254 nm. The product eluted at 22.4 min and was collected in a glass vial. After evaporation in vacuo, pure product was isolated as a white solid (3.60 mg, 30.5%).
2.8. Synthesis of [D5]N6-ethyldeoxyadenosine
This compound was produced analogously to N6-ethyldeoxyadenosine (N6-ethyl-dA), except that [ethyl-D5]EtNH3Cl was used. The product was isolated as a white solid (3.29 mg, 27.8%).
2.9. DNA Incubation with Acetaldehyde and Stabilization
The reaction of acetaldehyde with exocyclic amino groups of the DNA nucleobases forms unstable Schiff bases, which may be degraded during DNA hydrolysis [
10,
11]. To prevent degradation, the DNA was treated with the reducing agent NaBH
3CN following a previously reported procedure [
12]. Similarly, DNA hydrolysis was carried out as previously reported [
12]. In brief, CT-DNA (1 mg) was incubated with acetaldehyde (5 mmol) in Tris-10 Mm HCl/5 mM MgCl
2 buffer (pH ~7) at 37 °C for 24 h. Subsequently, 30 mg NaBH
3CN was added and the resulting solution incubated at room temperature (RT) overnight. NaBH
3CN and acetaldehyde negatively impact enzyme activity (unpublished data), resulting in low hydrolysis rates. For this reason, three different protocols of NaBH
3CN and acetaldehyde removal were evaluated and the one resulting in the best hydrolysis yield was selected (
Supplementary Information, SI). Treated DNA was precipitated and desalted via addition of cold IPA, washed with 70% IPA and 100% IPA sequentially, dried under a stream of N
2, and stored at −20 °C until analysis. The DNA recovery (~98%) for each sample was assessed by UV-Vis-spectrophotometry.
To evaluate concentration-dependent formation of DNA adducts due to acetaldehyde exposure, CT-DNA (1 mg) was incubated with increasing amounts of acetaldehyde (0, 1, 2.5, 5, 25, and 50 mmol) and processed as above. Acetaldehyde concentrations were selected to follow experimental procedures previously reported [
8,
13].
2.10. DNA Hydrolysis and Quantification
DNA was solubilized in 1 mL of 10 mM Tris-HCl/5 mM MgCl
2 buffer. DNA concentrations were assessed by UV-Vis-spectrophotometry. A total of 250 µg of DNA was hydrolyzed as reported [
11]. The digestion yields were assessed by quantifying dG by UPLC-UV (Ultimate 3000, Thermo Scientific, Waltham, MA, USA). The recovery of DNA adducts was evaluated by adding a mixture of isotopically labeled internal standards (100 fmol of [
15N
5]
N2-ethyl-dG, [
15N
5]
N6-methyl-dA, and [D
4]POB-dT) into the samples [
12].
2.11. Sample Purification and Enrichment
Hydrolyzed DNA samples were purified by HPLC fraction collection (FC). The system consisted of an HPLC (Ultimate 3000, Thermo Scientific, Waltham, MA, USA), equipped with a C18 column (4.6 × 250 mm, 100 Ǻ, 5 µm Luna-Phenomenex, Torrance, CA, USA). Two different FC methods were developed and optimized. The first method was used during the initial screening experiments: the instrument was operated at 25 °C with a multistep gradient using H2O and MeOH as mobile phase A and B, respectively. The eluent was held at 2% B and 0.5 mL/min for 5 min, brought to 1 mL/min in 1 min, then to 15% B in 24 min, to 35% in 5 min, and finally to 100% B in 5 min. This was followed by a wash at 100% B for 5 min and re-equilibration. Detection was accomplished using the UV-Vis detector set at 190 nm and 254 nm. The unmodified nucleobases were collected separately from the other fractions.
The second method was developed once the standards of the characterized adducts were synthesized to obtain the highest recovery of our analytes. The instrument was operated at 25 °C, performing a multistep gradient at a flow rate of 1 mL/min using H2O and MeOH as mobile phase A and B, respectively. The eluent was held at 2% B for 2 min, brought to 12% B in 10 min, then to 15% B in 3 min, to 20% B in 3 min, and finally to 100% B in 3 min. This was followed by a wash at 100% B for 5 min and re-equilibration. As reported above, detection was accomplished using the UV-Vis detector probing two different wavelengths at 190 and 254 nm. Unmodified nucleobases were collected separately from the other fractions. All collected fractions were subsequently dried under reduced pressure and stored at −20 °C until LC-MS analysis.
2.12. Oral-Cell DNA Collected from Volunteers Exposed to Known Amounts of Alcohol
Samples were collected as part of a study conducted at the University of Minnesota. The study was approved by the University of Minnesota Human Research Protection Programs Institutional Review Board. Volunteers were enrolled after signing a consent form and evaluation of the eligibility criteria. Medical history and alcohol-drinking history, both in the past 12 months and lifetime, were obtained through questionnaires. Oral rinse samples collected before and 2 h after alcohol exposure (resulting in a blood alcohol concentration (BAC) of 0.11%) were used to isolate DNA for the analysis of acetaldehyde-derived DNA adducts. Details of the study are reported in the
SI.
2.13. DNA Isolation and Purification from Oral Rinse Samples
A total of 18 oral-rinse samples collected from healthy volunteers, nine before and nine after (2 h) consumption of alcohol, were processed. Samples were centrifuged and the supernatant was removed. The pellet was resuspended in 1 mL of cell lysis solution and treated with proteinase K (24 h, RT), followed by treatment with RNase A (2 h at RT). Proteins were precipitated with 0.3 mL of protein precipitation solution. The supernatant was poured into an equal volume of ice-cold IPA (100%) to precipitate the DNA. Samples were then centrifuged. The supernatant was discarded, and the remaining DNA pellet was washed with 75% and 100% cold IPA. The liquid washes were discarded, and the residual IPA evaporated under a mild N2 stream. Dried samples were stored at −20 °C. The extraction yield was assessed by quantifying the DNA using a UV-Vis-spectrophotometer (BioPhotometer, Eppendorf, Hamburg, Germany). DNA hydrolysis and sample purification and enrichment were performed as reported above.
2.14. LC Conditions for MS Analysis
Adductomic methods were optimized using an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific, Waltham, MA, USA) interfaced to a nanoUPLC (UltiMate 3000 RSLCnano, Thermo Scientific, Waltham, MA, USA) with a NanoFlex ion source (Thermo Scientific, Waltham, MA, USA), operating in positive ionization mode with a voltage of 2.5 kV and an ion tube temperature of 300 °C. The UPLC system was equipped with a 5 µL loop and a reverse-phase column home-packed (silica emitter 230 × 0.075 mm, 15 um orifice, New Objective, Woburn, MA, USA) with C18 stationary phase (5 μm, 100 Ǻ, Luna-Phenomenex, Torrance, CA, USA). The mobile phase consisted of formic acid (0.05% v/v in H2O, phase-A) and ACN (100% v/v, phase-B).
For the untargeted screening, the eluent was held at 2% B for 2 min, brought to 20% B in 24 min, then to 60% B in 10 min, to 98% B in 1 min, and then maintained at 98% for 4 min. This was followed by a wash at 98% B for 4 min and column re-equilibration. For the targeted analysis, the eluent was held at 2% B for 6 min, brought to 35% B in 14 min, then to 98% B in 2 min, and kept at 98% B for 2 min, followed by column re-equilibration.
2.15. DDA-CNL/MS3 Gas-Phase Fractionation Method
Purified DNA extracted from rat liver was available from previous studies and was used as the matrix for our method development. DNA was enzymatically hydrolyzed and purified as reported above. A standard mixture of six isotopically labelled DNA adducts ([
15N
5]
N2-ethyl-dG, [
15N
5]
N6-methyl–dA, [D
4]
O6-POB-dG, [D
4]
O6-POB-dT, [D
4]
O6-PHB-dG, [
15N
5]8-OH-PdG (structures in
SI)) was prepared and spiked in the matrix previously reconstituted in 20 µL of LC-MS H
2O prior to LC-MS analysis.
The MS analysis was performed with Orbitrap detection (resolution of 60,000) in: (i) gas-phase fractionation mode with the mass range of interest split into four scan segments (m/z 197–310, m/z 305–380, m/z 375–450, and m/z 445–750) or (ii) in standard mode with a single scan segment (m/z 197–750). In each partial or full scan, quadrupole filtering was used with a maximum injection time of 200 ms and an automatic gain control (AGC) setting of 5.0 × 104.
For each scan segment, the top five ions were selected for MS
2 fragmentation with quadrupole isolation of 1.5
m/
z, using collision induced dissociation (CID) with a normalized collision energy of 30%, maximum injection time of 200 ms, and Orbitrap detection at a resolution of 30,000. An exclusion list of 95 ions (
SI) with a mass tolerance of 5 ppm was used, as was dynamic exclusion of 30 s and an intensity threshold of 2.0 × 10
3. MS
3 fragmentation was triggered upon observation of the accurate-mass neutral loss of 2′-deoxyribose (-dR: 116.0474 ± 0.0006
m/
z, 5 ppm) upon MS
2 fragmentation. MS
3 fragmentation was performed with high-energy collisional dissociation (HCD) with a normalized collision energy of 50%, maximum injection time of 250 ms, and Orbitrap detection at a resolution of 15,000. This gas-phase fractionation MS method was used for DNA-adduct profiling.
2.16. Targeted Mass Spectrometry-Based Approach
To attain the highest level of sensitivity in order to investigate the presence of the adducts, previously characterized in the in vitro experiment, in human oral-cell DNA, a targeted MS
2 analysis was performed with the parent ion masses listed in
Table 1, and the internal standard ion masses (
m/
z 301.1205,
m/
z 394.1911, and
m/
z 424.2191). Subsequently for the absolute quantitation of ethyl-adducts, a targeted MS
2 analysis was performed with eight parent ion masses (
m/
z 256.1292,
m/
z 261.1605,
m/
z 280.1404,
m/
z 285.1718,
m/
z 296.1353,
m/
z 301.1205,
m/
z 394.1911, and
m/
z 424.2191). The following parameters were set for the analysis: RF lens of 60%, quadrupole isolation window of 1.5
m/
z, HCD of 22%, AGC target of 5 × 10
4, maximum injection time of 50 ms, Orbitrap resolution of 60,000, and EASY-IC enabled. Frozen DNA samples were thawed and reconstituted in 20 µL H
2O and analyzed.
2.17. Method Validation
The ability of the targeted approach to quantify
N2-ethyl-dG,
N6-ethyl-dA, and
N4-ethyl-dC was evaluated, and the method was validated. The limits of detection (LOD) for the quantitation of
N2-ethyl-dG,
N6-ethyl-dA, and
N4-ethyl-dC were established using standard solutions of adducts. The limits of quantitation (LOQ), accuracy, and precision of the method were determined by analyzing CT-DNA spiked with different amounts of
N2-ethyl-dG (0, 2, 6, 10, 40, 100 fmol),
N6-ethyl-dA (0, 0.2, 0.6, 1, 4, 10 fmol), and
N4-ethyl-dC (0, 2, 6, 10, 40, 100 fmol). Each sample was analyzed in triplicate. Background levels of the adducts in CT-DNA were determined by analyzing three nonspiked samples; these amounts were subtracted from the amounts measured in the spiked samples. LODs and LOQs were calculated using the following equations: LOD = (3.3 × sd/S) and LOD = (10 × sd/S), where sd is standard deviation, S is the slope of the calibration curve, and the multipliers (3.3 and 10) are recommended by International Conference on Harmonization standards [
14]. Accuracy was determined by comparing added and measured amounts of the adducts at each level. Precision was determined as intraday coefficients of variation (% CV) for the triplicate samples. Recovery was determined by adding [
15N
5]
N2-ethyl-dG (10 fmol), [D
5]
N6-ethyl-dA (1 fmol), and [D
5]
N4-ethyl-dC (10 fmol) to CT-DNA, processed as described above and compared to CT-DNA samples with analytes added after processing.
2.18. Data Processing and Normalization
Putative DNA adducts were identified from LC-MS3 data using Xcalibur 3.0 (Thermo Scientific, Sunnyvale, CA, USA), where only ions which triggered an MS3 event and were unique or increasing in the exposed samples were considered. For relative quantification of a putative DNA adduct in a specific sample, the area of the full-scan extracted ion chromatogram (EIC) with a mass tolerance of 5 ppm was used, and the intensity was further normalized using the following: 1) amount of dG (µmol) determined in that specific sample and 2) area of the internal standard EIC.
2.19. Statistical Analysis
Statistical analyses were performed using SigmaPlot 12.5 (Systat Software, San Jose, CA, USA,
https://systatsoftware.com/products/sigmaplot/ (accessed on 15 January 2021)). The Student’s t-test was used to compare DNA-adduct levels between baseline and 2 h exposure. Statistical significance was set at
p ≤ 0.05.
4. Discussion
Our work presents a method for comprehensive analysis of acetaldehyde-derived DNA adducts in oral-cell DNA from volunteers exposed to alcohol. The method was used to screen DNA exposed to acetaldehyde using a high-resolution/accurate-mass data-dependent constant-neutral-loss-MS3 (DDA-CNL/MS3) DNA adductomic approach resulting in the identification of 22 acetaldehyde-derived DNA modifications—some of which have not been described before.
This work was done to expand upon previous studies that characterized reactions of acetaldehyde with DNA and focused on the major covalent binding occurring to dG, while only suggesting the occurrence of reactions with dC and dA [
7,
8,
13]. These previous studies were done using less-selective, and therefore often less-sensitive, technologies in comparison to the ones used in this study, limiting the ability to perform the simultaneous identification and quantitation of multiple modifications.
The LOD of the DDA-CNL/MS
3 method is limited either by the ion capacity of the Orbitrap to detect ions in the full-scan acquisition or the rate at which MS
2 spectra can be acquired. Background ion signal in DNA adductomic analyses is many orders of magnitude higher than the DNA-adduct signals which results in low injection times, and thereby limiting the number of analyte ions entering the Orbitrap for detection [
26]. To mitigate this issue, the sensitivity of the standard full-scan DDA-CNL/MS
3 method was enhanced by the implementation of a procedure called “gas-phase fractionation”(GPF), an approach which has been used in the fields of proteomics, metabolomics, and lipidomics to enhance the detection of low levels of analytes of interest [
27,
28,
29]. The use of GPF breaks the full-scan range of interest into multiple small ranges resulting in longer injection times by limiting the mass range of ions sampled in a given full-scan detection event. In this study, a comparison of the performance of the standard method with that of a GPF version was done where the full-scan range was divided into four segments, with overlaps of 5 Da. The GPF version outperformed the standard method as summarized in
Figures S1 and S2, with an average of 2083 (MS
2) and 240 (MS
3) triggered ions for the GPF method, and 1616 (MS
2) and 210 (MS
3) triggered ions for the standard method. Five and two spiked-in DNA adducts triggered MS
2 and MS
3 events, respectively, with the GPF method, compared to one spiked-in DNA adduct for both MS
2 and MS
3 events when the standard method was used. These results demonstrate that GPF can be a useful tool for increasing overall detection coverage of this method; therefore, this technique was used to perform in vitro screening.
The optimized method was used to comprehensively profile acetaldehyde-DNA adducts in CT-DNA exposed to acetaldehyde and treated with NaBH
3CN to reduce and stabilize any Schiff bases which formed. Following a rigorous data analysis (
Figure 1), 22 DNA adducts were detected and the high-quality spectra and the accurate masses obtained allowed for the assignments of chemical formulas and putative structures, including many dA, dC, dT, and dG adducts (
Table 1 and
SI). Several of the assigned DNA adducts were previously described, including
N2-ethyl-dG (
m/
z 296.1357) and 1,
N2-propano-dG (
m/
z 338.1459) which were detected in vivo [
6,
30,
31,
32]. Additionally, Wang and coworkers first demonstrated the generation of
N2-dimethyldioxane-dG (
m/
z 382.1727) and of the crosslink
m/
z 587.2324 [
8]. The crosslink
m/
z 589.2420 has also been reported as reduction product of
m/
z 587.2324 [
9,
33,
34,
35]. Our DNA adductomic screening analysis identified new DNA adducts including crosslinks involving dC and monoadducts involving dA, dT, and dC.
Acetaldehyde is ubiquitous and is a product of physiological processes; therefore, acetaldehyde-derived DNA adducts can be detected in samples not exposed to exogenous sources of this genotoxic compound. To characterize the role that exogenous acetaldehyde has on the formation of the 22 observed DNA adducts, the DNA-adduct ion signals were measured in CT-DNA exposed to increasing amounts of acetaldehyde (0, 1, 2.5, 5, 25, and 50 mmol). All adducts showed higher ion intensities as the dose of acetaldehyde augmented.
The 22 adducts were included in a more sensitive, targeted method for the investigation of acetaldehyde-derived DNA damage in human samples. Among the DNA adducts detected, those resulting from the reduction of Schiff bases formed upon reaction of acetaldehyde with the
N2 position of dG,
N6 position of dA, and
N4 of dC corresponded to the most intense signals. The
N2-ethyl-dG-attributed ion signal was the highest followed by those of the adducts of dA and dC, in accordance with the observations by Vaca et al., which reported a reactivity order of dG > dA > dC [
7]. The identities of these adducts were confirmed by comparison with synthetic standards. The
N2-ethyl-dG synthetic standard was readily available [
6], while
N6-ethyl-dA and
N4-ethyl-dC were synthesized. Isotopically labelled versions of the compounds were synthetized as well for quantitation using an isotope dilution method. Accuracy, precision, and limit of detection were determined for this newly developed quantitative method for
N2-ethyl-dG,
N6-ethyl-dA, and
N4-ethyl-dC, which was combined with the targeted detection of the other 19 adducts identified in the initial screening.
The method was tested to investigate alcohol-derived DNA adducts in oral-cell DNA isolated from oral rinses collected from healthy volunteers before and after consumption of an alcohol dose resulting in a 0.11% BAC. The increase in acetaldehyde concentration in saliva of the study participants after consumption of the alcohol dose was measured by LC-MS. Results from this analysis have been reported in a previous manuscript focusing on the method used for this analysis [
36]. In this study, levels of acetaldehyde in saliva were found to increase four- to seven-fold compared to baseline, in samples collected 1 h after the dose. This measurement confirmed that the dose administered indeed resulted in an increase in the levels of acetaldehyde in the saliva of our study participants.
The results of our work further confirmed the role of alcohol-derived acetaldehyde in the induction of DNA damage in the oral cavity, demonstrated by the increase in the levels of the three major acetaldehyde-derived DNA adducts
N4-ethyl-dC,
N6-ethyl-dA, and
N2-ethyl-dG and by the increase in the signal intensity and presence of the other identified adducts in the samples collected after alcohol exposure. These results confirm and expand upon our earlier studies that found a significant increase in the levels of
N2-ethyl-dG at exposures from alcohol doses resulting in a BAC of 0.03% [
6]. Additionally, the levels of
N2-ethyl-dG we measured in samples collected before the alcohol dose (1.30 pmol/μmol dG) were consistent with previously reported levels from a similar study measuring the same adduct in samples collected from volunteers before alcohol exposure (1.85 pmol/µmol dG) [
5].
This is to our knowledge the first application of a DNA adductomic approach for the comprehensive characterization of acetaldehyde-derived DNA adducts and the first application of a targeted adductomic approach for the investigation of alcohol-related DNA damage in the oral cavity. Previously published studies focused on the detection of alcohol-derived DNA damage using nonspecific methodologies like
32P-postlabelling or on the quantitation via LC-MS of
N2-ethyl-dG and
N2-propano-dG in peripheral white-blood-cell DNA of alcoholics [
37,
38,
39]. Indeed, our previous work on oral-cell DNA from volunteers exposed to increasing amounts of alcohol focused exclusively on the quantitation of
N2-ethyl-dG [
6].
Our method allowed for the detection of two new acetaldehyde-derived DNA adducts in oral-cell DNA. Furthermore, three putative crosslinks were detected in oral cavity cells. Crosslinks are known to be potentially highly mutagenic [
9,
34]. Recently, a study evaluating acetaldehyde crosslink repair assessed the presence of error-prone mechanisms of repair of these adducts, involving the Fanconi Anemia pathways and fork-convergence mechanisms [
34]. There is a need for further investigations on the relationship between acetaldehyde-DNA modifications and mutations in humans, especially in susceptible populations, and these studies will benefit dramatically from more comprehensive methods like the one we have developed.
In our work, DNA was treated with NaBH
3CN to stabilize the Schiff bases. The use of a reducing agent effectively stabilized the imines formed on dG, dA, and dC; however, the treatment may prevent or reduce the chances of detecting other DNA adducts, limiting the ability for this method to truly identify all DNA adducts generated by the reaction with acetaldehyde [
33,
34]. This may require a parallel screening of samples processed without the reducing agent. Therefore, the analysis of samples not treated with NaBH
3CN is currently ongoing.
Another potential limitation of the work presented is the unknown contribution of bacterial cells in the oral-rinse samples collected. Future investigations should involve a quantitative analysis of bacterial DNA contaminating the samples and evaluation of the use of antibacterial rinses before sample collection. Finally, only three adducts identified in our screening experiment were fully characterized by comparison with internal standards. Further studies will allow the full characterization of the other DNA modifications included in our targeted method, with priority given to crosslink DNA adducts detected in the oral DNA samples.