Reactions of Formaldehyde Plus Acetaldehyde with Deoxyguanosine

We investigated the reactions of formaldehyde plus acetaldehyde with dGuo and DNA in order to determine whether certain 1,N2-propano-dGuo adducts coul...
0 downloads 0 Views 174KB Size
Download PDF
Chem. Res. Toxicol. 2003, 16, 145-152

145

Reactions of Formaldehyde Plus Acetaldehyde with Deoxyguanosine and DNA: Formation of Cyclic Deoxyguanosine Adducts and Formaldehyde Cross-Links Guang Cheng, Yongli Shi, Shana J. Sturla, John R. Jalas, Edward J. McIntee, Peter W. Villalta, Mingyao Wang, and Stephen S. Hecht* University of Minnesota Cancer Center, Minneapolis, Minnesota 55455 Received September 6, 2002

We investigated the reactions of formaldehyde plus acetaldehyde with dGuo and DNA in order to determine whether certain 1,N2-propano-dGuo adducts could be formed. These adductss3-(2′-deoxyribosyl)-5,6,7,8-tetrahydro-8-hydroxypyrimido[1,2-a]purine-(3H)-one (1) and 3-(2′-deoxyribosyl)-5,6,7,8-tetrahydro-6-hydroxypyrimido[1,2-a]purine-(3H)-one (3a,b)shave been previously characterized as products of the reaction of acrolein with dGuo and DNA. Adduct 1 predominates in certain model lipid peroxidation systems [Pan, J., and Chung, F. L. (2002) Chem. Res. Toxicol. 15, 367-372]. We hypothesized that this could be due to stepwise reactions of formaldehyde and acetaldehyde with dGuo, rather than by reaction of acrolein with dGuo. The results demonstrated that adducts 1 and 3a,b were relatively minor products of the reaction of formaldehyde and acetaldehyde with dGuo and that there was no selectivity in their formation. These findings did not support our hypothesis. However, substantial amounts of previously unknown cyclic dGuo adducts were identified in this reaction. The new adducts were characterized by their MS, UV, and NMR spectra as diastereomers of 3-(2′-deoxyribosyl)6-methyl-1,3,5-diazinan[4,5-a]purin-10(3H)-one (10a,b). Adducts 10a,b were apparently formed by addition of formaldehyde to N1 of N2-ethylidene-dGuo, followed by cyclization. An analogous set of four diastereomers of 3-(2′-deoxyribosyl)-6,8-dimethyl-1,3,5-diazinan[4,5-a]purin-10(3H)one (12a-d) were formed in the reactions of acetaldehyde with dGuo. These products are the first examples of exocyclic dGuo adducts of the pyrimido[1,2-a]purine type in which an oxygen atom is incorporated into the exocyclic ring. Formaldehyde-derived adducts were the other major products of the reactions of formaldehyde plus acetaldehyde with dGuo. Prominent among these were N2-hydroxymethyl-dGuo (9) and the cross-link di-(N2-deoxyguaonosyl)methane (13). We did not detect adducts 1, 3a,b, or 10a,b in enzymatic hydrolysates of DNA that had been allowed to react with formaldehyde plus acetaldehyde. However, we did detect substantial amounts of the formaldehyde cross-links di-(N6-deoxyadenosyl)methane (17), with lesser quantities of (N6-deoxyadenosyl-N2-deoxyguanosyl)methane (18), di-(N2-deoxyguanosyl)methane (13), and N6-hydroxymethyl-dAdo (19). Schiff base adducts of formaldehyde and acetaldehyde were also detected in these reactions. These results demonstrate that the reactions of formaldehyde plus acetaldehyde with dGuo are dominated by newly identified cyclic adducts and formaldehyde-derived products whereas the reactions with DNA result in the formation of formaldehyde cross-link adducts. The carcinogens formaldehdye and acetaldehyde occur in considerable quantities in the human body and in the environment. Therefore, further research is required to determine whether the adducts described here are formed in animals or humans exposed to these agents.

Introduction 2

The 1,N -propano-dGuo adducts 1 and 2 have been detected in DNA isolated from human and untreated rodent tissues (1, 2). Adducts 1 and 2 are formed in the reactions of acrolein and crotonaldehyde, respectively, with dGuo or DNA (3). Adducts 3a,b are also produced in the reaction of acrolein with dGuo or DNA (3) but are less commonly detected in human tissues (1). Chung et al. proposed that 1,N2-propano-dGuo adducts in human tissues in part result from lipid peroxidation (4). Recently, they demonstrated that adducts 1 and 2 were formed in reactions of dGuo-5′-monophosphate with ω-3 polyun* To whom correspondence should be addressed. Telephone: (612) 624-7604. Fax: (612) 626-5135. E-mail: [email protected].

saturated fatty acids in the presence of Fe(II) and air (5). The acrolein adduct 1 was formed to a greater extent than the crotonaldehyde adduct 2. Interestingly, adduct 3, which could also be formed from acrolein, was not detected. The production of acrolein in these reactions was consistent with previous results and the formation of adduct 1, but the results could not explain the apparent absence of adduct 3. We have recently demonstrated that the reaction of acetaldehyde with dGuo and DNA produces adduct 2 (6). This is proposed to occur via Schiff base 4, the major dGuo, and DNA adduct of acetaldehyde. Schiff base 4 reacts with another molecule of acetaldehyde to produce intermediate 6, which then undergoes ring closure to 2. On the basis of this reaction, we hypothesized that the

10.1021/tx025614r CCC: $25.00 © 2003 American Chemical Society Published on Web 01/04/2003

146

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

formation of adduct 1 in reactions of ω-3 polyunsaturated fatty acids with dGuo might occur by a similar sequence. Formaldehyde could first form Schiff base 5, followed by reaction with acetaldehyde to give intermediate 7, which would then undergo ring closure to 1. This reaction appeared less likely to occur in the opposite sequence, i.e., by initial reaction of formaldehyde with N-1 of dGuo followed by addition of acetaldehyde and ring closure to produce 3. Therefore, in this mechanism, adduct 1 would be formed from sequential reactions of formaldehyde and acetaldehyde with dGuo rather than by reaction of acrolein with dGuo. This could potentially explain the selectivity for adduct 1 over 3, as observed in Chung’s recent study (5). Acetaldehyde is one of the many products of lipid peroxidation, and, alhough formaldehyde has not specifically been identified in these reactions, it is certainly a potential product (7). Therefore, we investigated the reactions of dGuo and DNA with combinations of formaldehyde and acetaldehyde. These reactions could have toxicological significance beyond testing of the hypothesis described above. Formaldehyde and acetaldehyde are mutagens and carcinogens (8, 9). Both of these aldehydes produce tumors of the nasal cavities when administered by inhalation to rats (8, 9). Formaldehyde and acetaldehyde occur ubiquitously in the human body and in the environment, but their cooperative reactions with dGuo or DNA have not been reported to our knowledge. The results provide some new insights about these reactions.

Experimental Section HPLC Analysis. HPLC was carried out with a Waters Corporation (Milford, MA) system equipped with a model 991 or 996 photodiode array detector. The following columns and solvent elution systems were used. System 1 was a 4.6 mm × 25 cm Supelcosil LC18-BD column (Supelco, Bellefonte, PA) with isocratic elution by 5% CH3OH in 40 mM ammonium acetate buffer (pH 6.6) for 10 min, then a gradient from 5 to 35% CH3OH in 60 min. The flow rate was 0.5 mL/min, and detection was by UV (254 nm). System 1 was used for analysis of all adducts and for collection of adducts for structural characterization. System 2 was a 4.6 mm × 25 cm 5 µm OD5 octadecyl column (Burdick & Jackson, Baxter, McGaw Park, IL) with elution by a gradient from 40 to 60% CH3OH in H2O over the course of 40 min. The flow rate was 1 mL/min, and detection was by UV (254 nm). System 2 was used for desalting of isolated adducts. MS Analysis. LC-APCI-MS and LC-ESI-MS were carried out with a Finnigan MAT LCQ Deca instrument (Thermoquest LC/ MS Division, San Jose, CA) interfaced with an Alliance 2690

Cheng et al.

Figure 1. Chromatogram obtained upon HPLC analysis of a reaction mixture of acrolein and dGuo. Structures of the numbered peaks are indicated. dR ) 2′-deoxyribose. HPLC multisolvent delivery system from Waters. The LC-MS system was also equipped with an SPD-10 A UV-Vis detector (Shimadzu Scientific Instruments, Columbia, MD). The HPLC column and elution conditions were the same as in System 1 above, except that the flow rate was 0.3 mL/min. APCI source parameters were as follows: voltage, 4.5 kV; current, 5.0 µA; capillary temperature, 200 °C; vaporizer temperature, 400 °C. The ESI source was set as follows: voltage, 2.0 kV; current, 10 µA; capillary temperature, 250 °C. APCI was used for characterization of the acrolein-dGuo adducts (Figure 1). All other studies were carried out using ESI. NMR Analysis. Spectra were obtained on an 800 MHz instrument (Varian, Inc., Palo Alto, CA). For analysis of small samples (10-20 µg), a 3 mm probe and 3 mm Shigemi tubes (Shigemi Inc., Allison Park, PA) were used. Chemicals and Enzymes. Formaldehyde (37% in H2O), acetaldehyde, and acrolein were obtained from Aldrich Chemical Co. (Milwaukee, WI) and used without purification. Formaldehyde contained 10-15% CH3OH. The purity of acetaldehyde was >99.5%. Acrolein (90%) was stabilized with 0.1-0.2% hydroquinone and may have contained up to 10% H2O and cyclic dimer. DNA, dGuo, and enzymes were obtained from Sigma Chemical Co. (St. Louis, MO). N2-Hydroxymethyl-dGuo (9), di(N2-deoxyguanosyl)methane (13), di-(N6-deoxyadenosyl)methane (17), (N6-deoxyadenosyl-N2-deoxyguanosyl)methane (18), N6hydroxymethyl-dAdo (19), and N2-ethyl-dGuo (20) were prepared as described (10, 11). N2-Methyl-dGuo (21) was prepared by reaction of formaldehyde with dGuo, followed by NaBH3CN reduction: MS, M + 1 m/z 282; UV, 253, 280 (sh) nm. Reactions. (1) Acrolein and dGuo. Acrolein (5.0 mg, 0.09 mmol) was allowed to react with dGuo (2.65 mg, 0.01 mmol) in 1 mL of 0.1 M phosphate buffer (pH 7.0) at 37 °C for 2.5 h. The reaction mixture was washed three times with 1 mL of CHCl3, and the aqueous phase was analyzed by HPLC. (2) Formaldehyde, Acetaldehyde, or Formaldehyde Plus Acetaldehyde and dGuo. All reactions were carried out in 1 mL of 0.1 M phosphate buffer (pH 7.0) at 37 °C. A cooled syringe was used to add the acetaldehyde, and the reaction vial was quickly sealed with a screw cap. For characterization of products, formaldehyde (68 mg, 169 µL, 2.25 mmol), acetaldehyde (99 mg, 126 µL, 2.25 mmol), or formaldehyde plus acetaldehyde (2.25 or 0.75 mmol formaldehyde; 2.25 mmol acetaldehyde) was allowed to react with dGuo (3.8 mg, 0.014 mmol) for 114 h. The reaction mixture was washed three times with 1 mL of CHCl3, and the aqueous phase was analyzed by HPLC.

Reactions of Formaldehyde Plus Acetaldehyde

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 147

Table 1. Spectral Data for Acrolein-dGuo Adducts

a

peak in Figure 1

LC-APCI-MS (m/z)

UV (λmax, nm)

1 2 3 4 5

264 (M + 1); 246 (M + 1 - H2O); 228 (M + 1 - 2H2O) 324 (M + 1); 306 (M + 1-H2O); 208 (BH+);a 190 (BH+ - H2O) 324 (M + 1); 306 (M + 1 - H2O); 208 (BH+); 190 (BH+ - H2O) 324 (M + 1); 208 (BH+), 164 (BH+ - CH2CHOH) 264 (M + 1); 246 (M + 1 - H2O); 228 (M + 1 - 2H2O)

219, 256, 289 258.0, 275 (sh) 258.0, 275 (sh) 259.1, 285 (sh) 219, 256, 289

BH+ ) M+ - 116.

Figure 2. Chromatogram obtained upon HPLC analysis of a reaction mixture of formaldehyde plus acetaldehyde and dGuo. Structures of the numbered peaks are indicated. dR ) 2′-deoxyribose. The concentration dependence of adduct formation was studied by allowing formaldehyde plus acetaldehyde (0.01-50 mM of each) to react with dGuo (0.017 mmol) for 138 h. The mixture was washed with CHCl3 as above and analyzed by LCESI-MS-SIM. The formation of adducts with time was studied by allowing formaldehyde plus acetaldehyde (10 mM of each) to react with dGuo (0.017 mmol) for 6, 12, 24, or 42 h, and analyzing as above. (3) Formaldehyde Plus Acetaldehyde and DNA. Formaldehyde plus acetaldehyde (0.001-50 mM of each) were allowed to react with calf thymus DNA (50 mg) in 3 mL of 0.1 M phosphate buffer (pH 7.0) at 37 °C for 139 h. The mixture was diluted with 10 mL of H2O and extracted twice with 10 mL of CHCl3/isoamyl alcohol (24:1). The DNA was precipitated by addition of ethanol, then washed with 70% aqueous ethanol and ethanol sequentially. For enzyme hydrolysis, the modified DNA (2.5 mg) was dissolved in 1 mL of 10 mM Tris-HCl/5 mM MgCl2 buffer (pH 7.0). The mixture was incubated at 37 °C for 10 min with DNAse I (1300 units), then with phosphodiesterase I (0.06 units) and alkaline phosphatase (380 units) for an additional 60 min. Enzymes were removed by centrifugation using a Centrifree MPS device (Millipore Corp., Bedford, MA). A portion of the DNA was treated with NaBH3CN prior to enzyme hydrolysis, as described previously (6). For purification of the hydrolysate, a 500 mg Vac 6 cm3 C18 Sep-Pak cartridge (Waters) was activated with two 5 mL portions of CH3OH, 5 mL of H2O, and 5 mL of 10 mM Tris-HCl/5mM MgCl2 buffer (pH 7.0). The DNA hydrolysate (1 mL) was loaded onto the Sep-Pak, and it was washed with 5 mL of the above Tris-HCl buffer, then sequentially with 5 mL of 15% CH3OH, 5 mL of 20% CH3OH, and twice with 5 mL of 100% CH3OH. Adducts were eluted in 100% CH3OH, except for 20 and 21, which were eluted with 30 and 20% CH3OH, respectively. The eluant was reduced to dryness under a stream of N2 and analyzed by LC-ESI-MS-SIM. (4) Formaldehyde and Deoxyribonucleosides. Formaldehyde solution (33 mg, 84 µL, 1.1 mmol) was allowed to react with dGuo (11.4 mg, 0.043 mmol), or dAdo (10.7 mg, 0.043 mmol), or dGuo plus dAdo (0.021 mmol of each) in 2 mL of 0.1 M phosphate buffer (pH 7.0) for 90 h. The reaction mixture was

washed twice with 2 mL of CHCl3, and the aqueous phase was analyzed by HPLC and LC-ESI-MS.

Results Acrolein was allowed to react with dGuo to produce standards for adducts 1 and 3. The chromatogram obtained upon HPLC analysis of the products is illustrated in Figure 1 along with structural assignments for each peak. UV and MS data are summarized in Table 1. Peaks 2 and 3 were identified as two diastereomers of adduct 3 by their MS, comparison of their UV spectra to published data, and their interconversion at room temperature (3). Peak 4 was identified as adduct 1 by its MS and by comparison of its UV spectrum and relative retention time to published data (3). Peaks 1 and 5 had UV spectra very similar to those of the corresponding 1,N2-7,8-bis-Gua adducts of crotonaldehyde (12), and their MS were consistent with the addition of two molecules of acrolein to dGuo, followed by depurination. The MS of peaks 1 and 5 each showed prominent peaks at m/z 246 and 228 corresponding to successive losses of two molecules of H2O. The MS of peaks 2 and 3 also had prominent M - H2O peaks, but no loss of H2O was observed in the MS of peak 4. These results indicated that peaks 1 and 5 had the structures illustrated (8a,b) in Figure 1, with the hydroxyl group of the 1,N2-propanoring in the same orientation as in peaks 2 and 3. This was confirmed by allowing the material in peaks 2, 3, and 4 each to react separately with acrolein. Peaks 2 and 3, which partially interconverted under the conditions, gave rise to peaks 1 and 5 while no further reaction of peak 4 with acrolein was observed. There are four possible diastereomers of adduct 8, but only two peaks were observed. These results establish the structures of peaks 1 and 5 as 8a,b, but the stereochemistry of the hydroxyl groups has not been determined.

148

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

Cheng et al.

Table 2. MS and UV Data for Certain Products of the Reaction of Formaldehyde Plus Acetaldehyde with dGuo

a

peak in Figure 2

LC-ESI-MS (m/z)

UV (λmax, nm)

1 2 3 4 5 6 7

595 (2M + 1); 356 (M + 59); 298 (M + 1); 182 (BH+)a 382 (M + 59); 324 (M + 1); 208 (BH+) 382 (M + 59); 324 (M + 1); 208 (BH+) 382 (M + 59); 324 (M + 1); 208 (BH+) 382 (M + 59); 346 (M + 23); 324 (M + 1); 208 (BH+); 178 (BH+ - CH2O) 382 (M + 59); 346 (M + 23); 324 (M + 1); 208 (BH+); 178 (BH+ - CH2O) 605 (M + 59); 569 (M + 23); 547 (M + 1); 431 (M - 116)

215, 255, 275 (sh) 212, 259, 275 (sh) 211, 258, 275 (sh) 213, 259, 285 (sh) 212, 258, 275 (sh) 212, 258, 275 (sh) 211, 256, 275 (sh)

BH+ ) M+ - 116. Table 3. 800 MHz NMR Data for Adducts 10 and 12 in DMSO-d6 (chemical shifts, ppm; J, Hz)

2

N-5H

6

C6-CH3

8-H

8-R

1′

2′-a,b

3′

4′

5′-a,b

3′-OH 5′-OH

peak 5 of Figure 2 (10a)

1H

7.92(s) 8.54(bs) 5.07(q) 1.33(d) 5.10(d) 5.77(d) 6.06(m) 2.15, 4.29(bs) 3.76(bs) 3.50, 3.45(d) 5.21 (J ) 4.8) (J ) 4.8) (J ) 9.8) (J ) 9.8) 2.48(m) (J ) 12.0) 13C 127.9 72.2 12.3 64.9 74.7 31.6 62.8 79.7 53.8

4.85

peak 6 of Figure 2 (10b)

1H

7.96(s) 7.86(s)

4.59a

13C

127.9

5.11(m) 1.36(d) 5.15(d) 5.80(d) 6.10(bs) 2.12, 4.33(bs) 3.80(bs) 3.54, 3.48(d) 4.72a (J ) 5.1) (J ) 9.0) (J ) 9.0) 2.45(m) (J ) 14.1) 72.3 12.1 65.0 74.6 31.8 62.8 79.7 53.9 5.31(m) 1.38(d) 6.08(m) 1.54(m) 6.12(m) 2.22(m), 4.34(bs) 3.81(bs) 3.55, (J ) 4.8) obsb 3.52(m)

4.91

peaks 12, 13 1H 7.97(s) 8.66(s) of Figure 3 (12a,b)

5.28

a Due to the absence of COSY cross-peaks to 3′-H and 5′-H, 3′-OH, and 5′-OH could not be distinguished from one another. b obs ) obscured by solvent.

Initial studies demonstrated that adducts 1 and 3 were relatively minor products of the reaction of formaldehyde plus acetaldehyde with dGuo. Therefore, the reaction time was increased to 114 h to facilitate product identification. A chromatogram obtained upon HPLC analysis of a reaction mixture of formaldehyde, acetaldehyde, and dGuo, 37 °C for 114 h, is presented in Figure 2. The numbered peaks were identified. Peaks 2-4 were identified by their retention times, UV spectra, and MS as acrolein-dGuo adducts 1 and 3a,b (Table 2). Peaks 2 and 3 are two diastereomers of adduct 3 (identical to peaks 2 and 3 of Figure 1) while peak 4 is adduct 1 (identical to peak 4 of Figure 1). We then investigated other products of this reaction, and identified the numbered peaks shown in Figure 2. On the basis of its UV, MS, and retention time (Table 2), peak 1 was identical to a major product of the reaction of formaldehyde with dGuo. This product has been characterized previously as N2-hydroxymethyl-dGuo (9) (11). The peaks eluting at 38, 50, and 55 min were also derived from formaldehyde reactions with dGuo, based on comparison of their retention times and MS to products of the reaction of formaldehyde with dGuo under the same conditions. They were not further characterized. UV and MS data for peaks 5 and 6 were essentially identical and are summarized in Table 2. The UV spectra were very similar to those of acrolein adducts 1 and 3, as well as other 1,N2- and N2-substituted dGuo adducts. The MS indicated that both compounds had a molecular weight of 324, corresponding to addition of formaldehyde plus acetaldehyde to dGuo, with loss of H2O. Each spectrum also showed a peak corresponding to loss of formaldehyde from the M+ - 116 fragment (BH+).

Proton and carbon NMR data for peaks 5 and 6 are summarized in Table 3. The data are fully consistent with two diastereomers 10a,b. There are no protons corresponding to N1-H or NH2 of dGuo, indicating that substitution has occurred at the 1 and N2 positions, consistent with the UV data cited above. An exchangeable N5-H proton of structure 10 is observed. The methyl

group protons on C-6 are a doublet, and coupling to the methine proton on C-6 was confirmed by COSY. COSY and HMQC analysis also confirmed coupling between the protons on C-8. Consideration of chemical shift data

Reactions of Formaldehyde Plus Acetaldehyde

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 149

Figure 3. Chromatogram obtained upon HPLC analysis of a reaction mixture of acetaldehyde and dGuo. Structures of the numbered peaks are indicated. dR ) 2′-deoxyribose.

favors 10 over the alternate structure 11. Comparison of chemical shift data for protons in the exocyclic ring of adducts 1 and 3, shown below, demonstrates that ring protons attached to carbon 8 are deshielded to a greater extent than those attached to carbon 6. Similarly, the methyl protons at position 8 of adduct 12, the characterization of which is discussed below, are deshielded to a greater extent than those at position 6. The methyl protons of adduct 11 would be expected to resonate at 1.54 ppm, similar to those in 12, not at 1.33 ppm as observed in 10. The methyl resonance of 10 is consistent with that of the 6-methyl protons in 12, 1.38 ppm. The proton attached to the carbon bearing the methyl group (e.g., position 8 of adduct 11) would be expected to resonate at about 6.1 ppm, similar to the corresponding proton in 12, not at 5.07 ppm, as observed in 10 for the proton at position 6. These chemical shift assignments were confirmed by the presence of a COSY cross-peak demonstrating coupling of N-5H and 6-H in one diastereomer of adduct 10. These observations exclude structure 11 as the main component of peaks 5 and 6, although it is possible that small amounts of this compound are present. The UV and MS of peak 7 are summarized in Table 2. These data are consistent with di-(N2-deoxyguanosyl)methane (13), which was previously characterized (10). The UV, MS, and retention time were the same as those of 13, prepared by reaction of formaldehyde with dGuo.

LC-ESI-MS data indicated that peaks 8-11 had molecular ions of m/z 338. This suggested that these might be analogues of adduct 10 formed by reaction of dGuo with 2 molecules of acetaldehyde and loss of H2O. Therefore, we investigated the reaction of acetaldehyde with dGuo. HPLC analysis of the acetaldehyde-dGuo reaction produced the chromatogram illustrated in Figure 3. Peaks 1-11 were identified in our earlier work (6, 13).

Peaks 12-15 of Figure 3 had retention times identical to peaks 8-11 of Figure 2. Peaks 12-15 of Figure 3 had identical UV spectra, with maxima at 209 and 259 nm and a shoulder at 280 nm. These spectra were similar to those of 10a,b. Peaks 12-15 had identical LC-ESI-MS, with base peaks of m/z 338 (M + 1), and prominent peaks at m/z 222 (BH+) and m/z 178 (BH+ - CH3CHO). These spectra were quite similar to those of 10a,b, and can be rationalized as shown below.

Proton NMR data were obtained on peaks 12 and 13 as a mixture. These data are summarized in Table 3. Assignments were confirmed by COSY. As discussed above, the data are completely consistent with the structure shown in Table 3, i.e., two diastereomers of adduct 12. Thus, peaks 12-15 of Figures 3 and 8-11 of Figure 2 were assigned as four diastereomers of adduct 12. Levels of the acrolein-derived adducts 1 and 3a,b, the newly identified cyclic adducts 10a,b, N2-hydroxymethyldGuo (9), and the formaldehyde-dGuo cross-link 13, formed in reactions of varying concentrations of formaldehyde and acetaldehyde (0.01-1 mM of each) with dGuo at 37 °C for 140 h, were estimated from peak areas obtained by LC-ESI-MS-SIM, as illustrated in Figure 4 for 1, 3a,b, 10a,b, and 13. The results of this analysis are presented in Figure 5, which shows that, at concentrations up to 1 mM, formaldehyde adducts 9 and 13 predominated, followed by adducts 10a,b and small amounts of the acrolein adducts 1 and 3a,b. At concentrations of 10-50 mM, N2-hydroxymethyl-dGuo (9) was the predominant product, followed by 13, then 10a,b and acrolein adducts 1, 3a,b. Treatment of these reaction mixtures with NaBH3CN established the presence of Schiff bases 4 and 5, as the corresponding reduced products N2-ethyl-dGuo (20) and N2-methyl-dGuo (21).

150

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

Cheng et al.

Figure 6. Time-dependent formation of adducts in the reaction of formaldehyde plus acetaldehyde (10 mM of each) with dGuo. Data are based on area of each adduct peak as determined by LC-ESI-MS-SIM (ordinate). Symbols: ([) N2-hydroxymethyldGuo (9); (]) di-(N2-deoxyguanosyl)methane (13); (4) cyclic adducts 10a,b; (b) acrolein adducts 1 and 3a,b. Figure 4. LC-ESI-MS-SIM analysis of a reaction mixture of formaldehyde plus acetaldehyde (50 mM of each) with dGuo. (A) SIM of m/z 324, which is M + 1 of acrolein adducts 1, 3a,b and cyclic adducts 10a,b; (B) m/z 547, which is M + 1 of di-(N2deoxyguanosyl)methane (13). Numbers next to the peaks refer to adduct structure numbers in the text and Figure 2.

Figure 5. Concentration-dependent formation of adducts in the reaction of formaldehyde plus acetaldehyde with dGuo. Abscissa represents concentrations of formaldehyde and acetaldehyde. Data are based on area of each adduct peak as determined by LC-ESI-MS-SIM (ordinate). Symbols: ([) N2hydroxymethyl-dGuo (9); (]) di-(N2-deoxyguanosyl)methane (13); (4) cyclic adducts 10a,b; (b) acrolein adducts 1 and 3a,b. (A) 0.01-1 mM; (B) 10-50 mM.

Since our initial study was carried out with a relatively long reaction time of 114 h, we examined adduct formation at shorter time periods, as illustrated in Figure 6. The formaldehyde-derived adducts 9 and 13 were present in the highest concentrations after 6-18 h, then gradually declined. Levels of the other adducts increased during the first 24 h of reaction.

Figure 7. Concentration-dependent formation of adducts in the reaction of formaldehyde plus acetaldehyde with DNA. Abscissa represents concentrations of formaldehyde and acetaldehyde. Data are based on area of each adduct peak as determined by LC-ESI-MS-SIM (ordinate). (A) ([) di-(N2deoxyguanosyl)methane (13); (9) di-(N6-deoxyadenosyl)methane (17); (2) (N6-deoxyadenosyl-N2-deoxyguanosyl)methane (18); (O) N6-hydroxymethyl-dAdo (19). (B) (0) N2-methylidene-dGuo (5); (]) N2-ethylidene-dGuo (4) detected after reduction with NaBH3CN as N2-methyl-dGuo (21) and N2-ethyl-dGuo (20), respectively.

The reaction of calf thymus DNA with mixtures of acetaldehyde and formaldehyde was then investigated. We could find no evidence for the presence of adducts 1, 3a,b, or 10a,b in enzymatic hydrolysates of the DNA. However, we did detect substantial amounts of di-(N6deoxyadenosyl)methane (17), with lesser amounts of (N6deoxyadenosyl-N2-deoxyguanosyl)methane (18), di-(N2deoxyguanosyl)methane (13), and N6-hydroxymethyldAdo (19) (Figure 7A). These adducts were identified by comparison of their HPLC retention times and MS to those of synthetic samples and their quantities were

Reactions of Formaldehyde Plus Acetaldehyde

Chem. Res. Toxicol., Vol. 16, No. 2, 2003 151

example of an exocyclic dGuo adduct of the “pyrimido[1,2-a]purine” type in which an oxygen atom is incorporated into the exocyclic ring. A likely mechanism of formation of 10a,b is outlined in Scheme 1. Reaction of Scheme 1. Formation of Adducts 10a,b

Figure 8. LC-ESI-SIM [m/z 515, M + H of di-(N6-deoxyadenosyl)methane (17)] analysis of an enzymatic hydrolysate of the reaction of acetaldehyde plus formaldehyde (1 µM of each) with DNA. The peak marked 17 corresponds to adduct 17.

estimated by LC-ESI-MS-SIM. Adduct 17 was readily detectable in reactions of DNA with formaldehyde and acetaldehyde at concentrations as low as 1 µM of each (Figure 8). Treatment of the DNA with NaBH3CN prior

to enzymatic hydrolysis allowed the quantitation of adducts 4 and 5, as the corresponding reduced adducts, N2-ethyl-dGuo (20) and N2-methyl-dGuo (21), respectively (Figure 7B).

dGuo with acetaldehyde produces Schiff base 4, known to be the major product of acetaldehyde-dGuo reactions (see Figure 3) (6). Reaction of 4 with formaldehyde gives intermediate 21 which cyclizes to 10a,b. In support of this mechanism, sequential reaction of acetaldehyde, then formaldehyde, with dGuo produced adducts 10a and 10b while the reverse sequence of addition did not yield 10a,b (data not shown). Adduct 11, with the methyl group attached to carbon 8, was not observed. This product would have resulted from reaction of dGuo with formaldehyde to produce the corresponding Schiff base 5, followed by addition of acetaldehyde at N1 giving 23, then cyclization. Intermediates 21 and 22 are formed and cyclize to the observed products 10a,b and 12a-d, respectively, while adduct 11, which would result from cyclization of intermediate 23, was not observed. Intermediates 21-23 have similar steric requirements and

Discussion Our results do not support the hypothesis that acrolein adduct 1 is selectively formed via intermediate 7 by sequential reactions of formaldehyde and acetaldehyde with dGuo. First, 1 is only a minor product of the reaction of formaldehyde plus acetaldehyde with dGuo, and we could not detect this adduct in the DNA reactions. Second, there appears to be little selectivity in the formation of 1, since 3a and 3b were also observed (Figure 4). The results suggest that adducts 1 and 3a,b, as detected in our study, were actually formed by the condensation of formaldehyde and acetaldehyde to give acrolein (14), which then reacted with dGuo. In support of this mechanism, we incubated formaldehyde plus acetaldehyde at 37 °C for 72 h, then added dGuo, and detected adducts 1 and 3a,b (data not shown). Nevertheless, our results do not exclude the possibility that sequential addition of formaldehyde and acetaldehyde to dGuo contributes to formation of adduct 1. While adducts 1 and 3a,b were relatively minor products of the reaction of formaldehyde plus acetaldehyde with dGuo, we did observe substantial amounts of the previously unknown adducts 10a,b. This is the first

there is little reason to believe that they should differ markedly in their propensity to cyclize. Therefore, reactions of formaldehyde Schiff base 5 that are more facile than condensation at N1 with acetaldehyde probably account for the relatively low yield of adduct 11, if formed. The new cyclic adducts 10a and 10b were not detected in DNA reacted with acetaldehyde plus formaldehyde. Previous studies demonstrate that Schiff base 4 is readily produced in reactions of acetaldehyde with DNA (6). Apparently, steric factors in the hydrogen bonding region of DNA are unfavorable for the addition of formaldehyde to Schiff base 4 producing the requisite intermediate 21. More sensitive methods are required to further investigate the presence of adducts 10a and 10b in DNA.

152

Chem. Res. Toxicol., Vol. 16, No. 2, 2003

Inspection of Figure 2 indicates that, in the reaction of formaldehyde plus acetaldehyde with dGuo, products derived from formaldehyde predominate. Thus, two major products were N2-hydroxymethyl-dGuo (9) and the formaldehyde cross-link adduct 13. Several other major peaks eluting at 38, 50, and 55 min were observed, but not characterized. These products were also formed in the reaction of formaldehyde with dGuo under the same conditions (data not shown). The only other major products were the new cyclic adducts 10a and 10b. These results are a reflection of the higher reactivity of formaldehyde than acetaldehyde with dGuo. Formaldehyde cross-links 13, 17, and 18 were major products of the DNA reactions. Among these, the deoxyadenosine cross-link 17 was formed in the highest yield, consistent with previous studies carried out with defined oligonucleotides (15, 16). The formation of this cross-link adduct was readily observed at concentrations of formaldehyde as low as 1 µM. Although specific formaldehydeDNA cross-links such as adducts 13, 17, and 18 have been previously characterized (10, 15, 16), their facile formation at relatively low formaldehyde concentrations has not been recognized. These results suggest that specific formaldehyde-DNA cross-links may be important in carcinogenesis and indicate that they might be detectable in vivo. Formaldehyde is produced endogenously from a variety of natural substrates including serine, glycine, methionine, and choline (8). Formaldehyde is a common metabolite of drugs and carcinogens. The endogenous concentration of formaldehyde in human blood has been estimated as 100 µM (8). Exogenous exposures to formaldehyde occur in numerous occupational and environmental settings. Cigarette smoke contains 12-106 µg of formaldehyde/cigarette (17). According to the International Agency for Research on Cancer (IARC), there is sufficient evidence for the carcinogenicity of formaldehyde in laboratory animals, and it is probably carcinogenic to humans (8). Acetaldehyde is also commonly found in the human environment (9). It is an intermediate in the metabolism of sugars and is the primary metabolite of ethanol. Its concentration in cigarette smoke ranges from 596 to 2133 µg/cigarette (16). IARC evaluates acetaldehyde as possibly carcinogenic to humans (9). Considering the commonplace human exposure to high levels of these aldehydes, it is not unreasonable to expect that they could occur individually or jointly in significant concentrations in humans. Therefore, further studies are required to assess the presence of their DNA adducts in human tissues. While DNA adducts derived from acetaldehyde have been quantified in humans in several studies (18-20), there are no reports on the presence of specific formaldehyde-DNA adducts.

Acknowledgment. This study was supported by Grants CA-85702 and ES-11297 from the National Institutes of Health and Grant RP-00-138 from the American Cancer Society. We appreciate the contributions of Drs. David Live and Beverly Ostrowski of the NMR facility.

Cheng et al.

References (1) Chung, F. L., Nath, R. G., Nagao, M., Nishikawa, A., Zhou, G. D., and Randerath, K. (1999) Endogenous formation and significance of 1,N2-propanodeoxyguanosine adducts. Mutat. Res. 424, 71-81. (2) Penn, A., Nath, R., Pan, J. S., Chen, L. C., Widmer, K., Henk, W., and Chung, F. L. (2001) 1,N2-propanodeoxyguanosine adduct formation in aortic DNA following inhalation of acrolein. Environ. Health Perspect.109, 219-224. (3) Chung, F. L., Young, R., and Hecht, S. S. (1984) Formation of cyclic 1,N2-propanodeoxyguanosine adducts in DNA upon reaction with acrolein or crotonaldehyde. Cancer Res. 44, 990-995. (4) Chung, F. L., Chen, H. J. C., and Nath, R. G. (1996) Lipid peroxidation as a potential source for the formation of exocyclic DNA adducts. Carcinogenesis 17, 2105-2111. (5) Pan, J., and Chung, F. L. (2002) Formation of cyclic deoxyguanosine adducts from omega-3 and omega-6 polyunsaturated fatty acids under oxidative conditions. Chem. Res. Toxicol. 15, 367-372. (6) Wang, M., McIntee, E. J., Cheng, G., Shi, Y., Villalta, P. W., and Hecht, S. S. (2000) Identification of DNA adducts of acetaldehyde. Chem. Res. Toxicol. 13, 1149-1157. (7) Esterbauer, H., Zollner, H., and Schaur, R. J. (1990) Aldehydes formed by lipid peroxidation: mechanisms of formation, occurrence, and determination. In Membrane Lipid Oxidation (VigoPelfrey, C., Ed.), Vol. I, pp 239-268, CRC Press, Boca Raton, FL. (8) International Agency for Research on Cancer (1995) Wood Dust and Formaldehyde. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 62, pp 217-362, IARC, Lyon, FR. (9) International Agency for Research on Cancer (1999) Re-evaluation of Some Organic Chemicals, Hydrazine and Hydrogen Peroxide (Part Two). IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol. 71, pp 319-335, IARC, Lyon, FR. (10) Chaw, Y. F., Crane, L. E., Lange, P., and Shapiro, R. (1980) Isolation and identification of cross-links from formaldehydetreated nucleic acids. Biochemistry 19, 5525-5531. (11) Beland, F. A., Fullerton, N. F., and Heflich, R. H. (1984) Rapid isolation, hydrolysis and chromatography of formaldehyde-modified DNA. J. Chromatogr. 308, 121-131. (12) Eder, E., and Hoffman, C. (1992) Identification and characterization of deoxyguanosine-crotonaldehyde adducts. Formation of cyclic adducts and 1,N2,7,8 bis-cyclic adducts. Chem. Res. Toxicol. 5, 802-808. (13) Wang, M., McIntee, E. J., Cheng, G., Shi, Y., Villalta, P. W., and Hecht, S. S. (2001) Reactions of 2,6-dimethyl-1,3-dioxane-4-ol (aldoxane) with deoxyguanosine and DNA. Chem. Res. Toxicol. 14, 1025-1032. (14) Nielsen, A. T., and Houlihan, W. J. (1968) The aldol condensation. In Organic Reactions (Cope, A. C., Ed.) Vol. 16, pp 1-87, John Wiley & Sons. (15) Huang, H., Solomon, M. S., and Hopkins, P. B. (1992) Formaldehyde preferentially interstrand cross-links duplex DNA through deoxyguanosine residues at the sequence 5′-d(AT). J. Am. Chem. Soc. 114, 9240-9241. (16) Huang, H., and Hopkins, P. B. (1993) DNA interstrand crosslinking by formaldehyde: nucleotide sequence preference and covalent structure of the predominant cross-link formed in synthetic oligonucleotides. J. Am. Chem. Soc. 115, 9402-9408. (17) International Agency for Research on Cancer (2002) Tobacco Smoke and Involuntary Smoking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 83, IARC, Lyon, FR (in press). (18) Fang, J. L., and Vaca, C. E. (1997) Detection of DNA adducts of acetaldehyde in peripheral white blood cells of alcohol abusers. Carcinogenesis 18, 627-632. (19) Vaca, C. E., Nilsson, J. A., Fang, J. L., and Grafstrom, R. C. (1998) Formation of DNA adducts in human buccal epithelial cells exposed to acetaldehyde and methylglyoxal in vitro. Chem.-Biol. 108, 197-208. (20) Matsuda, T., Terashima, I., Matsumoto, Y., Yabushita, H., Matsui, S., and Shibutani, S. (1999) Effective utilization of N2-ethyl-2′deoxyguanosine triphosphate during DNA synthesis catalyzed by mammalian replicative DNA polymerases. Biochemistry 38, 929-935.

TX025614R