2'-Deoxyguanosine Reacts with a Model Quinone Methide at Multiple

Aug 8, 2001 - Chemical Structure and Properties of Interstrand Cross-Links Formed by Reaction of Guanine Residues with Abasic Sites in Duplex DNA. Mic...
1 downloads 21 Views 97KB Size
Chem. Res. Toxicol. 2001, 14, 1345-1351

1345

2′-Deoxyguanosine Reacts with a Model Quinone Methide at Multiple Sites Willem F. Veldhuyzen, Yui-Fai Lam, and Steven E. Rokita* Department of Chemistry and Biochemistry, University of Maryland, College Park, Maryland 20742 Received June 20, 2001

Quinone methides and related intermediates have been implicated in a range of beneficial and detrimental processes in biology and effectively alkylate a variety of cellular components despite the ubiquitous presence of water. As a prerequisite to understanding the origins of their specificity, the major products generated by DNA and its components with an unsubstituted ortho quinone methide under aqueous conditions were recently characterized [Pande, P., Shearer, J., Yang, J., Greenberg, W. A., and Rokita, S. E. (1999) J. Am. Chem. Soc. 121, 6773-6779]. Investigations currently focus on the complete range of derivatives formed by deoxyguanosine (dG) and guanine residues in duplex DNA through product isolation and structure determination using reversed-phase chromatography and a range of one and twodimensional NMR techniques. Previous construction of a synthetic standard for dG alkylation is now shown to have yielded the N1-linked adduct rather than the N2-linked adduct. This later adduct has also now been characterized and confirmed to be the major product of reaction between the quinone methide and both duplex DNA and dG under neutral conditions. An N7 adduct of guanine has additionally been identified under these conditions and appears to result from spontaneous deglycosylation of the corresponding N7 adduct of dG. A combination of steric and electronic properties of duplex DNA likely contribute to the enhanced selectivity of the quinone methide for its guanine N2 position (7.8:3.2:1.0 for adducts of N2:N7:N1) relative to that of dG (4.7:3.5:1.0 for adducts of N2:N7:N1).

Introduction Structural characterization of DNA adducts provides one of the most useful tools for describing nucleic acid reactivity and hence contributes to our understanding of the mechanism by which numerous environmental carcinogens and certain anticancer antibiotics function. Although our ability to predict the target specificity of many complex intermediates remains limited, some trends have been established by study of simple electrophiles (1-5). In general, the sites of modification depend on the nature of the nucleophile-electrophile pairing and the accessibility of the target functional group in DNA. Electrostatic potential calculations suggest that guanine (G) N7, cytosine (C) N3, and adenine (A) N1 are most nucleophilic, and other sites with nucleophilic character include N3 and O6 of G, O2 of C, and N3 of adenine (A) (6). Relatively soft electrophiles such as dimethyl sulfate and methyl iodide participate in SN2-like processes and typically alkylate the most nucleophilic positions such as dG N7 (1, 3). In contrast, hard electrophiles such as those formed by N-alkyl-N-nitrosourea participate in SN1-like processes and often modify the most electronegative heteroatom, oxygen in dC, dG and the phosphate backbone (1-3). The intrinsic reactivity of each site within the nucleosides is not the sole variable affecting product distribution. Both the steric and electronic environments established by helical DNA are important determinants in the specificity of modification as well. While the steric properties of helical DNA most commonly lead to selec* To whom correspondence should be addressed.

tive suppression of reaction, electronic contributions by neighboring bases have the ability to activate certain sites for reaction (3, 7). Selectivity is also induced by preferential binding of electrophiles or their precursors to increase the local concentration of reactants (8). For example, nitrogen mustards related to chlorambucil generally react with most all guanine residues through the major groove (9), but reaction can be limited to a single region of this groove by conjugating the electrophile to a sequencing-directing and triplex-forming deoxyoligonucleotide (10). Alternative conjugation to minor groove-binding agents can redirect or restrict reaction to A N3 of the minor groove (11). Natural products that generate quinone methide or related intermediates predominantly alkylate the exocyclic amines of dG (N2) and to a lesser extent dA (N6) (12-17). The origins of this specificity are still under investigation and may result from selective binding and orientation as well as optimal pairing of the reactants’ electrophilic and nucleophilic properties. Even simple quinone methide models exhibit a propensity for reaction at these exocyclic amines of the purines (18-20). In contrast, the ring nitrogen N3 of the pyrimidine dC may be modified instead of its exocyclic N4 position (19, 21). Reaction is also not limited to these individual sites, and product profiles can respond to changes in quinone methide structure and reaction conditions. Loss of the aziridine ring of mitomycin to form a 2,7-diaminomitosene derivative increases formation of a GN7 adduct at the expense of its N2 adduct (22, 23). Likewise, paraquinone methides generated from di(tert-butyl)methylphenol derivatives (such as BHTOH) exhibit varying

10.1021/tx0101043 CCC: $20.00 © 2001 American Chemical Society Published on Web 08/08/2001

1346

Chem. Res. Toxicol., Vol. 14, No. 9, 2001

specificities for GN7, N1, and N2 as a function of their structure and stability (19). Our laboratory has developed a series of quinone methide-forming bioconjugates for sequence specific alkylation of DNA that are triggered by photochemistry (24), reduction (24), and fluoride (25), alternatively. A related model system was concurrently established to identify the preferred sites of modification on each nucleoside residue (20, 21). This utilized a simple quinone methide precursor (QMP) based on a silyl-protected phenol containing an ortho-substituted bromomethyl group (Scheme 1). Upon deprotection with fluoride, bromide is eliminated Scheme 1

to yield the ortho-quinone methide intermediate under mild aqueous conditions. The major product identified after reaction with DNA was assigned as the N2 adduct of dG (20), equivalent to that generated by mitomycin (12, 14) and an anthracycline derivative (13). Interestingly, the selectivity of QMP was not completely reflected in the intrinsic nature of the nucleosides. Modification of dC N3 dominated reaction of the nucleosides but was suppressed by Watson-Crick base pairing in reaction of duplex DNA (20). The exocyclic amine N2 of dG was least sensitive to the constraints of duplex DNA and consequently prevailed over the competing sites of reaction. We have since focused on the minor products of guanine alkylation in order to place these model studies in context with the reaction profiles determined for the paraquinone methide derivatives of BHTOH and mitomycin that exhibit a range of purine products (15, 16, 19, 22, 23).

Experimental Section General Materials and Methods. Solvents, chemicals and reagents of the highest commercial grade were used without further purification except when noted. All aqueous solutions were prepared with water purified by a standard filtration system to yield a resistivity of 18.0 MΩ. Silica gel (230-400 mesh) for column chromatography was purchased from EM Sciences. 1H and 13C spectra were recorded on an AMX 500 spectrometer (1H, 500.13 MHz; 13C, 125.77 MHz). All NMR chemical shifts (δ) are reported in parts per million (ppm) and were determined relative to the standard values for solvent. Coupling constants (J) are reported in hertz (Hz). Low and highresolution mass spectra were determined with a VG7070E mass spectrometer. UV absorption spectra were obtained with a Hewlett-Packard 8453 UV-vis spectrophotometer. Reverse phase C-18 chromatography was performed on analytical (Varian Microsorb-MV C-18, 300 Å particle size, 250 mm × 4.6 mm) and semipreparative (Alltech Econosphere C-18, 10 um particle size, 250 mm × 10 mm) columns using a Jasco PU-980 HPLC equipped with a Jasco MD-1510 UV-vis multiwavelength scanning detector. N2 Adduct of dG (1). QMP (21) (12 mg, 0.040 mmol) in DMF (180 µL) was combined with dG (6 mg, 0.02 mmol). Reaction was initiated by addition of aqueous KF (420 µL, 2.6 M) and incubated for 2 h at 37 °C. The N2 adduct was directly purified by semipreparative reverse-phase HPLC using a gradient of 10% acetonitrile in 27 mM triethylammonium acetate (TEAA, pH

Veldhuyzen et al. 4) to 15% acetonitrile in 26 mM TEAA (pH 4) over 60 min (5 mL/min). The synthesis and isolation was performed 5 times to obtain sufficient material for spectroscopic characterization (2 mg, 5% yield). 1H NMR (DMF-d7) δ 2.36 (ddd, 1H, J ) 13.0, 6.5, 3.0 Hz), 2.71 (m, 1H), 3.69 (dd, 1H, J ) 12.0, 4.5 Hz), 3.74 (dd, 1H, J ) 12.0, 4.5 Hz), 3.97 (m, 1H), 4.56 (m, 1H), 4.58 (s, 2H), 5.07 (bs, OH), 5.43 (bs, OH), 6.35 (dd, 1H, J ) 7.5, 6.5 Hz), 6.78 (td, 1H, J ) 7.5, 0.5 Hz), 6.94 (dd, 1H, J ) 7.5, 0.5 Hz), 7.12 (td, 1H, J ) 7.5, 1.5 Hz), 7.23 (bs, NH), 7.33 (dd, 1H, J ) 7.5, 1.5 Hz), 7.99 (s, 1H).13C NMR (DMF-d7) δ 40.8, 41.0, 63.1, 72.2, 84.2, 88.9, 115.9, 118.2, 119.5, 125.8, 129.1, 130.1, 136.2, 151.3, 154.0, 156.6, 157.7. UV (10 mM potassium phosphate pH 6.5): λmax 256, 280(s) nm. HRMS (FAB, glycerol) m/z 374.1451 (M + H+); calcd for C17H20N5O5 (M + H+): 374.1464. N1 Adduct of dG (2). QMP (21) (230 mg, 0.64 mmol) in DMF (5.0 mL) was combined with dG (130 mg, 0.46 mmol). Reaction was initiated by addition of aqueous KF (0.90 mL, 2.6 M) and incubated for 2 h at 37 °C as described previously (20). Adduct 2 was isolated in 18% yield (30 mg) after semipreparative reversed-phase HPLC using a gradient of 5% acetonitrile in 29 mM TEAA (pH 4) to 25% acetonitrile in 23 mM TEAA (pH 4) over 30 min (5 mL/min). 1H NMR (DMF-d7) δ 2.34 (ddd, 1H, J ) 13.5, 6.0, 2.5 Hz), 2.70 (ddd, 1H, J ) 13.5, 8.0, 5.5 Hz), 3.69 (dd, 1H, J ) 11.5, 4.5 Hz), 3.74 (dd, 1H, J ) 11.5, 4.5 Hz), 3.97 (m, 1H), 4.54 (m, 1H), 5.28 (s, 2H), 6.27 (dd, 1H, J ) 8.0, 6.0 Hz), 6.77 (t, 1H, J ) 7.5 Hz), 6.98 (d, 1H, J ) 7.5 Hz), 7.13 (t, 1H, J ) 7.5 Hz), 7.14 (d, 1H, J ) 7.5 Hz), 7.20 (bs, NH2), 8.05 (s, 1H). 13C NMR (DMF-d7) δ 39.3, 40.7, 63.1, 72.3, 83.8, 89.0, 116.0, 117.2, 120.0, 123.7, 129.0, 129.2, 136.5, 150.2, 154.9, 155.9, 157.8. UV (10 mM potassium phosphate pH 6.5): λmax 257, 275(s) nm. Tris-tert-butyldimethylsilyl Derivative of the dG N1 Adduct 2. Adduct 2 (8 mg, 0.02 mmol) and tert-butyldimethylsilyl chloride (65 mg, 0.43 mmol) were added to a solution of imidazole (60 mg, 0.88 mmol) in DMF (0.5 mL). The reaction mixture was stirred at room temperature for 4 h and then extracted with chloroform, washed with water and brine, dried (MgSO4), and concentrated under reduced pressure. Purification by silica gel flash chromatography (6:4 hexanes:ethyl acetate) yielded the desired product (14 mg, 88%) exhibiting three tertbutyldimethylsilyl groups by 1H NMR as expected for complete protection of the three hydroxyl groups. 1H NMR (CD3CN) δ 0.02 (s, 3H), 0.04 (s, 3H), 0.10 (s, 3H), 0.11 (s, 3H), 0.30 (s, 6H), 0.88 (s, 9H), 0.90 (s, 9H), 1.05 (s, 9H), 2.28 (ddd, 1H, J ) 13.5, 6.5, 4.0 Hz), 2.62 (dt, 1H, J ) 13.5, 6.0 Hz), 3.73 (m, 2H), 3.86 (m, 1H), 4.56 (m, 1H), 5.18 (AB quartet, 2H, JAB ) 16.5 Hz), 5.67 (s, NH2), 6.11 (t, 1H, J ) 6.5 Hz), 6.91 (t, 1H, J ) 7.5 Hz), 6.94 (d, 1H, J ) 7.5 Hz), 7.08 (d, 1H, J ) 7.5 Hz), 7.19 (t, 1H, J ) 7.5 Hz), 7.71 (s, 1H). 13C NMR (CD3CN) δ -5.34, -5.31, -4.7, -4.5, 18.5, 18.9, 39.9, 40.7, 63.9, 73.1, 83.9, 88.5, 119.9, 123.1, 127.3, 129.0, 129.8, 136.7, 150.1, 153.7, 154.4, 158.1. A signal corresponding to C5 (∼117.7 ppm) was obscured by the solvent signal but verified by HMBC (CD3CN). HRMS (FAB, glycerol) m/z 716.4092 (M + H+); calcd for C35H61N5O5Si3 (M + H+): 716.4059. N7 Adduct of Guanine (3). QMP (21) (18 mg, 0.060 mmol) was added to an aqueous solution of dG (5.7 mg, 0.020 mmol), KF (500 mM) and potassium phosphate (40 mM, pH 7), and 30% DMF (1 mL). The resulting mixture was incubated at 37 °C (24 h) and then directly purified by semipreparative reversedphase HPLC using a gradient of 10% acetonitrile in 27 mM TEAA (pH 4) to 15% acetonitrile in 26 mM TEAA (pH 4) over 60 min (5 mL/min). The synthesis and isolation was performed nine times to obtain sufficient quantities of the N7 adduct for spectroscopic characterization (6 mg, 10% yield). 1H NMR (0.2 M NaOD) δ 5.39 (s, 2H), 6.49 (td, 1H, J ) 7.5, 0.5 Hz), 6.66 (dd, 1H, J ) 7.5, 0.5 Hz), 6.79 (dd, 1H, J ) 7.5, 1.0 Hz), 7.12 (td, 1H, J ) 7.5, 1.0 Hz), 7.64 (s, 1H). 13C NMR (0.2 M NaOD) δ 46.6, 110.6, 114.1, 119.4, 125.5, 128.8, 129.8, 142.4, 159.5, 161.2, 164.9, 165.7. UV (10 mM potassium phosphate pH 6.5): λmax 280 nm. HRMS (FAB, glycerol) m/z 258.1002 (M + H+); calcd for C17H20N5O5 (M + H+): 258.0991.

Quinone Methide Reaction with Guanine

Chem. Res. Toxicol., Vol. 14, No. 9, 2001 1347 Table 1.

1H

and 13C NMR Data for the dG N2 Adduct (1) and the Guanine N7 Adduct (3)

dG N2 adduct (1)a

dGa

position δH (ppm) δC (ppm) δC (ppm)

Figure 1. Reversed-phase (C-18) chromatographic analysis of dG adducts formed from reaction of dG, QMP, and 400 mM KF in 85% aqueous DMF. For details, see the Experimental Section. Quinone Methide Alkylation of dG. Reactions were initiated by adding an acetonitrile solution of QMP21 (30 µL) to a solution of dG and KF in potassium phosphate (pH 7, 70 mL). The resulting mixture of QMP (25 mM), dG (0.5 mM), KF (500 mM), and buffer (10 mM) was incubated at 37° C (24 h) and directly separated by analytical reversed-phase HPLC using a gradient of 3% acetonitrile in 29 mM TEAA (pH 4.0) to 25% acetonitrile in 21 mM TEAA (pH 4) over 66 min (1 mL/min). The relative product yields represent averages of three independent determinations, and the alkylated products were assumed to have extinction coefficients proportional to their parent derivatives, dG and guanine, alternatively. Alkylation of Double-Stranded DNA. Reaction of calf thymus DNA (18 mM in nt, purified by phenol-chloroform extraction) was performed under identical conditions used for dG above except the concentration of QMP (21) was increased to 100 mM. Prior to HPLC analysis, the modified DNA was digested with alkaline phosphatase and phosphodiesterase as described previously (20). Products of alkylation were detected by analytical reversed-phase HPLC using a gradient of 3% acetonitrile in 29 mM TEAA (pH 4.0) to 25% acetonitrile in 21 mM TEAA (pH 4) over 66 min (1 mL/min). The relative product yields represent averages of four independent determinations, and the alkylated products were assumed to have extinction coefficients proportional to their parent deoxynucleosides and guanine, alternatively.

Results and Discussion Adducts Formed by the Nucleoside dG. Prior analysis of nucleic acid modification by QMP began by characterizing the major adduct of each residue and comparing their relative yields using deoxynucleosides to duplex DNA (20, 21). Additional investigations on the minor products formed by guanine residues was further expected to differentiate between reaction specificity dictated by the electrophile and that dictated by the DNA target. Our attention consequently turned to such products formed by dG and a model quinone methide conveniently generated by QMP and aqueous fluoride. Minor products were indeed evident by reverse-phase C-18 chromatography (Figure 1), but surprisingly, replacement of the original column (20) with another provided resolution of two products 1 and 2 that had previously coeluted and assigned to the dG N2 adduct (20). Most intriguingly, the synthetic conditions formerly used to prepare the dG adduct yielded 2, whereas DNA modification yielded 1 (see below). Structural characterization therefore began with examination of 1 and, subsequently, 2. Isolation and Structural Characterization of dG Adduct 1. Preparative isolation of 1 was initially made difficult by the great excess (15-fold) of 2 generated under the original synthetic conditions (85% aqueous DMF) (20). The formation of 2 was found to be dependent on DMF concentration and was suppressed 7-fold by use of

2 4 5 6 8 -CH210 11 12 13 14 15 a

7.99 4.58 7.33 6.78 7.12 6.94

154.0 151.3 118.2 157.7 136.2 40.8 156.6 125.8 130.1 119.5 129.1 115.9

154.8 151.8 117.9 157.8 136.2

guanine N7 adduct (3)b δH (ppm)

7.64 5.39 6.79 6.49 7.12 6.66

δC (ppm) 161.2 159.5 110.6 165.7 142.4 46.6 164.9 125.5 128.8 114.1 129.8 119.4

In DMF-d7. b In NaOD/D2O.

30% aqueous DMF. The organic solvent could not be omitted completely since it was necessary for solubilizing QMP. Under these conditions with dG, QMP (2 equiv) and KF (1.8 M), 1 was isolated from a semipreparative C-18 column in 5% yield with respect to dG and ∼24% with respect to total dG adducts. Despite this low yield, sufficient material was collected for complete spectroscopic analysis. The UV absorbance maximum of adduct 1 (256 nm) exhibited a small 4 nm bathochromic shift relative to dG (252 nm). Such a shift is consistent with the absorbance maxima of products measured under neutral aqueous conditions and formed by alkylation at N2 (256 nm), N1 (256 nm), or N7 (258 nm) (19). In contrast, alkylation at dG O6 results in a hypsochromic shift (247 nm) (26). Characterization of adduct 1 by NMR followed an equivalent approach to that previously published on the quinone methide adducts of dA and dC (20, 21). Assignment of 1H signals were based on their multiplicities and chemical shifts and confirmed by 2D COSY analysis. The 13C signals were then identified from their connectivities to 1H signals observed by HMQC and HMBC experiments (see Supporting Information) (27, 28). Aromatic carbons C12-C15 derived from QMP were distinguished by correlation with their corresponding aromatic hydrogens as detected by HMQC, and the carbons C10 and C11 were distinguished by their long range connectivities to aromatic hydrogens as detected by HMBC. The purine H8 helped to assign C8 as well as C4 and C5 by direct (HMQC) and long-range coupling (HMBC), respectively. The purine C4 was differentiated from C5 by a correlation between C4 and H1′ of the ribose ring (HMBC). No geminal or proximal hydrogens are present to discern the signals for purine C2 and C6. However, based on literature precedence for 13C NMR assignments of dG and its derivatives (29-31), C6 was assigned to be more downfield than C2. Due to the importance of these final signals in the differentiation of 1 and 2, their assignments were confirmed experimentally as described below. The 13C NMR chemical shifts (Table 1) of adduct 1 relative to dG are most consistent with alkylation at N2 rather than O6, N7, or N1. An O6 product would be expected to show significant perturbation of its purine carbon signals C2, C4, C5, C6, and C8 relative to dG (>2.0 ppm) (32). Similarly, modification at N7 would result in a downfield shift of C2 and C6 (>3.5 ppm) in addition to a large upfield shift in C5 (∼9 ppm) relative to dG (32), and modification at N1 would result an upfield shift in C4 (∼2-3 ppm) (32). Only alkylation at N2 yields

1348

Chem. Res. Toxicol., Vol. 14, No. 9, 2001

Veldhuyzen et al.

Figure 3. 2D COSY (280 K) of dG adduct 1 in DMF-d7.

Figure 2. HMBC of dG adduct 1 in DMF-d7.

little change in the 13C signals of the purine (e0.7 ppm) (33). Indeed, the chemical shifts of 1 deviate only slightly from those of dG (e0.7 ppm). The chemical shift of the benzylic hydrogens are also very sensitive to the position of their attachment to the purine, and a value of 4.58 ppm is most reminiscent of the 4.31 ppm attributed to the equivalent hydrogens of the N2 adduct of BHTOH and unlike the corresponding values of N1 (5.06 ppm) and N7 (5.23 ppm) adducts (19). However, unambiguous assignment of adduct 1 was best determined by the long-range 1H-13C connectivities observed by HMBC (Figure 2). Such coupling was inherent for the benzylic hydrogens (-CH2-) and C10, C11, and C12 of the phenolic moiety, but only the N2 derivative would also exhibit a unique and singular purine coupling to C2. Alternative alkylation at O6 or N1 would have displayed a correlation with C6, and alkylation at N7 or N3 would have displayed correlations to C5 and C8 or C2 and C4, respectively. The structure of 1 was further established by observing the predicted -NH- to -CH2- connectivity using 2D COSY NMR (DMF-d7). The -NH- proton signals for 1 are broad at 25° C but were suitably resolved at lower temperature (7° C) to detect their correlation (Figure 3). No equivalent connectivity was observed for adduct 2 under identical conditions (Supporting Information). Initial characterization of the synthetic standard 2 had also originally led to the same structural assignment as 1 due in part to the uncertainty of the C2 and C6 signals (20). The COSY experiment strongly suggests that 2 is not linked to dG through the N2 position. Structural Characterization of dG Adduct 2. Initial confusion over the assignment of C2 and C6 in 2 resulted from a lack of resolution between the quinone methide-forming moiety (C10) and a purine carbon in the 13C NMR when CD OD was used as solvent (20). This 3 was compounded by ascribing no further consequence to

HMBC data indicating a connectivity between C10 and the benzylic -CH2- group that was inherent from the QMP structure. However, the COSY experiment described above contradicted our earlier assumptions. Efforts to resolve and reassign the purine carbon resonances C2 and C6 of 2 were therefore initiated. By switching solvents from CD3OD to DMF-d7, each 13C signal became distinct and thus available for direct characterization. Identification of C2 by correlation to the protons of its attached -NH2 group were thwarted by their rapid exchange in DMF. The limited solubility of 2 prevented use of alternative solvents to circumvent this problem. Instead, the tris-tert-butyldimethylsilyl derivative of 2 was prepared to enhance its solubility for NMR studies in CD3CN (Supporting Information). Still, no correlation between the N2 protons and C2 could be observed by HMBC despite the relatively sharp signals of these protons. This result likely reflects the smaller coupling constant often associated with two vs three bond heteronuclear systems (34). The necessary correlations required for unambiguous assignments were finally obtained by heteronuclear NOE. Irradiation of the -NH2 protons (5.67 ppm) of the silylated derivative of 2 yielded an 80% enhancement of its 13C signal at 154.4 ppm (Supporting Information). In contrast, all other 13C signals were unaffected ((5%) by these conditions. Accordingly, the upfield signal of the silylated derivative at 154.4 ppm was assigned to C2 and by default the downfield signal at 158.1 ppm was assigned to its C6 (Table 2). This in turn indicated that the upfield signal of the nonsilylated parent 2 was associated with C2 as typical of other alkylated derivatives of dG (32, 33). Complete HMBC analysis of 2 in DMF-d7 detected correlations between the benzylic hydrogens and both newly assigned purine carbons, C2 and C6, in addition to those derived from QMP, C10, C11, C12, and C14 (Figure 4). These data are uniquely satisfied by alkylation of N1 rather than N2 of dG to form adduct 2. The UV absorbance maximum of 2 (257 nm) is also consistent with N1 alkylation, but does not differ from that resulting from N2 alkylation (19). Finally, the downfield shift of the benzylic hydrogens attached to N1 (5.28 ppm, 2) vs N2 (4.58 ppm, 1) mimic previous observations for N1 and N2 adducts of BHTOH (5.06 and 4.31 ppm, respectively) (19). Isolation and Characterization of dG Adduct 3. The trace quantities of 3 formed under conditions used

Quinone Methide Reaction with Guanine

Chem. Res. Toxicol., Vol. 14, No. 9, 2001 1349

Table 2. 1H and 13C NMR Data for dG N1 Adduct (2) and Its Tris-tert-butyldimethylsilyl Derivative dG N1 adduct (2)a position 2 4 5 6 8 -CH210 11 12 13 14 15 a

δH (ppm)

δC (ppm)

8.05 5.28 7.14 6.77 7.13 6.98

154.9 150.2 117.2 157.8 136.5 39.3 155.9 123.7 129.2 120.0 129.0 116.0

silyl derivative of the dG N1 adduct (2)b δH (ppm)

7.71 5.18 7.08 6.91 7.19 6.94

δC (ppm) 154.4 150.1 117.7 158.1 136.7 39.9 153.7 127.3 129.0 123.1 129.8 119.9

In DMF-d7. b In CD3CN.

Figure 5. HMBC of guanine N7 adduct 3 in 0.2 M NaOD/D2O.

Figure 4. HMBC of dG adduct 2 in DMF-d7.

to prepare 1 (30% aqueous DMF) or 2 (85% aqueous DMF) were insufficient for structural characterization. However, generation of 3 was enhanced by maintaining neutral pH with potassium phosphate in 30% DMF. This afforded an isolated yield of 10% for 3 relative to dG and accounted for almost 30% of all purine alkylation under these conditions. Preliminary examination of 3 by 1H NMR immediately suggested that deglycosylation had occurred since no signals for the deoxyribose group were evident. Mass spectral analysis (FAB) further confirmed this observation. Such a process is typical for intermediates containing an N7 alkyl substituent (8, 19, 35). The UV absorbance maximum of 3 (280 nm) is also equivalent to that observed for a depurinated N7 adduct formed by the related BHTOH (λmax ) 282 nm) (19). Alkylation of N7 was formally established by the same one- and two-dimensional NMR techniques used successfully to solve the structures of 1 and 2 above. The poor solubility of 3 and incomplete resolution of its 13C NMR signals precluded extensive use of DMF-d7. An NMR solvent of 0.2 M NaOD in D2O was found to be suitable

for all subsequent characterization. The 1H and 13C NMR signals of 3 were first assigned as described for the dG N2 adduct 1 (Table 1). However, the chemical shifts of C4 and C5 were not distinguishable by HMBC in this case due to the absence of the deoxyribose H1′ that would otherwise provide a unique coupling to C4. Assignment of the downfield signal (159.5 ppm) as C4 and the remaining signal (110.6 ppm) as C5 was instead based on extensive literature precedence for guanine N7 adducts (36-39). Similarly, C6 and C2 that also lack discernible couplings were again assigned by literature precedence (36-39) and are consistent with heteronuclear NOE data for the dG N1 adduct 2 in which C6 is downfield relative to C2. Long-range correlations were observed by HMBC between the benzylic hydrogens (-CH2-) and C10, C11, C12 and, most importanly, purine C5 and C8 (Figure 5). Only alkylation of N7 would generate this pattern, and neither alkylation of O6, N1, N2, or N3 would exhibit comparable coupling to C8. The facile deglycosylation leading to 3 is again expected for N7 alkylation. Characterization of dG Adduct 4. An additional adduct 4 was observed by HPLC after reaction of QMP and dG (Figure 1). This material was also isolated by semipreparative HPLC. During subsequent removal of solvent by lyophilization, 85% of 4 decomposed to 3 as indicated by reanalysis using HPLC. Examination of the resulting mixture by 1H NMR (DMF-d7) verified this decomposition and indicated an equivalent 85% loss of the deoxyribose hydrogens. Thus, 4 appears to be the initial N7 adduct of dG that spontaneously converts to the guanine derivative 3. The UV spectrum of 4 observed during HPLC elution exhibited an absorbance maximum at 260 nm (pH 4), closely corresponding to that of N7 methyl dG (257 nm, pH 3; 258 nm, pH 7) (40). Further characterization of 4 was prevented by its extreme lability.

1350

Chem. Res. Toxicol., Vol. 14, No. 9, 2001

Figure 6. Reversed-phase chromatographic analysis of nucleoside adducts formed from reaction of QMP with (A) dG (0.5 mM) and (B) calf thymus DNA (18 mM nucleotides) in potassium phosphate (10 mM, pH 7) and 30% CH3CN. Reactions with calf thymus DNA were digested with alkaline phosphatase and phosphodiesterase prior to HPLC analysis (20).

Alkylation Profile of dG. Identification of the adducts (1-4) was a prerequisite for returning to questions on the specificity of guanine alkylation at the deoxynucleoside and duplex DNA levels. Previous analytical conditions had focused on only the major adduct and had not resolved products 1 and 2 (20). The improved HPLC resolution illustrated in Figure 1 allowed for a more complete analysis of the derivatives formed by QMP and for direct comparison to the product profile formed by its more stable and sterically hindered para analogues (such as BHTOH) published earlier (19). Most dramatically, the dG N1 adduct 2 formed more readily than the N2 (1) or N7 (3, 4) adducts by g15-fold in unbuffered 85% aqueous DMF as indicated above. The N2 rather than N1 adducts of dG are customarily produced in greatest yield under neutral pH with quinone methide intermediates (12, 13, 19, 41). Thus, efficient formation of the N1 adduct 2 in this example is most likely due to deprotonation of N1 (pKa ) 9.2) (40) since the unbuffered conditions used to form 2 become mildly basic upon addition of potassium fluoride. Interestingly, the behavior of the dG anion is very much influenced by its electrophilic partner and reaction conditions. Deprotonation of dG appears to promote alkylation of its N2 position by benzyl bromide under polar protic conditions, although the specificity shifts to the N1 position with a polar aprotic solvent (42). Yield of the N1 adduct 2 decreased significantly below that of the common N2 adduct 1 when the reactivity of QMP was examined with dG under neutral conditions equivalent to those used for the DNA analysis below (pH 7, 30% aqueous CH3CN). Quantitative analysis of the products by HPLC indicated a ratio of 4.7:3.5:1.0 for the products N2 (1):N7 (3 + 4):N1 (2) (Figure 6A). Interestingly, this ratio does not significantly differ from that measured for the more stable para-quinone methide generated by BHTOH (5.5:2.8:1.0) (19). Despite the highly reactive nature predicted for unsubstituted ortho-

Veldhuyzen et al.

quinone methides (43-46), QMP produces an electrophilic intermediate that is not completely quenched under aqueous conditions and is instead capable of forming adducts with the cyclic and exocyclic nitrogens. The product profile is not dependent on the intrinsic nucleophilicity of each site and may instead be related to the thermodynamics more than the kinetics of reaction, a topic under current investigation. Alkylation Profile of dG in Duplex DNA. Product determination was reinvestigated for duplex DNA since our original analysis had unwittingly measured the combined yields of adducts 1 and 2 (20). An overall suppression of quinone methide reactivity was again expected and observed as a result of the reduced accessibility of nucleophiles in duplex DNA relative to its constituent deoxynucleosides. Modification of dG N2 still dominated the profile of DNA, and the relative reactivity of dC and dA was similar to that reported previously (Figure 6B) (20). The selectivity within dG residues of DNA could now also be determined, and the relative yield of the N2 adduct 1 was found to increase (7.8:3.2:1.0 for adducts of N2:N7:N1 in duplex DNA) over that formed by the deoxynucleoside. The diminished reactivity of dG N1 is consistent with its central position within the WatsonCrick hydrogen-bond array of helical DNA and far removed from the solvent accessible surface. In contrast, the N7 and N2 groups maintain exposure on the surface of the major and minor grooves, respectively (6). Accessibility alone cannot control reaction since the preference for alkylation of N2 was enhanced from deoxynucleoside to DNA. The highly ordered structure of duplex DNA also effects the electrostatic potential of its components (6), and reaction at dG N2 in particular may be assisted by the ability of the O2 group of the complementary dC to act as a general base (47). The hydrophobic nature of the minor groove may additionally promote reaction by increasing the local concentration of hydrophobic reactants such as QMP (20). Dominant formation of the N2 adduct of dG (1) with QMP has now been confirmed under conditions that resolve numerous products of dG and DNA alkylation. Reassignment of the NMR spectra of 1 was made possible by COSY and heteronuclear NOE experiments as well as by changing NMR solvents. In the future, related structures may be implied merely by the diagnostic chemical shift of their benzylic hydrogens since adducts formed with both ortho- and para-quinone methide intermediates exhibit equivalent trends (19). While the variable reactivity of dG N1 may be rationalized by its protonation state in part, the remaining processes controlling the specificity between conjugated electrophiles such as quinone methides and other nucleophiles of DNA may now be examined. Optimal pairing of nucleophilic and electrophilic properties likely represent only one determinant, and product stability and kinetics may play a substantial but as yet unappreciated role (16).

Acknowledgment. This work was supported in part by the National Institutes of Health (CA81571). We thank Ms. Carolyn Ladd for the mass spectral analyses and Dr. Gene Mazzola for technical advice and assistance. Supporting Information Available: Full HMQC and HMBC of the dG N2 (1), dG N1 (2), and guanine N7 (3) adducts.

Quinone Methide Reaction with Guanine Full HMBC and heteronuclear NOE data of the tris-tertbutyldimethylsilyl derivative of the dG N1 adduct (2). Full 2D COSY of the dG N2 (1) and dG N1 (2) adducts. This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Hathway, D. E., and Kolar, G. F. (1980) Mechanism of Reaction between Ultimate Chemical Carcinogens and Nucleic Acids. Chem. Soc. Rev. 9, 241-264. (2) Singer, B., and Grunberger, D. (1983) Reactions of Directly Acting Agents with Nuclei Acids. Molecular Biology of Mutagens and Carcinogens, Chapter 4, pp 45-141, Plenum, New York. (3) Hartley, J. A. (1993) Selectivity in Alkylating Agent-DNA Interactions. In Molecular Aspects of Anticancer Drug-DNA Interactions (Neidle, S., and Waring, M., Eds.) Vol. 1, pp 1-31, CRC Press, Boca Raton. (4) Loechler, E. L. (1994) A Violation of the Swain-Scott Principle, and Not SN1 versus SN2 Reaction Mechanisms, Explains Why Carcinogenic Alkylating Agents Can Form Different Proportions of Adducts at Oxygen versus Nitrogen in DNA. Chem. Res. Toxicol. 7, 277-280. (5) Barlow, T., and Dipple, A. (1998) Aralkylation of Guanosine with para-Substituted Styrene Oxides. Chem. Res. Toxicol. 11, 4453. (6) Pullman, A., and Pullman, B. (1981) Molecular Electrostatic Potential of Nucleic Acids. Q. Rev. Biophys. 14, 289-380. (7) Zhu, Q., and LeBreton, P. R. (2000) DNA Photoionization and Alkylation Patterns in the Interior of Guanine Runs. J. Am. Chem. Soc. 122, 12824-12834. (8) Warpehoski, M. A., and Hurley, L. H. (1988) Sequence Selectivity of DNA Covalent Modification. Chem. Res. Toxicol. 1, 315-333. (9) Mattes, W. B., Hartley, J. A., and Kohn, K. W. (1986) DNA Sequence Selectivity of Guanine N-7 Alkylation by Nitrogen Mustards. Nucleic Acids Res. 14, 2971-2987. (10) Belousov, E. S., Afonina, I. A., Podyminogin, M. A., Gamper, H. B., Reed, M. W., Wydro, R. M., and Meyer, R. B. (1997) SequenceSpecific Targeting and Covalent Modification of Human Genomic DNA. Nucleic Acids Res. 25, 3440-3444. (11) Wurtz, N. R., and Dervan, P. B. (2000) Sequence Specific Alkylation of DNA by Hairpin Pyrrole-Imidazole Polyamide Conjugates. Chem. Biol. 7, 153-161. (12) Tomasz, M., Chowdary, D., Lipman, R., Shimotakahara, S., Veiro, D., Walker, V., and Verdine, G. (1986) Reaction of DNA with Chemically or Enzymatically Activated Mitomycin C: Isolation and Structure of the Major Covalent Adduct. Proc. Natl. Acad. Sci. U.S.A. 83, 6702-6706. (13) Egholm, M., and Koch, T. (1989) Coupling of the Anthracycline Antitumour Drug Menogaril to 2′-Deoxyguanosine through Reductive Activation. J. Am. Chem. Soc. 111, 8291-8293. (14) Woo, J., Sigurdsson, S. T., and Hopkins, P. B. (1993) DNA Interstrand Cross-Linking Reactions of Pyrrole-Derived Bifunctional Electrophiles: Evidence for a Common Target Site in DNA. J. Am. Chem. Soc. 115, 3407-3415. (15) Palom, Y., Lipman, R., Musser, S. M., and Tomasz, M. (1998) A Mitomycin-N6-Deoxyadenosine Adduct Isolated from DNA. Chem. Res. Toxicol. 11, 203-210. (16) Ouyang, A., and Skibo, E. B. (2000) Iminium Ion Chemistry of Mitosene DNA Alkylating Agents. Enriched 13C NMR and Isolation Studies. Biochemistry 39, 5817-5830. (17) Paz, M. M., Sigurdsson, S. T., and Hopkins, P. B. (2000) Monoalkylation of DNA by Reductively Activated FR66979. Bioorg. Med. Chem. 8, 173-179. (18) Angle, S. R., and Yang, W. (1992) Nucleophilic Addition of 2′Deoxynucleosides to the o-Quinone Methides 10-(Acetyloxy) and 10-Methoxy-3,4-dihydro-9(2H)-anthracenone. J. Org. Chem. 57, 1092-1097. (19) Lewis, M. A., Yoerg, D. G., Bolton, J. L., and Thompson, J. A. (1996) Alkylation of 2′-Deoxynucleosides and DNA by Quinone Methides Derived from 2,6-Di-tert-butyl-4-methylphenol. Chem. Res. Toxicol. 9, 1368-1374. (20) Pande, P., Shearer, J., Yang, J., Greenberg, W. A., and Rokita, S. E. (1999) Alkylation of Nucleic Acids by a Model Quinone Methide. J. Am. Chem. Soc. 121, 6773-6779. (21) Rokita, S. E., Yang, J., Pande, P., and Greenberg, W. A. (1997) Quinone Methide Alkylation of Deoxycytidine. J. Org. Chem. 62, 3010-3012. (22) Kumar, G. S., Musser, S. M., Cummings, J., and Tomasz, M. (1996) 2,7-Diaminomitosene, a Monofunctional Mitomycin C Derivative, Alkylates DNA in the Major Groove. Structure and Base-Sequence Specificity of the DNA Adduct and Mechanism of Alkylation. J. Am. Chem. Soc. 118, 9209-9217.

Chem. Res. Toxicol., Vol. 14, No. 9, 2001 1351 (23) Maliepaard, M., de Mol, N. J., Tomasz, M., Gargiulo, D., Janssen, L. H. M., van Duynhoven, J. P. M., van Velzen, E. J. J., Verboom, W., and Reinhoudt, D. N. (1997) Mitosene-DNA Adducts. Characterization of Two Major DNA Monoadducts Formed by 1,10Bis(acetoxy)-7-methoxymitosene upon Reductive Activation. Biochemistry 36, 9211-9220. (24) Chatterjee, M., and Rokita, S. E. (1996) Inducible Alkylation of DNA Using an Oligonucleotide Quinone Conjugate. J. Am. Chem. Soc. 112, 6397-6399. (25) Li, T., and Rokita, S. E. (1991) Selective Modification of DNA Controlled by an Ionic Signal. J. Am. Chem. Soc. 113, 7771-7773. (26) Singer, B. (1972) Reaction of Guanosine with Ethylating Agents. Biochemistry 11, 3939-3947. (27) Bax, A., and Summers, M. F. (1986) 1H and 13C Assignments from Sensitivity-Enhanced Detection of Heteronuclear Multiple-Bond Connectivity by 2D Multiple Quantum NMR. J. Am. Chem. Soc. 108, 2093-2096. (28) Bax, A., and Subramanian, S. J. (1986) Sensitivity-Enhanced TwoDimensional Heteronuclear Shift Correlation NMR Spectroscopy. J. Magn. Reson. 67, 565-569. (29) Ochs, S., and Severin, T. (1994) Reaction of 2′-Deoxyguanosine with Glyceraldehyde. Liebigs. Ann. Chem. 851-853. (30) Ochs, S., and Severin, T. (1995) Reaction of 2′-Deoxyguanosine with Glucose. Carbohydr. Res. 266, 87-94. (31) Seela, F., Heckel, M., and Rosemeyer, H. (1996) 122. Xylose-DNA Containing the Four Natural Bases. Helv. Chim. Acta 79, 14511461. (32) Box, H. C., Lilga, K. T., Alderfer, J. L., French, J. B., and Potienko, G. (1979) 13C NMR Characterization of Alkyl Derivatives of Guanosine. J. Carbohydr. Nucleosides Nucleotides 6, 255-262. (33) Chang, C. J., Ashworth, D. J., Chern, L. J., Gomes, J. D., Lee, C. G., Mou, P. W., and Narayan, R. (1984) 13C NMR Studies of Methylnucleosides. Org. Magn. Reson. 22, 671-675. (34) Uzawa, J., and Uramoto, M. (1979) Assignment of Indirect Carbon-13-proton Couplings in the Carbon-13 Spectra of Some Purine and Pyrimidine Nucleosides and Their Analogs by LongRange Selective Proton Decoupling. Org. Magn. Reson. 12, 612615. (35) Maxam, A. M., and Gilbert, W. (1980) Sequencing End-Labeled DNA With Base-Specific Chemical Cleavages. Methods Enzymol. 65, 499-560. (36) Kjellberg, J., Hagberg, C. E., Malm, A., Noren, J. O., and Johansson, N. G. (1986) Studies on the Alkylation of Guanine. 2. The Synthesis of Acyclic Guanosine Analogs via the Precursor 7-Methyl-10-oxo-9, 10-dihydropyrimido[1,2-a]purine. Acta Chem. Scand. Ser. B 40, 310-312. (37) Bailey, S., and Harnden, M. R. (1988) Analogues of the Antiviral Acyclonucleoside 9-(4-Hydroxy-3-hydroxymethylbutyl)guanine. Part 2. Substitutions on C-1′ and C-3′ of the Acyclic N-9 Substituent. J. Chem. Soc., Perkin Trans. 2767-2776. (38) Crippa, S., Gennaro, P. D., Lucini, R., Orlandi, M., and Rindone, B. (1993) Characterization of Adducts of Nucleic Bases and Acrylic Monomers. Gazz. Chim. Ital. 123, 197-203. (39) Loeppky, R. N., Yu, L., Gu, F., and Ye, Q. (1996) DNA Guanine Adducts from 3-Methyl-1,2,3-oxadiazolinium Ions. J. Am. Chem. Soc. 118, 10995-11005. (40) Dawson, R. M. C., Elliott, D. C., Elliott, W. H., and Jones, K. M. (1986) Data for Biochemical Research, 3rd ed., Chapter 5, pp 103114, Oxford University Press, New York. (41) Tomasz, M., Chawla, A. K., and Lipman, R. (1988) Mechanism of Monofunctional and Bifunctional Alkylation of DNA by Mitomycin C. Biochemistry 27, 3182-3187. (42) Moon, K.-Y., and Moschel, R. C. (1998) Effect of Ionic State of 2′-Deoxyguanosine and Solvent on Its Aralkylation by Benzyl Bromide. Chem. Res. Toxicol. 11, 696-702. (43) Filar, L. J., and Winstein, S. (1960) Preparation and Behaviour of Simple Quinone Methides. Tetrahedron Lett. 9-16. (44) Wan, P., Barker, B., Diao, L., Fischer, M., Shi, Y., and Yang, C. (1996) Quinone Methides: Relevant Intermediates in Organic Chemistry. Can. J. Chem. 74, 1996. (45) Chiang, Y., Kresge, J., and Zhu, Y. (2000) Kinetics and Mechanisms of Hydration of o-Quinone Methides in Aqueous Solution. J. Am. Chem. Soc. 122, 9854-9855. (46) Modica, E., Zanaletti, R., Freccero, M., and Mella, M. (2001) Alkylation of Amino Acids and Glutathione in Water by o-Quinone Methides. Reactivity and Selectivity. J. Org. Chem. 66, 41-52. (47) Dannenberg, J. J., and Tomasz, M. (2000) Hydrogen-Bond Acid/ Base Catalysis: A Density Functional Theory Study of Protonated Guanine-(Substituted) Cytosine Base Pairs as Models for Nucleophilic Attack on Mitomycin in DNA. J. Am. Chem. Soc. 122, 2062-2068.

TX0101043