Time-Dependent Evolution of Adducts Formed between

for weak nucleophiles of DNA even when proximity effects do not guide reaction. On the basis ..... Results. Evolution of dA Adducts Generated by a Mod...
0 downloads 0 Views 150KB Size
1364

Chem. Res. Toxicol. 2005, 18, 1364-1370

Time-Dependent Evolution of Adducts Formed between Deoxynucleosides and a Model Quinone Methide Emily E. Weinert, Kristen N. Frankenfield, and Steven E. Rokita* Department of Chemistry and Biochemistry, University of Maryland, College Park, Maryland 20742 Received June 14, 2005

Highly electrophilic quinone methide (QM) intermediates often express a surprising selectivity for weak nucleophiles of DNA even when proximity effects do not guide reaction. On the basis of model studies with an unsubstituted ortho-QM, these observations can now be explained by the reversibility of QM alkylation and the time-dependent shift from kinetic to thermodynamic products. The persistent and most commonly identified QM adducts represent thermodynamic products that typically form in low yield by irreversible reaction with weak nucleophiles such as the N1 and N2 of dG and the N6 of dA under neutral conditions. In contrast, strong nucleophiles such as the N1 of dA and the N3 of dC generate relatively high yields of their QM adducts. However, these products dissipate over time as the QM is repeatedly regenerated and repartitioned over the available nucleophiles. The adduct formed by the N7 of dG undergoes a similar release of QM as well as deglycosylation at comparable rates. The kinetic products of QM alkylation serve as a reservoir for QM regeneration and transfer that are likely to prolong the cellular activity of an otherwise highly transient intermediate.

Introduction A well-practiced approach for assessing the toxicity and carcinogenicity of a new compound involves identifying its metabolic products in vitro and then assessing the effect of each individually in vitro and in vivo. An equivalent strategy is also followed when a series of DNA lesions are formed. First, their structures are determined, and then, each lesion is subsequently examined for its relative efficiency of repair and mutation. The significance of these latter investigations directly depends on the quality of information available on the range and yields of lesions formed. If lesions remain undetected, unidentified, or underestimated, the biological impact of the parent substance can be seriously misjudged. These concerns are particularly relevant to species that form reversible or transient products with DNA. For example, the mutagen malondialdehyde reacts with guanine reversibly allowing for its relatively efficient transfer to DNA through transient formation of diffusible deoxynucleoside adducts (1, 2). Similarly, the potent anticancer drug candidate Ecteinascidin 743 forms a covalent but reversible bond to the N2 (2-amino) of guanine in DNA (Scheme 1) allowing for its return to such sites after an initial response by excision repair (3). DNA lesions generated by highly reactive and electrophilic quinone methide (QM) and QM-related intermediates have been characterized during investigations on a wide variety of compounds including 2,6-di-tert-butyl-4methylphenol (BHT) (4), anthracycline antibiotics and their models (5, 6), mitomycin and its derivatives (710), and tamoxifen (11, 12). Most reports identify adduct formation at sites of low nucleophilicity such as the N2 of dG and N6 of dA (Scheme 1). In some cases, these results are easily rationalized by the effects of sequestering the reactive intermediate adjacent to the exocyclic * To whom correspondence should be addressed.

Scheme 1. Nitrogen Nucleophiles of DNA

amines, but this explanation does not easily extend to the QM formed by BHT since it would not likely associate with DNA in a specific orientation. Still, such minimal QM structures appeared to support a predominant alkylation of the exocyclic amines in DNA (4). A highly transient electrophile is typically expected to react through an early, or starting material-like, transition state and exhibit very little preference for its nucleophilic partner. Conversely, a stabilized electrophile is expected to react through a more productlike transition state and exhibit significant dependence on the nucleophilicity of its partner. The apparent specificity of simple QMs for weakly nucleophilic partners was therefore very unusual. Initial studies in our laboratory confirmed such a surprising specificity after prolonged incubations of a simple ortho-QM with deoxynucleosides or DNA (13, 14). A potential origin of this result was revealed by detailed analysis of reaction between a model QM and dA. Transient alkylation of N1, the most nucleophilic site of dA, was detected, and its product was shown to serve in turn as a further source of QM for transfer to sites reacting irreversibly (15). Thus, a kinetic product formed initially, but reversibly and ultimately, only the most stable adducts accumulated (Scheme 2). Density func-

10.1021/tx0501583 CCC: $30.25 © 2005 American Chemical Society Published on Web 08/30/2005

Kinetic and Thermodynamic Products of DNA Alkylation Scheme 2. QM Generation and Alkylation of Nucleosides by Reversible and Irreversible Reactions

tional theory had also been used to predict the lability of the dA N1 as well as dG N7 adducts (16) and is consistent with our observations. Our laboratory has since undertaken a kinetic study of reaction between a model QM and deoxynucleosides as described below to examine the role of product stability in the apparent specificity of QM reaction. Previous investigations with the model ortho-QM had not considered the potential for a time-dependent evolution of the alkylation products although some lability of products formed by other QMs had been suggested (4, 6).

Materials and Methods Materials. Solvents, starting materials, and reagents of the highest commercial grade were used without further purification. All aqueous solutions were prepared with distilled and deioinized water with a resistivity of 18.0 MΩ. Silica gel (230400 mesh) for column chromatography was purchased from EM Sciences. 1H and 13C spectra were recorded on a DRX 400 spectrometer (1H, 400.13 MHz; 13C, 100.62 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). High resolution mass spectra were determined with a JEOL SX102 mass spectrometer. Authentic samples of each deoxynucleoside adduct for use as chromatographic standards were prepared as described previously (13, 14). 2-(Acetoxymethyl)phenol (1). A solution of BF3‚Et2O (400 µL, 2.5 mmol) in acetic anhydride (3 mL) was added dropwise to a solution of 2-(hydroxymethyl)phenol (1.24 g, 10.0 mmol) in THF (5 mL) at 0 °C. The reaction was stirred at 0 °C for 4 h and then neutralized to pH 7 by dropwise addition of cold saturated NaHCO3. The resulting mixture was extracted with ether (3 × 50 mL). The organic layer was washed with brine (50 mL), dried with MgSO4, filtered, and concentrated under reduced pressure to yield a yellow oil. The desired product was purified by silica gel flash chromatography (hexanes:ethyl acetate step gradient of 17:3, 4:1, and finally 7:3) and isolated as a colorless oil (1.36 g, 83%). 1H NMR (CDCl3): δ 2.11 (s, 3H), 5.12 (s, 2H), 6.92 (m, 2H), 7.28 (m, 2H), 7.78 (s, 1H). 13C NMR (CDCl3): δ 21.1, 63.5, 118.0, 120.8, 121.9, 131.3, 132.4, 155.7, 173.9. 2-(Acetoxymethyl)-O-tert-butyldimethylsilylphenol (AcQMP). A solution of 1 (200 mg, 1.20 mmol) in DMF (6 mL) was combined under nitrogen with tert-butyldimethylsilyl chloride (908 mg, 6.02 mmol) and imidazole (816 mg, 12.2 mmol) at room temperature. The solution was stirred at room temperature overnight and then quenched by addition of H2O (20 mL). The mixture was extracted with ethyl acetate (3 × 20 mL). The combined organic phases were washed with brine, dried with MgSO4, and concentrated under reduced pressure to yield a crude product (oil). The desired compound was purified by silica gel flash chromatography (hexanes:ethyl acetate step gradient of 99:1, 49:1, and finally 19:1) and isolated as a colorless oil (320

Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1365 mg, 95%). 1H NMR (CDCl3): δ 0.23 (s, 6H), 0.99 (s, 9H), 2.06 (s, 3H), 5.10 (s, 2H), 6.81 (dd, J ) 8.0 and 1.1, 1H), 6.93 (td, J ) 8.0 and 1.1, 1H), 7.19 (td, J ) 8.0 and 1.5, 1H), 7.29 (dd, J ) 8.0 and 1.5, 1H). 13C NMR (CDCl3): δ -4.2, 18.2, 21.0, 25.6, 62.3, 118.5, 121.0, 126.4, 129.5, 130.4, 154.0, 171.0. HRMS (FAB) m/z 281.1599 (M + H+). Calcd for C15H24O3Si (M + H+): 281.1573. 2-(Bromomethyl)-O-tert-butyldimethylsilylphenol(BrQMP) (17). N-Bromosuccinimide (270 mg, 1.5 mmol) was added to a solution of 2-methyl-O-tert-butyldimethylsilylphenol (18) (300 mg, 1.4 mmol) in CCl4 (20 mL). The solution was heated to reflux, and then, AIBN (6.8 mg, 0.038 mmol) was added. The reaction was refluxed for 35 min, cooled, and filtered. The filtrate was washed with water, dried with MgSO4, and concentrated under reduced pressure. BrQMP 4 was purified by silica gel flash chromatography (hexanes:ethyl acetate, 97:3) to yield a colorless oil (0.24 g, 57%). 1H NMR (CDCl3): δ 0.28 (s, 6H), 1.04 (s, 9H), 4.51 (s, 2H), 6.79 (d, J ) 7.8, 1H), 6.90 (t, J ) 7.8, 1H), 7.17 (td, J ) 7.8 and 1.6, 1H), 7.31 (dd, J ) 7.8 and 1.6, 1H). Phenylhydrazine-QM Adduct [2-(N′-Phenylhydrazinomethyl)phenol]. A solution of BrQMP and phenylhydrazine in DMF was added to an aqueous solution of potassium fluoride, yielding final concentrations of 75 mM BrQMP, 75 mM phenylhydrazine, and 500 mM KF in a DMF:H2O mixture (80:20, 0.4 mL). The reaction was incubated (15 min) at 37 °C until all starting material had converted to a derivative with a new retention time by HPLC. This product was purified by preparative HPLC [3% CH3CN, 9.7 mM triethylammonium acetate (TEAA), pH 4, to 25% CH3CN, 7.5 mM TEAA, pH 4, over 66 min, 5 mL/min] and identified as the desired adduct. 1H NMR (DMSO-d6): δ 4.52 (s, 2H), 4.58 (s, 2H), 6.58-7.14 (m, 9H). 13C NMR (DMSO-d6): δ 54.5, 112.3, 115.3, 116.4, 118.7, 124.2, 127.7, 128.5, 128.6, 151.6, 155.5. HRMS (FAB) m/z 214.1106. Calcd for C13H14N2O (M+) 214.1111. General Methods. Preparative and analytical HPLC were performed on both a Jasco PU-908/MD1510 diode array instrument and a Jasco PU-2080 PLUS/UV-2077 PLUC fixed wavelength instrument. Analytical samples used a reverse phase C-18 analytical column (Varian, Microsorb-MV 300, 5 µm particle size, 250 mm × 4.6 mm) with a flow rate of 1 mL/min. Preperative samples used a semiprep column (Alltech, Econosphere C-18, 10 µm, 250 mm × 10 mm) with a flow rate of 5 mL/min. UV-vis spectra were measured on an HP 8543 series sprectrophotometer. Molar absorptivities () of the deoxynucleosides, 2-hydroxymethylphenol, and 2-(N′-phenylhydrazinomethyl)phenol were measured in 10 mM TEAA, pH 4, using serial dilutions. Adduct formation was quantified using HPLC. Areas of the deoxynucleoside-QM adducts were compared at λ260 relative to an internal standard (phenol) at λ260. Molar absorptivities of the deoxynucleoside adducts were estimated by the sum of 260 values for the individual deoxynucleoside and 2-hydroxymethylphenol. Time-Dependent Profile of Alkylation Products Formed by Each Deoxynucleoside Individually. AcQMP in DMF (30 µL) was combined with an aqueous solution (70 µL) of the other reaction components to yield a final concentration of 25 mM AcQMP, 4 mM phenol, 0.5 mM each deoxynucleoside (dA, dC, and dG alternatively), 10 mM potassium phosphate, pH 7, and 500 mM KF. The solution was incubated at 37 °C, and aliquots were analyzed at the indicated times by reverse phase HPLC using a linear gradient of 3% CH3CN, 9.7 mM TEAA, pH 4, to 25% CH3CN, and 7.5 mM TEAA, pH 4, over 66 min. Deoxynucleoside Competition Studies. The conditions described above for examing each deoxynucleoside independently were also used for the competitive studies containing dA, dC, dG, and T together. Each deoxynucleoside was present at 0.25 mM to yield a total deoxynucleoside concentration of 1.0 mM. Partitioning of the Unstable QM-dG N7 Adduct. dG in DMF (21 mM, 24 µL) was combined with an aqueous solution (50 µL) to yield final concentrations of dG (5.0 mM), phenol (4.0 mM), potassium phosphate (pH 7, 10 mM), and KF (500 mM).

1366

Chem. Res. Toxicol., Vol. 18, No. 9, 2005

Weinert et al.

Figure 1. Time-dependent profile of dA alkylation by an unsubstituted ortho-QM. Reaction products were separated and quantified by reverse phase (C-18) chromatography and monitored at A260. Data represent averages of two independent analyses and were fit to a single exponential. These lines are added to indicate product trends and do not represent kinetic modeling.

Figure 2. Time-dependent profile of dC alkylation by an unsubstituted ortho-QM. Reaction products were separated and quantified by reverse phase (C-18) chromatography and monitored at A260. Data represent averages of two independent analyses and were fit to two exponential functions. These lines are added to indicate product trends and do not represent kinetic modeling.

The reaction was initiated by adding BrQMP in DMF (6 µL) to a final concentration of 5.0 mM. The resulting mixture was incubated for 30 min (37 °C) and quenched by addition of phenylhydrazine in DMF (50 mM, 2 µL) or, as a control, just DMF (2 µL). Incubation at 37 °C was maintained, and aliquots were removed at the indicated times for analysis by reverse phase HPLC under analytical conditions (3% CH3CN, 9.7 mM TEAA, pH 4, to 28% CH3CN, and 7.2 mM TEAA, pH 4, over 78 min at 1 mL/min).

half-life of little more than 4 h. In contrast, less than 2% of dA reacted at its N6 position during the first 30 min; yet, the yield of this alternative product continued to increase slowly to a yield of 14% over more than 10 h and at the expense of the dA N1. This suggests a transfer of QM from the transient dA N1 product to the stable dA N6 product with an efficiency of approximately 50%. No decomposition of the N6 product was detected over the course of the 72 h incubation. Evolution of dC Alkylation. The N3 alkylated derivative of dC (Scheme 1) was the only product formed by the unsubstituted ortho-QM as previously predicted and detected (13, 18, 19), although reaction at the exocyclic amine (N4) had been additionally observed for a para-QM generated from a hydroxylated BHT (4). Reaction of dC was monitored in parallel to that described for dA above (Figure 2). The yield of the dC N3 adduct was equivalent to that for dA N1 after the first 30 min of reaction. However, the dC product continued to accumulate until all of the unprotected AcQMP had dissipated as indicated by HPLC analysis. The maximum yield of dC N3 adduct (19 nmol) represented approximately 40% reaction of the initial dC in excess QM. Slow decomposition of the dC N3 product was observed after 24 h, and less than 25% of the maximum remained after 72 h. However, no alternative product derived from dC was detected by HPLC. Slow release of the QM likely partitioned repeatedly between regenerating the dC N3 adduct and trapping with water irreversibly. Evolution of dG Alkylation. The product profile generated by dG and the unsubstituted ortho-QM under conditions equivalent to those above was considerably more complex than the profiles generated by either dA or dC. The combined yield of all dG adducts persisted at a level similar to the initial yield of the dA N1 adduct. However, this level was sustained only by formation of three stable and one unstable species (Figure 3). The N1 and N2 adducts of dG (Scheme 1) formed irreversibly and accumulated at similar rates. Neither site is strongly nucleophilic, and consequently, neither trapped the QM as effectively as dA N1 or dC N3.

Results Evolution of dA Adducts Generated by a Model QM. Our appreciation of the kinetic and thermodynamic products of DNA alkylation first arose while investigating the apparent selectivity of a simple ortho-QM for the weakly nucleophilic N6 of dA (15). Isotope labeling studies indicated that this stable product formed irreversibly from a bimolecular reaction with QM rather than from an intramolecular rearrangement of its N1-linked precursor. Reaction at the strongly nucleophilic N1 of dA was fast but reversible and ultimately regenerated the QM for terminal reaction with water and dA N6 (Scheme 2). These results provided an early suggestion that the product profile generated by deoxynucleoside alkylation with QM might be time dependent. Thus, dA was the first deoxynucleoside examined for a change in its product distribution over time. Subsequent studies also focused on dC and dG. The ortho-QM had not previously demonstrated a reactivity with T (13); therefore, this deoxynucleoside was only included in the final competition studies below. Reaction between dA and the unsubstituted ortho-QM was induced by fluoride-dependent deprotection of AcQMP, and the resulting products were monitored by reverse phase HPLC as described previously (15). A maximum yield of alkylation at the N1 position of dA was observed at the first time of analysis (30 min; Figure 1). Under these conditions, approximately 25% of the initial dA (0.5 mM) had reacted in competition with water for the large excess of QM (25 mM). This first product did not accumulate but rather decomposed quickly with a

Kinetic and Thermodynamic Products of DNA Alkylation

Chem. Res. Toxicol., Vol. 18, No. 9, 2005 1367

Figure 3. Time-dependent profile of dG alkylation by an unsubstituted ortho-QM. Reaction conditions and analysis are equivalent to those described for the reaction of dA in Figure 1. Again, the average of two independent analyses was fit to single (dG and guanine N7 adducts) and double (dG N1 and N2 adducts) exponential processes for highlighting the net change in product profile.

Figure 4. Competing pathways for decomposition of the dG N7 adduct were characterized by formation of their diagnostic guanine N7 and phenylhydrazine adducts. The parent dG N7 adduct was formed within 30 min by treatment with BrQMP in the presence of potassium fluoride, and subsequent QM formation was trapped by addition of phenylhydrazine (PH). The resulting adducts were separated and quantified by reverse phase HPLC. Data represent averages from a minimum of three independent analyses and were fit to single exponential processes for highlighting the net change of products.

The highly nucleophilic N7 of dG initially generated an adduct in yields that at first might be considered to be surprisingly low and only comparable to those of the weaker nucleophiles N1 and N2 of dG. Moreover, the concentration of dG N7 adduct even decreased as the dG N1 and N2 adducts increased with time. Both results are likely caused by the instability of the dG N7 adduct. Deglycosylation of this adduct was evident from its conversion to a stable guanine N7 adduct in a process common to alkylation or protonation of dG N7 (20). The dG N7 adduct also had the potential to release its QM in competition with deglycosylation. Regeneration of the QM would allow for its transfer from sites of reversible to irreversible reaction (dG N1 and N2) in analogy to the transfer of QM from dA N1 to dA N6 as characterized previously (15). To date, the potential for the dG N7 adduct to act as a QM source has been obscured by its competing deglycosylation. QM Regeneration vs Deglycosylation of the dG N7 Adduct. Ideally, the product of dG N7 alkylation formed by the unsubstituted ortho-QM should be monitored in isolation for its decomposition through the two competing paths. Previous attempts to purify the dG N7 product were not successful and yielded only the deglycosylated product (14). Consequently, partitioning between deglycosylation and QM regeneration was instead observed in situ directly after alkylation. This approach was best accommodated under conditions of maximum formation of the dG N7 adduct and minimal presence of residual QM or its equivalents from a precursor such as AcQMP. Unfortunately, these conditions are mutually exclusive since the greatest yield of dG N7 was measured at the shortest times (