Spectral Characterization of Fluorescently Labeled Catechol Estrogen

HPLC using fluorescence detection was conducted on a Waters 600E solvent delivery system equipped with a 700 WISP autoinjector, a Waters 990 photodiod...
0 downloads 0 Views 83KB Size
Chem. Res. Toxicol. 1998, 11, 1339-1345

1339

Spectral Characterization of Fluorescently Labeled Catechol Estrogen 3,4-Quinone-Derived N7 Guanine Adducts and Their Identification in Rat Mammary Gland Tissue Ryszard Jankowiak,*,† Dan Zamzow,† Douglas E. Stack,‡,§ Rosa Todorovic,‡ Ercole L. Cavalieri,‡ and Gerald J. Small† Ames Laboratory-USDOE, Iowa State University, Ames, Iowa 50011, and Eppley Institute for Research in Cancer, University of Nebraska Medical Center, Omaha, Nebraska 68198 Received May 26, 1998

The oxidation of carcinogenic 4-hydroxycatechol estrogens (CE) of estrone (E1) and estradiol (E2) to catechol estrogen 3,4-quinones (CE-3,4-Q) results in electrophilic intermediates that covalently bind to DNA to form depurinating adducts [Cavalieri et al. (1997) Proc. Natl. Acad. Sci. U.S.A. 94, 10937]. These DNA adducts, 4-OHE1-1-N7Gua and 4-OHE2-1-N7Gua, are nonfluorescent. To utilize laser-excited fluorescence methods, the catechol estrogen-derived metabolites and adducts were labeled with a fluorescent marker. The 4-OHEi-1-N7Gua adduct standards (i ) 1, 2) and 4-OHEi metabolites have been derivatized with 1-pyrenesulfonyl chloride and investigated by low-temperature spectroscopy under non-line-narrowing and linenarrowing conditions. Molecular modeling studies assisted in interpretation of the fluorescence spectra; energetically favored structures of the 4-OHE2-1-N7Gua-dipyrene adduct and 4-OHE2-dipyrene metabolite reveal unique conformations which, in agreement with fluorescence data, show a significant π-π interaction of pyrene labels with guanine and/or the aromatic ring of catechol estrogen. The conformation obtained for the 4-OHE2-1-N7Gua-dipyrene adduct appears to be conducive to mixing of its ππ* state with pyrene-guanine charge-transfer states, consistent with the experimentally observed strong electron-phonon coupling. Non-linenarrowed and line-narrowed spectra obtained at 77 and 4.2 K, respectively, are shown to distinguish 4-OHE2-1-N7Gua-dipyrene adducts from 4-OHE2-dipyrene metabolites. These standards have subsequently been used for the spectroscopic identification of depurinating DNA adducts formed in a tissue culture experiment where rat mammary gland tissue was treated with the estrogen quinone E2-3,4-Q. The depurinating adduct formed is 4-OHE2-1N7Gua.

Introduction Evidence that estrogens can exert their roles as tumor initiators is provided by data on their carcinogenicity in animal models, in particular the Syrian golden hamster (1-4). It has been established that the estrogens, estrone (E1)1 and 17β-estradiol (E2), are metabolized via two major pathways: 16R-hydroxylation and formation of 2-hydroxy- and 4-hydroxycatechol estrogens (CE). CE are normally inactivated by O-methylation catalyzed by * Corresponding author. E-mail: [email protected]. † Ames Laboratory, USDOE, Iowa State University. ‡ University of Nebraska Medical Center. § Present address: Department of Chemistry, University of Nebraska at Omaha, Omaha, NE 68182-0109. 1 Abbreviations: Ade, adenine; BP, benzo[a]pyrene; CE, catechol estrogen(s); CE-Q, catechol estrogen quinone(s); CE-SQ, catechol estrogen semiquinone(s); CuOOH, cumene hydroperoxide; DMF, dimethylformamide; DMSO, dimethyl sulfoxide; D-MEM/F-12, Dulbecco’s modified Eagle’s medium/nutrient mixture F-12; E1, estrone; E2, 17βestradiol; FLNS, fluorescence line narrowing spectroscopy; Gua, guanine; Gly, glycerol; HPLC, high-performance liquid chromatography; HRP, horseradish peroxidase; 2-OH-CE, 2-hydroxycatechol estrogen; 4-OHE2, 4-hydroxyestradiol; MM, molecular mechanics; NLN, non-line-narrowing; PAH, polycyclic aromatic hydrocarbon(s); PDA, photodiode array; RMG, rat mammary gland; SQ, semiquinones; S0 state, electronic ground state; S1 state, lowest excited singlet state; ZPLs, zero-phonon lines.

catechol-O-methyltransferases, but they can also be oxidized to semiquinones (CE-SQ) and quinones (CE-Q) that react directly with DNA and other cellular macromolecules (5, 6). The 2-OH-CE metabolite is not considered carcinogenic, because it fails to show up as a carcinogen in animals (6-8). 32P-postlabeling studies have shown that the noncarcinogenic 2-OH-CE generate more stable adducts than the carcinogenic 4-OH-CE (5). Liehr has demonstrated that 32P-postlabeling profiles of kidney DNA from Syrian hamsters exposed to natural and synthetic estrogens exhibited identical profiles irrespective of the estrogen compound (10). These studies indicate that stable adducts generated from estrogen exposure are not derived from estrogen metabolites. Whether tumor formation, induced by estrogen exposure in animal models, is the result of stable or depurinating adducts is still unclear. It is believed, however, that both 4-OH-CE and 16R-OH-E1 may damage DNA and lead to tumor development (8, 9). Recently, it has been shown that oxidation of 4-OH-CE to CE-3,4-Q results in the formation of electrophilic intermediates that react with DNA, both in vitro and in vivo, to form depurinating 4-OHEi-1-N7Gua (i ) 1,2) adducts (6, 9).

10.1021/tx980119+ CCC: $15.00 © 1998 American Chemical Society Published on Web 10/15/1998

1340 Chem. Res. Toxicol., Vol. 11, No. 11, 1998

NMR studies have shown that 4-OHEi-1-N7Gua adducts exist as a mixture of two conformational isomers, R and β (6). In the R conformer, the Gua moiety is mainly located below the estrogen plane (4-OHEi-1R-N7Gua), whereas in the β conformer the Gua moiety resides primarily above the estrogen plane (4-OHEi-1β-N7Gua). The structures of these adducts have also been elucidated by fast atom bombardment tandem mass spectrometry (6). The catechol estrogen 4-OHEi form 4-OHEi-1-N7Gua depurinating adducts, in the presence of oxidizing enzymes, e.g., horseradish peroxidase and/or cytochrome P450 (9). Binding at the N7-position of Gua results in cleavage of the glycosidic bond, leaving apurinic sites in DNA. Current methodologies for isolating, quantitating, and characterizing CE-derived DNA adducts at pico- or femtomolar level are underdeveloped. Several issues must be addressed to establish a protocol that can effectively identify both CE metabolites and adducts. First, isolation of CE from a biological matrix, especially fatty tissues, is difficult since large amounts of lipophilic material can complicate extraction procedures. Second, CE are prone to oxidative degradation. The UV absorbance of the catechol moiety at 290 nm is fairly weak and not suited for detection at low, biological levels. Because CE-3,4-Q-derived adducts such as 4-OHEi-1N7Gua are nonfluorescent, only electrochemical techniques can be used for quantitation and low-level detection of these catechol estrogens. This technique, however, requires extensive cleanup of the sample to reduce background noise. In addition, samples must be carefully handled to avoid oxidative degradation. Therefore, we have derivatized these adducts and their corresponding metabolites with a fluorescent marker. The use of a fluorescent marker attached to the CE at the phenol groups on the A-ring minimizes the complications described above. Such a marker assists in the selective isolation of CE from biological samples, prevents oxidation to o-quinone byproducts, and allows for selective detection and quantitation at low concentrations. Also, if a suitable fluorescent marker is chosen, the adducts and metabolites can be selectively identified using lowtemperature fluorescence spectroscopy. The latter has proven to be a valuable tool for the characterization of various polycyclic aromatic hydrocarbon (PAH)-DNA adducts and PAH metabolites. Laser-excited fluorescence spectroscopy under line-narrowing (FLN) and nonline-narrowing (NLN) conditions provides detailed spectral characterization [fingerprint identification (11-14)] and structural information of analytes of interest (1517). The issues involved in the development of a fluorescence-based protocol for detecting CE-derived DNA adducts in rat mammary tissue using a specific fluorescent marker, 1-pyrenesulfonyl chloride, are described. Derivatization is important since fluorescence provides high sensitivity, and the fluorescence spectra provide a means for adduct/metabolite identification in biological systems. In this paper, spectroscopic characterization of adduct standards and the depurinating DNA adducts formed in a tissue culture experiment where rat mammary gland (RMG) tissue was treated with the estrogen quinone E23,4-Q is provided. The results show that the depurinating adduct formed is 4-OHE2-1-N7Gua, proving that the fluorescence-based protocol developed can be used to further establish the role of estrogens in cancer.

Jankowiak et al.

Materials and Methods Caution: Catechol estrogen quinones are hazardous chemicals and should be handled carefully in accordance with NIH guidelines. Covalent Binding of Catechol Estrogen Quinones to DNA. E1-3,4-Q or E2-3,4-Q (1 mg/50 µL of DMSO) were mixed with 5 mL of 6 mM calf thymus DNA in 0.067 M sodiumpotassium phosphate, pH 7, and incubated for 2 h at 37 °C. DNA was precipitated with 2 volumes of ethanol, and then redissolved in 15 mM NaCl-1.5 mM sodium citrate. The DNA concentration was determined by the absorbance at 260 nm. The DNA supernatant was used for analysis of depurinating adducts. Binding of E2-3,4-Quinone to DNA in Rat Mammary Gland (Organ Culture). Mammary glands were cultured as previously described by Mehta et al. (18). Groups of four, 7-week-old female Sprague Dawley rats (Harlan Laboratories, Indianapolis, IN) were sacrificed, and the mammary region was shaved. The ventral surface was wetted thoroughly with 70% ethanol. The area around the fourth and fifth mammary glands, on both right and left sides, was excised and put in a Petri culture dish containing 20 mL of D-MEM/F-12 medium (4 mammary glands per dish), with 100 units/mL penicillinstreptomicin, 5 µg/mL prolactin, 5 µg/mL insulin, 1 µg/mL aldosterone, and 5 µg/mL hydrocortisone. The glands were incubated under a 95% air and 5% CO2 atmosphere at 37 °C. Every 24 h, the medium was replaced with fresh medium supplemented with the above hormones and antibiotics. After 48 h, each gland was treated with 200 nmol of E2-3,4-quinone dissolved in 10 µL of DMSO. After 2 h incubation at 37 °C under 95% air and 5% CO2, mammary tissue was separated from the medium, dried on filter paper, minced, and ground in liquid nitrogen. The tissue was weighed and Soxhlet-extracted for 24 h with hexane to remove mammary fat, and was further extracted with methanol containing 5% acetic acid for 48 h. The methanol-acetic acid extract was dried under reduced pressure, and the residue was dissolved in 0.5 mL of DMF. The crude sample was purified via preparative HPLC on a YMC ODS-AQ 10 × 250 mm column using a methanol/water gradient. Retention times of standard adducts were established, and fractions corresponding to the retention times of the standards were collected and dried under reduced pressure. The standards were subsequently derivatized for fluorescence studies as described below. The derivatized samples were analyzed by HPLC as described previously (8). Isolation of Depurinating Adducts by Preparative HPLC. Supernatants from in vitro binding reactions were evaporated to dryness under vacuum, and the residues were dissolved in 1.5 mL of dimethylformamide (DMF) and analyzed by HPLC. The extracted materials in DMF were initially purified by preparative HPLC on a YMC (Morris Plains, NJ) ODS-AQ 5 µm, 120 Å column (20 × 250 mm) at a flow rate of 8 mL/min, using a Waters (Milford, MA) 600E solvent delivery system equipped with a Waters 484 tunable absorbance detector operating at 290 nm. The appropriate retention times were first established by injecting a minimal amount of underivatized standard adduct. This was followed by injection of the 1.5 mL DMF solution containing the residue from the in vitro supernatant. Fractions from this purification were collected at retention times corresponding to the appropriate standard. These fractions were collected 1 min prior to and 1 min after the actual retention time to ensure complete capture of the underivatized adduct. The HPLC gradient used started with 40% methanol in water for 5 min, followed by a 15 min curvilinear gradient (CV6) to 55% methanol in water. Elution with 55% methanol in water was held for 15 min followed by a 10 min curvilinear gradient to 100% methanol. The 100% methanol elution was continued for 15 additional min for a total run time of 60 min. One hundred microliters of glacial acetic acid was added to each fraction immediately after elution from the HPLC column to help prevent oxidative degradation of the catechol adducts. The fractions were then dried under reduced

Catechol Estrogen 3,4-Quinone-Derived DNA Adducts

Chem. Res. Toxicol., Vol. 11, No. 11, 1998 1341 Lextra 100 XeCl excimer laser as the excitation source. For FLNS, several excitation wavelengths were used, each revealing a portion of the S1 excited-state vibrational frequencies of the fluorescence probe. NLN spectra were obtained using excitation at 308 nm. Samples were cooled in a glass cryostat with quartz optical windows. Fluorescence was dispersed by a McPherson 2061 1 m focal-length monochromator, and detected by a Princeton Instruments IRY 1024/G/B intensified photodiode array. For time-resolved detection, a Princeton Instruments FG-100 pulse generator was employed; the gate width was set to 200 ns, and various detector delay times (0-100 ns) were used. Spectral resolution used for the NLN and FLN measurements was 0.8 and 0.05 nm, respectively. Two solvent matrixes were used: ethanol or a mixture of glycerol/water (50/50 v/v). Ethanol was spectrophotometric grade from Aldrich; glycerol was purchased from Spectrum Chemical, Gardena, CA. Solutions (ca. 20 µL) were transferred to quartz tubes (2 mm i.d.) and sealed with a rubber septum. Analyte concentrations were about 10-6 M.

Figure 1. (A) Schematic structures of 1-pyrenesulfonyl chloride (left) and 4-hydroxycatechol estrogens of estrone (E1) and estradiol (E2). (B) Molecular structures of the dipyrene-labeled 4-OHE2 metabolite (left) and 4-OHE2-1-N7Gua adduct. pressure followed by derivatization with 1-pyrenesulfonyl chloride. Derivatization of Standards and Purification by Analytical HPLC. HPLC using fluorescence detection was conducted on a Waters 600E solvent delivery system equipped with a 700 WISP autoinjector, a Waters 990 photodiode array (PDA) detector, and a Waters 474 scanning fluorescence detector. For fluorescence experiments, the HPLC preparative fractions were derivatized with 1-pyrenesulfonyl chloride, and then analyzed by HPLC using fluorescence detection. The molecular structures of 1-pyrenesulfonyl chloride and 4-OHEi are shown in Figure 1A. At pH 8.5, only the dipyrene-labeled products were formed at positions O3 and O4; no binding to the exocyclic amino group was observed. Derivatization of metabolites and adducts was accomplished by adding 10 µL of the standard solution to 150 µL of dimethylformamide (DMF) containing 1 mg of 1-pyrenesulfonyl chloride. After addition of 50 µL of 0.2 M NaHCO3, the sample was heated at 50 °C for 30 min and passed through a 0.45 µm HPLC filter. A 20 µL aliquot was analyzed with fluorescence detection (λex, 350 nm; λem, 385 nm) using a YMC ODS-AQ 5 µm, 120 Å column (4.6 × 250 mm) at a flow rate of 1 mL/min with a methanol/water gradient. After elution for 5 min with 50% methanol in water, a 50 min curvilinear gradient (CV6) to 100% methanol was conducted and held for 20 min, for a total run time of 70 min. Under these conditions, the retention times for the 4-OHE1-1-N7Gua-dipyrene and 4-OHE2-1-N7Gua-dipyrene adducts were 57.2 and 58.5 min, respectively. Conditions under which the R and β rotational isomers could be separated were not found. The derivatized standards were collected and their structures confirmed by FABMS spectrometry (9). Isolated fractions from in vitro binding reactions were derivatized in a similar fashion. Twenty microliters aliquots were injected, and fractions corresponding to 4-OHEi-1-N7Guadipyrene adducts were collected. These fractions were dried and reinjected under the conditions described above, except that an acetonitrile/water gradient was employed to help further purify the 4-OHEi-1-N7Gua-dipyrene adducts. Peaks corresponding to adducts were collected and dried for the fluorescence studies. Low-Temperature Fluorescence Spectroscopy. Nonline-narrowing (NLN) fluorescence spectra at T ) 77 K and FLN spectra (S1rS0 excitation) at T ) 4.2 K were obtained using a Lambda Physik FL-2002 dye laser pumped by a Lambda Physik

Molecular Modeling and Molecular Dynamics Simulations. Structures based on molecular mechanics were optimized using HyperChem (version 5) with the MM+ force field developed for organic molecules. The Polak-Ribiere algorithm was used for molecular mechanics optimization. Stuctures were refined until the rms gradient of a minimum potential energy was less than 0.001 kcal/mol. To obtain structures with the lowest energy minima, a Monte Carlo conformational search was conducted using Spartan (19), version 4.1. Five bonds were selected on the basis of proximity of pyrene to the estrogen A-ring: the N7-C1 bond between Gua and the estrogen A-ring, the bond between sulfur and the A-ring (both pyrene groups), and the bonds between sulfur to the pyrene rings (see Figure 1B). The Monte Carlo search generated about 30 conformers with energies within ∼1 kcal/mol of each other. The lowest energy structure was subsequently used in the molecular dynamics simulations. Simulated annealing was conducted with HyperChem to further explore the conformational space of both the major 4-OHE2-1β-N7Gua-dipyrene adduct and the 4-OHE2-dipyrene metabolite. No constraints were used during the high-temperature searches. Starting and final temperatures were set to 0 K, and the structures were subjected to 60 (or 80) ps of molecular dynamics at 300 and 400 K. The heating time was set to 5 ps; various times (5-25 ps) were used for the cooling time. Often the optimized structures were subsequently used for further calculations. Optimization was restricted to the ground-state structures, since only minor geometry changes between S0 and S1 states are anticipated.

Results and Discussion Computer Modeling Studies. To guide the interpretation of fluorescence data presented below, the conformations of the dipyrene-labeled 4-OHEi-1-N7Gua adducts and 4-OHEi metabolites were investigated. The calculated structures of the 4-OHE2-1β-N7Gua-dipyrene adduct and the 4-OHE2-dipyrene metabolite are shown in Figure 2, frames A and B, respectively. The conformation of the major 4-OHE2-1β-N7Gua isomer reveals that the pyrene attached to the 4-OH group adopts a stacked type of conformation with Gua, while the plane of the pyrene attached to the 3-OH group is nearly parallel to the estrogen A-ring. Pyrene chromophores labeled as 1 and 2 in Figure 2 are attached at positions O3 and O4, respectively. The pyrene 2-Gua distance (center to center) is ∼5 Å, but interatomic distances as short as 3.5 Å exist. A similar conformation was obtained for the R-isomer (not shown). Frame B of Figure 2 shows the structure of the 4-OHE2-dipyrene metabolite, in which both pyrenes interact with the aromatic ring of the estrogen moiety. The stronger π-π interaction is for the

1342 Chem. Res. Toxicol., Vol. 11, No. 11, 1998

Figure 2. Optimized 0 K ground-state structures of the 4-OHE2-1-N7Gua-dipyrene-labeled adduct (isomer β, frame A) and the 4-OHE2-dipyrene-labeled metabolite (frame B). Molecular structures are shown without hydrogens and double bonds for clarity. Pyrene chromophores labeled as 1 and 2 are attached at positions O3 and O4, respectively.

Figure 3. NLN fluorescence spectra of estrogen dipyrene standards obtained at 77 K in ethanol glass for 308 nm excitation. Spectra a and b were obtained for the 4-OHE2-1N7Gua-dipyrene adduct at two different delay times of 0 and 60 ns, respectively. Spectra c and d were obtained for the 4-OHE2-dipyrene metabolite, at 0 and 60 ns delay times of the observation window. Spectrum e corresponds to the 4-OHE2-1N7Gua-dipyrene-labeled adduct in ethanol obtained after ∼5 min of UV irradiation at 308 nm (0 ns delay time).

pyrene attached to the 4-OH group; the pyrene 2-A-ring center to center distance is ∼3.5 Å. The average distance between the centers of the two pyrene labels is ∼8 and ∼7.5 Å for the 4-OHE2-1-N7Gua adduct and the 4-OHE2 metabolite, respectively. The relatively short pyrenepyrene distance suggests that energy transfer between the two pyrenes may be efficient (see below). Fluorescence Spectra of Derivatized Metabolites and DNA-Adduct Standards. Figure 3 shows the 77 K NLN fluorescence spectra of the 4-OHE2-1-N7Guadipyrene adduct (a and b) and the 4-OHE2-dipyrene metabolite (c and d) in an ethanol glass. Spectra a and c were obtained with 0 ns delay while the delay time for spectra b and d was 60 ns. The spectra are normalized so that the intensities of the most intense feature in each spectrum are equal. The actual overall intensity of spectrum b relative to a, and c relative to d, is about a factor of 0.02. Fluorescence origin bands in spectra a-d

Jankowiak et al.

are located at 383.1, 374.2, 380.0, and 374.2 nm, respectively. The dependence of the relative intensity of the two fluorescence origin bands for the adduct and metabolite on delay time establishes that the fluorescence lifetime associated with the lower energy origin band is significantly shorter than that of the higher energy origin band (for both the adduct and metabolite). The same experiments which led to Figure 3 were performed on the E1 adduct and metabolite. The spectra obtained for the adduct and metabolite are identical to those shown in Figure 3. The inability of low-temperature fluorescence spectroscopy to distinguish between the E1 and E2 species is not surprising since the structural differences in the D-ring between 4-OH catechol estrogens of estrone (E1) and estradiol (E2) are far removed from the fluorescent labels. Nevertheless, Figure 3 shows that the dipyrene-labeled Gua adducts can be distinguished from their corresponding precursor metabolites, a finding that is important for fluorescence-based studies of the depurinating 4-OHE2 adducts formed in vivo, vide infra. We consider next the question of the nature of the pyrenyl species responsible for the two fluorescence bands seen for both the adduct and the metabolite. Here we are guided by the structures shown in Figure 2 and our earlier NLN and FLN studies of benzo[a]pyrene diolepoxide (BPDE)-DNA adducts in oligonucleotides, polynucleotides, and calf thymus DNA (11, 15-17). Those studies proved that as the pyrene fluorophore moves from an external to a partially base-stacked to an intercalated configuration in DNA, its fluorescence origin band shifts to the red and broadens. Both the shifts and broadenings are due to an increase in the π-π interaction between pyrene and the DNA bases. Increasing π-π interaction leads to increasing charge-transfer character for the fluorescent state, which translates to an increasing permanent dipole change for the optical transition from the ground state. Increasing dipole moment change leads to a larger geometry change of the “host” molecules around the fluorophore upon its excitation (in solid-state spectroscopy this would be referred to as an increase in electron-phonon coupling). Increasing electron-phonon coupling leads to a loss of intensity for the sharp zerophonon lines (ZPL) in FLN spectra (11). With the above in mind, we return to Figure 3 and note that the NLN spectra yield widths of ∼200 and ∼430 cm-1 for the 374.2 and 383.1 nm origin bands for the -N7Gua adduct and widths of ∼200 and ∼320 cm-1 for the 374.2 and 380.0 nm bands of the metabolite. The results of references 11, 16, 17, and references cited therein, suggest that the 374.2 nm bands are due to species whose pyrenyl moieties experience little in the way of π-π interaction with Gua, in the case of the adduct, or the A-ring of the estrogen, in the case of the metabolite. We confirmed this by performing FLN experiments on the adduct and metabolite. The FLN spectra exhibited pronounced ZPL lines consistent with weak electron-phonon coupling (results not shown). Also, experiments were performed on 1-pyrenesulfonyl chloride, which exhibits a single fluorescence origin band at ∼376 nm (77 K) with a width of ∼200 cm-1 and FLN spectra with pronounced ZPLs (results not shown), consistent with weak electron-phonon coupling. Before discussing further the 374.2 nm bands, it is important to first present our conclusions for the 383.1 and 380.0 nm bands which we view as marker bands. The results shown in Figure 3, when viewed in terms of those for DNA-BPDE adducts and the structures shown

Catechol Estrogen 3,4-Quinone-Derived DNA Adducts

Figure 4. Non-line-narrowed (0,0) fluorescence origin bands of the 4-OHE2-1-N7Gua-dipyrene adduct (spectra a and c) and 4-OHE2-dipyrene metabolite (spectra b and d) obtained in different matrixes; T ) 77 K, λex ) 308 nm. Curves a,b and c,d were obtained in glycerol/water and ethanol matrixes, respectively. Spectra c and d were obtained by a subsequent dilution of the glycerol/water samples (a,b) with ethanol by a factor of 50.

in Figure 2, suggest that the breadth and location of the red-most fluorescence band of the adduct and metabolite are due to a strong π-π interaction of pyrene with Gua and the A-ring of the catechol estrogen, respectively. Strong support for this is provided by the FLN spectra shown later for CE-3,4-Q-derived adducts from rat mammary gland tissue. The spectra are consistent with strong π-π interactions. We conclude that the red-most origin band at 383.1 nm for the -N7Gua adduct and at 380.0 nm for the metabolite are due to structures which, respectively, exhibit a strong π-π interaction between pyrene and Gua and pyrene and ring A of the estrogen. The structures shown in Figure 2 are consistent with this. What structures or species are responsible for the 374.2 nm fluorescence origin band observed for samples of the adduct and metabolite (Figure 3)? Their wavelength, relatively narrow width, and weak electron-phonon coupling indicate that the pyrene moiety responsible for the fluorescence is not involved in strong π-π coupling. Emission at ∼374 nm is in the region expected for a pyrene chromophore whose interactions are dominated by the solvent. It was observed that different samples of the adduct and metabolite exhibited different intensities of the 374.2 nm band relative to the red-most origin. This suggested that a photodecomposition product might contribute to its intensity. Spectrum e of Figure 3, obtained with zero-delay, is that of an -N7Gua sample which was irradiated (308 nm) at room temperature for several minutes prior to cooling to 77 K. It should be compared with spectrum a. The significantly higher intensity of the 374.2 nm band relative to the 383.1 nm band in spectrum e provides strong support for the above suggestion.2 However, Figure 4 suggests that a conformation of the 4-OHE2-1-N7Gua-dipyrene adduct and the -dipyrene metabolite, which exhibit little π-π interaction, may also contribute to the 374.2 nm band. The NLN spectra a (adduct) and b (metabolite) were obtained in a glycerol/water glass and exhibit only one fluorescence origin at 374.0 nm. Dilution of the glycerol/water matrix 2

No attempt was made to identify the photoproduct since we are primarily interested in the red-most fluorescence origins.

Chem. Res. Toxicol., Vol. 11, No. 11, 1998 1343

Figure 5. Non-line-narrowed fluorescence spectra of the 4OHE2-1-N7Gua-dipyrene adduct standard (a) and the HPLC fraction (b) obtained from the RMG tissue sample (in ethanol, T ) 77 K, λex ) 308 nm).

with ethanol, however, results in the appearance of the 380.0 (metabolite) and 383.1 (adduct) nm origin bands. Spectra c and d for the adduct and metabolite, respectively, were obtained with a dilution factor of 50. That the glycerol/water solvent disrupts the π-π interactions indicated by the structures of Figure 2 has a precedent, namely, BPDE-DNA adducts for which the pyrene is in a partially base-stacked configuration (11, 17). Finally, we considered the possibility that the 374.2 and 380.0 nm origins (metabolite) and 374.2 and 383.1 nm origins (adduct) are the result of dual fluorescence due to the existence of two energetically inequivalent pyrenes. We think this possibility to be unlikely since the Fo¨rster energy transfer radius Ro ) 10 Å (20) and the center to center pyrene-pyrene distances (R) are ∼8 Å for the adduct and ∼7.5 Å for the metabolite. The fluorescence lifetime of monomer pyrene in solution is ∼200 ns. The energy transfer rate constant is proportional to (Ro/R)6. Thus, one would expect the 374 nm bands to exhibit a lifetime of ∼40 ns, which is considerably shorter than our estimate of a lower limit of 100 ns based on the dependence of the 374.2 nm band’s intensity on the gate delay time. An additional argument against dual fluorescence is that the high-resolution vibronically excited FLN spectra for the 374.2 nm species of the adduct and metabolite are identical (results not shown), which is not expected given that their structures (Figure 2) are so different. Identification of the E2-3,4-Q-Derived Adduct in Rat Mammary Gland (RMG) Tissue. As described under Materials and Methods, preparative HPLC was used to isolate analytes in a region which contains the 4-OHE2-1-N7Gua adduct. The sample was then subjected to the 1-pyrenesulfonyl derivation procedure. This was followed by higher resolution HPLC to isolate a fraction believed to be the dipyrenyl derivative of the above adduct. (The same procedure was employed in ref 8 where mass spectrometry was used to identify the above adduct in RMG tissue.) The 77 K NLN fluorescence spectrum (zero-delay) of the fraction is b in Figure 5. The spectrum of the standard is a. The prominent features of the two spectra are identical. Thus, NLN fluorescence spectroscopy at 77 K can be used to identify the 4-OHE2-1-N7Gua adduct formed in vivo from E2-3,4Q.3 Both spectra show the weak origin band at 374.2 3 We remind the reader that derivatization with 1-pyrenesulfonyl chloride cannot distinguish between the E2 and E1 adducts. Work is in progress to find fluorescence label-derivatization schemes which will.

1344 Chem. Res. Toxicol., Vol. 11, No. 11, 1998

Jankowiak et al.

Figure 6. Vibronically excited FLN spectra obtained for the 4-OHE2-dipyrene metabolite (spectra a and b) and the 4-OHE21-N7Gua-dipyrene adduct (spectra c and d) in ethanol. Spectra a,c and b,d were obtained at λex of 367.0 and 370.0 nm, respectively; T ) 4.2 K. The FLN peaks are labeled with their excited-state vibrational frequencies, in cm-1.

Figure 7. (0,0)-band excited FLN spectra of the 4-OHE2dipyrene metabolite (curve a), the 4-OHE2-1-N7Gua-dipyrene adduct standard (curve b), and the HPLC fraction from the RMG tissue experiment (curve c); T ) 4.2 K, λex ) 382.0 nm. The FLN peaks are labeled with their ground-state vibrational frequencies, in cm-1.

nm discussed earlier in the context of the standard spectra for the adduct and the 4-OHE2 metabolite. The weak features (dashed arrows) at 370 and ∼377 nm in spectrum b for the sample from RMG tissue are most likely due to species, other than the adduct of interest, in the HPLC fraction. Their FLN spectra (not shown) indicate that they possess the pyrenyl fluorophore and that the respective pyrene moieties are not involved in significant π-π interactions. Vibronically excited FLN spectroscopy has proven to be a powerful technique for characterization of PAHDNA adducts (12-20), as reviewed in (11). From a fingerprint point of view, this type of spectroscopy is most informative when the electron-phonon coupling associated with the optical transition is weak, meaning that the spectra exhibit sharp and pronounced zero-phonon lines (ZPL). Figure 6 shows spectra for the 4-OHE2dipyrene metabolite (a and b) and the 4-OHE2-1-N7Gua adduct (c and d). Spectra a,c and b,d were obtained at zero-delay with excitation wavelengths of 367.0 and 370.0 nm, respectively. These wavelengths are suitable for generating informative FLN spectra. The spectra for the metabolite (a,b) exhibit only weak ZPLs superimposed on broad features. (The numbers, e.g., 570, label the excited vibrational frequencies in cm-1.) The weakness of the ZPLs indicates strong electron-phonon coupling which we interpret in terms of a strong π-π interaction between pyrene and the A-ring of estrogen, vide supra. The spectra for the adduct exhibit no ZPLs, indicating even stronger electron-phonon coupling. We interpret this stronger coupling to π-π interaction of the pyrene in close proximity to Gua (Figure 2). The spectra shown in Figure 6 are for the synthesized standards of the adduct and metabolite. The same experiment was performed on the adduct from RMG tissue. The spectra obtained (not shown) are very similar to c and d of Figure 6 except for the appearance of sharp ZPLs due to the unknown species responsible for the shoulder at ∼377 nm in spectrum b of Figure 5. To avoid this interference, it is necessary to excite directly into the origin absorption band of the adduct near 383 nm. Spectra b and c of Figure 7 are for the synthesized dipyrene adduct standard and the derivatized adduct

from RMG tissue obtained with λex ) 382.0 nm. The two spectra are identical, revealing stronger electron-phonon coupling than the coupling observed for the metabolite. The appearance of weak ZPLs in spectrum a is consistent with spectra a and b of Figure 6. The numbers labeling the ZPL in Figure 7 correspond to ground-state vibrational frequencies, since origin excitation was used.

Conclusions and Final Remarks A procedure for derivatization of the nonfluorescent 4-OHEi-1-N7Gua adducts (i ) 1, 2) and 4-OHEi metabolites with the fluorescent label 1-pyrenesulfonyl chloride was described. Results were presented which show that non-line-narrowing laser-excited fluorescence spectroscopy at 77 K and fluorescence line-narrowing spectroscopy at 4 K can be used to distinguish the labeled adducts from the labeled metabolites. The fluorescence spectra of synthesized standards were used to confirm that the 4-OHE2 adduct is formed in culture in rat mammary gland tissue from exposure to the E2-3,4-Q. Given the high sensitivity of laser-excited fluorescence spectroscopy, we believe that derivatization with fluorescent labels is a promising approach for future in vivo studies of depurinating adducts formed from catechol estrogens. With derivatization by 1-pyrenesulfonyl chloride, however, the 4-OHE1 and 4-OHE2 adducts (as well as the metabolites) could not be distinguished on the basis of their low-temperature spectra because the pyrenyl moieties are far removed from the D-ring of the catechol estrogen. Currently, we are exploring other derivatization schemes which may allow for distinction. The E1and E2-labeled adducts (metabolites) can be separated by HPLC and capillary electrophoresis. It was recently shown that capillary electrophoresis can be combined with NLN and FLN spectroscopy for on-line structural characterization of PAHs and PAH-DNA adducts (13, 21). Thus, we plan to use these combinations in future studies of labeled adducts formed in vivo from catechol estrogens. In previous laser-excited fluorescence studies of PAHDNA adducts, we have found that molecular modeling and dynamics simulations are very helpful in interpreta-

Catechol Estrogen 3,4-Quinone-Derived DNA Adducts

tion of fluorescence spectra (22, 23). The results from modeling and dynamics simulations presented here were important for interpretation of the spectra of the 1-pyrenesulfonyl chloride-labeled adducts and metabolites. Although the spectra, themselves, indicate the extent of π-π interaction between the pyrene fluorophore and Gua and/or the A-ring of the catechol estrogen, the simulated structures (in vacuo) (Figure 2) provide insight on how the π-π interaction occurs. As in previous work with BPDE-DNA adducts, we showed that the strong π-π interactions indicated by the structures shown in Figure 2, which are consistent with the spectra obtained using an ethanol glass, are disrupted in the glycerol/water glass. It is important, therefore, to use a variety of glassforming solvents in studies of the type described here. Although 1-pyrenesulfonyl chloride is a useful fluorescent probe, it suffers from the disadvantage that it introduces two pyrene fluorophores to the adducts or metabolites which complicates interpretation of the spectra. Thus, an emphasis in our future research will be on labeling with a single fluorophore.

Acknowledgment. This research project was supported by the National Institutes of Health (Grant POI CA 49210-05). Partial support from the Office of Health and Environmental Research of the U.S. Department of Energy is also acknowledged. Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract W-7405-Eng-82.

References (1) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (1979) 21, 279-362. (2) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (1987) Suppl. 7, 272-310. (3) Kirkman, H. (1959) Estrogen-induced tumors of the kidney in the Syrian hamster. Natl. Cancer Inst. Monogr. 1, 1-59. (4) Li, J. J., Li, S. A., Klicka, J. K., Parsons, J. A., and Lam, L. K. T. (1983) Relative carcinogenic activity of various synthetic and natural estrogens in the Syrian hamster kidney. Cancer Res. 43, 5200-5204. (5) Dwivedy, I., Devanesan, P., Cremonesi, P., Rogan, E., and Cavalieri, E. (1992) Synthesis and characterization of estrogen 2,3- and 3,4-quinones. Comparison of DNA adducts formed by the quinones versus horseradish peroxidase-activated catechol estrogens. Chem. Res. Toxicol. 5, 828-833. (6) Stack, D., Byun, J., Gross, M. L., Rogan, E. G., and Cavalieri, E. (1996) Molecular characteristics of catechol estrogen quinones in reaction with deoxyribonucleotides. Chem. Res. Toxicol. 9, 851859. (7) Lier, J. G., Fang, W. F., Sirbasku, D. A., and Ari-Ulubelen, A. (1986) Carcinigenecity of catechol estrogens in Syrian hamster. J. Steroid Biochem. 24, 352-356. (8) Li, J. J., and Li, S. A. (1987) Estrogen carcinogenesis in Syrian hamster tissues. Role of metabolism. Fed. Proc. 46, 1858-1863.

Chem. Res. Toxicol., Vol. 11, No. 11, 1998 1345 (9) Cavalieri, E. L., Stack, D. E., Devanesan, P. D., Todorovic, R., Dwivedy, I., Higginbotham, S., Johansson, K. D., Patil, K. D., Gross, M. L., Gooden, J. K., Ramanathan, R., Ceruy, R. L., and Rogan, E. L. (1997) Molecular origin of cancer: catechol estrogen3,4-quinones as endogenous tumor initiators. Proc. Natl. Acad. Sci. U.S.A. 94, 10937-10942. (10) Liehr, J. G., Avitts, T. A., Randerath, E., and Randerath, K. (1986) Estrogen-induced endogeneous DNA adduction: possible mechanism of hormonal cancer. Proc. Natl. Acad. Sci. U.S.A. 83, 53015305. (11) Jankowiak, R., and Small, G. J. (1998) Analysis of PAH-DNA AdductssFluorescence Line-Narrowing Spectroscopy, an invited book chapter In Handbook of Environmental Chemistry (3J) Neilson, A. H., Ed.; Springer, pp 119-145. (12) Jankowiak, R., Ariese, F., Zamzow, D., Luch, A., Kroth, H., Seidel, A., and Small G. J. (1997) Conformational studies of stereoisomeric tetraols derived from syn- and anti-dibenzo[a,l]pyrene diolepoxides. Chem. Res. Toxicol. 10, 677-686. (13) Zamzow, D., Lin, C.-H., Small, G. J., and Jankowiak, R. (1997) Application of capillary electrophoresis-fluorescence line narrowing system for on-line spectral characterization of closely related analytes. J. Chromatogr. A787, 73-80. (14) Todorovic, R., Ariese, F., Devanesan, P., Jankowiak, R., Small, G. J., Rogan, E., and Cavalieri, E. (1997) Determination of benzo[a]pyrene- and 7,12-dimethylbenzo[a]anthracene-DNA adducts formed in rat mammary glands. Chem. Res. Toxicol. 10, 941947. (15) Marsh, G. A., Jankowiak, R., Suh, M., and Small, G. J. (1994) Sequence dependence of benzo[a]pyrene diol epoxide-DNA adduct conformer distribution: a study by laser-induced fluorescence/ polyacrylamide gel electrophoresis. Chem. Res. Toxicol. 7, 98109. (16) Suh, M., Jankowiak, R., Ariese, F., Mao, B., Geacintov, N. E., and Small, G. J. (1994) Flanking base effects on the structural conformation of the (+)-trans-anti-benzo[a]pyrene diolepoxide adduct to N2-dG in sequence-defined oligonucleotides. Carcinogenesis 15, 2891-2898. (17) Suh, M., Ariese, F., Small, G. J., Jankowiak, R., Liu, T.-M., and Geacintov, N. E. (1995) Conformational studies of the (+)-trans, (-)-trans-, (+)-cis-, and (-)-cis adducts of anti-benzo[a]pyrene diolepoxide to N2-dG in duplex oligonucleotides using polyacrylamide gel electrophoresis and low-temperature fluorescence spectroscopy. Biophys. Chem. 56, 281-296. (18) Mehta, R. G., Eerny, W. L., Madigan, M. J., and Moon, R. C. (1983) Modification of the mouse mammary gland organ culture technique. J. Tissue Culture Methods 8, 27-30. (19) Spartan version 4.1, Wavefunction, Inc., 18401 Von Karman Ave., #370, Irvine, CA 92715. (20) Berlman, I. B. (1973) Energy Transfer Parameters of Aromatic Compounds, Academic Press, New York and London. (21) Jankowiak, R., Zamzow, D., Ding, W., and Small, G. J. (1996) Capillary electrophoresis-fluorescence line narrowing system for on-line structural characterization of molecular analytes. Anal. Chem. 68, 2549-2553. (22) Ariese, F., Small, G. J., and Jankowiak, R. (1996) Conformational studies of depurinating DNA adducts from syn-dibenzo[a,l]pyrene diolepoxide. Carcinogenesis 17, 829-837. (23) Jankowiak, R., Ariese, F., Zamzow, D., Luch, A., Kroth, H., Seidel, A., and Small, G. J. (1997) Conformational studies of stereoisomeric tetrols derived from syn- and anti-dibenzo[a,l]pyrene diol epoxides. Chem. Res. Toxicol. 10, 677-686.

TX980119+