Metabolic Profiling of Endocrine-Disrupting Compounds by On-Line

SPE columns and filter units were produced in-house and are described ...... Blair , R. M., Fang , H., Branham , W. S., Hass , B. S., Dial , S. L., Mo...
0 downloads 0 Views 204KB Size
Download PDF
Chem. Res. Toxicol. 2007, 20, 1825–1832

1825

Metabolic Profiling of Endocrine-Disrupting Compounds by On-Line Cytochrome P450 Bioreaction Coupled to On-Line Receptor Affinity Screening Sebastiaan M. Van Liempd,† Jeroen Kool,‡ John H. Meerman,†,§ Hubertus Irth,† and Nico P. Vermeulen*,† LACDR-DiVisions of Molecular Toxicology and Biomolecular Analysis, Department of Chemistry and Pharmaceutical Sciences, Vrije UniVersiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands, and Kiadis Pharma, Groningen, The Netherlands ReceiVed March 7, 2007

We present a fully automated and hyphenated bioanalytical method for metabolic profiling of potentially harmful xenoestrogens. The system consists of an on-line cytochrome P450 bioreactor coupled to a reversed-phase, gradient high-performance liquid chromatograph. A C18 solid-phase extraction (SPE) unit is used as an interface between the P450 bioreactor and the HPLC column. The HPLC column is linked on-line to a high-resolution screening (HRS)-estrogen receptor R affinity detection (ERAD) assay. In effect, the P450 bioreactor produces metabolites that are subsequently trapped on-line by SPE and separated by HPLC. The separated metabolites are then screened on-line, at the moment of elution, for affinity toward estrogen receptor R (ERR) using the HRS-ERAD assay. The SPE method was optimized with methoxychlor (MXC) and its metabolites mono- and bis-OH-MXC. After optimization, the P450bioreactor–SPE–HPLC system was made generally applicable to the biocatalysis and trapping of polar to highly apolar compounds. The precision of the P450-bioreactor–SPE–HPLC system is high (relative standard deviation e15%), and the HRS-ERAD assay is also very sensitive (having lower limits of detection of 250 ng for bis-OH-MXC and 240 ng for mono-OH-MXC). Finally, bioactivation of 2-hydroxy4-methoxybenzophenone (benzophenone-3) into ERR-binding metabolites by P450 was studied using the validated P450-bioreactor–SPE–HPLC–ERAD system in combination with atmospheric pressure chemical ionization MS. This resulted in the detection of three ERR-binding metabolites, of which at least one, a hydroxylated metabolite initially detected only by ERR affinity, had not been described previously. The hyphenated P450-bioreactor–SPE–HPLC–HRS-ERAD methodology presented here will be of great interest in on-line research of metabolic activation of endocrine-disrupting compounds. Introduction 1

Endocrine-disrupting compounds (EDCs), defined as exogenous compounds that interfere with endocrinological processes, are used in a wide variety of applications. For example, they have been used as flame retardants, coatings, and pesticides (1, 2). In general, these pollutants degrade slowly in the environment and therefore bioaccumulate in the food chain. Furthermore, they have long half-lives in humans (3). A disruption of endocrine homeostasis may lead to undesirable effects in human populations or individuals, ranging from changes in reproductive behavior to increased vulnerability to cancer (4, 5). Meanwhile, it is known that the endocrinedisrupting effects of some of these compounds are potentiated after metabolism by cytochrome P450 enzymes (6–11). Me* To whom correspondence should be addressed. Tel: +31(0)20 598 7590. Fax: +31(0)20 598 7610. E-mail: [email protected]. † Vrije Universiteit. ‡ Kiadis Pharma. § Present address: LACDR-Division of Pharmacogenomics, University of Leiden, Einsteinweg 55, 2333 CC Leiden, The Netherlands. 1 Abbreviations: benzophenone-3, 2-hydroxy-4-methoxybenzophenone; BP3, benzophenone-3; CID, collision-induced dissociation; CYP, cytochrome P450 subtype indication; EDC, endocrine-disrupting compound; ERR, estrogen receptor R; ERAD, estrogen receptor R affinity detection; G6P, glucose-6-phosphate; GDH, glucose-6-phosphate dehydrogenase; HRS, high-resolution screening; MXC, methoxychlor; NADPH, β-nicotinamide adenine dinucleotide phosphate; RLMs, rat liver microsomes; RSD, relative standard deviation; SPE, solid-phase extraction.

tabolism by P450 normally results in hydroxylation of the compounds, increasing their availability for phase-II metabolism and facilitating their excretion from the body. However, the compounds can also be transformed into metabolites that have affinities for receptors or other proteins, often resulting in pharmacological or toxicological activity (12, 13). In the case of methoxychlor (MXC) (7), 2-hydroxy-4-methoxybenzophenone (benzophenone-3 or BP3) (10), brominated biphenyls (11), or bisphenol A (8), metabolites having greater affinities for estrogen receptor R (ERR) are formed by P450 activity. This study presents an on-line, hyphenated, high-resolution screeningbased bioanalytical system for bioactivation and screening of estrogen-receptor affinities of EDCs. MXC, extensively used as a pesticide, and BP3, widely used in sun creams as a UVblocker, were used for optimization and validation of the system. ERR is one of the most prominent receptors involved in endocrine disruption (14). Usually, when complex mixtures of compounds (e.g., mixtures of P450-derived metabolites or environmental samples) are analyzed with respect to ERR interactions, first the mixture itself is screened for ERR affinity. When ERR affinity is detected, the mixture is separated chromatographically and then purified by fraction collection, and each individual fraction is analyzed again in order to identify ERR-binding compounds (2, 9, 10). However, with the on-line high-resolution screening (HRS)-ERR affinity detection (ERAD) system employed in this study, we were able to circumvent time-

10.1021/tx7000724 CCC: $37.00  2007 American Chemical Society Published on Web 11/01/2007

1826 Chem. Res. Toxicol., Vol. 20, No. 12, 2007

Van Liempd et al.

Figure 1. Simplified schematic representations of the on-line SPE–HPLC–P450-bioreactor system coupled to the on-line ERR affinity detection assay. The Superloop labeled “MeOH” contained 40% (v/v) MeOH in water and the one labeled “Tracer” contained the fluorescent ligand coumestrol. The wash solution contained 250 mM NaOH and 0.5% (w/v) SDS. For clarity, Superloops containing NADPH, quadruple bioreactor reaction coils, and quadruple SPE columns are not shown.

consuming fractionation steps. The HRS-ERAD platform is based on gradient HPLC separation with on-line, postcolumn detection of ERR-binding compounds (15–17). When a compound elutes from the HPLC column, it is introduced into the on-line ERAD assay, where it is allowed to bind to ERR in a microliter-scale reaction coil. In a second reaction coil, coumestrol, a fluorescent ERR-binding tracer molecule, is added to the reaction mixture. When coumestrol binds to ERR, fluorescence intensity is enhanced. If an eluting compound shows affinity for ERR, the fraction of receptor-bound coumestrol is decreased, leading to a negative (below-baseline) peak in the on-line ERR bioassay. When the flow is split directly after the HPLC column, part of the eluent can be introduced into the on-line bioassay and part can be directed to a UV detector or a mass spectrometer for parallel detection or identification. This HRS-ERAD setup is especially useful for on-line analysis of complex mixtures (13). In order to produce potential ERR-binding metabolites, we recently developed a novel, on-line, solid-phase extraction (SPE)–HPLC-coupled bioreactor that uses P450 as a catalyst (17, 18). This on-line system is capable of sample cleanup, extraction, and preconcentration of a metabolite mixture by means of microfiltration and SPE. A sophisticated switch-valve system makes the SPE-trapped metabolites directly available for gradient HPLC. Recently, use of a simplified version of the bioreactor for the conversion of ethoxyresorufin in resorufin by rat liver microsomes containing P450 was reported (18). A more complex setup coupling the P450 bioreactor to an ERR bioassay on-line was subsequently developed for metabolic profiling of tamoxifen and raloxifene (17). However, as a result of several improvements in SPE trapping in the SPE–HPLC-coupled P450 bioreactor presented in this study, the accuracy of the HRSERAD system improved significantly. The ultimate aim of the present study was the validation and optimization of a hyphenated P450-bioreactor–SPE–HPLC–HRSERAD system for analysis of EDCs and its application to the on-line profiling of P450-generated ERR-binding metabolites of MXC and BP3. β-Naphthaflavone (BNF)-induced rat liver microsomes (RLMs), which contain high levels of the P450 CYP1A subtypes, were used as a source of P450 (19). In the P450-bioreactor setup presented in this study, we introduced an improved SPE trapping unit that allows trapping of polar to highly apolar compounds. This makes the system generally applicable to metabolic profiling of a wide variety of compounds. The novel bioanalytical HRS platform in which an online P450 bioreactor is combined with on-line ERR affinity detection is a powerful tool for the analysis of potential estrogenic enhancement of endocrine disruptors by P450 activity.

Materials and Methods Materials. MXC, 2,2-bis(p-hydroxyphenyl)-1,1,1-trichloroethane (bis-OH-MXC), BP3, 2,4-dihydroxybenzophenone (BP1), BNF, glucose-6-phospate (G6P), HPLC-grade methanol (MeOH), dichloromethane, and glucose-6-phosphate dehydrogenase (GDH) were obtained from Sigma-Aldrich (Zwijndrecht, The Netherlands). Riedel de Häen (Seelze, Germany) supplied sodium hydroxide, sodium chloride, magnesium chloride, potassium dihydrogen phosphate, dipotassium hydrogen phosphate, and sodium thiosulfate. HPLC-grade acetonitrile (ACN) was purchased from J. T. Baker (Meppel, The Netherlands). β-Nicotinamide adenine dinucleotide phosphate (NADPH) tetrasodium salt and ethylenediaminetetraacetic acid (EDTA) were purchased from Applichem (Lokeren, Belgium). All water used was purified with a Barnstead Nanopure type-1 water deionizer setup. Biomaterials. BNF-induced RLMs were prepared according to Rooseboom et al. (20) and contained 11 µM P450, as measured with a carbon monoxide UV difference spectrum. The P450 reaction buffer consisted of 50 mM potassium phosphate (pH 7.4), 5 mM MgCl2, and 2 mM EDTA. The buffer used in the HRS ERR affinity assay (EB) consisted of 10 mM potassium phosphate (pH 7.4) and 150 mM NaCl. The regenerating system for the P450 reaction contained 10 mM NADPH, 30 mM G6P, and 3 units/mL GDH dissolved in P450 reaction buffer. The ERR ligand binding domain (LBD) was produced with recombinant Escherichia. coli BL21 (DE3)-expressing His6-ER-LBD, according to Eiler et al. (21). The concentration of ERR LBD was 250 nM, as measured by determination of estradiol binding ability in a saturation radioligand binding assay described by Eiler et al. (21). Synthesis of 2-(p-Hydroxyphenyl)-2-(p-methoxyphenyl)-1,1,1trichloroethane (Mono-OH-MXC). An aliquot (1.5 equiv) of a 1 M solution of boron tribromide in dichloromethane was slowly added to a solution of MXC in dichloromethane at room temperature under a flow of dry nitrogen gas, and the mixture was stirred for 2 h. Water was added to quench the reaction, and the mixture was extracted twice with methylene chloride. The combined methylene chloride extracts were dried with MgSO4, and the solvent was concentrated in vacuo. The residue contained both mono-OH-MXC and bis-OH-MXC. Pure mono-OH-MXC was obtained after chromatography on a silica-gel column with 3:17 diethyl ether/hexane. Identification of mono-OH-MXC: Rf ) 0.29 for silica gel TLC using 3:17 diethyl ether/hexane; 1H NMR (400 MHz, CDCl3) δ 7.50 (d, 2H), 7.45 (d, 2H), 6.90 (d, 2H), 6.80 (d, 2H), 4.95 (s, 1H), 3.80 (s, 3H). Components of the On-Line P450 Bioreactor System. All pumps used in the bioreactor and HRS-ERAD systems (Figure 1) were Knauer K-500 HPLC pumps. Injections were performed with a Gilson 234 autoinjector (50 µL injection loop). The HPLC columns, SPE columns, and reaction coils were thermostated with Shimadzu CTO-10AC column ovens. Hydraulically driven 50 and 100 mL syringes (Superloops) were purchased from GE Healthcare (Roosendaal, The Netherlands). Knitted 0.0625 in. × 0.75 mm PTFE reaction coils, PEEK tubing, six-way dead-end switch valves, and two-position six-port switch valves were obtained from VICI Jour (Amstelveen, The Netherlands). The autoinjector, pumps, and switch valves were operated by ScreenControl software running

Metabolic Profiling of Endocrine Disruptors under Windows 2000 (Kiadis, Groningen, The Netherlands). Flow splitters were made of 0.0625 in. × 0.5 mm PEEK tubing with 50 µm i.d. × 375 µm o.d. fused-silica inserts. Polyethersulfone (PES) membrane filters (0.22 µm) were purchased from Sterlitech (Kent, WA). SPE columns and filter units were produced in-house and are described elsewhere (18). SPE Optimization. SPE optimization was carried out by flushing a mixture containing 50 µM MXC or bis-OH-MXC and 440 nM P450 out of the reaction coil with 1:9, 2:8, 4:6, and 6:4 (v/v) MeOH/ water solutions at a flow rate of 1.0 mL/min for 4 min, without regenerating the system. Just before the SPE column, the MeOH/ water flow was diluted with a make-up flow of water at 1.5 mL/ min in a 100 µL mixing coil made of 0.0625 in. × 0.25 mm i.d. PEEK material. The maximum MXC and bis-OH-MXC signals were determined by direct injection of the separate compounds into the UV detector. Trapping was monitored by eluting the SPE column with 9:1 ACN/water over the UV detector (220 nm) and comparing this signal with signals from direct injections into the UV detector. On-Line Metabolite Generation. Metabolites were formed in the P450 bioreactor unit of the bioreactor system (Figure 1). After injection, the substrate (MXC or BP3), microsomes, and regenerating system (the latter two contained in chilled Superloops) were mixed in a reaction coil in a 1:8:1 ratio at a total flow rate of 500 µL/min until the total reaction volume reached 500 µL. Following incubation, the reaction mixture was flushed over the 0.22 µm PES filter with a degassed 40% MeOH solution at a flow rate of 1.0 mL/min for 4 min. Just before the SPE column, the MeOH solution was diluted by adding a 1.5 mL/min flow of water, leading to a total flow of 2.5 mL/min over the SPE column with a MeOH concentration of 16%. After reconcentration of analytes on the SPE column, the filter unit and reaction coil were rinsed by a series of cleaning steps. First, the filter was back-flushed with water at a flow rate of 1.0 mL/min for 1 min and then with a 250 mM NaOH/ 0.5% w/v SDS wash solution at the same flow rate for 1 min. Finally, the system was rinsed with water at a flow rate of 1.0 mL/ min for 4 min, which proved to be sufficient for total removal of the wash solution. P450 Bioreactor Conditions. For one reaction procedure, the 500 µL reaction coils (left side of Figure 1) were thermostated at 37 °C and filled with microsomes, substrate (MXC or BP3), and regenerating system. The final coil concentrations of MXC in the reaction runs were 25, 50, and 100 µM, whereas P450 concentrations were varied from 220 to 880 nM. BP3 was metabolized at a concentration of 50 µM by 44 nM P450. The concentrations of the regenerating system components (1.0 mM NADPH, 3.0 mM G6P, and 0.3 unit/mL GDH) were kept constant for all reactions. Next, the coils were successively emptied over the filter onto the SPE columns. The injection and coil-emptying procedures were programmed such that incubation times in the different coils were 7, 15, 24, and 33 min. The 33 min incubations were performed without regenerating the system and were considered to be t ) 0 incubations. Amounts of the MXC and BP3 metabolites formed were determined using calibration curves obtained by injecting 0.3, 0.6, 1.3, 2.5, 5.0, and 10 µM standard solutions of mono- and bis-OH-MXC and BP1 into the bioreactor system without microsomes and regenerating system. All measurements were carried out in triplicate, except for those on the reaction of 100 µM MXC with 440 nM P450, which were carried out in duplicate. Chromatography. When all of the coils were emptied onto the SPE columns, the trapped compounds were further separated by gradient HPLC. Analytes were separated in the chromatographic unit with the following gradients (all concentrations v/v in water): For MXC and its P450-generated metabolites, a linear gradient was applied, starting at 40% ACN, increasing to 90% ACN in 30 min, holding constant for 10 min, and decreasing back to 40% ACN in 10 min. BP3 and its metabolites were separated in a gradient that increased from 25 to 70% ACN in 30 min, stayed constant for 10 min, and decreased back to 25% ACN in 10 min. All separations were carried out on a 150 × 4.6 mm i.d. Luna C18 (2) column protected by a 2.0 × 3.0 mm i.d. C18 guard column (Phenomenex,

Chem. Res. Toxicol., Vol. 20, No. 12, 2007 1827 Amstelveen, The Netherlands). The HPLC flow rate for all separations was 250 µL/min. After the HPLC column, the eluent was split in a 9:1 ratio between a UV detector (Agilent Technologies, Amstelveen, The Netherlands) and the on-line ERR affinity assay, respectively. MXC and its metabolites were detected with UV at 230 nm, while BP3 and its metabolites were monitored at 280 nm. The HPLC column and SPE cartridges were thermostated at 37 and 22 °C, respectively. Tandem Mass Spectrometry. To identify metabolites of BP3, reaction mixtures (50 µM BP3, 880 nM P450, incubation time of 1 h) were trapped on the SPE columns as described above. Subsequently, the SPE columns were transferred to the LC-MS/ MS setup and placed in front of the HPLC column. The gradients used were the same as those in the on-line ERR affinity assay. MS was performed using an LCQ Deca mass spectrometer (Thermo Finnigan, Breda, The Netherlands) in full scan (m/z 100–500) and negative atmospheric pressure chemical ionization (APCI) modes. The most abundant ion was selected for collision-induced dissociation (CID) at a normalized collision energy of 40. Eluent from the HPLC column was sprayed into the mass spectrometer at -10 kV. The temperature of the heated capillary was set at 150 °C; the sheath and auxiliary gas flows were 80 and 10, respectively. A UV detector was placed in series with the MS. Alignment of the UV spectra and ERR affinity traces obtained from the bioreactor system with those obtained prior to MS was used as a basis for assigning masses to compounds responsible for ERR affinity signals. ERr-Affinity Detection System. The present homogeneous ERR affinity detection assay was performed according to Kool et al. (13). The HRS-ERAD system consisted of Superloops, 0.0625 in. × 0.25 mm i.d. Tefzel reaction coils (VICI Jour), an Agilent 1100 series fluorescence detector, and an Agilent 1100 series UV detector. The HPLC eluent was split so that 1/10 of the flow (25 µL/min) was directed to the assay. In the four-way junction, this flow was combined with a 150 µL/min flow of the ERR solution (10 nM ERR in EB, SL 5) and a make-up flow. The make-up flow consisted of an H2O/ACN gradient opposite to the HPLC gradient, in order to keep the ACN concentration everywhere in the assay constant at 15%. ERR and the eluted compounds were allowed to bind in the first reaction coil (25 µL). This mixture was combined with a 150 µL/min flow of coumestrol solution (0.43 µM in EB). The final equilibrium between ERR, ligand, and coumestrol was established in the second reaction coil (50 µL). Detection took place directly after the second coil using a fluorescence detector set to λex ) 340 nm and λem ) 410 nm.

Results and Discussion Growing environmental awareness demands new tools for the analysis of pollutants. Therefore, we validated and optimized an on-line SPE–HPLC-coupled P450 bioreactor linked to an HRS ERR affinity detection system (Figure 1). This hyphenated system was optimized and validated using methoxychlor, a common halogenated pesticide, in order to make it applicable for the analysis and profiling of ERR-binding metabolites of endocrine-disrupting compounds. Subsequently, the applicability of this fully automated on-line system was tested by using the system to analyze novel ERR-binding metabolites of the UV blocker benzophenone-3. Optimization of the SPE Trapping System with MXC. In previous publications, SPE trapping conditions for hydrophilic, basic compounds such as tamoxifen and raloxifene and acidic compounds such as resorufin were described (17, 18, 22). EDCs, in contrast, are highly apolar, lipophilic compounds (logPow,MXC ) 5.1 and logPow,BP3 ) 3.7) and were not trapped well under those conditions, tending instead to stick to the lipid bilayers of liver microsomes and to flow-path components such as PEEK tubing and reaction coils. Therefore, a novel SPE trapping method was developed, in which analyte was efficiently redissolved by flushing the system with a high concentration

1828 Chem. Res. Toxicol., Vol. 20, No. 12, 2007

Van Liempd et al.

Figure 2. (A) ERR affinity traces obtained after a 24 min incubation of 50 µM MXC with 880 nM P450 (thick line) and after a 15 min incubation of 50 µM MXC with 220 nM P450 (thin line, LLD for bis-OH-MXC) in the bioreactor. (B) UV absorption chromatogram for the 24 min incubation. The reaction volume was 500 µL.

of an organic modifier. This SPE method was optimized by monitoring redissolution and trapping of the apolar compound MXC and its less-apolar metabolite bis-OH-MXC at different concentrations of MeOH. Using lower concentrations of MeOH [1:9 and 2:8 (v/v) MeOH/water] resulted in the incomplete removal (99%) from the system. Since a 40% (v/ v) MeOH/water solution was sufficient for the complete removal of the compounds, this was chosen as the rinsing solution. However, high concentrations of MeOH decreased the breakthrough volume of the SPE material, meaning that MXC and bis-OH-MXC could not be retained on the SPE column. In order to circumvent this problem without a decrease in extraction efficiency, a modification of the SPE trapping mechanism was introduced, in which the initial 40% (v/v) MeOH solution was diluted with a make-up flow of water in a mixing coil just before the SPE column, decreasing the MeOH concentration to 16% (v/v). In this way, both the apolar MXC and the less-apolar bis-OH-MXC could be trapped completely (>99%). Moreover, with this newly introduced SPE make-up system, the precision of the bioreactor system was dramatically improved. In the previous setup, the relative standard deviation (RSD) values ranged from 18% to as much as 43% (17). With the new setup, however, the RSD values were less than 15% (as described in the next section) and therefore met or exceeded the current criteria for validation of biological methods (23). A possible explanation for this phenomenon is that the more rigorous redissolution method employed in this study may have made the system more resistant to variabilities in the metabolite mixtures, such as variabilities in lipid and protein concentrations (24). It seems that with the previous, water-based redissolution methods, varying amounts of the analytes adhered to other constituents of the system (protein, lipid bilayers, tubing) rather than to SPE material. This variable loss of analyte is circumvented with the new organic modifier-based method, leading to a higher precision for the system. Validation of the P450 Bioreactor System. Validation of the P450 bioreactor system was performed by monitoring the amount of product formed through conversion of MXC by BNFinduced RLMs as a function of time. The resulting relative ERR affinity traces (Figure 2A) show distinct negative peaks at retention times which, according to the UV trace (Figure 2B),

correspond to those for elution of compounds from the HPLC column. By comparison with retention times for synthetic reference compounds, we were able to distinguish the primary metabolite mono-OH-MXC (30.1 min) and the secondary metabolite bis-OH-MXC (22.6 min) of MXC (37.7 min). Although the absolute ERR affinity values (EC50 or Kd values) of the eluted metabolites were not determined, the affinities of mono- and bis-OH-MXC for ERR were significantly higher than that of MXC. This follows from the fact that the ERR affinity signals of mono- and bis-OH-MXC were much stronger than that of MXC, whereas the amounts of the three compounds in the ERAD assay were in the same range (1.6, 2.6, and 1.7 nmol, respectively). These results agreed with ERR response data found in the literature: the reported EC50 values for MXC, monoOH-MXC, and bis-OH-MXC are 7900, 198, and 79.4 nM, respectively (25, 26). Moreover, with the achiral gradient HPLC separation method used, it was not possible to establish the ratio of the amounts of the mono-OH-MXC enantiomers formed, yet it is known that both enantiomers of mono- and bis-OH-MXC have significantly higher affinities for ERR than does MXC itself (27). Furthermore, Figure 2 shows that the MXC metabolites MM1 (28.5 min), MM2 (27.2 min), and MM3 (19.1 min) also show higher affinities for ERR than does MXC, assuming that the UV absorption coefficients for all three compounds are approximately equal. MM1 or MM2 could be the catechol of MXC previously described by Hu et al. (7), and MM3 (19.1 min) might be a tris-OH-MXC catechol species (7, 28), given its relatively short retention time. However, no further experiments were carried out in order to confirm the identities of these metabolites. A critical point in biocatalysis studies is that products should be formed by enzymatic rather than nonenzymatic processes. In this study, product formation was indeed generated by enzymatic processes, as demonstrated by the metabolite formation curves (Figure 3) obtained by integrating the peaks of UV chromatograms acquired at different incubation times. These curves show that formation of MXC metabolites is NADPHand P450-dependent. The NADPH dependence is demonstrated by the fact that the t ) 0 data points were actually obtained from 33 min P450 incubations containing no NADPH. The P450 dependence is shown by the fact that increasing the P450 concentration increased the amount of product formed (Figure 3A–C).

Metabolic Profiling of Endocrine Disruptors

Chem. Res. Toxicol., Vol. 20, No. 12, 2007 1829

Figure 3. Metabolite formation curves depicting mono-OH-MXC, bis-OH-MXC, and cumulative metabolite formation (i.e., mono- plus bis-OHMXC production) as a function of time in the bioreactor system. The upper graphs (A–C) represent metabolite formation at P450 concentrations of 220 nM (0), 440 nM (4), and 880 nM (3) with a fixed MXC concentration of 50 µM. The lower graphs (D–F) represent product formation at MXC concentrations of 25 µM (9), 50 µM (4), and 100 µM (1) at a fixed P450 concentration of 440 nM. The total incubation volume was 500 µL, and n ) 3 for each data point.

The metabolite formation curves (Figure 3) were also used to determine product formation rates. The maximum rate of mono-OH-MXC production under the present conditions was 4.8 ( 0.1 nmol in 24 min at 100 µM MXC and 440 nM P450. Graphs of mono-OH-MXC formation versus time at various concentrations of both P450 (Figure 3A) and MXC (Figure 3D) showed nonlinearity. The graphs of bis-OH-MXC formation versus time were linear (r2 g 0.99) over a 24 min time span at P450 concentrations of 440 and 880 nM with a fixed MXC concentration (Figure 3B) and yielded formation rates of 62 ( 5 and 124 ( 16 pmol/min, respectively. Formation of bis-OHMXC was negligible at a P450 concentration of 220 nM. Plots of bis-OH-MXC formation versus time at 25 and 100 µM MXC and 440 nM P450 were also linear (r2 g 0.99) and gave production rates of 47 ( 3 and 130 ( 3 pmol/min, respectively (Figure 3E). Under normal circumstances, overall product formation is proportional to enzyme concentration (29). However, this is not the case when the cumulative formation of products from MXC is considered (i.e., mono- plus bis-OH-MXC formation, Figure 3C,F). Linear (r2 ) 0.99) cumulative product formation as a function of time was obtained only at 100 µM MXC and 440 nM P450, corresponding to a formation rate of 329 ( 24 pmol/ min. The nonlinearity at other concentrations can be explained by the fact that MXC and its metabolites, especially the catechol metabolites (possibly MM1 or MM2), covalently bind to rat liver microsomal proteins, including P450s, at a Vmax of approximately 20 pmol min–1 (mg of protein)–1 (30–32). Consequently, an increase in protein concentration may yield a cumulative amount of metabolic products that is lower than expected, as a result of covalent binding of the metabolites to protein. Interestingly, it was found that the metabolite yields and formation curves obtained with the present P450 bioreactor system were similar to those found using rat or human microsomes or human recombinant CYP1A isoforms (7, 28, 33). Finally, the sensitivity and repeatability of the method were determined. The lower limit of detection (LLD) of the ERR

affinity assay for mono-OH-MXC was 240 ng, as shown in the upper ERR affinity trace (thin line) in Figure 2A. This amount was obtained after incubation of 50 µM MXC with 220 nM P450 for 7 min. The LLD for bis-OH-MXC was 250 ng, which was obtained after incubation of 50 µM MXC with 220 nM P450 for 15 min. Previous publications (16, 34) showed that the sensitivity of the on-line ERR affinity assay used here is in the same range as those of other ER binding assays, such as fluorescence polarization and solid-phase scintillation proximity, yet those comparisons were not made for MXC. Moreover, when MXC was tested for ERR affinity using radiometric competitive binding assays, the compound was classified as a nonbinder because it was not able to replace the trace compound at a concentration of 100 µM (35, 36). However, with the HRS method employed in this study, a clear ERR affinity signal could be observed for MXC at concentrations lower than 50 µM (Figure 2). To determine the precision of the P450 bioreactor system, we calculated pooled RSDs of all data points for monoand bis-OH-MXC product formation, obtaining values of 12 and 15%, respectively. The present studies with MXC showed that the nature, amount, and rate of formation of the metabolites were comparable to those previously obtained with nonautomated off-line methods (7, 28, 33). MXC was found to be bioactivated by BNF-induced RLMs into five ERR-binding metabolites, of which mono- and bis-OH-MXC could be identified unambiguously. Interestingly, all five of these MXC metabolites showed higher affinities for ERR than did the parent compound. Metabolic Profiling of BP3. After optimization and validation of the on-line, hyphenated P450-bioreactor–ERAD system via analysis of MXC, the optimized system was used to investigate the on-line bioactivation of BP3 to ERR-binding metabolites by BNF-induced RLMs. To that end, 50 µM BP3 was metabolized by 44 nM P450 over a range of incubation times. The ERR affinity trace clearly showed negative peaks at 24.9, 25.5, and 26.9 min (Figure 4A), indicating the formation of three ERR-binding metabolites. The remaining parent com-

1830 Chem. Res. Toxicol., Vol. 20, No. 12, 2007

Van Liempd et al.

Figure 4. (A) ERR affinity trace obtained after reaction of 50 µM BP3 with 44 nM P450 for 24 min in a 500 µL reaction coil of the on-line bioreactor system. (B) Corresponding UV and LC-APCI-MS selective ion traces obtained after incubation of 50 µM BP3 with 1.8 µM P450 for 1 h. Note that the ERR affinity trace is obtained with a smaller amount of P450 enzyme and a shorter incubation time than the UV and selective ion trace chromatograms. Ions at m/z 227, 213, and 243 correspond to the parent compound (BP3), the demethylated metabolite (BP1), and the monohydroxylated metabolites (BM1 and BM2), respectively. (C) Metabolite formation curve showing the amount of the major metabolite BP1 formed as a function of time in the bioreactor system.

pound, BP3 (36.7 min), did not show any sign of ERR affinity. The metabolite formation curve of the third metabolite eluting from the column is shown in Figure 4C. It was determined that metabolite formation from BP3 was NADPH-dependent in the same way as that from MXC. Interestingly, neither UV nor selected ion trace (SIT) chromatograms (Figure 4B) showed any peaks at the retention times of the first two ERR-binding compounds (BM1 and BM2), even at the longest incubation time (24 min), whereas the ERR affinity trace showed distinct negative peaks (Figure 4A) at these retention times. This reflects the sensitivity of the ERR affinity assay. In order to optimize product formation in the bioreactor system for LC-MS analysis, the maximum incubation time was increased to 1 h and the P450 concentration to 1.8 µM. This resulted in the formation of amounts of products sufficient for detection with UV and identification with LC-MS (Figure 4B). Alignment of the ERR affinity response with the UV and LC-MS data showed that the first two peaks (BM1 and BM2) in the UV and SIT chromatograms (Figure 4B) matched mono-hydroxylated metabolites of BP3, according to the [M + OH – H]- ions at m/z 243. CID mass spectra of the selected ions at m/z 243 for both BM1 and BM2 confirmed that these ions were derived from BP3 metabolites, on the basis of the fragments at m/z 227, which correspond to the [M – H – OH]- ions of the metabolites. Although the present CID mass spectra could not reveal the regioselectivity of the hydroxyl groups introduced, previous research demonstrated the formation of 2,2′-dihydroxy-4methoxybenzophenone as a hydroxylated BP3 metabolite (9). However, according to Blair et al. (37), this metabolite has negligible affinity for the estrogen receptor. Therefore, it is unlikely that either BM1 or BM2 corresponds to 2,2′- dihydroxy-

4-methoxybenzophenone, since both BM1 and BM2 showed significant ERR affinity responses even when present in amounts undetectable by UV absorption (Figure 4A). This leads us to conclude that the hydroxyl groups are introduced at positions not described previously. This discovery of several new metabolites on the sole basis of ERR affinity demonstrates the effectiveness of the present hyphenated P450-bioreactor–ERAD system. The third major metabolite that eluted from the HPLC column most probably corresponds to BP1, the demethylated metabolite of BP3, given the demethylated-deprotonated ion [M – CH3 – H]- at m/z 213 and a comparison of its retention time to that of the chemical standard. Formation of BP1 is in agreement with previous findings (9, 10). From the literature (38) it is also known that BP1 has affinities for rainbow trout and human ERR that are 27 and 10 times higher, respectively, than those of the parent compound, BP3.

Conclusion In conclusion, we have developed an automated, hyphenated HRS-based system, consisting of an SPE–HPLC-coupled P450 bioreactor connected on-line with an ERR affinity assay, which is suitable for metabolic profiling of EDCs. Compared with a previous version of the system, improvements in SPE trapping led to an improved efficiency in analyzing compounds having a wide variety of polarities and to improved accuracy and wider applicability of the HRS-ERAD system. The system was optimized, validated, and tested using the EDCs MXC and BP3 and several of their ERR-binding metabolites. Sensitivity and precision of the P450-bioreactor–ERAD system are very good, meeting standards for modern bioassays. The hyphenated HRS

Metabolic Profiling of Endocrine Disruptors

system presented here allows on-line bioactivation, metabolite collection and concentration, and ERR affinity detection in a single run and proved effective in the metabolic profiling of two important EDCs. It thus constitutes a valuable addition to the nonautomated methods normally used in this type of research. Acknowledgment. The PEEK filter unit was designed and fabricated by D. J. van Ieperen and R. Boegschoten at the FEW fine mechanical workshop of the Vrije Universiteit. The hardware and computer software for the analytical system presented here were provided by Kiadis B.V. (Groningen, The Netherlands). The E. coli cells expressing ERR LBD were a kind gift of Dr. Marc Ruff and Dr. Dino Moras (Institut de Genetique et de Biologie Moleculaire et Cellulaire, Illkirch, France). Synthesis of bis-OH-MXC was supervised and partly performed by Maikel Wijtmans of the Medicinal Chemistry group at the Vrije Universiteit. Financial support for this project was provided by Senter-Novem/BTS (Grant BTS00091) and Merck Research Laboratories (Drug Metabolism Department).

References (1) Petrovic, M., Eljarrat, E., Lopez De Alda, M. J., and Barcelo, D. (2004) Endocrine disrupting compounds and other emerging contaminants in the environment: a survey on new monitoring strategies and occurrence data. Anal. Bioanal. Chem. 378, 549–562. (2) Koester, C. J., and Moulik, A. (2005) Trends in environmental analysis. Anal. Chem. 77, 3737–3754. (3) Massart, F., Parrino, R., Seppia, P., Federico, G., and Saggese, G. (2006) How do environmental estrogen disruptors induce precocious puberty? MinerVa Pediatr. 58, 247–254. (4) Mantovani, A. (2006) Risk assessment of endocrine disrupters: the role of toxicological studies. Ann. N.Y. Acad. Sci. 1076, 239–252. (5) Fenton, S. E. (2006) Endocrine-disrupting compounds and mammary gland development: early exposure and later life consequences. Endocrinology 147, S18–S24. (6) Charles, G. D., Bartels, M. J., Gennings, C., Zacharewski, T. R., Freshour, N. L., Bhaskar Gollapudi, B., and Carney, E. W. (2000) Incorporation of S-9 activation into an ER-R transactivation assay. Reprod. Toxicol. 14, 207–216. (7) Hu, Y., and Kupfer, D. (2002) Metabolism of the endocrine disruptor pesticide-methoxychlor by human P450s: pathways involving a novel catechol metabolite. Drug Metab. Dispos. 30, 1035–1042. (8) Yoshihara, S., Makishima, M., Suzuki, N., and Ohta, S. (2001) Metabolic activation of bisphenol A by rat liver S9 fraction. Toxicol. Sci. 62, 221–227. (9) Nakagawa, Y., and Suzuki, T. (2002) Metabolism of 2-hydroxy-4methoxybenzophenone in isolated rat hepatocytes and xenoestrogenic effects of its metabolites on MCF-7 human breast cancer cells. Chem.– Biol. Interact. 139, 115–128. (10) Takatori, S., Kitagawa, Y., Oda, H., Miwa, G., Nishikawa, J., Nishihara, T., Nakazawa, H., and Hori, S. (2003) Estrogenicity of metabolites of benzophenone derivatives examined by a yeast twohybrid assay. J. Health Sci. 49, 91–98. (11) van Lipzig, M. M., Commandeur, J. N., de Kanter, F. J., Damsten, M. C., Vermeulen, N. P., Maat, E., Groot, E. J., Brouwer, A., Kester, M. H., Visser, T. J., and Meerman, J. H. (2005) Bioactivation of dibrominated biphenyls by cytochrome P450 activity to metabolites with estrogenic activity and estrogen sulfotransferase inhibition capacity. Chem. Res. Toxicol. 18, 1691–1700. (12) Fura, A. (2006) Role of pharmacologically active metabolites in drug discovery and development. Drug DiscoVery Today 11, 133–142. (13) Kool, J., Ramautar, R., van Liempd, S. M., Beckman, J., de Kanter, F. J., Meerman, J. H., Schenk, T., Irth, H., Commandeur, J. N., and Vermeulen, N. P. (2006) Rapid on-line profiling of estrogen receptor binding metabolites of tamoxifen. J. Med. Chem. 49, 3287–3292. (14) Fisher, J. S. (2004) Are all EDC effects mediated via steroid hormone receptors? Toxicology 205, 33–41. (15) Schenk, T., Breel, G. J., Koevoets, P., van den Berg, S., Hogenboom, A. C., Irth, H., Tjaden, U. R., and van der Greef, J. (2003) Screening of natural products extracts for the presence of phosphodiesterase inhibitors using liquid chromatography coupled online to parallel biochemical detection and chemical characterization. J. Biomol. Screening 8, 421–429.

Chem. Res. Toxicol., Vol. 20, No. 12, 2007 1831 (16) Oosterkamp, A. J., Villaverde Herraiz, M. T., Irth, H., Tjaden, U. R., and van der Greef, J. (1996) Reversed-phase liquid chromatography coupled on-line to receptor affinity detection based on the human estrogen receptor. Anal. Chem. 68, 1201–1206. (17) van Liempd, S. M., Kool, J., Niessen, W. M., van Elswijk, D. E., Irth, H., and Vermeulen, N. P. (2006) On-line formation, separation, and estrogen receptor affinity screening of cytochrome P450-derived metabolites of selective estrogen receptor modulators. Drug Metab. Dispos. 34, 1640–1649. (18) van Liempd, S. M., Kool, J., Reinen, J., Schenk, T., Meerman, J. H., Irth, H., and Vermeulen, N. P. (2005) Development and validation of a microsomal online cytochrome P450 bioreactor coupled to solidphase extraction and reversed-phase liquid chromatography. J. Chromatogr. A1075, 205–212. (19) Weaver, R. J., Thompson, S., Smith, G., Dickins, M., Elcombe, C. R., Mayer, R. T., and Burke, M. D. (1994) A comparative study of constitutive and induced alkoxyresorufin O-dealkylation and individual cytochrome P450 forms in cynomolgus monkey (Macaca fascicularis), human, mouse, rat and hamster liver microsomes. Biochem. Pharmacol. 47, 763–773. (20) Rooseboom, M., Commandeur, J. N., Floor, G. C., Rettie, A. E., and Vermeulen, N. P. (2001) Selenoxidation by flavin-containing monooxygenases as a novel pathway for beta-elimination of selenocysteine Se-conjugates. Chem. Res. Toxicol. 14, 127–134. (21) Eiler, S., Gangloff, M., Duclaud, S., Moras, D., and Ruff, M. (2001) Overexpression, purification, and crystal structure of native ER R LBD. Protein Expression Purif. 22, 165–173. (22) Poole, C. F., Gunatilleka, A. D., and Sethuraman, R. (2000) Contributions of theory to method development in solid-phase extraction. J. Chromatogr. A885, 17–39. (23) Shah, V. P., Midha, K. K., Findlay, J. W., Hill, H. M., Hulse, J. D., McGilveray, I. J., McKay, G., Miller, K. J., Patnaik, R. N., Powell, M. L., Tonelli, A., Viswanathan, C. T., and Yacobi, A. (2000) Bioanalytical method validation—a revisit with a decade of progress. Pharm. Res. 17, 1551–1557. (24) Dills, R. L., Wu, R. L., Checkoway, H., and Kalman, D. A. (1991) Capillary gas chromatographic method for mandelic and phenylglyoxylic acids in urine. Int. Arch. Occup. EnViron. Health 62, 603–606. (25) Sonneveld, E., Jansen, H. J., Riteco, J. A., Brouwer, A., and van der Burg, B. (2005) Development of androgen- and estrogen-responsive bioassays, members of a panel of human cell line-based highly selective steroid-responsive bioassays. Toxicol. Sci. 83, 136–148. (26) Gaido, K. W., Maness, S. C., McDonnell, D. P., Dehal, S. S., Kupfer, D., and Safe, S. (2000) Interaction of methoxychlor and related compounds with estrogen receptor R and β, and androgen receptor: structure-activity studies. Mol. Pharmacol. 58, 852–858. (27) Miyashita, M., Shimada, T., Nakagami, S., Kurihara, N., Miyagawa, H., and Akamatsu, M. (2004) Enantioselective recognition of monodemethylated methoxychlor metabolites by the estrogen receptor. Chemosphere 54, 1273–1276. (28) Kupfer, D., Bulger, W. H., and Theoharides, A. D. (1990) Metabolism of methoxychlor by hepatic P-450 monooxygenases in rat and human. 1. Characterization of a novel catechol metabolite. Chem. Res. Toxicol. 3, 8–16. (29) Shou, M., Lin, Y., Lu, P., Tang, C., Mei, Q., Cui, D., Tang, W., Ngui, J. S., Lin, C. C., Singh, R., Wong, B. K., Yergey, J. A., Lin, J. H., Pearson, P. G., Baillie, T. A., Rodrigues, A. D., and Rushmore, T. H. (2001) Enzyme kinetics of cytochrome P450-mediated reactions. Curr. Drug Metab. 2, 17–36. (30) Bulger, W. H., and Kupfer, D. (1989) Characteristics of monooxygenase-mediated covalent binding of methoxychlor in human and rat liver microsomes. Drug Metab. Dispos. 17, 487–494. (31) Morrell, S. L., Fuchs, J. A., and Holtzman, J. L. (2000) Effect of methoxychlor administration to male rats on hepatic, microsomal iodothyronine 5′-deiodinase, form I. J. Pharmacol. Exp. Ther. 294, 308–312. (32) Zhou, L. X., Dehal, S. S., Kupfer, D., Morrell, S., McKenzie, B. A., Eccleston, E. D., Jr., and Holtzman, J. L. (1995) Cytochrome P450 catalyzed covalent binding of methoxychlor to rat hepatic, microsomal iodothyronine 5′-monodeiodinase, type I: does exposure to methoxychlor disrupt thyroid hormone metabolism? Arch. Biochem. Biophys. 322, 390–394. (33) Stresser, D. M., and Kupfer, D. (1998) Human cytochrome P450catalyzed conversion of the proestrogenic pesticide methoxychlor into an estrogen. Role of CYP2C19 and CYP1A2 in O-demethylation. Drug Metab. Dispos. 26, 868–874. (34) Schobel, U., Frenay, M., van Elswijk, D. A., McAndrews, J. M., Long, K. R., Olson, L. M., Bobzin, S. C., and Irth, H. (2001) High resolution screening of plant natural product extracts for estrogen receptor R and β binding activity using an online HPLC-MS biochemical detection system. J. Biomol. Screening 6, 291–303. (35) Kuiper, G. G., Lemmen, J. G., Carlsson, B., Corton, J. C., Safe, S. H., van der Saag, P. T., van der Burg, B., and Gustafsson, J. A. (1998)

1832 Chem. Res. Toxicol., Vol. 20, No. 12, 2007 Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor β. Endocrinology 139, 4252–4263. (36) Matthews, J., Celius, T., Halgren, R., and Zacharewski, T. (2000) Differential estrogen receptor binding of estrogenic substances: a species comparison. J. Steroid Biochem. Mol. Biol. 74, 223–234. (37) Blair, R. M., Fang, H., Branham, W. S., Hass, B. S., Dial, S. L., Moland, C. L., Tong, W., Shi, L., Perkins, R., and Sheehan, D. M.

Van Liempd et al. (2000) The estrogen receptor relative binding affinities of 188 natural and xenochemicals: structural diversity of ligands. Toxicol. Sci. 54, 138–153. (38) Kunz, P. Y., Galicia, H. F., and Fent, K. (2006) Comparison of in vitro and in vivo estrogenic activity of UV filters in fish. Toxicol. Sci. 90, 349–361.

TX7000724