Article pubs.acs.org/ac
Direct Coupling of Thin-Layer Chromatography with a Bioassay for the Detection of Estrogenic Compounds: Applications for EffectDirected Analysis Sebastian Buchinger,* Denise Spira, Kathrin Bröder, Michael Schlüsener, Thomas Ternes, and Georg Reifferscheid German Federal Institute of Hydrology, Am Mainzer Tor 1, D-56068 Koblenz, Germany S Supporting Information *
ABSTRACT: The present study investigated the hypothesis that the coupling of high-performance thin-layer chromatography with the yeast estrogen screen (planar-YES, p-YES) can be used as a screening tool for effect-directed analysis. Therefore, the proposed method was challenged for the first time with several real samples from various origins such as sediment pore water, wastewater, and sunscreens. It was possible to detect and quantify estrogenic effects in all investigated sample types, even in the presence of demanding matrixes. Furthermore, the specific agonistic effect of the estrogen receptor activation could be detected in samples exhibiting cytotoxic effects and at cytotoxic levels of analyzed estrogenic compounds, which is not possible with the classic YES. The analysis of samples by the p-YES results in profiles of estrogenic activity. By means of this profiles samples can be compared qualitatively and quantitatively with respect to different compositions of bioactive compounds in mixtures. In conclusion, the p-YES approach seems to have a high potential to be used as a valuable screening tool for various applications in effect-directed analysis.
T
identified phytotoxic and fungitoxic acetylenes from Conyza canadensis by a preparative HPLC of a dichloromethane extract from the aerial parts of this plant. Bioactive fractions were identified according to Dayan et al.12 and analyzed by means of GC/MS and NMR for a structural elucidation of the compounds.13,14 Fabel et al. coupled HPLC with a gassegmented online acetylcholinesterase inhibition assay for the direct identification of potential neurotoxic effects in fractions of mixtures.15 Such air-segmented online systems allow a highthroughput screening of, e.g., fractions from an HPLC, but suffer from some limitations as summarized by Weller:3 “The disadvantages of liquid chromatography are the use of organic solvents and other additives, which are rarely compatible with bio- or biochemical assays. Particularly difficult are gradientbased separations, in which the solvent composition is heavily changing.” As an alternative separation method to liquid chromatography, thin-layer chromatography (TLC) could be used. The separation efficiency of TLC is lower compared to HPLC but was improved by the development of new materials for the stationary phase16,17 and new techniques for plate development like automated multiple development (AMD) which allows a
he identification of bioactive compounds in mixtures is one of the main tasks in a number of fields in chemical analysis such as environmental chemistry and ecotoxicology, food chemistry, drug development, and the risk characterization of, e.g., personal care products. The fundamental question is, which of the sample constituents can cause a desired or undesired biological effect. The answer to this question is especially challenging in environmental analysis because of the vast number of unknown compounds in environmental samples. Due to the rapid development of chemical analysis it is now possible to detect a multitude of compounds at very low concentrations (down to the picogram per liter level). However, the identification of compounds with a clear (eco)toxicological relevance is still a major challenge. The effect-directed analysis (EDA) is a common and highly valuable strategy to link chemical analysis with biological effects.1−3 A number of studies describe the successful application of EDA for the identification of, e.g., mutagenic compounds,4−7 xenoestrogens,4,8−10 or aryl hydrocarbon receptor agonists.11 The basic principle of EDA is the separation of a compound mixture in different fractions with a subsequent analysis of the fractions for biological activity.1 Bioactive fractions can be separated further or are directed to compound identification by chemical analysis. The methods for sample fractionation are commonly based on high-performance liquid chromatography (HPLC). As one of a number of examples for such bioassay guided analysis Queiroz et al. © 2013 American Chemical Society
Received: April 15, 2013 Accepted: June 25, 2013 Published: June 25, 2013 7248
dx.doi.org/10.1021/ac4010925 | Anal. Chem. 2013, 85, 7248−7256
Analytical Chemistry
Article
Chromatography was performed on 10 cm × 20 cm silica gel 60 F254 HPTLC plates (1.05642, Merck, Darmstadt, Germany). Standard solutions or samples were sprayed on the TLC plates as 6 mm bands by using an ATS 4 application device. An automated two-step development was performed with the AMD2. The first step was performed with 100% methanol p.A. from the starting line at 10 mm up to 20 mm, followed by a drying under vacuum for 5 min. The second step was carried out by using a mobile phase according to Cimpoiu et al.31 The following solvent mixture was automatically prepared by the AMD2: chloroform/acetone/petroleum fraction 55:20:25 (v/ v/v). The final migration distance on the TLC plate was 85 mm, and the final drying time under vacuum was adjusted to 5 min. A picture of the developed TLC plate was taken under UV light (366 nm) in order to detect fluorescent compounds which might be present in the sample. Planar-Yeast Estrogen Screen (p-YES). A general overview about the various process steps of the procedure is given in Figure 1. The used yeast strain was generated by McDonnell et al.27,32 It is derived from the yeast strain Saccharomyces cerevisiae BJ3505 (protease-deficient, MATα, PEP4::HIS3, prb-1-δ1.6R, HIS3-δ200, lys2-801, trp1-δ101, ura3-52gal2can1). For the detection of compounds with an estrogenic potential, i.e., agonists of the estrogen receptor α, the TLC plate was immersed in a suspension of the yeast cells in the test medium. The cell density of the suspension was adjusted to 150 ± 50 formazin attenuation units (FAU). The FAU calibration was done according to ISO 702733 at a wavelength of 600 nm. In order to increase the viscosity of the test medium for an enhanced adhesion of the cells on the surface of the TLC plate and to ensure a sufficient moisture during the incubation period Gelrite (Sigma-Aldrich) was added to a final concentration of 0.375 g/L. A total volume of 200 mL of cell culture with Gelrite was prepared for the immersion chamber. The developed and vacuum-dried plates were immersed into the cell suspension with the chromatogram immersion device III for 3 s with the adjusted speed of 2 cm/s. After removing the plates from the immersion device they were dried on the rear site with paper towels. Subsequently, the TLC plates were placed in a plastic box with closed lid which in turn was placed in a temperature-controlled water bath at 30 °C with a saturated humid atmosphere (92% relative humidity). The yeast cells were exposed for 3 h under the described conditions. After the incubation the substrate 4-methylumbelliferyl-β-Dgalactopyranoside (MUG) (CAS 6160-78-7) was applied to the TLC plate. A 100-fold MUG stock suspension (50 mg/mL in ethanol p.A.) was added to the lacZ buffer to a final concentration of 0.5 mg/mL. The MUG stock suspension was stored in aliquots at −20 °C. The spraying solution was prepared immediately before use. The mixture was steadily sprayed 20 times on each plate (10 × 20 cm) under a fume hood. For the enzymatic reaction the plates were incubated in a plastic bowl without lid in an incubator at 37 °C for at least 15 min. The relative humidity of the atmosphere in the incubator after the incubation time of 15 min was approximately 30%. Signal Detection. The fluorescence of the cleaved MUG substrate, which is a measure for the estrogenic potential of a compound or sample, was observed at an excitation wavelength of λex = 366 nm using the UV cabinet 2. Pictures of the plates were taken at λex = 366 nm with the TLC visualizer. In addition to the photographic signal detection the fluorescence measure-
sequential development of a thin-layer plate with solvents of lower elution strength resulting in a gradient elution of the analytes.18,19 With the overpressured layer chromatography (OPLC) a technique is available which combines the principles of TLC and high-performance liquid chromatography (reviewed in refs 20 and 21). Several studies demonstrated the direct coupling of TLC with bioassays for the detection of acute toxic effects.22−24 Recently, OPLC was used to identify antibacterial compounds in Matricaria recutica L. by a combination with bioautography using Bacillus subtilis and Pseudomonas syringae.25 The spectrum of possible biological end points which can be detected directly on the surface of a TLC plate was broadened beyond acute bacterial toxicity by the implementation of a cholinesterase inhibition assay.23 Initial work for the coupling of high-performance thin-layer chromatography (HPTLC) and a modified version of the yeast estrogen screen (YES)26,27 for the detection of estrogenic compounds was reported by Müller et al.28 The yeast estrogen screen is a yeast-based reporter gene assay in which the reporter gene lacZ is controlled by an estrogen-dependent promoter. Thus, the expression of lacZ which encodes the enzyme βgalactosidase is a measure for the estrogenic activity of a sample. Although the first steps for a combination of the yeast estrogen screen with TLC were already shown,28 a proof of principle for the whole assay procedure including plate development, bioassay, and a subsequent analysis by mass spectrometry is still missing. After the implementation of a method for the coupling of TLC with the yeast estrogen screen29,30 the present work aims to test the hypothesis that the described approach is a valuable and robust screening tool for effect-directed analysis. Therefore, the method is challenged with real samples which cover various fields of applications, i.e., analysis of wastewater and sediment as well as the characterization of personal care products, e.g., sunscreens.
■
MATERIAL AND METHODS Chemicals. All chemicals were obtained from Sigma-Aldrich in reagent grade. The following compounds were used as model compounds with known estrogenic activity: 17α-ethinylestradiol (EE2, CAS 57-63-6), 17β-estradiol (E2, CAS 85-05-7), estrone (E1, CAS 53-16-7), bisphenol A (BPA, CAS 85-05-7), and for the determination of recovery deuterated bisphenol A (BPA-d16, CAS 96210-87-6). All standard compounds were dissolved in ethanol p.A. or methanol p.A. Buffer and Media. Growth medium contained the following: yeast nitrogen base 6.7 g/L, α-D-glucose 20 g/L, Llysine 36 mg/L, L-histidine 24 mg/L. Test medium contained the following: yeast nitrogen base 31.4 g/L, α-D-glucose 138.79 g/L, L-lysine 168.75 mg/L; L-histidine 112.5 mg/L, ampicillin sodium salt 375 mg/L, streptomycin sodium salt 375 mg/L, CuSO4·5H2O 35 mg/L. lacZ buffer contained the following: Na2HPO4·2H2O 10.67 g/L; KCL 0.75 g/L; MgSO4·7H2O 0.25 g/L; NaH2PO4 5.5 g/L; pH: 7; sodium dodecyl sulfate 1 g/L. High-Performance Thin-Layer Chromatography. The following equipment was used for the chromatographic separation and the sample application: automatic TLC sampler ATS4, automated multiple development system AMD2, chromatogram immersion device III, TLC visualizer, TLC scanner 4, TLC lab glass sprayer (all devices from Camag, Muttenz, Switzerland). 7249
dx.doi.org/10.1021/ac4010925 | Anal. Chem. 2013, 85, 7248−7256
Analytical Chemistry
Article
99.7% confidence intervals were calculated additionally and were used for the determination of the limit of detection (LOD) and limit of quantification (LOQ) with the calibration curve approach according to Hubaux and Vos34 and Funk and Dammann.35 The ED50, i.e., the dose of a compound which induces a half-maximal effect, corresponds to the parameter x0. Mass Spectrometry. The HPLC system consisted of a G1367E autosampler, a G1312B binary HPLC pump, a G1379B degasser, and a G1314B column oven (all Agilent 1260 SL series, Waldbronn, Germany). The detection was performed on a API 5500 QTrap mass spectrometer (Applied Biosystems, Foster City, CA, U.S.A.). Separations were performed using a Synergi-Luna-HST column (2 mm i.d., length 50 mm, particle size 2.5 μm) and a SecurityGuard (both Phenomenex, Torrance, CA, U.S.A.) at 20 ± 2 °C. The flow rate was 0.4 mL/min. The HPLC gradient was established by mixing two mobile phases: phase A, Milli-Q water with 40 mg/L ammonia acetate, and phase B, pure methanol. Chromatographic separation was achieved with the following gradient: 0−0.8 min, 10% B; 0.8−1.2 min, 10% → 98% B; 1.2−4.8 min, 98% B; 4.8−4.9 min, 98% → 10% B; 4.8− 8 min, 10% B (Supporting Information Table S-1). An amount of 10 μL of each sample was directly injected without any further sample treatment. The tandem mass spectrometer was operated in negative ion mode using nitrogen as collision gas and multiple reaction monitoring (MRM) for quantification. Parameters adjusted were collision gas (CAD), medium; curtain gas (CUR), 55 psi; ion source gas 1 (GS1), 40 psi and ion source gas 2 (GS2), 50 psi; source temperature (TEM), 550 °C; entrance potential (EP), −10 V. The ionspray voltage (IS) was adjusted to −4500 V, and the interface heater (ihe) was set on. Two MRM transitions were monitored for identification and quantification of bisphenol A-d16. Parameters such as declustering potential, collision energy, and cell exit potential were optimized in the autotuning routine of the Analyst 1.6.1 software (see Supporting Information Table S-2 for an overview of all MRM parameters). The LOQ was defined as a signal-to-noise ratio of 10:1 and the LOD as a signal-to-noise ratio of 3:1. The LOQ of bisphenol A-d16 was 5 ng/mL.
Figure 1. Scheme of the planar-YES. After performance of the thinlayer chromatography the TLC plate with the separated compounds is dried and subsequently immersed in a suspension of yeast cells (S. cerevisiae) which are genetically modified to express the reporter gene lacZ (encoding the enzyme β-galactosidase) in dependence of an estrogenic stimulus. The yeast cells which adhere on the surface of the TLC plate are exposed to the separated compounds in a humid atmosphere. Estrogenic compounds enter the cells and induce the heterologously expressed estrogen receptor which in turn activates lacZ. Finally, a nonfluorescent substrate (MUG) is sprayed on the TLC plate for development. At areas on the TLC plate where the βgalactosidase is present the substrate is cleaved to a fluorescent product which is detectable by UV irradiation.
■
RESULTS AND DISCUSSION
Characterization of Analytical Parameters with Model Compounds. After the initial implementation of the coupling of TLC with the yeast estrogen screen (p-YES)29 the estrogenic model compounds E2, EE2, and BPA were characterized with the proposed method in terms of LOD, LOQ, and retention factor. The results are shown in Supporting Information Figure S-1 (E2 only) and summarized in Table 1 for all three compounds. As indicated by the different Rf values obtained by the p-YES a separation of the steroidal compounds E2 and EE2
ment was done with a TLC scanner 4 in combination with the software WinCATS 1.4.6 (Camag) for a quantitative evaluation. The excitation wavelength for the fluorescence detection of 4methylumbelliferone (4-MU) on the TLC plate was set to λex = 320 nm (deuterium lamp). A 440 nm cutoff filter was used to detect emitted fluorescence light above 440 nm only. The slit dimension for the scan was set to 6 mm × 0.3 mm. To create dose−response curves arbitrary units (AU) of the determined peak areas were used. Determination of Analytical Parameters. For each investigated model compound the measured data points of the dose−response relationships were collected and fitted with a four-parametric logistic equation [y = y0 + a/(1 + (x/x0)b)] with the software package SigmaPlot 10.0. The upper and lower
Table 1. Analytical Parameters of the p-YES for Estrogenic Model Compoundsa compd
no. of dilution levels
LOD
LOQ
ED50
Rf
E2 EE2 BPA
8 10 6
1.3 pg 0.5 pg 70 ng
2.8 pg 1.6 pg 190 ng
7.5 pg 17 pg 272 ng
0.67 ± 0.02 0.73 ± 0.03 0.66 ± 0.06
a
LOD, limit of detection; LOQ, limit of quantification; ED50, effect dose 50%; Rf, retention factor; n = 3. 7250
dx.doi.org/10.1021/ac4010925 | Anal. Chem. 2013, 85, 7248−7256
Analytical Chemistry
Article
substantially decreased the signal intensity of BPA-d16. The relative recovery was decreased to 15% after the p-YES and to 9% after the chromatographic development and the p-YES. This drop might be explained by the uptake of BPA-d16 by the yeast cells which are immobilized on the surface of the TLC plate for the detection of the specific estrogenic effect. The uptake of analytes by the cells is a prerequisite for the detection of the biological activity of the analytes. Consequently, the compound is extracted from the TLC plate by the yeast cells. The fate of the analytein the present case BPA-d16after the cellular uptake is unknown, but an intracellular metabolization or a possibly irreversible binding to cellular structures could explain the decreased recovery after the p-YES. Nevertheless, the experiment demonstrated the direct accessibility of compounds for a subsequent analysis by mass spectrometry even after the performance of the p-YES. To avoid the loss in recovery by the p-YES the compound elution should be done from a second TLC plate which is run in parallel. Characterization of Estrogenic Effects in Environmental Samples (Sediment Pore Water). As an example for the characterization of an environmental sample with the described method freeze-dried pore waters of sediments taken from the German river Elbe were analyzed by the p-YES. Ethanolic extracts of the freeze-dried sediment pore waters (see the Supporting Information) were applied on a TLC plate and analyzed by the p-YES. Figure 2a and Supporting Information Figure S-2A show the TLC plate after chromatographic development under UV light (366 nm). Lane 1 contains a process blank, i.e., the ethanolic solution of 100 mL of freezedried tap water. A mixture of E1 (38 pg), E2 (8 pg), and EE2 (8 pg) was applied to lane 6 as a positive control. All analyzed pore waters showed fluorescence signals at various Rf values. The fluorescence image after the p-YES is shown in Figure 2b (grayscale) and Supporting Information Figure S-2B (color). In lane 2, 50 μL of the sample pw316 was applied. No fluorescence signal is detected after the exposition of the yeast cells and the administration of MUG. In case of the other samples fluorescence signals were visible at Rf = 0.70. At this Rf no fluorescence signal was visible before the p-YES. A comparison of the quality of the fluorescence signals directly after the chromatography (Figure 2a) and the fluorescence signals of the cleaved MUG (Figure 2b) shows that the signals which are caused by fluorescent compounds present in the sample are sharper and more distinct than the signals after the bioassay. This is possibly caused by diffusion processes during the exposure of the cells and the subsequent enzymatic reaction. Further research should aim to increase the quality of the fluorescence signals after the bioassay. Additionally, the chromatographic conditions should be optimized in future with respect to compound separation and as well the composition of the mobile phase in order to avoid the use of chlorinated alkanes. However, the presented findings clearly demonstrate the specific detection of cleaved MUG which indicates the presence of estrogenic compounds in the samples. The p-YES supports a compound identification in two ways. First, as demonstrated with the model compound BPA-d16, the analyte can be eluted from the stationary phase of the TLC plate for a subsequent analysis by mass spectrometry. Second, the chromatographic behavior of a compound allows a comparison with the mobility of a reference substance. A comparison of the retention behavior between the unknown estrogenic compounds in the samples and the three reference
is possible with the applied chromatographic conditions. However, BPA showed a mobility similar to E2. In all cases an analysis of the estrogenic activity after the chromatographic run was possible. In accordance to former publications it was found that the estrogenic potency of the xenoestrogen BPA is lower by orders of magnitude compared to the steroidal estrogens E2 and EE2. The quotient of the EC50 for E2 to the EC50 of a given compound is frequently used to express the relative estrogenic potency of a compound. For the present study the effect dose 50% (ED50) was used for this calculation instead of the EC50, i.e., the ratio of the absolute amounts applied to the TLC plate in nanograms which result in a halfmaximal response instead of the ratio of concentrations. On the basis of the shown results the estrogenic potency of EE2 is 0.44 and that of BPA 2.8 × 10−5 in relation to E2. These data are in a general agreement with reported values for EE2 (0.936) and BPA (6.2 × 10−5 37). In the case of E2 and EE2 the LOQ of the p-YES is about two to 3-fold higher compared to the classic YES which is performed in a 96-well plate. The LOQ for E2 and EE2 of approximately 10 ng/L are usually achieved with the YES.36 This concentration corresponds to a total amount of 0.8 pg for a typical sample volume of 80 μL per well. Recovery of Analytes after the p-YES. Deuterated bisphenol A (BPA-d16) was used as a xenoestrogenic model compound to investigate the direct accessibility of analytes after separation and bioautography for a subsequent chemical analysis by mass spectrometry. The TLC silica matrix around the fluorescence signal was scratched from the TLC plate, extracted with methanol (see the Supporting Information), and analyzed by a tandem mass spectrometer with electron spray ionization after performing the whole assay procedure including TLC development, application of the yeast cells, and signal detection after MUG cleavage by the β-galactosidase. Bisphenol A-d16 was clearly identified by two MRM transitions (Supporting Information Table S-2). In addition to the qualitative analysis the recovery of BPA-d16 from the silica matrix was assessed. In order to characterize possible impacts of the various steps of the process on the recovery, BPA-d16 was applied on a TLC plate, eluted from the same spot, and analyzed by LC/MS. The determined peak area of this signal served as reference for the subsequent experiments. The procedure was repeated with the same amount of BPA-d16 after a chromatographic run, i.e., the BPA-d16 was eluted from the TLC plate after migration. The determined peak area of this sample was 95% with respect to the reference (Table 2). It can be concluded that the chromatographic run alone does not impact the recovery of compound and that there is no loss of BPA-d16 during the development of the TLC plate. However, the position of the analyte on the plate, i.e., its retention behavior, has to be known. Whereas the chromatographic run did not decrease the recovery, the p-YES procedure Table 2. Recovery of BPA-d16 in Dependence of Various Process Steps sample I (reference) ii iii iv process blank solvent blank
TLC development
pYES
no yes no yes
no no yes yes
signal intensity (AU) 2.84 2.70 4.39 2.39
