Liquid Chromatography-Bioassay-Mass Spectrometry for Profiling of

Article pubs.acs.org/ac

Liquid Chromatography-Bioassay-Mass Spectrometry for Profiling of Physiologically Active Food Gertrud E. Morlock*,† and Ines Klingelhöfer† Chair of Food Science, Justus Liebig University Giessen, Interdisciplinary Research Centre (IFZ) and Institute of Nutritional Science, Heinrich-Buff-Ring 26-32, 35392 Giessen, Germany S Supporting Information *

ABSTRACT: Complex samples like food contain thousands of single compounds. In the past, only known target compounds were looked for; however, most bioactive compounds in food are unknown. On the contrary, nontarget analyses face the challenge of determining the thousand peaks’ identities, but it remains largely unclear which peaks are bioactive. Here, we show a novel effect-directed food profiling, as food and food supplements can be unknowingly physiologically active. By the combination of planar chromatography, using water-wettable reversed phase high-performance thin-layer chromatography (HPTLC RP18 W) plates, with detection by specific microorganisms, endocrine compounds in food were quantitatively detected as sharpbounded zones and further characterized by mass spectrometry. This analytical workflow allowed frequent food intakes to be identified as risky with regard to estrogen-effective compounds, in discussion for their potential involvement in foodborne pathogenesis and for use in personalized health care. Using this accelerated workflow with its comprehensive detection potential, unknown endocrine compounds can be discovered. Exemplarily, the discovery of up to six endocrine disrupting compounds was shown in seven propolis samples and in four spices. For example, microorganisms quantitatively detected an estrogen-effective compound in the range of 0.07−0.24% in seven propolis samples, which was assigned to be caffeic acid phenethyl ester by mass spectrometry. This streamlined nontarget analysis detected modes of action, followed by targeted characterization of newly discovered effective compounds. Also, drug discovery or analysis of traditional medicines may profit from this effect-directed profiling of complex samples.

A

processing or contaminants etc., and largely out of the analytical focus. As planar chromatography is a liquid chromatography (LC) technique,12 we introduce a quantitative LC-bioassay-mass spectrometry (LC-bioassay-MS) workflow for detecting small changes induced by a food ingredient on a cell-, receptor-, or enzyme-based effect level. Thus, information is obtained on the full range of effective compounds in a single measurement, with a procedure that detects modes of action and does not preselect analytes. Exemplarily, this effect-directed food profiling was demonstrated for endocrine disrupting compounds (EDCs). EDCs are important bioactive target compounds due to the increasing xeno-pollution but also with regard to physiologically active food, inclusive of drinking water. For performing the bioassay, recombinant Saccharomyces cerevisiae cells were used, carrying the DNA sequence of the human estrogen receptor (hERα) together with the reporter gene lac-Z.9,10 EDCs induced lac-Z gene expression, which encoded ß-galactosidase, reacting at pH > 7 with 4-methylumbelliferyl-ß-D-galactopyr-

fter food intake, foodomics pursues largely the output side, the unique biochemical and small-molecule fingerprint that food components induce in the metabolome.1 On the contrary, still too little is known about the active principle on the input side, meaning the intake of bioactive food ingredients. Stimuli in food might be responsible for diseases, when special food is preferred for daily intake. Over the past decades, the success of foodomics has relied on advances in mass spectrometry instrumentation2 and bioinformatics, which made it possible to detect thousands of compounds simultaneously3 or to improve food safety.4,5 This nontarget analysis faces the challenge of either determining the identities of all the thousand peaks found or defining bioinformatically selected peaks as marker compounds. Despite these dedicated efforts, it still remained a secret which of the thousand peaks are bioactive, except the few ones known. On the other hand, food analysts use a targeted platform, in which only a specified list of metabolites or ingredients is measured like in residue analysis.6,7 Even excellent multiresidue methods6−8 can only contribute to a certain extent in identifying noxious matter. Only known bioactive target compounds are looked for; however, most bioactive components of food are unknown, either natural ingredients or degradation products of food © 2014 American Chemical Society

Received: May 8, 2014 Accepted: July 28, 2014 Published: July 28, 2014 8289

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295

Analytical Chemistry

Article

capsules (250 mg/capsule), Börner, Bremen, Germany (P6), and Sanhelios Aagaard propolis throat lozenges (30 mg/suck lozenges), Börner (P7). For all samples, voucher specimens were deposited at the Chair of Food Science, JLU Giessen, Germany. Sample Solutions. Liquid samples (P1−P5) were used directly or diluted 1:10 with ethanol. The content of one P6 capsule was cold extracted with 1 mL of ethanol (20 °C) for 3 min using a vortex and centrifuged at 15 000g for 5 min. P7 lozenges were pestled, and 100 mg of the crushed material was extracted analogously to P6. Spice samples (0.5 g) were pestled and suspended each with 5 mL of methanol−ethanol 1:1 (V/ V). The suspensions were heated to 60 °C in sealed laboratory vessels for 30 min, followed by 20 min of ultrasonification and centrifugation at 15 000g for 5 min. The supernatants were stored at −20 °C. For removal of chlorophylls, the supernatant (0.5 mL) was diluted with bidistilled water (0.25 mL) and vortexed. Petrolether (40−60 °C, 0.5 mL) was added. This 3:2 mixture was vortexed and centrifuged at 15 000g for 5 min. The resulting upper (apolar) and lower phases (polar) were investigated separately. Standard Solutions. Stock solutions (2 mg/mL) were prepared for E2, E3, EE2, and BPA with ethanol and for E1 and NP with methanol. CAPE was dissolved in methanol (0.1 mg/ mL). All sample and standard solutions were stored at −20 °C until use. Respective standard solutions were prepared by alcoholic dilution to 50 ng/μL (S1), 2.5 ng/μL (S2), 50 pg/μL (S3), 1 pg/μL (S4), and 0.1 pg/μL (S5). HPTLC Analysis. Standard and sample solutions (0.1−20 μL) were applied as 6.5 mm bands on the HPTLC plate RP18 W using the Automatic TLC Sampler 4 (ATS4; CAMAG, Muttenz, Switzerland). The standard mixture was applied by overspraying of the individual standard solutions on the same start zone, allowing flexibility for individual mixtures and volumes. For recovery, 0.1 μL of P2 sample was sprayed on the plate and oversprayed with 0.1−2 μL CAPE solution. The start zones were dried for 1 min in a stream of cold air using a hair dryer. Development was performed in a Twin Trough Chamber (20 × 10 cm, CAMAG) using n-hexane−toluene− ethyl acetate 4:1.5:1 (V/V/V) up to a migration distance of 70 mm.16 The plate was dried with cold air of a hair dryer for 2 min. The chromatogram was documented at UV 366 nm illumination using the TLC Visualizer (CAMAG). For the different detections, the plate was cut into segments of, e.g., 5 cm × 10 cm using the TLC Plate Cutter (CAMAG). One segment was immersed into the Neu’s reagent followed by dipping in polyethylene glygol (PEG) 4000 solution, using the Chromatogram Immersion Device (CAMAG) at an immersion speed of 3.5 cm/s and an immersion time of 1 s. Plate drying and documentation at UV 366 nm followed. HPTLC-pYES. The plate was automatically immersed into the cell suspension prepared16 using the Chromatogram Immersion Device (CAMAG; speed 3.5 cm/s, immersion time of 5 s). It was horizontally incubated in a plastic box (almost 100% humidified air) at 30 °C for 3 h and dried with cold air using a hair dryer for 2 min. The plate was immersed (parameters as before) into the MUG substrate solution (dissolved in DMSO (20 mg/mL) and added to the reaction buffer (citric acid 6 g/L, disodium hydrogen phosphate 10 g/L, pH 12) to a final concentration of 0.5 mg/mL). After a second incubation period of 60 min at 37 °C, the substrate reaction was terminated by immersion of the plate into a solution of glycine and NaOH (0.1 M, pH 12). The blue MU fluorescence

anoside (MUG) to produce 4-methylumbelliferone (MU), detected at UV 366 nm as blue fluorescence. Consequently, EDCs are detected as a blue fluorescent MU zone.11 The initial application of the liquid yeast estrogen screen (LYES) bioassay on a high-performance thin-layer chromatography (HPTLC) plate resulted in diffuse zones for 17βestradiol (E2, bandwise applied, but not developed) after several hours of incubation with aqueous culture medium.11 The peak width obtained for a nonchromatographed E2 (0.5 ng) zone was already 3 cm.11 It will yet increase during chromatography and impair a reliable assignment of estrogeneffective compounds. So far, method optimization studies for a combination of L-YES with planar chromatography remained unsatisfactory.13−15 In contrast, sharply bounded estrogeneffective compounds without any remarkable diffusion could recently be obtained in our previous study using a fundamentally altered HPTLC-planar (p)YES workflow.16 In a Nature paper in 1946,17 the combination of planar chromatography with bioassay detection was first introduced; however, the challenge of diffuse zones over long incubation times was first solved in 2014.16 This outcome is of general interest for all aqueous bioassay applications used in combination with planar chromatography and will impact the whole field of bioassay applications, making it an attractive tool for medics, pharmacists, biologists, biotechnologists, food technologists, food chemists, and environmental scientists. In this study, the novel workflow was applied to effect-directed profiling of complex food samples leading to positive results. Owed to the peak sharpness, HPTLC-pYES was demonstrated to be suited for quantitation of discovered estrogen-effective unknowns after assignment by HPTLC-MS.

■

EXPERIMENTAL SECTION Chemicals and Materials. Methanol, ethanol, glycerol, nhexane, toluene, ethyl acetate, dimethyl sulfoxide (DMSO), glycine, sodium hydroxide (NaOH), and 4-methylumbelliferylß-D-galactopyranoside (MUG) were obtained from Roth, Karlsruhe, Germany (all of HPLC grade). Bidistilled water was produced in-house (Heraeus Destamat Bi-18E, Thermo Fisher Scientific, Schwerte, Germany). Ammonium formate, agar, copper sulfate, trypan blue, D-glucose, and yeast nitrogen base were bought from Fluka Sigma-Aldrich, Steinheim, Germany. The Saccharomyces cerevisiae BJ3505 strain (protease-deficient, MATα, PEP4: HIS3, Prb1Δ1.6R, HIS3Δ200, lys2-208, trp1Δ101, ura3-52) was generated by McDonnell et al.9,10 Citric acid, disodium hydrogen phosphate, and HPTLC plates RP-18 W, 20 cm × 10 cm, were from Merck Millipore, Darmstadt, Germany. Estrone (E1, 95%), estriol (E3, 95%) (both from Cayman Chemical Company, Ann Arbor, MI, USA), 17ß-estradiol (E2, 98.5%), 17α-ethinylestradiol (EE2, 98%) (both from Dr. Ehrenstorfer, Augsburg, Germany), bisphenol A (BPA, 97%), 4-n-nonylphenol (NP, 98%) (both from Alfa Aesar, Karlsruhe, Germany), and caffeic acid phenethyl ester (CAPE, ≥99%, Roth) were purchased. Parsley and dill (both freeze-dried) were obtained from Ostmann, Dissen, Germany, and cumin and oregano (crushed) were from Fuchs, Dissen Germany. Propolis samples were bought at local European markets: KAPI propolis (30% in ethanol), Pčelarska kuća Kovačević, Belgrad, Serbia (P1), propolis (30% solution), Serbia (P2), KAPI propolis extra (25% in ethanol), Sinefarm, Belgrad, Serbia (P3), propolis Ø (62% in ethanol, w/w), Hansosan, Garbsen, Germany (P4), propolis (drops), allcura Naturheilmittel, Wertheim, Germany (P5), Bakanasan propolis 8290

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295

Analytical Chemistry

Article

Figure 1. Schematic representation of an accelerated LC-bioassay-MS workflow to complement foodomics. Within 5 h, 20 samples can be analyzed in parallel with regard to effective compounds, exemplarily shown for estrogens in complex samples (step 1 is simultaneous development of many samples in parallel by HPTLC, step 2 is bioassay application for effect-directed detection (here: pYES), and step 3 is bioactive substance characterization by MS); physiolocially active compounds were discovered in food (here: reaction with human estrogen receptor down to the ng/L range).

was measured at 366/>400 nm using the mercury lamp of the TLC Scanner 3 (CAMAG). Fluorescence images of the blue MU fluorescence were captured at UV 366 nm (Reprostar 3 documentation system, CAMAG). HPTLC-MS. HPTLC was coupled via the TLC-MS Interface (CAMAG) to a single quadrupole mass spectrometer (CMS, Advion, Harlow, UK). Zone elution was performed with the oval elution head (4 mm × 2 mm) using methanol−ammonium formate buffer (10 mM, pH 4, 98:2, V/V) at a flow rate of 0.1 mL/min. Full scan mass spectra (m/z 100−450) were recorded in the negative ionization mode (ESI−) using a capillary temperature of 250 °C, voltage of 170 V, source voltage offset of 25 V (for fragmentation 35 V), source voltage span of 25, source gas temperature of 250 °C, and an ESI voltage of 3 kV.

Figure 2. Sharp-bounded zones after 4 h of aqueous bioassay application. Separation and bioassay detection of natural estrogens E1, E2, E3, and synthetic estrogen EE2 and xenoestrogen NP on tracks 1− 4 (densitogram depicted for track 1) showing the narrow and sharpbounded zones; the blue MU fluorescence of the zones indicate bioactive EDCs formed upon initial hER binding and expression of the lac-Z gene encoding ß-galactosidase.

■

RESULTS AND DISCUSSION Principle of the Complementary Nontarget Analysis for Foodomics. For effect-directed profiling of complex food samples, a three-step workflow was employed (Figure 1): the simultaneous separation of many samples in parallel (HPTLC separation, step 1), the effect-directed detection of compounds (bioassay application, step 2), and their targeted characterization (mass spectrometry, step 3). This nontarget analysis of complex samples discovered relevant compounds depending on the bioassay selected. Instead of using polar normal phase layers, water-wettable lipophilic plates (HPTLC RP18 W) were used, which allowed wettening of the chromatogram by immersion into the aqueous bioassay suspension. These lipophilic surfaces substantially reduced the zone diffusion, owed to the fact that water had no elution power on such layers. The compounds thus remained in the form of narrow, sharp-bounded zones (Figure 2) obtained after all in all 4 h of aqueous incubation of the chromatogram. For good detectabilities on these layers, the novel workflow was adjusted with regard to enzyme kinetics and the very low pH value of these plates.16 Although the direct link to the effective zone was obtained, the following aspect has to be considered for its further characterization. HPTLC peak capacity is not that of HPLC or ultra (U)HPLC, and separation number is limited to about 30− 40 compounds within a sample track. Thus, compounds coeluting with the bioactive zone might be expected in crude sample extracts and noticed in the mass spectrum. These coelutions can be overcome and separated using an orthogonal, short monolithic column integrated into the transfer line of the

TLC-MS Interface to the MS, together with an adjusted eluent (with regard to the orthogonal column). In order to identify the bioactive compound among the separated peaks of the second dimension, the heart-cut HPTLC-HPLC compounds can be collected separately in sampler vials (with 300 μL inserts), and these microfractions can be subjected to the HPTLC bioassay. Although this is an additional step, it is still a fast protocol and only necessary in case of coelution. This way, extensive, preparative sample preparation steps are not necessary. Comprehensive Information on Estrogen-Effective Compounds on the Food Intake Side. Various types of estrogen-effective compounds were detected in samples so far. Natural EDCs include, for example, E1, E2, and E3. Foodborne EDCs are phytoestrogens such as genistein, 18 coumestrol,19 and ß-sitosterol.20 Xenoestrogens comprise pesticides,21 fungicides,22 preservatives,23 plasticizers like NP and BPA,24−26 or drugs like EE2,19 which can also bind to the hERα or ß and thereby activate or inhibit the endocrine system of mammalians. With the present nontarget analysis, estrogeneffective molecules present in complex food samples were detected as blue fluorescent MU zone and fluorescence measurement was performed at UV 366/>400 nm. Even at high concentrations (zone amounts), clearly separated endocrine disrupting compound (EDC) peaks were obtained in the densitogram (Figure 2, track 1 with highest concentration). Not until now were such narrow, sharp8291

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295

Analytical Chemistry

Article

(Figure 4B). Additionally, the fluorescence measurement was performed at UV 366/>400 nm (Figure S-1, Supporting Information). In the propolis samples, three to six EDC zones were repeatedly detected (Figure 4, bands a−f). The investigated propolis samples consistently showed the estrogen-effective zones (b) and (d), whereas they differed in the zones (a), (c), (e), and (f). For multidetection, the developed HPTLC plate was cut into small segments. For information on phenolic compounds, as some are known EDCs, one plate section was derivatized with Neu’s reagent (Figure S-2B, Supporting Information). The second plate section was subjected to the pYES assay (Figure S2C, Supporting Information). High sample application volumes or analyte concentrations reduced the zone resolution (Figure S-2, Supporting Information, versus Figure 4). In all images, the propolis samples looked overloaded and the clear differentiation of the three unknown EDCs in the upper hRF range (Figure S-2C,D, zones d−f, Supporting Information) was challenging. However, it clearly revealed weak EDC responses in the lower hRF range, like zone a in samples 1−3 (Figure S-2 C, Supporting Information). Two different propolis types with regard to the discovered EDCs were distinguished (Figure S-2 D, Supporting Information). Figures S-2C,D, Supporting Information, showed dark zones with fluorescent edges (e.g., zone d or e), which were obtained in case of a too high fluorescence intensity (response). This effect of fluorescence quenching, known from fluorescence measurement of overloaded HPTLC zones, was caused by reabsorption or formation of dimers. The fluorescence quenching of the inner darkened zone part was considerably reduced at lower concentrations or reduced sample application volumes (Figure 4). Negative Control. Each discovered EDC zone showed the same characteristic light blue fluorescent hue of the MU substrate. However, in complex mixtures like raw food extracts, coeluting natively blue fluorescent compounds could look similar. Thus, the characteristic hue has to be inspected carefully during evaluation. In the food examples given, the EDC profiling differed only partially to natively blue fluorescent compounds (e.g., it differed for bands a and f in Figure 4A versus Figure 4B). As a negative control, the pYES was performed without yeast cells in the medium for the different propolis samples P1−P4 (other samples were comparable in the EDC profile). Without the yeast cells, there was no blue fluorescence, i.e., EDC zone, detectable (Figure S-3B versus S3A, Supporting Information). This proved that the characteristic blue MU fluorescence was formed upon initial hER binding of an EDC and expression of the lac-Z gene encoding ß-galactosidase, which cleaves MU from MUG. Optionally, substrates other than MUG reacting to a visual color, e.g.,

bounded zones (only 1−3 mm broad) achieved in direct bioautography. In our previous methodological optimization study,16 the limit of detection (LOD) was demonstrated to reach 0.5 pg/zone for E2, obtained by medium-working yeast cells. Such good LODs allow the direct detection of ECDs at a physiologically active concentration level in complex food samples. Without enrichment steps, normally used for sample preparation in trace analysis, the ng/kg-range can directly be reached for rectangular applications27−32 of up to a 500 μL sample volume, which is the focus of another study. It could be shown that the working range covered 2 to 3 decades16 which is advantageous for screening of food samples. For the proof-ofprinciple, commercially available propolis and spice samples were investigated after a minimal sample preparation. Propolis samples contained up to six different EDCs, and in the polar phase extract of parsley, oregano, dill, and cumin, two EDCs were discovered (Figure 3, marked with ∗). In the following section, the results obtained are discussed in detail.

Figure 3. Unknown estrogens discovered in food, exemplarily shown for propolis and spice extracts. HPTLC-pYES chromatograms at UV 366 nm revealed EDC zones with a characteristic blue hue (marked with a ∗); the blue MU fluorescence of each EDC zone is formed upon initial hER binding and expression of the lac-Z gene encoding ßgalactosidase; applied tracks: (1, 2) propolis extracts (0.3 μL; two different profiles are exemplarily depicted) as well as spice extracts of (3) parsley (20 μL), (4) cumin (20 μL), (5) oregano (5 μL), and (6) dill (5 μL).

Estrogenic Properties of Propolis Samples. Commercially available propolis fluids were diluted and investigated directly, without further sample preparation. Capsule and tablet formulations of propolis were extracted with ethanol. All propolis samples were applied 2-fold on the HPTLC plate, separated, and documented at UV 366 nm (Figure 4A). The developed HPTLC plate was subjected to the direct bioautography (pYES) and again documented at UV 366 nm

Figure 4. Chromatograms at UV 366 nm of commercially available propolis samples. Samples P1−P4 applied at low concentrations (0.3 μL, applied 2-fold) (A) after chromatography (native fluorescence) and (B) after direct bioautography (HPTLC-pYES) showing effective zones a−f. 8292

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295

Analytical Chemistry

Article

regions. For the first time, reliable and verified quantitative results were obtained for HPTLC-direct bioautography. This outcome proved the capabilities for quantitation of the substantially improved HPTLC-pYES assay.16 Confirmation of Unknown EDCs by HPTLC-MS. Exemplarily, the unknown effective zone in the propolis sample, preliminarily assigned as CAPE based on chromatographic and spectral characteristics, was verified through HPTLC-ESI-MS via the elution head-based TLC-MS Interface (Figure S-7, Supporting Information). Therefore, the sample was applied twice (one section for bioassay, second for MS) and analyzed. In the negative ionization mode, the CAPE standard zone showed the deprotonated molecule at m/z 283 [M − H]− (Figure 5A) with a specific fragmentation pattern at m/z 179, 161, and 135 for higher voltages applied (Figure 5B). The fragmentation pattern was investigated because highresolution MS (HRMS) was not available. The fragments obtained were according to the literature (Figure S-8, Supporting Information).33 The unknown EDC (Figure 4, band b) in the propolis sample P2 showed also a mass signal at m/z 283, confirming the deprotonated CAPE molecule [M − H]− in the sample zone (Figure 5C). The obtained fragmentation pattern of the unknown EDC in the propolis sample P2 (Figure 5D) was comparable to the CAPE standard solution. This confirmation of the unknown EDC zone to be CAPE was repeated three times by HPTLC-MS, as it takes less than 1 min for the complete elution of a zone. A further mass signal was observed at m/z 269 [M* − H]−, which fit to the deprotonated molecule of benzyl caffeate (Figure S-8, Supporting Information). This mass signal was assigned to the adjacent band above zone b (Figure 4). Thus, it appeared in the CAPE mass spectrum as the positioning of the elution head of 2 mm × 4 mm dimension was performed manually. An automated TLC-MS Interface taking the peak coordinates would ease and improve the positioning.12 Although the proofof-principle through fragmentation was successful using a single-quadrupole MS, HRMS is conditio sine qua non for the routine workflow. Further phytoestrogens, i.e., apigenin, chrysin, quercetin, and luteolin, were reported to be present in propolis samples.33−35 Single standard solutions of these phytoestrogens (0.01% in methanol) were applied (2 μL) and analyzed by HPTLC-pYES. All showed an estrogenic effect (not depicted; the individual estradiol equivalents are the focus of further investigations). However, these phytoestrogens, detected as blue fluorescent EDC zones, were not at the same hRF value as those blue EDCs in the propolis samples analyzed (Figure 4). Hence, these phytoestrogens were not responsible for the intense estrogenic effect of propolis, although these phytoestrogens might be present in the sample at concentrations below the LOD of the applied HPTLC-pYES assay. Except for CAPE, the other blue unknown ECDs (Figure 4, bands a and c−f; especially band d showed an intense signal) were still unclear and the recording of HRMS spectra would be helpful but was not available. Another important aspect is the different hER receptor affinity of these phytoestrogens. Most of the phytoestrogens have a higher affinity to hERß. The affinity to this receptor was by a factor of 5−30 higher if compared to hERα,36 which was used in this study. Hence, future studies integrating the hERß in comparison to hERα would be of interest for food of plant origin. For the Aliivibrio f ischeri bioassay, the recording of mass spectra directly from the bioactive zone (inclusive of the

chlorophenolred-ß-D-galactopyranoside, can be employed for the pYES assay. Estrogenic Properties of Spice Samples. Aside from propolis samples, commercially available spices were screened for EDCs and discovered in parsley, oregano, dill and cumin. Again, the EDC zones detected by HPTLC-pYES (Figure S-4B, Supporting Information) had to be compared with the natively blue fluorescent zones (Figure S-4A, Supporting Information), both documented at UV 366 nm. The discovered EDC zones had a characteristic light blue fluorescent hue and their fluorescence intensity increased, if compared to respective natively blue fluorescent zones. Hence, this HPTLC image comparison between native fluorescence versus pYES fluorescence and a negative control (as for Figure S-3, Supporting Information) is crucial for correct interpretation of the results obtained. Oversprayed and Overlapped Application to Discover Matrix Influence on hRF. HPTLC offers the possibility to discover any retention (hRF value) shift caused by a matrix in complex raw extracts. Sample and standard solutions can be applied in an oversprayed or overlapped mode. Thus, the standard components migrate completely or partially in the sample matrix, and any matrix effects can be recognized. Exemplarily, this was shown for the propolis sample P2 (Figure S-5, Supporting Information). CAPE migrated in the sample by overspraying (Figure S-5B, track 1 + 2, Supporting Information) or migrated partially by overlapping (Figure S5D, track 2, Supporting Information). A matrix influence on the hRF value was not observed. The unknown EDC zone in the propolis sample (track 1) was confirmed to be CAPE with regard to its chromatographic property. For further confirmation, mass spectra were recorded. Quantitation of CAPE in Propolis Samples. The accuracy of HPTLC-pYES was investigated via the standard addition method. CAPE standard zones between 10 and 150 ng/band were oversprayed with propolis samples. The sample tracks spiked with CAPE were developed and detected by the pYES assay. The mean recovery of CAPE determined twice over 7 different concentrations ranged from 10 to 150 ng/band was 95.4% (%RSD 15.2%; n = 7). A five point calibration was performed (Figure S-6, Supporting Information) to determine the amount of CAPE in 7 commercially available propolis samples. The calculated CAPE content ranged from 710 to 2387 μg/g, referred to the propolis weight (Table 1, n = 2). Our quantitative HPTLC-pYES results (detected by microorganisms) were in accordance with the results obtained by HPLC-MS/MS,33 for which the CAPE content ranged from 1112 to 2525 μg/g, for propolis from different geographical Table 1. Quantitation of CAPE Found in 7 Propolis Samples by HPTLC-pYES propolis sample P1 P2 P3 P4 P5 P6 P7 a

(30%) (30%) (25%) (62%) (not specified) (250 mg/capsule) (30 mg/lozenge)

CAPE content in sample [μg/mL]

CAPE content [μg/g] referred to propolis weight (n = 2)

481 476 471 348 380 359a 22b

2028 2009 2387 710 380c 1435 1089

μg/capsule. bμg/pastille. cμg/mL. 8293

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295

Analytical Chemistry

Article

into the waste. This will take ca. 10 min instead of less than 1 min per elution; however, the same zone can be used for the bioassay detection and for HRMS.

■

CONCLUSIONS HPTLC exists in the shade of the powerful HPLC; however, it has been proven that hyphenated HPTLC deserves its place in modern analytical laboratories. The wealth of information gained from using this technique has been key to progress on many of our recent projects. This novel workflow for effectdirected food profiling is nontargeted at the beginning (detecting all effective compounds in a complex sample depending on the bioassay selected) and then highly targeted, when only the effective compounds are further characterized by HRMS, derivatization reagents, and spectroscopic techniques (NMR, FT-IR, and SERS). All these hyphenations were performed at the analytical scale, directly from the zone of interest on the HPTLC plate.27 Thus, this streamlined food profiling would be suited as a complementary tool for foodomics. It can contribute due to its nontargeted specificity (effect-directed detection), good detectability (down to fg/zone levels), and reduced sample preparation (taking raw sample extracts), as demonstrated for the HPTLC-pYES analysis of EDCs in spices and propolis. Due to the parallel chromatography, up to 20 samples can simultaneously be analyzed within 5 h (calculated as 15 min/sample). If compared to the sum parameter obtained by Petri dish, cuvette, or microtiter plate assays, this strategy allows the direct link to single effective compounds in complex mixtures at low detectabilities. All compounds showing the effect are individually targeted in a complex sample, such as foodborne noxious matters, (process) contaminants, residues, degradation products, and metabolites. However, it was also pointed out that HPTLC image comparison between native fluorescence versus pYES fluorescence and a negative control was crucial for correct interpretation of the results obtained. Effect-directed food profiling can complement the scientific understanding with regard to foodborne pathogenesis and development of functional food. This powerful profiling will help to solve some of the new challenges that modern food safety and quality control have to face.33 Finally, also drug discovery or analysis of traditional medicines may profit from this effect-directed profiling of complex samples.

■

ASSOCIATED CONTENT

S Supporting Information *

Figures S-1−S-8. This material is available free of charge via the Internet at http://pubs.acs.org.

■

Figure 5. HPTLC-ESI‑-MS of an unknown ECD. Confirmation of the unknown’s identity to be CAPE: mass spectra of a CAPE standard zone eluted via the elution head-based TLC-MS Interface into the ESIMS (A) without and (B) with fragmentation revealed the same daughter ions as the unknown ECD zone in sample P2 (C) without and (D) with fragmentation.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel. +49-(0)641-99-39141. Fax +49-(0)641-99-39149. Author Contributions

bioassay medium) was already demonstrated.27 For HPTLCpYES, this is focus of another study. A useful tool is the integration of an additional short monolithic column and a valve in the TLC-MS Interface outlet line to the MS, allowing heart-cut HPTLC/pYES-HPLC-HRMS of discovered estrogeneffective zones27 and the direction of the eluted polar medium

†

G.E.M. and I.K. contributed equally to this work. Both authors discussed the outline, experiments, and results. G.E.M. wrote the manuscript. I.K. performed the experiments. Notes

The authors declare no competing financial interest. 8294

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295

Analytical Chemistry

■

Article

(30) Vega-Herrera, M. A.; Morlock, G. E. J. Planar Chromatogr. 2007, 20, 411−417. (31) Morlock, G. E.; Prabha, J. J. Agric. Food Chem. 2007, 55, 7217− 7223. (32) Morlock, G. E.; Morlock, L. P.; Lemo, C. J. Chromatogr., A 2014, 1324, 215−223. (33) Medana, C.; Carbone, F.; Aigotti, R.; Appendino, G.; Baiocchi, C. Phytochem. Anal. 2008, 19, 32−39. (34) Falcão, S. I.; Vale, N.; Gomes, P.; Domingues, M. R. M.; Freire, C.; Cardosoe, S. M.; Vilas-Boasa, M. Phytochem. Anal. 2013, 24, 309− 318. (35) Chernetsova, E. S.; Bromirski, M.; Scheibner, O.; Morlock, G. E. Anal. Bioanal. Chem. 2012, 403, 2859−2867. (36) Kuiper, G. G. J. M.; Lemmen, J. G.; Carlsson, B.; Corton, J. C.; Safe, S. H.; van der Saag, P. T.; van der Burg, B.; Gustafsson, J.-Å. Endocrinology 1998, 139 (10), 4252−4263.

ACKNOWLEDGMENTS Thanks is owed to M. Schulz and M. Oberle, Merck Millipore, Darmstadt, Germany, for supplying the plates and W. Schwack, University of Hohenheim, Stuttgart, Germany, for delivery of the McDonnell yeast cells.

■

REFERENCES

(1) Medina, S.; Domínguez-Perles, R.; Ferreres, F.; Tomás-Barberán, F. A.; Gil-Izquierdo, Á . TrAC 2013, 52, 88−99. (2) Hu, C.; Xu, G. TrAC 2013, 52, 36−46. (3) Reemtsma, T. J. Chromatogr., A 2009, 1216, 3687−3701. (4) Giacometti, J.; Josic, D. TrAC, Trends Anal. Chem. 2013, 52, 16− 22. (5) Wright, C. J. Chromatogr., A 2009, 1216, 316−319. (6) Alder, L.; Greulich, K.; Kempe, G.; Vieth, B. Mass Spectrom. Rev. 2006, 25 (6), 838−865. (7) Bousova, K.; Senyuva, H.; Mittendorf, K. J. Chromatogr., A 2013, 1274, 19−27. (8) Reemtsma, T.; Alder, L. J. Chromatogr., A 2013, 1271, 95−104. (9) McDonnell, D. P.; Nawaz, Z.; Densmore, C.; Weigel, N. L.; Pham, T. A.; Clark, J. H.; O’Malley, B. W. J. Steroid Biochem. Mol. Biol. 1991, 39 (3), 291−297. (10) McDonnell, D. P.; Nawaz, Z.; O’Malley, B. W. Mol. Cell. Biol. 1991, 11 (9), 4350−4355. (11) Mueller, M. B.; Dausend, C.; Weins, C.; Frimmel, F. H. Chromatographia 2004, 60 (3/4), 207−211. (12) Morlock, G.; Schwack, W. TrAC, Trends Anal. Chem. 2010, 29 (10), 1157−1171. (13) Buchinger, S.; Spira, D.; Broeder, K.; Schluesener, M.; Ternes, T.; Reifferscheid, G. Anal. Chem. 2013, 85 (15), 7248−7256. (14) Spira, D.; Reifferscheid, G.; Buchinger, S. J. Planar Chromatogr. 2013, 26 (5), 395−401. (15) Schönborn, A.; Andrea, G. A. J. Planar Chromatogr. 2013, 26 (5), 402−408. (16) Klingelhöfer, I.; Morlock, G. E. J. Chromatogr., A, DOI: 10.1016/ j.chroma.2014.07.083. (17) Goodall, R. R.; Levi, A. A. Nature 1946, 158, 675−676. (18) Nishihara, T.; Nishikawa, J.; Kanayama, T.; Dakeyama, F.; Saito, K.; Imagawa, M.; Takatori, S.; Kitagawa, Y.; Hori, S.; Utsumi, H. J. Health Sci. 2000, 46 (4), 282−298. (19) Soto, A. M.; Sonnenschein, C.; Chung, K. L.; Fernandez, M. F.; Olea, N.; Serrano, F. O. Environ. Health Perspect. 1995, 103 (Suppl7), 113−122. (20) Breithofer, A.; Graumann, K.; Scicchitano, M. S.; Karathanasis, S. K.; Butt, T. R.; Jungbauer, A. J. Steroid Biochem. Mol. Biol. 1998, 67 (5−6), 421−429. (21) Vinggaard, A. M.; Breinholt, V.; Larsen, J. C. Food Addit. Contam. 1999, 16 (12), 533−542. (22) Boehmler, G.; Kohnen, R.; Borowski, U.; Ruehe, A. J. Verbraucherschutz Lebensmittelsicherh. 2006, 1 (4), 325−331. (23) Lemini, C.; Jaimez, R.; Avila, M. E.; Franco, Y.; Larrea, F.; Lemus, A. E. Toxicol. Ind. Health 2003, 19 (2−6), 69−79. (24) Gutendorf, B.; Westendorf, J. Toxicology 2001, 166 (1−2), 79− 89. (25) Folmar, L. C.; Hemmer, M. J.; Denslow, N. D.; Kroll, K.; Chen, J.; Cheek, A.; Richman, H.; Meredith, H.; Grau, E. G. Aquat. Toxicol. 2002, 60 (1−2), 101−110. (26) Fang, H.; Tong, W.; Branham, W. S.; Moland, C. L.; Dial, S. L.; Hong, H.; Xie, Q.; Perkins, R.; Owens, W.; Sheehan, D. M. Chem. Res. Toxicol. 2003, 16, 1338−1358. (27) Morlock, G. In Instrumental Methods for the Analysis of Bioactive Molecules; 246th ACS National Meeting and Exposition, Indianapolis, Indiana, September 8−12, 2013; Patil, B. S., Jayaprakasha, G. K., Pellati, F., Eds.; American Chemical Society: Washington, DC, 2013. (28) Morlock, G. E.; Schuele, S.; Grashorn, S. J. Chromatogr., A 2011, 1218, 2745−2753. (29) Hošt’álková, A.; Klingelhöfer, I.; Morlock, G. E. Anal. Bioanal. Chem. 2013, 405, 9207−9218. 8295

dx.doi.org/10.1021/ac501723j | Anal. Chem. 2014, 86, 8289−8295