An Isomer-Specific Approach to Endocrine-Disrupting Nonylphenol in

Article pubs.acs.org/JAFC

An Isomer-Specific Approach to Endocrine-Disrupting Nonylphenol in Infant Food Klaus Günther,*,†,‡ Torsten Rac̈ ker,‡ and Roswitha Böhme† †

Institute of Nutritional and Food Sciences, Food Chemistry, Rheinische Friedrich-Wilhelms-Universität Bonn, Endenicher Allee 11-13, D-53115 Bonn, Germany ‡ Research Centre Jülich, Institute of Bio- and Geosciences (IBG-2), D-52425 Jülich, Germany S Supporting Information *

ABSTRACT: Nonylphenols (NPs) are persistent endocrine disruptors that are priority hazardous substances of the European Union Water Framework Directive. Their presence in the environment has caused growing concern regarding their impact on human health. Recent studies have shown that nonylphenol is ubiquitous in commercially available foodstuffs and is also present in human blood. The isomer distribution of 4-nonylphenol was analyzed by gas chromatography − mass spectrometry in 44 samples of infant food. Our study shows that the distribution of nonylphenol isomers is dependent on the foodstuff analyzed. Although some isomer groups prevail, different distributions are frequent. Variations are even found in the same food group. Nonylphenol is a complex mixture of isomers, and the estrogenic potentials of each of these isomers are very different. Consequently, to determine the potential toxicological impact of NP in food, an isomer-specific approach is necessary. KEYWORDS: nonylphenol, endocrine disruptors, isomer-specific, infant food, isomer distribution

■

INTRODUCTION Nonylphenol polyethoxylates (NPEs) are nonionic surfactants that are widely used in detergents, paints, pesticides, and herbicides and as emulsifying, dispersing, and flotation agents. The starting compound for NPEs is nonylphenol (NP, 4-NP), a compound commonly used as an antioxidant [as tris-nonylphenylphosphite (TNPP)]1 or as an antifogging agent (as nonylphenol tetraethoxylate) in food-packaging polymers. In 1984, it was confirmed that degradation processes of NPEs in the environment lead to extraordinarily high concentrations of nonylphenol in sewage sludge.2 Subsequent studies revealed that nonylphenol is persistent, toxic, and estrogenically active3 and ubiquitous in many environmentally relevant matrices,4−6 including food,7−13 human tissue samples,14 and blood.12,15,16 Although nonylphenol has been identified as a priority hazardous substance of the European Union Water Framework Directive17 and the use of NPEs in Europe has been restricted, the import of textiles from non-European countries (e.g., from Asia) is a major source of nonylphenol and nonylphenol ethoxylates in European wastewater treatment plants.18 Endocrine-disrupting effects of nonylphenol include harmful impacts on reproductive,19 immune,20 and nervous systems.21 Young animals and humans are especially sensitive to endocrine disruptors in their development phases.22 Food is a major contamination source of nonylphenol for humans. As technical syntheses are based on nonene obtained from trimerization of propene, nonylphenol consists of a very complex mixture of isomers with differently branched nonyl groups, and the isomer composition varies depending on the manufacturer.4,23,24 Furthermore, in the environment, isomerspecific degradation alters the isomer mixture again.25−27 This results in a very complex contamination problem concerning the endocrine-disrupting potential of nonylphenol. Theoretically, there are 211 possible constitutional isomers, and © 2017 American Chemical Society

approximately 100 have been observed in environmentally relevant matrices.24 The toxicological assessment of nonylphenol contamination is, therefore, a major challenge because the estrogenic activity of the isomers is heavily dependent on facets of the structure of the nonyl side chain, such as the degree of branching and bulkiness.25,28−30 Hence, it is necessary to consider the nonylphenol problem from an isomer-specific viewpoint, including in food and health studies. Until now, only total nonylphenol concentrations were analyzed in foodstuffs. In this study, the distribution of NP isomers in commercially available foodstuffs for infants was evaluated. NP isomers were extracted from the foodstuffs using a combination of steam distillation, liquid/liquid extraction, and high-performance liquid chromatography. After derivatization, the extracts were analyzed by gas chromatography and mass spectrometry (GC− MS). The developed method is highly selective and sensitive for nonylphenol isomers using the elution times of the NP isomers and the diagnostic mass-to-charge ratios of the resonance-stabilized benzylcarbenium ions differently substituted in the α-position depending on the isomer structure. This developed advanced analytical procedure could be used for the isomer-specific analysis of nonylphenol in the ultratrace range in the different very matrix-variable foodstuffs of a complete food basket. The important question of our investigation was whether the nonylphenol isomer distribution in foodstuffs is equal or different. The answer to this question is of far-reaching significance. If the distribution is equal, it will be sufficient to Received: Revised: Accepted: Published: 1247

November 2, 2016 January 11, 2017 January 15, 2017 February 3, 2017 DOI: 10.1021/acs.jafc.6b04916 J. Agric. Food Chem. 2017, 65, 1247−1254

Article

Journal of Agricultural and Food Chemistry Table 1. Categories and Samples of Baby Food and Toddler Food infant food

category

samples

baby food

breast milk infant formula follow-on formula complete meal/meat vegetable pap

3rd/4th and 19th/20th week of breastfeeding hypoallergenic infant formulas 1 and 2, infant formula 3 hypoallergenic follow-on formulas 1 and 2, follow-on formula 3 beef with rice and potatoes; tomatoes, spaghetti, and pork; ham noodles with vegetables; broccoli, potatoes and turkey; rice and creamy vegetables with chicken early carrots A and B oat flakes chocolate semolina pudding, mixed grain pap, vanilla semolina pudding, biscuit milk pudding, milk pudding with semolina, banana milk pudding tropical fruits in apple juice, peaches with honey, baby apples peach passion fruit with yogurt, apples and bananas with curd rice with apples, apple cereal pap zwieback mild apple juice baby tea, balm tea with grape juice cornflakes, cooked potatoes, carrots, bananas whole milk (UHT), eggs, chicken/turkey bologna, rock salmon butter, refined sugar tap water

baby food beikost commercial infant food (CIF)

vegetarian menu cereals (non-milk) milk pudding (ready-to-eat)

toddler food other toddler food

fruit purée fruit purée with dairy products cereal fruit purée biscuits/zwieback (rusk) juice tea vegetable origin animal origin fats/oils, confectionery beverages

CB (Chrompack, Frankfurt, Germany); column B, FS-OV-1-CB (Chromatographie Service). Column B was also used for the GC-FID analysis of the different NP isomers for the technical 4-NP standard using an HP 5890 Series II gas chromatograph (Hewlett-Packard) with a flame ionization detector (Agilent Technologies, Palo Alto, CA). Helium was used as the GC carrier gas (column head pressure of 19 psi, gas flow velocity of 45 cm/s at 70 °C). Rapid large-volume injections for GC−MS were performed with a CTC A200 SE autosampler (CTC Analytics, Zwingen, Switzerland) into a programmable temperature vaporization injection system (Optic 2-200, ATAS Cambridge, Cambridge, U.K.) with a wide-bore liner packed with silanized diatomite (ATAS Port 1, 3.5 mm inside diameter). Food Samples. The examined foodstuffs included baby food (up to 1 year of age) and food for infants from 1 to 3 years of age (toddlers). Baby food samples included breast milk, infant formula, follow-on formula, and mainly commercial infant food (CIF) in jars. Samples of toddler food were normal foodstuffs that were mostly of vegetable or animal origin. The different food samples are listed in Table 1. Infant formulas, follow-on formulas, cornflakes, zwieback, and bologna were packed in plastic film bags. Two 35-year-old women provided the breast milk samples. Food Sample Preparation and Extraction. The examined foodstuffs for infants, purchased from German supermarkets, were stored at room temperature or 4 °C until they were opened for analysis. While liquid, pastelike, or powdered homogeneous food could be extracted without pretreatment, solid foodstuffs had to be homogenized immediately before being extracted using a blender or an Ultra-Turrax. To obtain regular servings of eggs, carrots, and potatoes, these were peeled and cooked. The analytical procedure, which can be applied to all kinds of food matrices, was as follows: sample preparation and spiking with an internal standard → concurrent steam distillation and solvent extraction → HPLC cleanup with fluorescence detection → derivatization → GC−MS and separation and quantitation of nonylphenol isomers. The extraction of nonylphenol isomers from the food samples was performed by using a Veith-Kiwus steam distillation apparatus for concurrent steam distillation and solvent extraction with a 2 L distilling flask at the bottom and a reflux condenser at the top. The extraction unit was thoroughly cleaned before each steam distillation and extraction by successive refluxing with nitric acid (and removing the acid with water), methanol, and hexane. The extraction solvent, a solution of sodium chloride (40 g) in water (600 mL) with hydrochloric acid (2 mL), was purified by boiling and extracting with “organic-phase” cyclohexane/isooctane (20 mL, 1:1, v/v) for 5 h

analyze the total concentration of all the isomers in the future for toxicological assessment and only the development of fast overall analysis methods for official food control will be necessary. If the distribution is different, however, it will be necessary to develop high-resolution techniques for the determination and structural elucidation of all single isomers in typical food baskets of different countries. Then, the crucial isomers should be synthesized, and by using different in vitro and in vivo test systems, the estrogenic effect should be determined and normalized to the most potent isomer. Consequently, a system of estrogenic equivalency factors (EEFs) could be designed in a manner similar to that of the toxic equivalency factors (TEFs) in the case of dioxins or dioxin-like polychlorinated biphenyls.31

■

MATERIALS AND METHODS

Chemicals. The following nonylphenols (NPs and their sources) were used: 4-n-nonylphenol (4-n-NP) (Dr. Ehrensdorfer, Augsburg, Germany), technical nonylphenol 85% (CAS Registry No. 84852-153) (Fluka, Buchs, Switzerland), and 4-(1-methyloctyl)-phenol (4-NP2) (K. H. Dötz, University of Bonn, Bonn, Germany). Stock solutions of these NPs prepared in cyclohexane or methanol (internal standard 4n-NP) were stored at 4 °C. N-Methyl-N-tert-butyldimethylsilyltrifluoroacetamide (MTBSTFA) was from Chromatographie Service (Langerwehe, Germany). Instruments. For grinding and homogenizing solid food samples, a model 38BL41 blender (Waring) or an Ultra-Turrax18/10 blender (IKA-Werk, Staufen, Germany) was used. The high-performance liquid chromatography (HPLC) system from Merck-Hitachi (Darmstadt, Germany) contained an L-6200 gradient pump, a T-6300 column thermostat, a D-6000 data interface, a Rheodyne syringe loading sample injector (model 7125, Cotati), and an F-1080 fluorescence detector. The column used for normal-phase HPLC was a Hypersil aminopropylsilica (APS) column [125 mm × 4 mm (inside diameter)], with a particle size of 3 μm (Chromatographie Service). The silylated samples were analyzed using a Finnigan MAT GCQ GC/MS/MS system (Thermo Finnigan, San Jose, CA). The precolumn was deactivated fused silica [10 m × 0.25 mm (inside diameter)] (J&W, Folsom, CA). For the separation of nonylphenol isomers, two different capillary columns (both 30 m × 0.25 mm, 0.25 μm film thickness) were used: column A, WCOT Fused Silica CP-Sil-8 1248

DOI: 10.1021/acs.jafc.6b04916 J. Agric. Food Chem. 2017, 65, 1247−1254

Article

Journal of Agricultural and Food Chemistry in the steam distillation apparatus. The organic layer was then replaced with the same amount of fresh cyclohexane/isooctane mixture, and the food sample (10−250 g) was suspended in the aqueous extraction solvent. For liquid samples, 525 g (mild apple juice) and 1000 g (tap water as a sample and as part of the extraction solvent) were used. After 20 μL of 4-n-NP in methanol (0.04 mg/L) had been added as the internal standard, distillation and extraction were performed for 5 h. Both the organic layer and the aqueous layer distilled into the VeithKiwus apparatus were collected in a volumetric flask and separated. After being dried with Na2SO4, the organic solution was evaporated almost to dryness using a nitrogen stream. The sample residue was redissolved in cyclohexane (500 μL) and stored in the dark at 4 °C until it was analyzed (duplicate analysis). Before each food sample extraction, procedural blanks were run. Cleanup by HPLC. Co-extracted, interfering organic compounds were removed in a cleanup step by normal-phase HPLC (column, aminopropylsilica; for solvents and the elution program, see below). Detector wavelengths were as follows: λexcitation = 259 nm, and λemission = 300 nm. For determination of the time window in which 4-NP and the internal standard(s) eluted (peak maximum ± 1 min), a 4-n-NP standard solution (0.01 mg/L in cyclohexane) was injected before the sample solution. Aliquots of the sample solution (100 μL) were injected two times using a 100 μL sample loop. The fraction containing 4-NP eluted at a retention time of approximately 12.5 min. After the fractions had been collected with the separated 4-NPs and the internal standard (6 mL), the eluent was evaporated in a nitrogen stream and the residue stored at 4 °C. The elution program for HPLC cleanup was as follows: eluent A, hexane; eluent B, hexane and 2-propanol (80:20, v/v). The linear elution program started with 97% A and 3% B (1 min), then changed to 75% A and 25% B (within 10 min at a flow rate of 1.0 mL/min), and then changed to 50% A and 50% B (within 2 min at a flow rate of 2 mL/min). The system was returned to the initial composition over a 10 min period and equilibrated for 1 min at the initial flow rate. GC−MS Analysis. First, the evaporated HPLC fractions were derivatized by adding MTBSTFA in acetonitrile (1:100, v/v, 400 μL) and instrument standard 4-NP2 (5 μL, at a concentration of 0.41 mg/ L) and transferring the solutions to 500 μL crimp top vials. After their contents had been mixed, the closed vials were heated at 60 °C for 10 min to complete the derivatization. The silylated compounds were separated by GC−MS using the following temperature program: 70 °C for 3 min, a temperature ramp of 20 °C/min to 165 °C, to 250 °C increasing at a rate of 4 °C/min, to 280 °C at a rate of 40 °C/min, and held at 280 °C for 5 min. The interface temperature was set to 280 °C, and the ion source temperature was 180 °C. The detector was operated in electron impact (EI) ionization mode and full scan mode (m/z 100−340). An autosampler was used to perform rapid largevolume injections (100 μL) of the derivatized samples. Vaporization of the solvent via the split exit of the injector followed (vent rate of 200 mL/min, total vent time of 60 s, at a constant temperature of 70 °C). Following vaporization of the solvent, the temperature of the injector was increased from 70 to 275 °C (33 min) at a rate of 5 °C/s (splitless time of 2 min). Calibrations, Response Factors, and Quantitation. A total of 390 μL of derivatization solution (MTBSTFA) was added to 5 μL of a solution containing technical 4-NP (four different concentrations for calibration, 0.85, 1.06, 5.30, and 10.59 mg/mL), 4-n-NP (0.465 mg/ mL), and 5 μL of a 4-NP2 solution (0.41 mg/mL) to yield silylated standards for the calibration of the GC−MS system. These were used to generate approximate response factors f i and fo for all 4nonylphenol isomer peaks in relation to the internal standard 4-nNP. To calculate the relative response factors (RRFs) f i, the unknown amounts of every single 4-NP isomer peak i had to be determined by GC−MS analysis of the standard solutions. The RRFs f i and fo were used to quantitate the 4-NP isomer peaks in the derivatized samples (internal standard method). Each HPLC fraction was injected twice, and the average result of both fractions was used. To quantitate the 4NP isomers, areas in reconstructed ion current (RIC) chromatograms were integrated for peaks having the same retention times as the calibration standard as well as concurrent mass spectra.

The mass spectrometric fragmentation of nonylphenols leads preferentially to the benzylic cleavage forming a stabilized tropylium ion derivative, a seven-membered ring. Thereby, alkyl groups located at the α-carbon atom can be cleaved, depending on the substitution pattern. The following masses were used for the RIC (reconstructed ion current) of the silylated nonylphenol isomers and/or groups (see below): m/z 277 ([M − C4H9]+) for isomers 1, 6a, and 9, m/z 263 ([M − C5H11]+) for isomers 2, 4, 6b, 7, and 11, and m/z 249 ([M − C6H13]+) for isomers 3, 5, 8, and 10. As there was no separation of isomer 6 in the TIC (total ion current), it was measured as the sum of both RICs. Factors f i and fo were used to calculate the 4-NP concentrations from the sum of the integrated areas in relation to the internal standard. From this total amount (sample), the previously determined total blank value was subtracted (6 ± 3.5 ng of 4-NP). Duplicate analysis was performed, and the concentrations were calculated in relation to the fresh weight of the sample (unless otherwise stated). Determination of the Isomer Distribution of 4-NP. To determine the nonylphenol isomer distribution, a derivatized solution of 4-NP (technical mixture) and internal standard 4-n-NP was measured by a GC-FID system. As the signals produced by a FID are proportional to the carbon content, the same amounts of portions of all the isomers should lead to the same signals. The internal standard 4-n-NP, a pure single compound not present in the technical 4-NP mixture, could be separated well. Thus, the area values of the single isomers could be converted into concentrations using eq 1 and subsequently set into proportion among themselves. c(isomer n) =

c(IS)A(isomer n) A(IS)

(1)

where c(isomer n) is the concentration of the corresponding 4-NPisomer (milligrams per liter), c(IS) is the concentration of internal standard 4-n-NP (milligrams per liter), A(isomer n) is the area of 4-NP peak n in area units (au) determined by GC-FID, and A(IS) is the area of the internal standard peak 4-n-NP in area units (au) determined by GC-FID. As this analytical method allows the separation of nonylphenol into only 11 peaks, there might be several co-eluting isomers with very similar structures. Therefore, we use the designation “isomer groups” (see also Figures 1−6). Although some structures of nonylphenol isomers from technical mixtures could be identified23,32 and more isomers have been separated from environmental matrices,26 for the unambiguous identification of the single nonylphenol structures in food, better separation techniques (e.g., two-dimensional chromatography) and a comparison with the synthesized nonylphenol standards would be necessary. Syntheses of such substances have been performed by several groups.32−34

■

RESULTS AND DISCUSSION Recoveries and Limits of Detection. For the validation of the complete analytical procedure, every food sample (duplicate extraction) and the appropriate procedural blank were spiked with internal standard 4-n-NP, whose recoveries were determined by GC−MS in relation to the instrument standard 4-NP2. For the procedural blanks, there were efficient recoveries of 90−100% with up to 5% relative standard deviations (RSDs), while the different food samples showed high variations with recoveries of 16−127% and RSDs of 1− 35%. Low recoveries (45% of isomer group 4. The sum of the percentage of isomer groups 1 and 4 for potatoes is >60%. Cornflakes stand out with approximately 25% of isomer group 10, and the corresponding sum of isomer groups 3, 6, and 10 amounts to 1250

DOI: 10.1021/acs.jafc.6b04916 J. Agric. Food Chem. 2017, 65, 1247−1254

Article

Journal of Agricultural and Food Chemistry

Figure 3. Further examples for commercial infant food with a different distribution of 4-NP isomers.

Figure 4. Distribution of 4-NP isomers in commercial infant food identified as fruit purée and fruit purée with dairy products.

group 10 (approximately 30%), while in the other food, its content varies between 15 and 19%. Apple cereal pap contains mainly isomer groups 2 and 10 (sum of 35%). Banana milk pudding and peaches with honey have, as principal isomer groups, isomers 3, 6, and 10 (the sum of these groups is 50 and 47%, respectively). Figure 4 depicts nonylphenol in fruit purée and fruit with dairy products or rice. The most abundant isomer group is isomer 10 in four of these foodstuffs [apples and bananas with curd (21%) and tropical fruits in apple juice (17%)]. Baby apples are different from the rest, as they have a higher percentage of isomer group 3 (20%; 10−13% in other food) with a sum of isomers 3 and 10 of 37%. Isomer group 6

50%. Isomer groups 4, 6, and 10 prevail in the bologna (sum of >50%). Figure 2 shows the isomer distribution in two commercial infant foodstuffs (early carrots and apple juice) and two homemade purées (bananas and carrots). Early carrots contrast with the other food as their content of isomers 4 and 8 is >60%. Compared to “carrots”, there are great differences, although they belong to the same vegetable group. Instead, carrots contain more of isomer group 10 (24%) and