Development of Three Stable Isotope Dilution Assays for the

Jan 15, 2014 - Quantitation of (E)‑2-Butenal (Crotonaldehyde) in Heat-Processed ... oil or 33.9−34.4 mg/kg of linseed oil after heat-processing fo...
1 downloads 0 Views 642KB Size
Article pubs.acs.org/JAFC

Development of Three Stable Isotope Dilution Assays for the Quantitation of (E)‑2-Butenal (Crotonaldehyde) in Heat-Processed Edible Fats and Oils as Well as in Food Michael Granvogl* Lehrstuhl für Lebensmittelchemie, Technische Universität München, Lise-Meitner-Straβe 34, D-85354 Freising, Germany ABSTRACT: Three stable isotope dilution assays (SIDAs) were developed for the quantitation of (E)-2-butenal (crotonaldehyde) in heat-processed edible fats and oils as well as in food using synthesized [13C4]-crotonaldehyde as internal standard. First, a direct headspace GC−MS method, followed by two indirect methods on the basis of derivatization with either pentafluorophenylhydrazine (GC−MS) or 2,4-dinitrophenylhydrazine (LC−MS/MS), was developed. All methods are also suitable for the quantitation of acrolein using the standard [13C3]-acrolein. Applying these three methods on five different types of fats and oils varying in their fatty acid compositions revealed significantly varying crotonaldehyde concentrations for the different samples, but nearly identical quantitative data for all methods. Formed amounts of crotonaldehyde were dependent not only on the type of oil, e.g., 0.29−0.32 mg/kg of coconut oil or 33.9−34.4 mg/kg of linseed oil after heat-processing for 24 h at 180 °C, but also on the applied temperature and time. The results indicated that the concentration of formed crotonaldehyde seemed to be correlated with the amount of linolenic acid in the oils. Furthermore, the formation of crotonaldehyde was compared to that of its homologue acrolein, demonstrating that acrolein was always present in higher amounts in heat-processed oils, e.g., 12.3 mg of crotonaldehyde/kg of rapeseed oil in comparison to 23.4 mg of acrolein/kg after 24 h at 180 °C. Finally, crotonaldehyde was also quantitated in fried food, revealing concentrations from 12 to 25 μg/kg for potato chips and from 8 to 19 μg/kg for donuts, depending on the oil used. KEYWORDS: (E)-2-Butenal, crotonaldehyde, acrolein, stable isotope dilution assay, fat and oil, food-borne toxicant



INTRODUCTION During high-temperature cooking with fats and oils, oxidative decomposition leads to the formation of several volatile compounds with significant reactivity, e.g., unsaturated aldehydes as well as fatty acids and alcohols.1−3 Some of the volatile aldehydes are desirable aroma-active compounds such as (E,E)- and (E,Z)-2,4-decadienal4−6 or undesirable toxicologically relevant compounds, which may be harmful to humans, e.g., the α,β-unsaturated 2-alkenals 2-propenal (acrolein)7−9 and (E)2-butenal (crotonaldehyde). Crotonaldehyde is a reactive and volatile monounsaturated aldehyde, which belongs to the group of environmental pollutants due to its formation, e.g., by combustion of organic material. Therefore, sources of crotonaldehyde are cigarette smoke (9−48 μg/cigarette in the inhaled smoke),10 wood-burning fireplaces (6−116 mg/kg),11 as well as gasoline and diesel exhausts of automobiles (2−114 mg/ km).12 Beside its intake by inhalation, the general population may also be exposed to crotonaldehyde by the daily diet. It has been reported in different types of food such as fruits and vegetables (0.008−0.1 mg/kg),13,14 mussels (11.5 mg/kg),15 heat-processed oils (emission rates of 0.38−3.06 mg/kg and h),16 wine (0.04−0.72 mg/L),17 or whisky (0.04−0.21 mg/ kg).18 Crotonaldehyde was found to be mutagenic in cell systems,19−22 revealed irritating properties,23 and induced liver tumors in rats when it was orally administered with tap water.24 Furthermore, the aldehyde is able to form 1,N2-etheno25 as well as 1,N2-propano26−28 DNA adducts with guanine and to conjugate with glutathione29,30 or proteins, e.g., human serum albumin.31 Crotonaldehyde is considered as a possible human © 2014 American Chemical Society

carcinogen by the US Environmental Protection Agency (EPA)32 and is listed in group 3 (“not classifiable as to its carcinogenicity to humans”) by the International Agency for Research on Cancer (IARC).33 Exact quantitative data are the prerequisite for studies on the intake of toxicologically relevant food constituents. However, the analysis of crotonaldehyde is a challenge due to its low molecular weight and reactivity as well as the lack of a chromophore. A proper tool to enhance the selectivity and, thus, the sensitivity of compounds with low molecular weight is a derivatization step. Reagents such as 2,4-dinitrophenylhydrazine (2,4-DNPH) followed by HPLC analysis34,35 or pentafluorophenylhydrazine (PFPH)9,36 prior to gas chromatography are well-known in the analysis of (short-chain) aldehydes. However, these analytical methods require extensive workup procedures, because the derivatives need to be isolated prior to chromatography. Another method for the isolation of substances with a low boiling point is the static or dynamic headspace analysis. However, although no sample cleanup is necessary prior to headspace GC analysis, commonly one drawback of this method is the possible coelution with other volatiles, especially for target molecules having low and, therefore, less selective molecular masses. In the past few decades, stable isotope dilution assays have been established as appropriate methods to quantitate volatile and/or unstable molecules, e.g., in aroma analysis,37,38 but also in Received: Revised: Accepted: Published: 1272

October 31, 2013 December 19, 2013 January 15, 2014 January 15, 2014 dx.doi.org/10.1021/jf404902m | J. Agric. Food Chem. 2014, 62, 1272−1282

Journal of Agricultural and Food Chemistry

Article

vortexer (1 min), and the organic layer was separated and, finally, dried over anhydrous sodium sulfate. Aliquots (2 μL) of the solutions were analyzed by means of a Thermo Quest gas chromatograph (Thermo Finnigan, Egelsbach, Germany), equipped with a fused silica DB-5 column (30 m × 0.25 mm i.d., 0.25 μm film thickness) (J&W Scientific, Agilent Technologies, Köln, Germany) using the cold-on-column technique at 60 °C. After 2 min, the temperature was raised at 10 °C/ min to 240 °C and, finally, held for 10 min isothermally. The flow of the carrier gas helium was 2.5 mL/min. A flame ionization detector was used at 250 °C. Identification was carried out using a reference mixture of fatty acid methyl esters. Sample Preparation. To check possible matrix effects during the quantitation of crotonaldehyde by headspace analysis of fats and oils, first, either crotonaldehyde or [13C4]-crotonaldehyde (100 ng each, dissolved in pentane/diethyl ether (70/30, v/v)) was added to unheated extra virgin olive oil (1 g in 20 mL vials), which did not contain crotonaldehyde above the detection limit. After 20 min of continuous stirring at room temperature, the samples were subjected to mass spectrometric analysis as described below. Oil Heating Process. Oil samples (5 g) were weighed into glass tubes (200 mm × 15 mm i.d.) covered with aluminum foil and were heated in a metal block. Heating was controlled by a thermocouple placed inside the oil. Samples were singly heated at 100, 140, 180 (regular deep-frying temperature), and 220 °C for 24 h each. In a further series of experiments, the samples were heated at 180 °C for 2, 4, 8, 16, 24, and 55 h. Immediately after the samples were cooled to room temperature, [13C4]-crotonaldehyde (0.5−150 μg, dissolved in pentane/diethyl ether (70/30, v/v), depending on the amounts determined in preliminary experiments) and [13C3]-acrolein (0.5−250 μg) were added, the glass tubes were closed and kept at −20 °C prior to analysis. Preparation of Deep-Fried Food. Dough Recipe for Donuts. After the yeast (42 g) was dissolved in warm water, wheat flour (500 g), whole milk (150 mL), sugar (50 g), eggs (100 g), and water (25 mL) were mixed until a homogeneous dough was obtained. The dough was then divided into pieces of about 50 g and incubated at 30 °C for 2 h. Deep-frying (4 min at 180 °C) was performed as mentioned below for potatoes. Deep-Frying of Potatoes. Thin slices (peeled, 1.5 mm cross section) were fried for 2.5 min at 180 °C in an electrical deep-fryer (Kenwood DF320, Heusenstamm, Germany) containing 2.3 L of oil, which was preheated for 10 min prior to frying. The potato-to-oil mass ratio was about 1:7. After the potatoes were cooled to room temperature, an aliquot (10 g) of the chips was weighed into a headspace vial (20 mL), [13C4]-crotonaldehyde and [13C3]-acrolein (each 150 ng) were added, and the mixture was stirred for equilibration (15 min). To provide the most similar conditions, the first five batches of the respective fried food were rejected, and only the sixth batch was used for analysis. Quantitation of Crotonaldehyde by Headspace HighResolution Gas Chromatography−Mass Spectrometry (HRGC−MS) (Method I). Aliquots (250−2000 μL) of the headspace volume were injected into the hot PPKD injector (Thermo Finnigan) of a Trace 2000 series gas chromatograph (Thermo Finnigan) using a Combi Pal autosampler (CTC Analytics, Zwingen, Switzerland) held at 20 °C and equipped with a 2 mL gastight syringe (SGE Analytic Science, Darmstadt, Germany). After each injection, a possible carryover into the syringe was eliminated by an automatic syringe flush (helium). The headspace autosampler conditions were set as follows: incubation temperature, 30 °C; syringe temperature, 35 °C; syringe injection volume, 2 mL; syringe injection speed, 0.1 mL/s. The effluent was quantitatively transferred into a cold trap (initial temperature, −150 °C for 0.1 min; heating rate, 15 °C/s; final temperature, 240 °C for 3 min) (SGE Analytic Science) using a moving column stream switching system (Thermo Finnigan). After the cooling was turned off, the trapped compounds were transferred onto a fused silica DB-5 MS column (30 m × 0.25 mm i.d., 1.0 μm film thickness) (J&W Scientific). The oven was heated from 10 °C to 60 °C at 6 °C/min and then to 220 °C at 10 °C/min. The effluent was monitored by an ion trap mass spectrometer, Saturn 2000 (Varian, Darmstadt), running in the chemical ionization mode (70 eV ionization energy) with methanol as the reagent gas. Crotonaldehyde and [13C4]-crotonaldehyde were

the analysis of food-borne toxicants such as acrolein, 9 acrylamide,39 furan,40 or glycidamide.41 However, to the best of my knowledge, no further method for the quantitation of crotonaldehyde using a stable isotopically labeled internal standard has been reported yet (after submission of this paper, I became aware of a similar approach of a Belgian group, which is not yet published), except our recently presented preliminary studies using synthesized deuterium or 13C-labeled crotonaldehyde.42,43 Because it is a common trend in high-temperature cooking to use vegetable oils with a high degree of unsaturation, the aims of the present study were (i) to develop a stable isotope dilution analysis using a 13C-labeled internal standard for the quantitation of crotonaldehyde, which also enables the simultaneous determination of acrolein using [13C3]-acrolein, (ii) to compare the data obtained by three different techniques, namely, a direct headspace GC−MS analysis (method I) and two indirect analyses using liquid−liquid extraction combined with a highvacuum distillation after derivatization by means of GC−MS (with pentafluorophenylhydrazine, method II) and LC−MS/ MS (with 2,4-dinitrophenylhydrazine, method III), and (iii) to apply these methods on thermally treated oils differing in their fatty acid composition as well as on fried foods.



MATERIALS AND METHODS

Safety. Acrolein, [13C3]-acrolein, crotonaldehyde, and [13C4]crotonaldehyde are hazardous and must be handled carefully. Food Samples. Potatoes, wheat flour, eggs, whole milk, sugar, and vegetable fats and oils (coconut oil, extra virgin olive oil, rapeseed oil, safflower oil, and linseed oil) were purchased at local supermarkets. Chemicals. [13C2]-Acetaldehyde (99%), crotonaldehyde, 2,4DNPH (99% purity, stabilized with 50% water), and PFPH (97%) were obtained from Sigma-Aldrich (Steinheim, Germany). Acrolein and sodium methylate were from Fluka (Steinheim). [13C3]-Acrolein (99%) was obtained from Isotec (Ohio, IL). All other reagents were of analytical grade. Synthesis of the Stable Isotopically Labeled Standard [13C4]Crotonaldehyde. To a solution (8 mL) of [13C2]-acetaldehyde in dichloromethane (500 mg in 8 mL) were added, first, a solution (8 mL) of pyrrolidine in dichloromethane (48.3 mg in 10 mL) under stirring at 4 °C and, second, a solution (8 mL) of pure acetic acid in dichloromethane (81.5 mg in 10 mL) using a dropping funnel. After a reaction time of 30 min, the mixture was extracted with a solution (2 × 15 mL) of saturated sodium hydrogen carbonate, and the organic phase was dried over anhydrous sodium sulfate. For purification, the solution was first subjected to high-vacuum distillation, and then the solvent was distilled off using a Vigreux column (60 cm × 1 cm i.d.). The residue was dissolved in pentane (1.5 mL), and the obtained solution was applied onto a silica gel column (water-cooled, 45 g of silica gel 60 (0.040−0.063 mm) containing 7% water) using the following elution: pentane (P)/diethyl ether (E) = 100/0 (v/v; 100 mL, fraction 1), P/E = 90/10 (v/v; 100 mL, fraction 2), P/E = 80/20 (v/v; 100 mL, fraction 3), and P/E = 70/30 (v/v; 100 mL, fraction 4). The target compound [13C4]-crotonaldehyde (isotopic purity >97%) was obtained in fraction 4. The concentration of [13C4]-crotonaldehyde was determined using methyl octanoate as the internal standard using a response factor of 0.90 calculated from data obtained by GC-FID analysis of a mixture of unlabeled crotonaldehyde and methyl octanoate. Fatty Acid Composition Analysis. The fatty acid composition was determined by GC-FID after transesterification with methanol/sodium methylate to obtain the corresponding fatty acid methyl esters closely following the IUPAC standard method.44 Aliquots of the oil (100 mg) were accurately weighed into screwcapped centrifuge tubes, and tetrahydrofuran (1 mL) was added. After the oil was dissolved by shaking, sodium methylate (2 mL, 0.5 mol/L) was added, and the solution was heated for 1 h at 50 °C. Then, water (5 mL) and hexane (5 mL) were added, the suspension was mixed using a 1273

dx.doi.org/10.1021/jf404902m | J. Agric. Food Chem. 2014, 62, 1272−1282

Journal of Agricultural and Food Chemistry

Article

251 (analyte) and m/z 255 (labeled standard) for the PFPH derivatives were used for determination) and by means of their retention indices. Instrumental Conditions for HPLC−MS/MS (Method III). Mass spectra were recorded by means of a triple-quadrupole tandem mass spectrometer (TSQ Quantum Discovery) (Thermo Electron, Dreieich, Germany) coupled to a Surveyor high-performance liquid chromatography system (Thermo Finnigan) equipped with a thermostated (20 °C) autosampler and an 120 mm × 2.0 mm i.d., 3 μm Nucleosil C18 120 Å HPLC column (Macherey-Nagel, Düren, Germany) kept at 30 °C connected to a 4 mm × 2.0 mm i.d. C18 precolumn (Phenomenex, Aschaffenburg, Germany). The solvent system was composed of formic acid in water (0.1% by weight; solvent A) and formic acid in acetonitrile (0.1% by weight; solvent B). A linear gradient was applied by increasing the concentration of solvent B from 30% to 100% within 15 min. Separation was performed at a flow rate of 0.2 mL/min. For quantitation, mass spectrometry was performed with positive electrospray ionization (ESI+) in multiple reaction monitoring (MRM) mode with a spray needle voltage of 3.3 kV. The temperature of the capillary was 300 °C. The sheath gas and auxiliary gas (nitrogen) were adjusted to 35 and 10 arbitrary units, respectively. The collision cell was operated at a collision gas (argon) pressure of 0.20 Pa. First, the derivatives of crotonaldehyde and [13C4]-crotonaldehyde were characterized by means of their molecular masses obtained in the full-scan mode. Clear molecular ions were obtained [M + H]+: m/z 251 for derivatized crotonaldehyde and m/z 255 for derivatized [13C4]crotonaldehyde. These were then subjected to MS/MS (collision energy 20 V). First, the most intense transitions of the precursor ions were determined, and second, the yields of the product ions were optimized by performing a series of runs with different collision energies and flow rates of the sheath gas and auxiliary gas. Finally, the mass transitions m/z 251 to m/z 234 and m/z 251 to m/z 157 for crotonaldehyde as well as m/z 255 to m/z 238 and m/z 255 to m/z 161 for [13C4]-crotonaldehyde were selected for quantitation, including the signal-to-noise ratio in oil samples as criteria. Next, the 2,4-DNPH derivatives of acrolein and [13C3]-acrolein were characterized by means of their molecular masses obtained in the fullscan mode. Again, clear molecular ions were obtained [M + H]+: m/z 237 for derivatized acrolein and m/z 240 for derivatized [13C3]-acrolein. These were then subjected to MS/MS (collision energy 20 V). First, the most intense transitions of the precursor ions were determined, and second, the yields of the product ions were optimized by performing a series of runs with different collision energies and flow rates of the sheath gas and auxiliary gas. Finally, the mass transitions m/z 237 to m/ z 220 and m/z 237 to m/z 143 for acrolein as well as m/z 240 to m/z 223 and m/z 240 to m/z 146 for [13C3]-acrolein were selected for quantitation, including the signal-to-noise ratio in oil samples as criteria. On both mass filter quadrupoles, the resolution setting was 0.8 full width at half-maximum, the scan time for each transition and single reaction monitoring (SRM) was 0.20 s, the scan width was 0.8 amu, and the source collision-induced dissociation (CID) voltage was 12 V. Quantitation of Acrolein by HRGC−MS after Derivatization with PFPH. Quantitation of acrolein after derivatization with PFPH was performed by high-resolution gas chromatography−mass spectrometry using [13C3]-acrolein as the internal standard as previously described.9 Calibration Curves. For calibration, seven mixtures of crotonaldehyde and [13C4]-crotonaldehyde as well as of acrolein and [13C3]acrolein in defined concentrations (molar ratios 10:1 to 1:10) were analyzed for all methods as described above, and the response factors were calculated from the results as previously described.41 Calibration was always performed prior to sample measurement. Limit of Detection (LoD) and Limit of Quantitation (LoQ). Calculation of the LoD and LoQ was carried out on the basis of a correlation between the intensity of the respective ions and the background noise with a minimum ratio of 3:1. Unheated oil samples were spiked with known amounts of crotonaldehyde (or acrolein for method III) in decreasing concentrations (100, 50, 25, 20, 15, 10, 5, 2.5, 1.25, and 0.5 μg/kg) and were analyzed in triplicate using all three methods. A control run without addition of crotonaldehyde and [13C4]-

located by means of their molecular masses obtained in the full scan mode (the mass range from m/z 50 to m/z 250 was recorded; m/z 71 for the analyte and m/z 75 for the labeled standard were used for determination) and by means of their retention indices. Crotonaldehyde concentrations were calculated from the area counts obtained from the mass chromatograms using the following equation:

m(crotonaldehyde) = R f ×

A(crotonaldehyde) × m([13C4 ]‐crotonaldehyde) A([13C4 ]‐crotonaldehyde)

where A(crotonaldehyde) is the area of unlabeled crotonaldehyde, A([13C4]-crotonaldehyde) is the area of [13C4]-crotonaldehyde, and m([13C4]-crotonaldehyde) is the amount of added [13C4]-crotonaldehyde. The response factor (Rf) was determined by analysis of defined mixtures of crotonaldehyde and [13C4]-crotonaldehyde (10:1, 5:1, 3:1, 1:1, 1:3, 1:5, and 1:10, w/w) as described below using the resulting response curve showing a very good linearity (coefficient of determination (R2) of 0.99). Applied concentration ranges varied from 0.2 to 200 μg of crotonaldehyde depending on the amounts determined in preliminary experiments. Quantitation of Acrolein by Headspace HRGC−MS. Quantitation of acrolein was performed by headspace high-resolution gas chromatography−mass spectrometry using [13C3]-acrolein as the internal standard as previously described.9 Quantitation of Crotonaldehyde by HRGC−MS after Derivatization with PFPH (Method II) or by High-Performance Liquid Chromatography−Tandem Mass Spectrometry (HPLC−MS/ MS) after Derivatization with 2,4-DNPH (Method III). Preparation of Reference Substances. First, the reference derivatives of both crotonaldehyde and labeled crotonaldehyde were made up by adding these aldehydes into olive oil to consider possible matrix interferences, which might also occur during the analysis of real samples. Therefore, unheated extra virgin olive oil (10 g) was spiked singly either with crotonaldehyde or with [13C4]-crotonaldehyde (5 μg each, dissolved in pentane/diethyl ether (70/30, v/v)). After the addition of PFPH solution (0.5 mL; 2 mg of PFPH were dissolved in 1 mL of 0.1 mol/L HCl, method II) or 2,4-DNPH solution (1 mL; first, 100 mg of 2,4DNPH were dissolved in 10 mL of 2.0 mol/L HCl using an ultrasonic bath and, then, diluted with 90 mL of methanol, method III), the mixture was stirred for 60 min (PFPH) or for 180 min (2,4-DNPH) at room temperature using a magnetic stirrer. In the same way, acrolein and [13C3]-acrolein were derivatized with PFPH9 or 2,4-DNPH. Then, the water-in-oil emulsion was extracted with dichloromethane (50 mL) by continuous liquid−liquid extraction. The PFPH derivatives as well as the 2,4-DNPH derivatives of crotonaldehyde and [13C4]crotonaldehyde were separated from the bulk of the fat by the solventassisted flavor evaporation (SAFE) technique45 at 48 °C under vacuum (5 × 10−3 Pa). The distillate was dried over anhydrous sodium sulfate and concentrated to a final volume of ∼1 mL using a Vigreux column (100 cm × 3 cm i.d.) at 48 °C. Aliquots (2 μL) of the solution were analyzed by GC−MS or LC−MS/MS as described below. Analyses of Oil Samples. Oil samples were prepared as described above for the oil-heating process. To aliquots (approximately 5 g) was added PFPH (method II) or 2,4-DNPH (method III) solution, and the mixture was stirred for 20 min at room temperature. Further workup was done as described above for the reference substances. Instrumental Conditions for HRGC−MS (Method II). Using a Combi Pal autosampler (CTC Analytics), an aliquot (2 μL) of the sample was injected into the hot PPKD injector of a Trace 2000 series gas chromatograph and transferred onto a DB-1701 column (30 m × 0.25 mm i.d., 0.25 μm film thickness) (J&W Scientific). The following temperature program was used: 40 °C for 2 min and then at 8 °C/min to 230 °C for 10 min. The effluent was monitored using an ion trap mass spectrometer Saturn 2000 running in the chemical ionization mode (70 eV ionization energy) with methanol as the reagent gas. The derivatives of crotonaldehyde and [13C4]-crotonaldehyde were characterized by means of their molecular masses obtained in the full scan mode (the mass range from m/z 50 to m/z 300 was recorded; m/z 1274

dx.doi.org/10.1021/jf404902m | J. Agric. Food Chem. 2014, 62, 1272−1282

Journal of Agricultural and Food Chemistry

Article

crotonaldehyde (or acrolein and [13C3]-acrolein) was performed to ensure the absence of the respective mass signals in the oil samples. HRGC−Sector Field Mass Spectrometry. To obtain mass spectra in electron impact (EI) as well as in chemical ionization (CI) mode for the PFPH derivatives of crotonaldehyde and [13C4]-crotonaldehyde, HRGC−MS analyses were performed by means of a Hewlett-Packard gas chromatograph, type 5890 series II (Hewlett-Packard, Waldbronn, Germany), coupled to a sector field mass spectrometer, type MAT 95 S (Finnigan MAT, Bremen, Germany), running either in EI mode (70 eV ionization energy) or in CI mode using isobutane as the reagent gas (115 eV ionization energy). The samples were injected by the cold oncolumn technique at 40 °C onto a DB-FFAP fused silica capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness) (J&W Scientific). The oven temperature was held for 2 min isothermally and then raised by 8 °C/min to 230 °C for 10 min.

traces. This prompted us to develop two additional methods to confirm the reliability of the results obtained by the headspace method similar to the development of a derivatization method using 2-mercaptobenzoic acid for the quantitation of acrylamide,39 also having an m/z signal of 72 (MS-CI) with a low specificity, or using pentafluorophenylhydrazine (PFPH) for the quantitation of acrolein (m/z 57, MS-CI).9 Preliminary trials showed a good conversion of crotonaldehyde by derivatization with the nucleophiles PFPH or 2,4DNPH, and the respective adducts were analyzed. Both hydrazines were found to convert crotonaldehyde within 60 min (PFPH) or 180 min (2,4-DNPH) (120 min for acrolein using 2,4-DNPH) with high yields into stable hydrazones, which were detected either by GC−MS (PFPH) or by LC−MS/MS (2,4-DNPH). For method development, first, the derivatives were synthesized by reacting crotonaldehyde as well as [13C4]crotonaldehyde singly with either PFPH (Figure 2A) or 2,4DNPH (Figure 2B) to obtain stable derivatives. Samples were analyzed by GC−MS (CI mode) or by LC−MS/MS (ESI+ mode).



RESULTS AND DISCUSSION Synthesis of the Stable Isotopically Labeled Internal Standard [13C4]-Crotonaldehyde. Due to the fact that the deuterated labeled standard, which was used in preliminary trials,42 showed a heterogeneous labeling pattern ([2H3−4]crotonaldehyde), first, a new 13 C-labeled standard was synthesized using [13C2]-acetaldehyde. By Aldol condensation, 4-fold-labeled [13C4]-crotonaldehyde was obtained (Figure 1),

Figure 1. Reaction scheme for the synthesis of the internal standard [13C4]-crotonaldehyde. The symbol “■” indicates 13C-labeling.

revealing a ratio between [13C4]-(E)-2-butenal (=[13C4]crotonaldehyde) and [13C4]-(Z)-2-butenal of 95:5. Stereochemistry was confirmed by comparison of the retention index using commercially available (E)-2-butenal. Mass spectrometry experiments in the EI and CI modes have proven the successful incorporation of the labeling into the internal standard [13C4]crotonaldehyde. The isotopic purity of [13C4]-crotonaldehyde was >97% (data not shown). Development of the Analytical Procedure. Method I (Headspace GC−MS). In the first experiment, crotonaldehyde and [13C4]-crotonaldehyde (2 μg, dissolved in pentane/diethyl ether (70/30, v/v)) were singly placed into two 20 mL headspace vials, and after 20 min of equilibration, the solutions were analyzed by headspace GC−MS. The mass chromatogram of crotonaldehyde obtained by chemical ionization showed a clear peak at m/z 71. At the same retention index, the labeled isotopologue revealed a peak at m/z 75. Next, either crotonaldehyde or [13C4]-crotonaldehyde (100 ng each, dissolved in pentane/diethyl ether (70/30, v/v)) was singly added to unheated extra virgin olive oil (1 g in 20 mL vials). The analyte and the labeled analogue were characterized by selected ion monitoring at m/z 71 (for crotonaldehyde) and m/z 75 (for [13C4]-crotonaldehyde) as well as by their retention indices. The results showed that the fatty matrix neither contained coeluting compounds showing a signal either at m/z 71 or at m/z 75 at the respective retention index nor contained crotonaldehyde above the detection limit. Method II (GC-MS after Derivatization) and Method III (LC−MS/MS after Derivatization). The crucial factor in headspace analysis of low molecular weight compounds such as crotonaldehyde is, however, the low specificity of the mass

Figure 2. Derivatization scheme used for the conversion of crotonaldehyde (analyte) and [13C4]-crotonaldehyde (internal standard) into stable pentafluorophenylhydrazones (method II, A) or 2,4dinitrophenylhydrazones (method III, B) prior to analysis. The symbol “■” indicates 13C-labeling.

PFPH Derivatives of Crotonaldehyde and [13C4]-Crotonaldehyde (Method II, GC−MS). As PFPH itself has a lower retention time compared to the crotonaldehyde derivatives, the MS detector was switched off during its elution to avoid any contamination of the ion trap. The mass spectra (EI mode) of the PFPH derivatives of crotonaldehyde and [13C4]-crotonaldehyde are given in Figure 3A and 3B. Various ions were obtained, which are postulated to be [C5F3]+, [C5F4]+, [C5F5]+, and [C6F5NH2]+.9 The fragment m/z 183 in the spectrum is attributable to the cleavage of the hydrazone bond between both nitrogen atoms, whereas the fragments m/z 155, 136, and 117 may be formed by rearrangement of the aromatic ring. Intense [M]+ ions of the unlabeled (m/z 250) and the labeled (m/z 254) 1275

dx.doi.org/10.1021/jf404902m | J. Agric. Food Chem. 2014, 62, 1272−1282

Journal of Agricultural and Food Chemistry

Article

Figure 3. Mass spectra (GC−MS, EI) of the PFPH derivatives of crotonaldehyde (A) and [13C4]-crotonaldehyde (B) as well as mass spectra (LC−MS/ MS, ESI+) of the 2,4-DNPH derivatives of crotonaldehyde (C) and [13C4]-crotonaldehyde (D). The symbol “■” indicates 13C-labeling.

2,4-DNPH Derivatives of Crotonaldehyde and [13C4]Crotonaldehyde (Method III, LC−MS/MS). The mass spectra (ESI+, full scan mode) of the 2,4-DNPH derivatives of crotonaldehyde and [13C4]-crotonaldehyde are shown in Figure 3C and 3D. The following molecular ions were obtained [M + H]+: m/z 251 for derivatized crotonaldehyde and m/z 255 for derivatized [13C4]-crotonaldehyde. Subjection to MS/MS revealed the following mass transitions, which were used for quantitation: m/z 251 to m/z 234 and m/z 251 to m/z 157 (crotonaldehyde) as well as m/z 255 to m/z 238 and m/z 255 to m/z 161 ([13C4]-crotonaldehyde). Even if m/z 157 and m/z 161 did not reveal the highest intensities, they showed the best signal-to-noise ratios in real samples besides m/z 234 and m/z 238. 2,4-DNPH Derivatives of Acrolein and [13C3]-Acrolein (Method III, LC−MS/MS). Next, the 2,4-DNPH derivatives of acrolein and [13C3]-acrolein were characterized by means of their molecular masses obtained in the full scan mode, revealing the following molecular ions [M + H]+: m/z 237 for derivatized acrolein and m/z 240 for derivatized [13C3]-acrolein (data not shown). Subjection to MS/MS resulted in the following mass transitions for quantitation: m/z 237 to m/z 220 and m/z 237 to m/z 143 (acrolein) as well as m/z 240 to m/z 223 and m/z 240 to m/z 146 ([13C3]-acrolein).

crotonaldehyde derivatives confirmed the successful derivatization reactions of crotonaldehyde and its isotopologue using PFPH. In contrast to the assumption that two peaks (syn and anti forms of the enamine) might be generated, double peaks were not detected in the chromatogram, probably because the two isomers did not separate on the stationary phase used (Figure 4). Next, unheated oil samples (10 g) were spiked with crotonaldehyde and [13C4]-crotonaldehyde (each 5 μg, dissolved in pentane/diethyl ether (70/30, v/v)), and the solutions were derivatized with PFPH. The best results to remove interfering compounds from the derivatives were obtained by high-vacuum distillation, the so-called SAFE technique.45 Thus, mass spectra were obtained for the oil extracts identical to those shown in Figure 3A and 3B obtained from the model solutions, proving the workup procedure to be suitable for the quantitation of crotonaldehyde. Last, similar to the development of method I, unheated extra virgin olive oil was analyzed without addition of either crotonaldehyde or [13C4]-crotonaldehyde. Again, no signals for m/z 251 and m/z 255 (both MS-CI) were detected in the respective unspiked samples, revealing again that crotonaldehyde is, if at all, only present below its detection limit and that [13C4]-crotonaldehyde is an appropriate internal standard. 1276

dx.doi.org/10.1021/jf404902m | J. Agric. Food Chem. 2014, 62, 1272−1282

Journal of Agricultural and Food Chemistry

Article

Figure 4. Extracts of GC−mass chromatograms (MS-CI) obtained for the analysis of crotonaldehyde (about 0.1 mg/kg) and acrolein (about 0.3 mg/ kg) in a heat-processed olive oil sample using method I (without derivatization, A) and method II (derivatization with PFPH, B), respectively. The symbol “■” indicates 13C-labeling.

derivatized with 2,4-DNPH and subjected to SAFE distillation.45 In this way, mass spectra were obtained for the oil extracts identical to those of the reference compounds from the model solutions, proving the workup procedure to be suitable for the quantitation of crotonaldehyde and acrolein.

Again, unheated oil samples (10 g) were spiked with crotonaldehyde and [13C4]-crotonaldehyde (each 5 μg, dissolved in pentane/diethyl ether (70/30, v/v)) as well as with acrolein and [13C3]-acrolein (each 5 μg, dissolved in methanol/ dichloromethane (50/50, v/v)). Then, the solutions were 1277

dx.doi.org/10.1021/jf404902m | J. Agric. Food Chem. 2014, 62, 1272−1282

Journal of Agricultural and Food Chemistry

Article

Last, similar to the development of methods I and II, unheated extra virgin olive oil was analyzed without addition of crotonaldehyde, [13C4]-crotonaldehyde, acrolein, or [13C3]acrolein. Again, no signals for m/z 251, 255, 237, or 240 were detected in the respective unspiked samples, revealing again that crotonaldehyde and acrolein are, if at all, only present below their detection limits and that [13C4]-crotonaldehyde and [13C3]-acrolein are appropriate internal standards. Calibration Curves, LoDs, and LoQs. A calibration curve was generated for each method by analyzing seven model mixtures in unheated extra virgin olive oil containing defined amounts of labeled and unlabeled crotonaldehyde in different molar ratios (10:1 to 1:10). The mixtures were worked up for each method as described above and were analyzed by headspace GC−MS (method I), GC−MS (method II), or LC−MS/MS (method III). The [M + H]+ mass ratios of the area counts of the analyte (m/z 71 for method I and m/z 251 for methods II and III) and the labeled standard (m/z 75 for method I and m/z 255 for methods II and III) were plotted against the concentration ratios. From these data, three response curves were obtained showing a good linearity (R2 > 0.99). For all three methods, limits of detection were calculated on the basis of a signal-to-noise ratio of 3:1. For this purpose, unheated extra virgin olive oil samples were spiked with different amounts of crotonaldehyde and acrolein (each 0.5−100 μg/kg, methods I−III), worked up, and measured by headspace GC− MS, GC−MS, or LC−MS/MS. The results allowed the calculation of LoDs for crotonaldehyde of 3 μg/kg for method I, 2 μg/kg for method II, and 1.5 μg/kg for method III as well as LoQs of 9 μg/kg for method I, 6 μg/kg for method II, and 4.5 μg/kg for method III. Interestingly, all three methods revealed nearly the same LoDs, pointing out method III as the most sensitive one due to MS/MS detection. For acrolein, LoDs were determined to be 3 μg/kg for method I and 2 μg/kg for methods II and III and LoQs to be 9 μg/kg for method I and 6 μg/kg for methods II and III. Thus, the main goal of this method development, using selective quantifier fragments of m/z 251 (analyte) and m/z 255 (internal standard) to check the correctness of the results obtained by the headspace GC−MS method (method I) using less selective quantifier signals of m/z 71 (analyte) and m/z 75 (internal standard), was achieved. Generation of Acrolein and Crotonaldehyde during Thermal Processing of Oils. To estimate the influence of the fatty acid composition on the amounts of crotonaldehyde formed after heating, five different oil samples varying in their fatty acid compositions were heated at 180 °C for 24 h, and analyses were carried out using all three methods. First, the fatty acid composition of the fat and oil samples was determined (Table 1). In agreement with the literature, coconut oil was dominated by lauric and myristic acid, while olive oil mainly consisted of oleic acid. Also in rapeseed oil, oleic acid was the most prominent fatty acid, while safflower oil showed the highest amount of linoleic acid, and linseed oil was highest in linolenic acid. To check the influence of heat-processing on the fatty acid composition, the five fats were heated for various times (up to 96 h) at 180 °C and were analyzed again, revealing only marginal decreases, particularly for the unsaturated fatty acids (data not shown). Typical chromatograms of heated oil samples obtained by method I (Figure 4A), method II (Figure 4B), and method III (Figure 5) are exemplarily shown for a thermally processed olive oil sample, showing clear peaks for acrolein and crotonaldehyde and their respective internal standards, [13C3]-acrolein and

Table 1. Fatty Acid Composition of the Unheated Fat and Oil Samples concna (g/100 g of fat) sample

C12:0

coconut oil olive oil rapeseed oil safflower oil linseed oil frying fat

50