Article pubs.acs.org/JAFC
Development of Stable Isotope Dilution Assays for the Quantitation of Amadori Compounds in Foods Michael Meitinger, Sandra Hartmann, and Peter Schieberle* Deutsche Forschungsanstalt für Lebensmittelchemie, Lise-Meitner-Straße 34, D-85354 Freising, Germany S Supporting Information *
ABSTRACT: During thermal processing of foods, reducing carbohydrates and amino acids may form 1-amino-1-desoxyketoses named Amadori rearrangement products after the Italian chemist Mario Amadori. Although these compounds are transient intermediates of the Maillard reaction, they are often used as suitable markers to measure the extent of a thermal food processing, such as for spray-dried milk or dried fruits. Several methods are already available in the literature for their quantitation, but measurements are often done with external calibration without addressing losses during the workup procedure. To cope with this challenge, stable isotope dilution assays in combination with LC-MS/MS were developed for the glucose-derived Amadori products of the seven amino acids valine, leucine, isoleucine, phenylalanine, tyrosine, methionine, and histidine using the respective synthesized [13C6]-labeled isotopologues as internal standards. The quantitation of the analytes added to a model matrix showed a very good sensitivity with the lowest limits of detection for the Amadori compound of phenylalanine of 0.1 μg/ kg starch and 0.2 μg/kg oil, respectively. Also, the standard deviation measured in, for example, wheat beer was only ±2% for this analyte. Application of the method to several foods showed the highest concentrations of the Amadori product of valine in unroasted cocoa (342 mg/kg) as well as in dried bell pepper (3460 mg/kg). In agreement with literature data, drying of foods led to the formation of Amadori products, whereas they were degraded during roasting of, for example, coffee or cocoa. The study presents for the first time results on concentrations of the Amadori compounds of tyrosine and histidine in foods. KEYWORDS: LC-MS/MS, stable isotope dilution assay, [13C6]-labeled Amadori rearrangement products, 1-amino-1-desoxyfructosyl amino acids
■
INTRODUCTION 1-Amino-1-desoxyketoses, named Amadori rearrangement products after the Italian chemist Mario Amadori, have been known for decades as reaction products of free amino acids with reducing carbohydrates. They are known to be formed in particular upon drying or thermal processing of foods. Their formation starts with the generation of a Schiff base between the aldehyde function of the open-chain carbohydrate and the amino group of the amino acid followed by a rearrangement into the 1-amino-1-desoxyketosyl amino acid and finally a cyclization into the pyranose moiety of the carbohydrate (Figure 1). Because Amadori products are more stable than the first formed glycosylamines, they are often used as quantitative markers to evaluate the extent of the Maillard reaction induced by food processing, for example, during drying of malt, vegetables, or fruits. Nevertheless, they are transient intermediates of the Maillard reaction and may further degrade, forming, for example, aroma-active compounds, such as Strecker aldehydes,1 and some also show taste activity, for example, the umami-tasting Amadori compound from glucose and glutamic acid.2 Several methods have already been suggested for the analysis of Amadori compounds.3−9 For quantitation by means of GCFID,3 they must be extracted, isolated by means of cation exchange chromatography, derivatized into the stable oximes, and finally silylated.3,4 However, such preseparation and derivatization steps may lead to a degradation during workup and also do not allow their fast quantitation in a high number of © 2014 American Chemical Society
Figure 1. Pathway leading to the formation of Amadori reaction products from glucose and α-amino acids.
samples. In addition, a direct measurement after extraction with aqueous buffer was carried out by means of an amino acid analyzer using postcolumn reaction with ninhydrin.5 However, peak resolution was poor, and a coelution with free amino acids may lead to incorrect results. HPLC separation with different methods of detection is another option, and one possibility is the application of an N,N-diethylaminoethyl-modified silica gel Received: Revised: Accepted: Published: 5020
March 26, 2014 May 10, 2014 May 12, 2014 May 12, 2014 dx.doi.org/10.1021/jf501464g | J. Agric. Food Chem. 2014, 62, 5020−5027
Journal of Agricultural and Food Chemistry
Article
of cold acetone (45 mL) at −25 °C and, after centrifugation at 3500 rpm and 0 °C for 10 min, the precipitate was collected and the mother liquor was concentrated and treated again with cold acetone at −25 °C to gain a second portion of the target compound. The combined precipitates were dissolved in methanol (5 mL) at ∼45 °C and treated again with cold acetone (45 mL) at −25 °C before centrifugation. This step was repeated until the crystals were nearly white. The material was dried under vacuum in a desiccator overnight and then freezedried for 2 days. Yields varied between 8 and 40%. The products formed were assigned as 1-desoxyfructosyl amino acids using the following abbreviations, for example, Fru-Leu for the Amadori product from glucose and leucine, Fru-Phe for the Amadori product from glucose and phenylalanine, and so on. Synthesis of the Carbon-13 [13C6]-Labeled Amadori Products. The respective carbon-13-labeled internal standards were synthesized using the same procedure but substituting glucose by [13C6]-glucose in the reaction with each of the amino acids. The assignment of the synthesized compounds was done by adding [13C6], for example, [13C6]-Fru-Phe for the carbon-13-labeled Amadori product synthesized from labeled glucose and phenylalanine. HPLC-MS. The removal of glucose and the respective amino acid as well as byproducts was controlled by HPLC-MS using a combination of a P-4000 pump (Thermo Separation Products, Waltham, MA, USA), a 150 × 2 mm TSK Gel Amide-80 column (Tosoh Bioscience, Tokyo, Japan), and an ion trap MAT LCQ mass spectrometer (Finnigan, Bremen, Germany). Spectra were generated in the ESI+ mode in parallel with UV detection at 210 nm. Quantitative Nuclear Magnetic Resonance (qNMR) Spectroscopy. To cope with the amount of the remaining solvent and water in the labeled as well as unlabeled Amadori products, these were analyzed by quantitative NMR (qNMR).11−13 Each analyte (100 mg) was dissolved in D2O (Euriso-Top, Saarbrücken, Germany) in a 5 mL volumetric flask, and 600 μL of this stock solution was transferred into a 178 × 5 mm NMR tube (Norell, Landisville, NJ, USA) and analyzed using an AV III System (Bruker, Rheinstetten, Germany) at 400.13 MHz. The solutions were diluted with acetonitrile, placed in amber bottles, flushed with argon, and stored at −24 °C. 1 H NMR spectra were measured at 25 °C with a Bruker AV III system equipped with a 5 mm multinuclear observe probe head (BBFO, plus). The probe was tuned and matched with the sample in place, and the optimal 90° pulse width for the quantitation setup was investigated for every sample. The qNMR spectrum was determined by acquiring 16 scans with a relaxation delay of 40 s. Before Fourier transformation, the free induction decay was multiplied with a 0.3 Hz exponential line broadening factor and zero-filled. Phasing and integration of the peaks were done manually after automatic baseline correction. If necessary, the integrals were adjusted with the software functions SLOPE and BIAS. The concentration of each compound was calculated with the ERETIC 2 function of Topspin 3.0 (Bruker). Calibration of the qNMR was based on a solution of L-tyrosine of known concentration using the signal at 7.1 ppm (m, 2H). The signal of the methylene group at carbon-1 of the former glucose at 3.3 ppm (m, 2H) was used for the quantitation of the Amadori products FruLeu, Fru-Ile, Fru-Val, and Fru-Met. In the cases of Fru-Phe, Fru-Tyr, and Fru-His the aromatic protons were used for quantitation (Fru-Phe, 7.4 ppm, m; Fru-Tyr, 7.0 ppm, m; and Fru-His, 7.7 ppm, s) because the free amino acids were confirmed to be absent by HPLC measurements. Isolation and Quantitation of Amadori Compounds in Foods. Either ground solid material (∼2 g) or beverages (∼20 mL) were exactly weighed in flasks and extracted with methanol (200 mL) after addition of the labeled internal standards. The mixtures were stirred for 1 h, then filtered and concentrated to ∼5 mL by means of a rotary evaporator. An extraction time of 1 h was needed for reproducible results, but a longer extraction did not change the results. To separate nonpolar substances, a 140 Å Strata C18-T SPE cartridge (1000 mg; volume, 6 mL) (Phenomenex, Aschaffenburg, Germany) was conditioned with methanol (10 mL). After application of the extract, the eluate (10 mL) was concentrated to dryness,
as the stationary phase and a postcolumn reaction with 2,3,5triphenyltetrazolium chloride. The resulting 1,3,5-triphenylformazans are detected photometrically at 480 nm, and in this case, free amino acids do not interfere with the quantitation.6 Another approach follows the analysis of N-(2-furoylmethyl) amino acids formed by the controlled degradation of the Amadori products after acid hydrolysis, which can then be measured by either HPLC-UV or HPLC-MS, respectively.7 High-performance cation or anion exchange chromatography in combination with either electrochemical detection8 or tandem mass spectrometry9 was recently used, but for electrochemical detection a special analytical equipment is required. Most of the methods that have been developed so far use external calibration. However, quantitative data obtained by methods based on external calibration are difficult to correct for losses during extraction or degradation of the analyte, but such losses during workup can be completely compensated for by stable isotope dilution assays. There is only one study available on the quantitation of the Amadori product of lysine bound in a peptide using a stable isotope dilution assay.10 Therefore, the aim of this study was to develop stable isotope dilution assays for seven Amadori compounds generated by a reaction of glucose with the amino acids valine, leucine, isoleucine, methionine, and phenylalanine as well as tyrosine and histidine. For this purpose, all seven unlabeled Amadori products as well as six carbon-13-labeled isotopologues were synthesized, and the new method was applied to the quantitation of the seven Amadori compounds in various food samples using LC-MS/ MS.
■
MATERIALS AND METHODS
Materials. The following compounds were obtained from commercial sources: D-glucose (99.5%), malonic acid, N,N-dimethylformamide, and methanol (both Chromasolv) (Sigma-Aldrich, Steinheim, Germany); L-leucine, L-isoleucine, L-methionine, L-phenylalanine, L-valine, L-tyrosine, L-histidine (all for biochemistry), acetonitrile, and acetone (both Suprasolv) (Merck, Darmstadt, Germany); deuterium oxide (99.96% deuterium) (Euriso-top, SaintAubin Cedex, France); and [13C6]-D-glucose (99%) (Cambridge Isotope Laboratories, Andover, MA, USA). Foods. Unroasted cocoa beans, 5 days fermented in wooden boxes, as well as roasted cocoa beans of the same batch (origin.. Sulawesi) were kindly provided by a food manufacturer. The green coffee beans (Arabica, Colombia) were supplied by Nestle PTC (Orbe, Switzerland). The beans were roasted in-house by means of a BRZ 4 drum roaster (Probat, Emmerich, Germany). The initial temperature of 185 °C was held for 3.5 min. Then, the air supply was opened, and the beans were roasted for another 2 min before cooling in the air stream of the collection chamber. Barley and wheat malt as well as wheat beer were obtained from a local brewery. The beer was produced from the same malt. Bell pepper and tomatoes were bought at a local market. The bell pepper powder was produced by drying of the vegetables in an oven at 50 °C for 7 days followed by grinding with a mill. Tomato powder was produced from fresh tomatoes by first shredding with a blender and then heating at 95 °C for 5 min followed by boiling for 3 h at 85−90 °C. The homogenized tomato paste was finally freezedried. Syntheses of Amadori Rearrangement Products. D-Glucose (500 mg, 2.8 mmol) was dissolved in a 1:1 mixture of methanol/N,Ndimethylformamide (100 mL) and refluxed at 80 °C singly in the presence of each of the seven amino acids (3.8 mmol) with stirring for 3 h. Then, malonic acid (100 mg, 1.0 mmol) was added, and the mixture was refluxed for another 2 h at 90 °C until a light brownish color appeared. Undissolved components were filtered off, and the organic phase was concentrated to ∼5 mL by means of a rotary evaporator. The Amadori products were precipitated by the addition 5021
dx.doi.org/10.1021/jf501464g | J. Agric. Food Chem. 2014, 62, 5020−5027
Journal of Agricultural and Food Chemistry
Article
Figure 2. Mass spectra (ESI+) of fructosyl leucine (Fru-Leu; A), [13C6]-fructosyl leucine (Fru-Leu; B), fructosyl tyrosine (Fru-Tyr; C), and [13C6]fructosyl tyrosine (Fru-Tyr; D). redissolved in acetonitrile (10 mL), and filtered through a Whatman Spartan 13/0.45 RC membrane filter (GE Healthcare, Wien, Austria). HPLC-MS/MS. The samples were analyzed by LC-MS/MS using a Thermo Finnigan Quantum TSQ mass spectrometer (Finnigan, Bremen, Germany) connected to a 150 × 2 mm TSK Gel Amide 80 HPLC column (Tosoh Bioscience). Spectra were generated in the ESI-positive mode with selected reaction monitoring. The flow conditions were as follows: 0.2 mL/min acetonitrile/0.1% aqueous formic acid (9 + 1) for 3 min with a gradient to 0.1% aqueous formic acid/acetonitrile (9 + 1) within 12 min, then held for 5 min. The spray voltage was 3500 V, and the capillary temperature was 290 °C. Determination of the Limit of Detection (LoD) and Limit of Quantitation (LoQ). Decreasing amounts of the synthesized, unlabeled Amadori products were added to either wheat starch or sunflower oil, respectively, and the mixtures were analyzed as described above for the food samples. The solutions were diluted until the analyte peaks were
