Analysis of α-dicarbonyl compounds and volatiles formed in Maillard reaction model systems

In this study, production of three α-dicarbonyl compounds (α-DCs) including glyoxal (GO), methylglyoxal (MGO), and diacetyl (DA) as well as volatile flavor compounds was analyzed using Maillard reaction (MR) model systems. A total of 16 model systems were assembled using four amino acids and four reducing sugars, and reactions were performed at 160 °C and pH 9. Determination of α-DCs was conducted using a gas chromatography/nitrogen phosphorous detector (GC-NPD) after derivatization and liquid-liquid extraction. α-DC levels in MR model systems were 5.92 to 39.10 μg/mL of GO, 3.66 to 151.88 μg/ml of MGO, and 1.10 to 6.12 μg/mL of DA. The highest concentration of total α-DCs was found in the fructose-threonine model system and the lowest concentration in the lactose-cysteine model system. Volatile flavor compounds were analyzed using solid-phase micro-extraction (SPME) followed by GC-mass spectrometry (GC-MS). Different volatile flavor compound profiles were identified in the different MR model systems. Higher concentrations of α-DCs and volatile flavor compounds were observed in monosaccharide-amino acid MR model systems compared with disaccharide-amino acid model systems.

caramel sauce, and bread. Pyrazines, representative compounds of the roasted flavor, are formed by reaction of amines and α-dicarbonyls through Strecker degradation 13,14 . Formation of furaneols, representative compounds of the sweet flavor, occurs through the 2, 3-enolization pathway leading to 1-deoxyosones as intermediates 15 .
MRPs that consist mainly of low-molecular volatile compounds are difficult to sample and analyze, and individual compound analysis has been conducted in such cases. Studies of MRPs with sensory characteristics have also been performed. However, studies comparing desirable aroma compounds and undesirable toxic compounds are not common. In this study, we compared α-dicarbonyl compounds and aroma compounds produced in reducing sugar-amino acid model systems. This work puts down the bases for studies on desirable preference factor like flavors and undesirable toxic substances.
preparation of Maillard reaction model system solutions. Equimolar solutions (0.1 M) of four reducing sugars (glucose, fructose, lactose, and maltose) and four amino acids (lysine, serine, threonine, and cysteine) were prepared in distilled water and pH was set to 9. Solutions were placed in swing-top bottles and reactions were performed at 160 °C for 2 h in an oven (OF-22, Jeiotech Co., Seoul, Korea) to instigate the Maillard reaction. All solutions were cooled under running cold water to room temperature immediately in order to stop the reaction progress. All samples were stored at 4 °C until they were analyzed.
Analysis of α-Dicarbonyl compounds. Quantification and analysis of α-DCs produced in the Maillard reaction model systems was performed using o-phenylenediamine dihydrochloride derivatization reported in a previous study 1 . A 3 mL volume of each sample and 2 mL of o-phenylenediamine dihydrochloride were placed in 20 mL vials. The mixtures were set to pH 12 and stirred for 2 h at 600 rpm for derivatization of α-DCs to quinoxalines. Then, 5 mL of ethyl acetate was added to the mixtures and shaken for extraction. Extracted solutions were analyzed using GC-NPD. The concentrations used to perform standard calibration curve were 0.5, 1, 5, 10, 50, 100 µg/ml.
An Agilent 6890 N gas chromatograph with a nitrogen phosphorous detector was used for analysis of α-DCs. A DB-WAX column (30 m × 250 × 0.25 μm; J&W Scientific, Folsom, CA) was used for separation. Helium was used as carrier gas at constant flow of 1.5 mL/min. Injection was set to the splitless mode at 260 °C. Oven temperature was held at 40 °C for 2 min and then increased to 170 °C at 20 °C/min and held for 15 min. Detector temperature was set at 300 °C. Nitrogen was used as make-up gas at a flow rate of 5 mL/min. Analysis of aroma compounds. Aroma compounds were analyzed using SPME (Solid phase micro-extration). A 5-mL volume of sample was added to a 20 mL headspace vial containing 0.75 g of sodium chloride. 5 μL of methyl cinnamate (internal standard, 100 μg/mL), 10 μL of alkane standard (10 μg/mL), and a magnetic stirring bar were added and the vial sealed with a PTFE cap. Subsequently, samples were stirred for 30 min at 70 °C to reach equilibrium. Volatile flavor compound extraction was carried out by injecting a fiber into the vial for adsorption at 70 °C for 10 min. Then, the fiber was inserted into the GC injector port and held for 10 min to desorb the volatile flavor compounds.
An Agilent 7820 A gas chromatograph and a 5977E mass spectrometer were used for volatile flavor compound detection. Separation of volatile aroma compounds was performed using a DB-WAX column (60 m × 250 × 0.25 μm). The oven was held at 40 °C for 5 min, then raised to 185 °C at 5 °C/min and held for 20 min, and then raised to 200 °C at 10 °C/min and held for 5 min. Helium was used as carrier gas at a constant flow rate of 0.8 mL/min. Injection mode was set to splitless at 230 °C. The detector was set to the scan mode and the scan range was from 20 to 550 m/z. Volatile flavor compounds were identified based on Kovats index on DB-WAX, co-injection, and mass spectrum from the NIST library. The quantification of each flavor compound was displayed as peak area ratio (PAR, peak area of each compound/that of internal standard). statistical analysis. All samples were analyzed in triplicate and the analysis results are presented as mean ± standard deviation.

Results and Discussion
Analysis of α-DCs in Maillard reaction model systems. Reaction model solutions were prepared with glucose and lysine at four temperatures (100, 120, 140 and 160 °C) and four pH levels (3,5,7, and 9) to simulate Maillard reaction model system conditions. Level of different α-DCs produced in the reactions, as shown in Table 1 was 0.36 to 16.78 μg/mL of GO, trace to 50.81 μg/mL of MGO, and N.D. to 2.27 μg/mL of DA. Higher pH and temperature resulted in greater production of total α-DCs including GO, MGO, and DA. Conditions yielding the maximum concentration of total α-DCs were 160 °C and pH 9. These conditions were then used to study the reducing sugar-amino acid model systems. As shown in Table 2, a total of 16 model systems were used with combinations of four amino acids (lysine, serine, threonine, and cysteine) and four reducing sugars (glucose, fructose, maltose, and lactose). Under the above conditions, basic amino acids containing hydroxyl groups have higher reactivity with α-DCs than acidic and nonpolar amino acids 16 . Levels of α-DCs in MR model www.nature.com/scientificreports www.nature.com/scientificreports/ systems were 5.92 to 39.10 μg/mL of GO, 3.66 to 151.88 μg/mL of MGO, and 1.10 to 6.12 μg/mL of DA. The highest concentration of α-DCs was detected in the fructose-threonine model system and the lowest concentration in the lactose-cysteine model system. In addition, DA concentration was 1.66 to 6.12 μg/mL in lactose model systems, compared with other reducing sugar model systems. These results indicated that α-DCs were produced mainly from monosaccharides rather than from disaccharides. Hollnagel and Kroh reported that monosaccharides forms more α-DCs than disaccharides 2 . Different from our results among monosaccharides glucose formed more α-DCs than fructose 2 . In the report they suggested that fructose forms more cyclic compounds rather than fragmentation products such as α-DCs. However, Maillard reaction conditions were different and it could give rise to the conflicting results.

Conclusions
In this study, desirable volatile flavor compounds and undesirable α-dicarbonyl compounds formed during the Maillard reaction were analyzed using reducing sugar-amino acid model systems. The levels of α-DCs produced in MR model systems were 1.10 to 151.88 μg/mL. The highest concentration of total α-DCs was found in the fructose-threonine model system and the lowest concentration of total α-DCs was found in the lactose-cysteine model system. Different volatile flavor compound profiles were identified from different MR model systems.
Higher concentrations of α-DCs and volatile flavor compounds were observed in monosaccharide-amino acid Maillard model systems than in disaccharide-amino acid model systems. As a future work, the balanced study between desirable flavor compounds and toxic compounds such as α-DCs are supposed to be carried out in our laboratory. 12 Table 5. Aroma compounds produced in maltose (Mal)-amino acid (lysine (Lys), serine (Ser), threonine (Thr), and cysteine (Cys)) Maillard reaction model systems.  Table 6. Aroma compounds produced in lactose (Lac)-amino acid (lysine (Lys), serine (Ser), threonine (Thr), and cysteine (Cys)) Maillard reaction model systems.