Investigations on the Key Odorants Contributing to the Aroma of Children Soy Sauce by Molecular Sensory Science Approaches

To investigate the key odor-active compounds in children’s soy sauce (CSS), volatile components were extracted by means of solvent extraction coupled with solvent-assisted flavor evaporation (SE-SAFE) and solid-phase microextraction (SPME). Using gas chromatography-olfactometry (GC-O) and gas chromatography-mass spectrometry (GC-MS), we identified a total of 55 odor-active compounds in six CSSs by comparing the odor characteristics, MS data, and retention indices with those of authentic compounds. Applying aroma extract dilution analysis (AEDA), we measured flavor dilution (FD) factors in SE-SAFE isolates, ranging from 1 to 4096, and in SPME isolates, ranging from 1 to 800. Twenty-eight odorants with higher FD factors and GC-MS responses were quantitated using the internal standard curve method. According to their quantitated results and thresholds in water, their odor activity values (OAVs) were calculated. On the basis of the OAV results, 27 odorants with OAVs ≥ 1 were determined as key odorants in six CSSs. These had previously been reported as key odorants in general soy sauce (GSS), so it was concluded that the key odorants in CSS are the same as those in GSS.


Introduction
Soy sauce (SS) originated in China about 2700 years ago [1]. As a kind of condiment, SS was mainly manufactured in Asian countries, but it was consumed in various places around the world. In recent years, with the rapid development of children's food, many children's soy sauces (CSSs) have been supplied in the Chinese market. These CSSs are claimed to have more nutritional elements, to be manufactured by a special process, and to be more suitable for consumption by children; their prices are much higher than those of general SS (GSS). Odor is one of the important sensory properties of CSS; to our knowledge, there have been no reports to date on the flavor constituents of CSS, nor is there a Chinese standard for CSS.
To date, reports about the flavor constituents of SS have focused on GSS. From 1887, researchers began to investigate the volatile compounds in SS [2], and to date, there have been many reports about the volatiles in SS [3][4][5][6][7][8][9][10]. Among the volatile compounds identified, not all of them contribute to the overall odor profiles of SS. Gas chromatographyolfactometry (GC-O) analysis has been used as an effective method to screen the odor-active compounds from the volatiles in food extracts. Volatile components in Korean SS were

SE-SAFE for Volatile Components in CSS
CSS samples (100 mL) were extracted with redistilled dichloromethane (50 mL × 3) at room temperature by stirring vigorously for 1.5 h × 3, and the obtained extracts were merged together. The volatiles were isolated from the combined extracts via high vacuum distillation using SAFE (Edwards TIC Pumping Station from BOC Edwards, England). The extract containing neutral and basic volatile components was obtained by washing the distillate from SAFE with 0.05 mol/L sodium carbonate solution (100 mL × 2) and saturated sodium chloride (50 mL × 3), respectively. The alkaline aqueous phase was acidified to a pH value of 2 using 0.5 mol/L HCl solution, and then the mixture was extracted with dichloromethane (50 mL × 3) to obtain the isolate containing acidic volatile compounds [4]. Both extracts were dried over anhydrous sodium sulfate for about 12 h and concentrated to approximately 3-5 mL with Vigreux columns (50 cm × 1 cm) (Beijing Jingxing Glassware Co., Ltd., Beijing, China) at 45 • C, and then they were further concentrated to 0.3 mL using gentle nitrogen streams. These concentrates were used for GC-O and GC-MS analyses.

SPME for Volatile Constituents in CSS
The volatile compounds in CSS were also extracted by means of SPME, as described previously with some modifications [11]. A 2-cm (coated with 50/30 µm DVB/CAR/PDMS) SPME fiber (Supelco, Bellefonte, PA, USA) was preconditioned before extraction experiments in accordance with the manufacturer's instructions. A mixture of 16 mL CSS and 2 g sodium chloride was placed in a 40-mL static headspace amber glass bottle fitted with a stir bar and a polytetrafluoroethylene (PTFE)-faced silicon septum. The extraction conditions for SPME obtained by optimizing experiments were as follows: equilibrium and extraction temperatures of 45 • C, an equilibrium time of 20 min, and an extraction time of 40 min. After the extraction experiment, the fiber was transferred to the injector port of GC for a 5-min desorption at 250 • C to conduct the GC-O and GC-MS analyses.

GC-O Analysis
GC-O was performed by means of an Agilent 7890 GC combined with an olfactory detection port (ODP3, Gerstel, Germany) and an FID (Agilent Technologies, USA). The GC effluent at the end of the capillary column was split into a 1:2 ratio by volume using a Y-type splitter and two uncoated deactivated fused silica capillaries between the FID and ODP. To maintain the nose sensitivity, the sniffing port was coupled with humidified air. The temperatures of the GC injector port, the FID, the transfer line of ODP3 and the olfactory port were 250 • C, 280 • C, 250 • C and 220 • C, respectively. The extracts were analyzed on both a DB-Wax column and a Hp-5MS column (Agilent, both are 30 m × 0.25 mm × 0.25 µm). When the DB-Wax column was used, the oven temperature was held at 40 • C for 2 min, increased to 80 • C at a rate of 8 • C/min, increased to 100 • C at a rate of 4 • C/min, then rose to 230 • C at a rate of 6 • C/min, and finally held at 230 • C for 5 min. When the Hp-5MS column was used, the oven temperature was held at 40 • C for 2 min, increased to 100 • C at a rate of 4 • C/min, ramped to 230 • C at a rate of 10 • C/min and finally held at 230 • C for 5 min. Ultra-high purity helium was used as the GC carrier gas at a constant flow rate of 1 mL/min. All concentrated fractions (1 µL) or SPME isolates were injected in splitless mode. During GC-O analyses, three trained evaluators (two females and one male, who had been trained to sniff the aromas of reference compound solutions with different concentrations in the laboratory for at least 3 months) from Beijing Key Laboratory of Flavor Chemistry at Beijing Technology and Business University sniffed the odors of the effluent from the sniffing port. When evaluators detected the odor, they needed to record the retention time (RT) and the odor characteristics. Analyses were carried out three times by each evaluator.

GC-MS Analysis
GC-MS analyses for identification were conducted with an Agilent 7890B GC connected to an Agilent 5975 mass selective detector. The parameters, columns and temperature program for GC were the same as those employed in the GC-O analyses described above. Mass spectra in election ionization mode at 70 eV were recorded at 150 • C; the ion source temperature was kept at 230 • C. Detection was carried out in full-scan mode, and mass range was from 33 to 350 amu.

Odor-Active Compound Identification
A series of normal alkanes were analyzed using GC-O and GC-MS under the conditions described in Section 2.4.1 and Section 2.4.2, and RTs of normal alkanes were measured. Retention indexes (RIs) of the detected odor-active compounds were computed on the basis of their RTs and the RTs of normal alkanes. If the concentrations of odor-active compounds were higher than the detection limits of the mass selective detector, their MS data were obtained, and they were positively identified by comparing their MS data, RIs and odor characteristics with those of standard compounds and data in NIST2014. If the concentrations of odor-active compounds were lower than the detection limits of the mass selective detector, and their MS data were not available, they were positively identified by comparing their RIs and odor characteristics with those of standard compounds.

Quantitation of Selected Odor-Active Compounds in CSS
The odor-active compounds giving peaks in GC-MS chromatograms and having FD factors ≥32 in SE-SAFE isolates or FD factors ≥25 in SPME extracts were quantitated using the internal standard curve method; 2-octanol was used as an internal standard.
Firstly, a series of solutions of the mixture of internal standard and authentic compounds were prepared and analyzed by GC-MS under the conditions described in Section 2.4.2 except that selective ion monitoring mode was used. The standard curves were obtained by plotting the ratios of the peak areas of the authentic compounds relative to that of 2-octanol against their concentration ratios. Then 2-octanol (300 µL, 37.15 µg/mL) was added into 100 mL CSS, and its final concentration was 111.45 µg/L. The volatiles in CSS were extracted via SE-SAFE according to the method described in Section 2.3.1; the extracts were concentrated to 1 mL and analyzed by GC-MS. Finally, the concentrations of selected odor-active compounds in CSS were calculated on the basis of GC-MS analysis results and standard curves.

Odor Evaluation
SE-SAFE and SPME were used for isolating the volatile constituents from CSS. In order to confirm if the odorants contributing to the characteristic odor of CSS had been extracted, the odors of the isolates obtained were evaluated by three well-experienced evaluators. The results showed that both the liquid extract obtained by SE-SAFE and the fiber of SPME had the same overall aroma profile as CSS. They had caramel, cooked potato, smoky, sour and floral notes. The odor intensity of isolates obtained via SE-SAFE was stronger than that of SPME fiber. That is, the extraction methods used were appropriate.

Odor-Active Compounds Detected Using GC-O
The volatile isolates of six CSSs obtained via SE-SAFE and SPME were analyzed by means of GC-O; the odor-active regions were detected. To identify the structures of the odor-active compounds, their odor characteristics, mass spectra data and RIs were compared with the data obtained from the published literature and authentic standards. The results are listed in Table 1.
Ten ester compounds (2,5,6,8,10,11,28,33,41 and 53) were detected as odoractive compounds in six CSSs. Of the 10 ester compounds, nine esters were ethyl esters. All of them had been identified as volatile compounds in Chinese SS [9,12], Japanese SS [7], Thai SS [13] or Korean SS [14,15]; most of them had also been identified as odoractive compounds in SS; for example, ethyl propanoate, ethyl 2-methylpropanoate, ethyl butanoate, ethyl 2-methylbutanoate, ethyl 3-methylbutanoate and ethyl phenylacetate had been found in Japanese SS as aroma-active compounds [7,16]; ethyl acetate and ethyl propanoate had been identified as odorants in Chinese SS [8]. However, as odor-active compounds in SS, γ-dodecalactone and ethyl 2-hydroxy-4-methylpentanoate (EHMP) had not been reported. As SS volatiles, γ-dodecalactone was only identified in SS manufactured using Bacillus species and fused yeast [15], and EHMP only in Chinese SS [12] by MS. EHMP was a very important flavor compound; it occurred in fresh fruits, grape brandies, wines, etc. When this ethyl ester was mixed with C4−C10 alkanoic acids, it could enhance natural, ripe and tropical fruit flavors. It may have contributed greatly to the fruity odor of SS [17]. All of the esters identified were thought to be a result of two pathways. The first was the metabolism of yeasts. In the production process of SS, a variety of microorganisms, including yeast, lactic acid bacteria, Aspergillus oryzae, etc., were used. During SS fermentation, some esters were formed enzymatically through the metabolism of yeasts. The second pathway was the reaction of alkanol with organic acid during sterilization and storage; because the reaction was non-enzymatic catalysis, the reaction rate was slow, and the number of esters formed was less. The production of esters depended on many factors, such as aeration, concentrations of organic acids, alcohols and their precursors, etc. [18]. Nine carboxylic acids (23,29,31,35,37,40,42, 43 and 55), including four linearchain carboxylic acids, four branched-chain carboxylic acids and one aromatic acid, were identified as odorants in six CSSs; acetic acid, butanoic acid, 3-methylbutanoic acid and 3-methylpentanoic acid were the common substances in six samples. All of these organic acids have been reported as volatiles and odor-active compounds of GSS in the published literature [4,9,10,14], and they were formed as microorganism metabolic products. For example, the metabolism of lactic acid bacteria led to the production of acetic acid, propionic acid, butanoic acid, etc. [19]. The precursors of 2-methylpropionic acid, 3-methylbutanoic acid and phenylacetic acid were valine, leucine and phenylalanine, respectively; these acids could be produced as yeast metabolic products by transamination and decarboxylation oxidation [20].
Seven aldehydes (3, 12, 14, 21, 32, 36 and 54) were identified as odor-active compounds. 2(3)-methylbutanal and benzeneacetaldehyde belonged to Strecker aldehyde. Not only could they be formed through the Strecker degradation of isoleucine (or leucine) and phenylalanine, but also were derived from the corresponding amino acid catabolism by the Ehrlich pathway [20]. Hexanal, octanal, nonanal and (E,Z)-2,6-nonadienal were lipid-derived compounds. Soybean seeds contained more than 20% soybean oil, which contained monounsaturated and polysaturated fatty acids, such as oleic acid, linoleic acid, arachidonic acid, etc. [27]. These four aliphatic aldehydes could be derived from unsaturated fatty acid (UFA) through an oxidation reaction. Vanillin could be produced by microorganisms, such as bacteria, fungi, yeast or engineered microbial cells; its precursor was ferulic acid, present in the cell wall of wheat (6.6 g/kg), which was one of the materials of SS, or lignin, which exists in soybeans and wheat. The bioconversion of ferulic acid into vanillin occurs in both aerobic and anaerobic conditions [28].
Seven ketones (7,9,15,44,47,49 and 50) were identified as odor-active compounds; all of them have been found in GSS. There were two main pathways for the formation of 2,3-butanedione and 2,3-pentanedione. The first was that they were generated during the Maillard reaction. 2,3-butanedione was formed through the sugar degradation pathway, and its precursor was glucose. 2,3-pentanedione was produced by the sugar degradation pathway and through the further interaction of sugar degradation products with amino acids, and its precursors were glucose and L-alanine [29]. The second pathway was yeast fermentation. 2,3-butanedione was formed by decomposition of the α-acetolactic acid synthesized by yeast, and 2,3-pentanedione from α-aceto-α-hydroxybutyric acid [30]. 1octen-3-one, belonging to the lipid-derived compound, was formed via the autoxidation of UFAs [31]. The formation of both methylcyclopentenolone and maltol were associated with the Maillard reaction. Methylcyclopentenolone has been identified in volatile compounds of the glucose-tyrosine model system and the glucose-histidine model system [32], and maltol has been formed directly from the Amadori product which was the intermediate of the Maillard reaction [33]. Both DMHF and HEMF could be produced not only by the Maillard reaction but also could be biosynthesized by yeasts [2].
There were five alcohols (4, 13, 30, 38 and 46) identified as odor-active compounds. Ethanol, 3-methylbutanol and phenethyl alcohol were the metabolites of yeast; ethanol was formed by the EMP pathway and both 3-methylbutanol and phenethyl alcohol were derived from amino acid catabolism via the Ehrlich pathway [20]. 2-furanmethanol was a known thermal degradation product of ribose during the Maillard reaction. Linalool was identified in Japanese CSS, though not in Chinese CSS. It might come from the kombu, which is only used in Japanese CSS, because some kombu contains linalool [34].
Four phenols (45, 48, 51 and 52) were detected as aroma-active compounds; they could be formed by two pathways. Firstly, they were synthesized by different yeasts from some phenolic acids present in materials used for manufacturing SS, for example, 4-ethylphenol from p-coumaric acid and 4-ethylguaiacol from ferulic acid [35]. Secondly, they were produced by lignin pyrolysis; for instance, guaiacol and 2,6-dimethoxyphenol could be obtained from coconut shell pyrolysis [36]. Before wheat and soybeans were used for manufacturing SS, they were roasted. Lignin underwent pyrolysis, and some phenols were produced.
Four sulfur-containing compounds (1, 19, 24 and 39) were identified as odor-active compounds, and they were the common odorants in six CSS samples. Methanethiol was only detected in the isolate obtained via SPME; because its boiling point was about 6 • C, it was removed easily when the isolate obtained by solvent extraction was concentrated to recover the solvent. It arose from the degradation of methionine or cysteine derivatives. Dimethyl trisulfide came from the oxidation of methanethiol. Methional was a Strecker aldehyde, and it could originate from Strecker or microbiological degradation of methionine. Methionol was formed through the decarboxlation of 4-methylthio-2-oxobutyric acid, which was transamination product of methionine [20,37]. These four sulfur-containing compounds have been found in GSS.

The FD Factor of Odor-Active Compounds in Six CSSs
To screen more important odor-active compounds from 55 odorants identified in six CSSs, their FD factors were measured via GC-O, combined with AEDA. The results obtained are listed in Table 2.

Quantitation of the Odor-Active Compounds with FD Factors ≥32 or 50
To calculate OAVs, a total of 28 compounds with FD factors ≥32 (in SE-SAFE isolates) or ≥25 (in SPME isolates) were quantitated by constructing standard curves; the results gained are shown in Tables 3 and 4.
Of the 28 odor-active compounds, acetic acid had the highest concentration (57,948-406,726 µg/L) in all CSSs; the result was similar to Wang's data relating to odorants in GSS [10]. Four odorants, including ethyl butanoate, ethyl 2-methylbutanoate, ethyl 3methylbutanoate and octanal, had lower concentrations in six CSSs, and their values were less than 1 µg/L.
The odorants quantitated could be grouped into eight categories according to their chemical structures, that is, alcohols, carboxylic acids, esters, aldehydes, ketones, phenols, pyrazines and sulfur-containing compounds. Among these eight categories, the total concentrations of carboxylic acids in all of six CSSs were higher than those of the other seven categories, and the values ranged from 62,711 µg/L to 413,936 µg/L. The value of the total concentrations of all the quantitated odorants in J3 (542,622 µg/L) was the highest, and that in J1 (74,373 µg/L) was the lowest.

OAVs of Odor-Active Compounds in Six CSSs
To evaluate further the contributions of the 28 odor-active compounds to the aromas of the six CSSs and to screen for the key odorants, their OAVs were calculated based on their obtained concentrations and odor detection thresholds in water, and the results are shown in Table 5.
According to the OAV results, 27 odorants identified in different CSSs were further screened as key odorants contributing to the characteristic aroma of CSS. Except for EHMP, the other odor-active compounds had been identified as key odorants of GSS. Therefore, according to the results obtained, it was concluded that the key odorants of CSS should be same as those of GSS. The question of whether CSSs contain more nutritional components requires further study.