Characterization and Evaluation of Aroma Quality in Doubanjiang, a Chinese Traditional Fermented Red Pepper Paste, Using Aroma Extract Dilution Analysis and a Sensory Profile

Doubanjiang, a Chinese traditional fermented red pepper paste, is eaten worldwide for its unique flavor. The objective of this study was to evaluate the aroma quality of doubanjiang using solvent-assisted flavor evaporation (SAFE) and headspace solid-phase microextraction (HS-SPME) coupled with gas chromatography-olfactometry (GC-O) and aroma extract dilution analysis (AEDA). A total of 165 volatile compounds, belonging to 13 chemical classes, were identified. Esters and hydrocarbons were the predominant groups. Thirteen aroma-active compounds were detected by AEDA of SAFE and HS-SPME, and their odor activity values (OAVs) were calculated by dividing their concentration by their odor threshold in water. Among them, ethyl isovalerate, β-damascenone, 3-isobutyl-2-methoxypyrazine (IBMP), and sotolone had the highest OAVs (>1000). In addition, sotolone, methional, β-damascenone, 3-isobutyl-2-methoxypyrazine, ethyl isovalerate, phenylethyl alcohol and linalool had high flavor dilution (FD) factors. Sotolone, β-damascenone and 3-isobutyl-2-methoxypyrazine were identified for the first time in doubanjiang and played significant roles in its aroma quality.


Introduction
Doubanjiang is a traditional fermented red pepper paste with a Chinese protected geographical indication (PGI) that is produced with red pepper (Capsicum annuum L.), broad bean (Vicia faba L.), wheat flour, and salt [1]. It has been appreciated as a uniquely flavored seasoning for several centuries in China [2,3] and has become popular worldwide [4,5]. In general, the traditional manufacturing process consists of three phases: (1) fermentation of broad beans with salt (12-14% w/w) to make doubanjiang-meju [6], (2) fermentation of red peppers (approx. 1-2 cm) with salt (14-16% w/w) to yield red pepper moromi, and (3) aged fermentation for more than six months in the natural environment of a mix of doubanjiang-meju with red pepper moromi at a ratio of 4:6 to improve flavor and taste characteristics. The average of annual production value of doubanjiang is $1.5 billion with the export value exceeding $100 million, according to local government statistics in 2016.
The volatile composition of doubanjiang is quite complex. Several studies on the identification and quantitation of volatile compounds in doubanjiang have been reported [3,[7][8][9]. However, the key

Volatile Compounds Identified in Doubanjiang
Volatile compounds in doubanjiang were extracted using SAFE and HS-SPME and identified using GC-MS. A total of 168 volatile compounds belonging to 13 chemical classes were identified. As shown in Table 1, hydrocarbons and esters were the predominant groups, each containing 30 and 26 volatile compounds, respectively, followed by alcohols, ketones, and acids, with each group including 27, 19, and 13 volatile compounds, respectively. For the SF-A sample, the most abundant volatile group (>5% average relative areas) was alcohols (17.86%), followed by phenols (10.08%) and esters (10.02%), while for the XH-B and XH-C commercial samples, esters (9.41% and 15.04%, respectively) and alcohols (6.14% and 7.01%, respectively) were major components ( Table 2). This difference might be due to the different manufacturing conditions, as the SF-A sample was produced by a different company than XH-B and XH-C.   585 µg), an internal standard compound, were used as relative quantification analysis. Mean ± standard deviation.
β-Damascenone, exhibiting fruity and sweet notes, is an important odorant in modern perfumery and a major contribution to wine aromas [22]. β-Damascenone has a very low odor threshold in water (2 ng/L) ( Table 4) and may be formed by acid-catalyzed hydrolysis of certain plant-derived metabolites under heat treatments and low pH during the manufacturing process [23]. Moreover, synergistic phenomena have also been identified, for instance, reducing the orthonasal detection thresholds of linalool and raising the detection threshold of 3-isobutyl-2-methoxypyrazine in the presence of β-damascenone [22]. To our knowledge, this is the first time that β-Damascenone is identified in doubanjiang. 3-Isobutyl-2-methoxypyrazine has been reported in bell and chili peppers, grapes and wines, and fermented red pepper food such as gochujang, a Korean fermented red pepper paste [15]. The odor threshold of IBMP has an extremely low value of 2 ng/L in water, contributing to a characteristic bell pepper, green pepper and gochujang-like aroma [15]. l-Leucine was speculated to be a precursor in regulating IBMP biosynthesis in grape berry [24]. 3-Isobutyl-2-methoxypyrazine was also identified for the first time in doubanjiang to our knowledge.
Sotolone, β-damascenone and 3-isobutyl-2-methoxypyrazine were identified for the first time in doubanjiang and validated by the mass spectrum, RI (DB-wax and/or DB-5ms columns), aroma properties and authentic compounds. Odor activity values (OAVs) were used to estimate the contribution of individual aroma compounds to the overall flavor of doubanjiang. As shown in Table 4, the concentrations of the selected key aroma volatiles exceeded their estimated odor threshold, suggesting that they may be responsible for the doubanjiang aroma. Ethyl 3-methylbutanoate showed the highest OAV, exceeding its threshold (0.01 µg/L in water) by a factor of 6229, followed by β-damascenone, 3-isobutyl-2-methoxypyrazine and sotolone, with OAVs of 3345, 2490 and 1602, respectively. These results correlated well with the conclusion that ethyl 3-methylbutanoate had a high OAV in doubanjiang [3]. β-Damascenone was the key odorant in red wine and Chinese liquor, with a high OAV, and significantly imparted the overall aroma of Chinese liquor [16]; sotolone showed the highest OAVs in soy sauce [19].

The effects of Selected Key Odorants on the Sensory Profiles of Doubanjiang
Sensory evaluation tests were used to access the effect of selected key odorants on the overall aroma of doubanjiang. The effects of spiking certain volatiles alone on flavor descriptors are shown in Table 5. Significant difference was observed compared with the unspiked control, which further confirmed that these key active-aromas played an important role in the flavor of doubanjiang. Doubanjiang has a quite complex food matrix, from which flavor release not only depends on the nature of the aroma compounds but also the interaction between sensory attributes of aroma compounds or with the nonvolatile constituents of the food matrix. For example, spiking acetic acid, 3-methylbutanoic acid and methional separately from doubanjiang did not impart the overall aroma, although methional has a relatively high OAV (160), whereas adding sotolone to doubanjiang not only imparted a caramel-like aroma but also increased the fruity aroma. Table 5. Effects of key aroma-active compounds on doubanjiang sensory attributes.

Doubanjiang Samples
Three commercial doubanjiang samples (labeled as SF-A, XH-B, and XH-C), manufactured by two doubanjiang companies located in Pixian city (Sichuan province, China-Latitude 30.79, Longitude 103.89), were used in this study. SF-A was produced in 2015 (aged 3 years), while XH-B and XH-C were produced by the same company in 2014 (aged 2 and 5 years, respectively). All three samples were collected from the company directly. The label on these products indicated that they contained ca. 47.8% red pepper, 26.1% broad beans, 5% wheat flour, and 16.7% salt.

Solvent-Assisted Flavor Evaporation (SAFE)
Five gram of each doubanjiang sample was mixed with 100 mL double distilled water (ddH2O), shaken at 250 rpm for 1.5 h, and filtered through a filter paper (Whatman, United Kingdom) under vacuum using a Büchner funnel. Volatiles from the total filtration containing internal standard (3-heptanol, 48.585 µg) were isolated using a solvent-assisted flavor evaporation (SAFE) unit (ACE Glass Inc., Vineland, NJ, USA) at 25 • C for 1.5 h under vacuum (10 −6 torr), according to the Seo et al. [25] method with some modifications. After distillation, the extract was obtained by sequential extraction with dichloromethane solvent (15 mL, 15 mL, and 20 mL) at 250 rpm for 1 h. After extraction, removal of the water was carried out by freezing at −20 • C overnight and followed by dehydrating over anhydrous sodium sulfate. A gentle N 2 stream method was used to concentrate the extract to 200 µL. Extractions were performed in duplicate.

Headspace Solid-Phase Microextraction (HS-SPME)
The volatile constituents of doubanjiang were extracted using two fibers, namely 75 µm carboxen/PDMS (CAR/PDMS) and 50/30 µm DVB/CAR/PDMS (Supelco Co., Bellefonte, PA, USA) with ca. 80% and 60% global perception similarity to doubanjiang, respectively, according to the direct GC-O method [15,17]. Briefly, 5.0 g of doubanjiang sample mixed with 10 mL of deodorized distilled water (DDW) was placed into a 20-mL vial, and balanced at 40 • C for 30 min in a water bath under stirring. The SPME fiber was then inserted into the vial and exposed to the headspace to extract volatiles at 40 • C for 30 min. After extraction, the fiber was inserted into the injection port of gas chromatograph (GC) for thermal desorption with the same parameters described by our previous method [26]. Extractions were performed in triplicate.

GC-Mass Spectrometry (GC-MS) Conditions
An Agilent 6890N gas chromatograph coupled with a mass spectrometer (MS 5973N, Agilent Technologies, Santa Clara, CA) was used in this study. First, one µL of SAFE extracts or SPME fiber extracts was injected in splitless mode onto both DB-wax column (60 m length, 0.25 mm i.d., 0.25 µm film thickness, J&W Scientific, Folsom, CA, USA) and DB-5ms column (60 length, 0.25 mm i.d., 0.25 µm film thickness, J&W Scientific, Folsom, CA, USA), respectively. Helium was used as the carrier gas with flow rate of 1 mL/min. The temperature program parameters of mass spectrometer were the same as our newly published paper [26].

Gas Chromatography-Olfactometry (GC-O)
The GC-olfactometry (GC-O) system was composed of a Varian 3800 GC (Varian Instrument Group, Walnut Creek, CA) coupled to a sniffing port (ODO II, SGE International, Ringwood, Australia) and a flame ionization detector (FID). Analytical conditions were performed following our published paper [27]. In brief, the conditions of volatile compounds extracted by both SAFE and HS-SPME were the same as GC-MS. Two different columns, namely, DB-wax column (30 m length, 0.25 mm i.d., 0.25 µm film thickness, J&W Scientific, Folsom, CA, USA) and DB-5ms column (30 length, 0.25 mm i.d., 0.25 µm film thickness, J&W Scientific, Folsom, CA, USA),respectively, were used for GC-O. Each SPME fiber was injected into the GC injector for 10 min. The GC conditions were the same as GC-MS: oven temperature program was from 40 • C to 200 • C at a rate of 5 • C/min with initial 5 min and final 10 min, respectively. The flow rate of the carrier gas helium was 1.4 mL/min. Injector and detector temperatures were 280 • C and 250 • C, respectively. GC-O was performed in triplicate.
To determine the most aroma-active components, aroma extract dilution analysis (AEDA) was carried out by serially diluting (1:3) the SAFE extracts using the same diluent. The flavor dilution (FD) factor is determined as the highest dilution where a volatile odor can still be perceived [27]. The HS-SPME extracts were stepwise diluted (10 mL DDW mixed with 5 g, 2.5 g, 1.25 g and 0.625 g of doubanjaing, respectively, were put into a 20-mL vial) and then balanced at 40 • C for 30 min. The sample dilution (SD) factor is determined as the highest sample dilution where an odorous substance can be perceived [28].

Compound Identification, Quantitation of the Key Odorants and Calculation of Odor Activity Values (OAVs)
Compounds were identified based on a comparison of the WILEY 7.0 mass spectral library, retention indices (RIs) and authentic standard compounds [26]. The retention indice (RI) values were calculated on DB-wax and DB-5ms columns using C 10 -C 22 and C 8 -C 20 as the external reference, respectively.
The selective ion monitoring (SIM) mode was used to quantitate the concentration of a compound in the SAFE concentrate mentioned above with 8-point external standard curves. The identified key aroma compounds were confirmed by authentic chemicals, and the standard curve for individual aroma compounds was obtained by calculating the response factor of known concentrations on GC-MS. The unit of concentration in the SAFE concentrate (µg/L) was converted to micrograms per kilogram (µg/kg, doubanjiang) according to the weight of sample used. For each compound, the OAVs were calculated by dividing the compound concentration in the samples by the mean values of its estimated odor threshold. Relative quantification measurement was also performed for all the volatile compounds, using 3-heptanol as the internal standard.

Sensory Evaluation
To access the effects of selected key aroma-active compounds on the overall aroma of doubanjiang, sensory evaluation was performed according to a previous method [11] by eight panelists (4 females, 4 males, aged 22-38) recruited from the Institute of Agro-products Processing Science and Technology, Sichuan Academy of Agricultural Sciences. Briefly, a total of 8 attributes containing ethanol (alcohols), acetic acid (sour), methional (cooked potato), 4-ethyl-2-methoxyphenol (burnt), phenylacetaldehyde (sweet), soy sauce (caramel-like), and ethyl acetate (fruity) [11,29] were used to characterize the sensory properties of the doubanjiang samples. The doubanjiang was spiked with 2-3 times concentration of certain key aroma-active compounds, and the panelists were asked to evaluate the odor intensity compared to unspiked samples at room temperature (25 ± 2 • C) using a line scale of 0-9, in which 0 was the lowest intensity without perception and 9 was the highest intensity. All the panelists had experience in the theory and practical application of doubanjiang, and could differentiate differences in doubanjiang's flavor.

Statistical Analysis
Univariate statistical analysis (one-way ANOVA) was used to determine significance between spiked and unspiked individual key aroma-active compounds using the statistical software OriginPro 8 (OriginLab Corporation, Northampton, MA, USA).

Conclusions
In summary, a total of 165 volatile compounds were identified using SAFE and/or HS-SPME. Thirteen most important odorants were detected, including eight by SAFE-GC-O and eight by HS-SPME-GC-O; of these, five aroma-active compounds were identified by both methods. Sotolone, β-damascenone and 3-isobutyl-2-methoxypyrazine, contributing to caramel-like, fruity, and red pepper aromas, respectively, were identified for the first time in doubanjiang. These compounds played key roles in the aroma quality of doubanjiang. In addition, methional, ethyl 3-methylbutanoate, phenylethyl alcohol, 2-methoxyphenol, 3-methylbutanal, and phenylacetaldehyde also played important roles in the aroma quality of doubanjiang. These results will be useful to improve the aroma quality of this traditional food.