Volatile Profile Characterization of Winter Jujube fromDifferent Regions via HS-SPME-GC/MS and GC-IMS

Institute of Food Science and Technology, Chinese Academy of Agricultural Sciences (CAAS)/ Key Laboratory of Agro-Products Processing, Ministry of Agriculture and Rural Affairs, 100193 Beijing, China Gembloux Agro-Bio Tech Department, University of Liége, 5030 Gembloux, Belgium Institute of Agro-Products Processing Science and Technology, Xinjiang Academy of Agricultural and Reclamation Science, 832000 Shihezi, China


Introduction
Winter jujube (Ziziphus jujuba Mill. cv. Dongzao) is a unique late-ripening jujube cultivar which originated in China [1]. It is widely distributed in China, including Hebei, Shandong, Shaanxi, Shanxi, and Xinjiang provinces after years of domestication and cultivation [2]. As a representative variety of fresh jujube, it is recognized by its delicious taste and pleasant aroma [1,3]. As a vital factor considered by consumers, sensory quality can be affected by aroma characteristics, which was correlated with VOCs. e VOCs of fresh winter jujube were composed by alkanes, alcohols, esters, and amines [4]. Geographic region and variety will cause the change of VOCs as reported for different varieties of red jujube [5,6]. ere were significant differences in contents of alcohols, acids, and aromatic compounds among different red jujube varieties [6]. However, there is no systematic study on the geographical differentiation of winter jujube, especially on the VOCs.
At present, headspace solid phase microextraction gas chromatography-mass spectrometry (HS-SPME-GC/MS) and electronic nose (E-nose) are widely applied for the analysis of VOCs in jujube [4,6]. HS-SPME-GC/MS combines the high separation ability of SPME and the superior identification ability of MS for VOCs [7]. E-nose distinguishes samples via the values of VOCs on different metal-oxide sensors. It defects in the identification of differential components [8]. In contrast to these analytical methods, IMS is a complementary way to detect compounds with low concentration (ppbv levels) with an advantage of fast response [9]. In IMS, vapors from the sample are firstly transferred into the ionization region by carrier gas. Secondly, the product ions formed from the interaction between neutral sample molecules and buffer gas, or bath gas molecules in the drift tube will be transported by the electric field into the separation region. irdly, the product ions with different drift velocities will reach the detector with different times. At last, the drift time will be used for the qualitative analysis [9]. GC-IMS combines the separation characteristics of GC and the fast response of IMS [10]. GC-IMS has been proved successful in regional identification and varieties classification based on differentiation of VOCs [11][12][13]. GC-IMS was also used to monitor the changes of VOCs in winter jujube during cold storage [14]. Besides, the combined analysis of GC-IMS and GC-MS has been successfully applied in the differentiation of VOCs from food matrices [15][16][17]. However, few investigations on the fingerprinting of volatile profiles in winter jujubes from different regions have been done by GC-IMS.
Principal component analysis (PCA) is an effective way to classify samples using unsupervised statics and has been widely employed in chemometrics and bioinformatics today [18]. Variable influence on projection (VIP) (greater than 1.0) has been applied to identify primary and specialized metabolites that are responsible for the discrimination of fruit and vegetables [19]. PCA and VIP scores would be effective for differentiation and selection of potential aroma maker of winter jujubes.
In this study, the geographic differentiation of winter jujubes was investigated by GC-IMS and E-nose based on VOCs. Furthermore, the potential markers of winter jujube from each specific region were identified and quantified by GC-IMS and GC-MS.

Winter Jujube Preparation and Chemicals.
Winter jujubes from eight different regions were purchased from China's famous jujube trading market (Beiyuanchun Jujube Market in Xinjiang, Xinzheng Jujube Market in Henan, and Cuierzhuang Jujube Market in Hebei). e information of jujube samples is shown in Table 1. 50 kg of winter jujube samples was collected from each region. Fresh fruits free from pests and diseases were washed, and pits were removed, sliced, and frozen by liquid nitrogen and then stored at −40°C for testing as soon as possible.

Sample Preparation.
About 100 g of frozen winter jujube slices was ground for 60 s with Joyoung juicer (JYL-CO20, Joyoung Co., Ltd., Shandong, China). Accurate 2.0 g of winter jujube pulp was put into a 20 mL vial sealed with a magnetic screw cap and septum before testing.

GC-IMS Analysis.
A FlavourSpec instrument (G.A.S. Gesellschaft für analytische Sensorsysteme mbH Dortmund, Germany) was used for GC-IMS analysis. Procedures were referred to previous studies with small modifications [20]. e sample vial was incubated at 50°C for 20 min, and then 500 μL of headspace was injected at 85°C in splitless mode. Tritium (6.5 keV) was used as the ionization source in this project. A FS-SE-54-CB-1 (15 m × 0.53 mm ID) column was used for separation at 60°C. Linear pressure program of the column was as follows: 2 mL/min for 2 min, ramped up to 10 mL/min over 8 min, then reached to 100 mL/min over 10 min, and at last got to 150 mL/min over 5 min. Nitrogen of 99.99% purity was the drift gas at a flow rate of 150 mL/ min and the drift tube was operated at 45°C. All standards were run under the same test procedure to supplement GC × IMS Library Search for qualitative analysis. e spectrogram was analyzed by Laboratory Analytical Viewer (LAV), where retention time and drift time were analyzed for the qualification of VOCs. Reporter plug-in was used to compare spectrogram differences among samples from the two-dimensional and three-dimensional view. Gallery Plot plug-in was used to compare the differences of volatile fingerprints visually. PCA plug-in was used for classification analysis.

E-Nose Analysis.
A commercial PEN 3.5 E-Nose (Airsense Analytics, GmBH, Schwerin, Germany) containing ten metal-oxide semiconductors was used to distinguish the overall flavor perception of winter jujubes from eight regions.
e detailed procedure of sample preparation was referred to Chen et al. [6]. Sample preparation was the same as that in GC-IMS analysis.

HS-SPME-GC/MS Analysis.
HS-SPME conditions and GC-MS analysis were referred to Chen et al. with small modifications [6]. e vials (the same as those in GC-IMS analysis) were equilibrated at 50°C for 40 min. Polydimethylsiloxane/divinylbenzene (PDMS/DVB) fiber was used for extraction and desorbed at 250°C for 3 min in splitless mode. DB-WAX capillary column (30 m × 0.25 mm × 0.25 μm) was used for separation. e oven temperature program was as follows: 40°C for 3 min, heating to 120°C (5°C/min), then rising to 200°C (10°C/min), and holding at 200°C for 5 min. Helium was the carrier gas at a flow rate of 1.0 mL/min. Electron Impact mode with the ion source temperature set at 200°C and the ionization energy of 70 eV was performed in the MS detector. e acquisition was full-scan mode and mass acquisition range was 35-550 m/z. e standards were qualitatively analyzed under HS-SPME-GC/MS analysis. Retention index (RI) of the compounds was calculated based on the retention time of C4-C20 alkanes mixture and used as additional support for the identification of compounds. A 2 mg/L solution of cyclohexanone as the optimized internal standard was used for the relative quantitative analysis of VOCs. NIST 17 database was used for identification of the compounds.

Determination of VOCs from Eight Regions of Winter
Jujube by GC-IMS. VOCs of winter jujube samples from eight regions of China were analyzed by GC-IMS, and the result is shown in Figure 1(a). e red baseline on the left is the reaction ion peak (RIP). RIP represents the total number of ions in the ionization chamber, described as H + (H 2 O)n (n represents the number of water molecules) [17]. e points on the right of RIP represent the signal (monomer, dimer, and even trimer) of volatiles extracted from the samples.
Most of the signals appeared in the retention time of 100-600 s and the drift time of 6.0-14.0 ms. e red color represents the higher signal intensity (higher concentration) of the substance, while the white color indicated weaker intensity (lower concentration) [21]. In IMS, the formation of product ions mainly depends on the affinity of analytes to   protons. Product ions could also be affected by the concentration of the analyte, the chemical properties, and the temperature of the drift tube [20].
Varieties of fruits have distinct aroma depending on the composition, concentration, aroma description, and odor threshold of VOCs [22]. e composition of VOCs varied in winter jujubes from different regions. In order to identify the specific differences and make a quantitative comparison, all peaks were identified and numbered for fingerprint comparison. e direct comparison of the components in each sample was shown in Figure 1(b). VOCs in the black frame were the common components among eight regions of winter jujubes. VOCs in the yellow and red frame were the ones present only in SXYC and XJAKS. VOCs in the green frame were the differential ones in eight regions of winter jujube. e corresponding data of retention time (RT) and drift time (DT) are presented in Table 2. e X-axis represented the code of the compound, corresponding to the information in Table 2.
Alcohols were formed by sugar catabolism and/or acid degradation in the food matrix [23]. Benzyl alcohol contributes to a slightly fruity aroma, while linalool, a monoterpenoid, contributes to floral, lavender, lemon, and rose flavor [23,24]. Aldehydes were thought to be mainly produced via lipid oxidation and decomposition, contributing most to the overall aroma among all categories because of relatively low odor thresholds. Hexanal derived from linoleic acid contributes to flavor of fruity, grass, and green, with a low odor threshold of 1.1 ng/L [25]. 2-Nonenal is a plant metabolite and derives from linoleate decomposition, with flavor of paper [25]. 2-Heptenal is a plant metabolite with soapy and fatty aroma and existed in white pomelo peel [26]. Acids like octanoic acid, hexanoic acid, and pentanoic acid might be related to cheese aroma [27]. Ethyl decanoate is present in fruits like cherry and pineapple with fruity aroma [26]. Styrene was reported to have sweet, balsamic, and almost floral odor that was extremely penetrating [28]. Esters endow fruity aroma for fruits and methyl myristoleate   Journal of Food Quality "RI cal" means RI calculated by the retention time of alkanes. "RI lit" means RI listed in the literature. Compounds with " * " mean variables with VIP score >1. "nd." means not detected. 10 Journal of Food Quality was found to be with aroma of honey and iris [29]. Mutual comprehensive effect of VOCs is the basis of the formation of the overall aroma of winter jujubes. Most of these common volatiles were found to be aroma attributes of fruity, grass, or green, which was in agreement with Pu et al. [4]. As is shown in Figure 2, PCA of GC-IMS data showed that SXYC and XJAKS were individually apart from the other winter jujubes. All the jujubes studied in this study were distributed in the middle latitude of the northern hemisphere, among which SXYC had the smallest latitude and XJAKS has the smallest longitude (Table 1). e special geographical location and associated climatic factors of SXYC and XJAKS might be the reason why the jujube aroma of these two regions was different from others. SXYC and XJAKS had their own specific VOCs that are framed with yellow and red color separately in Figure 1(b). e substances identified in the yellow frame, including acetone, 6methyl-5-hepten-2-one, 2-nitrophenol, 2-methoxyphenol, and 2-nitrophenetole, were the specific chemicals in SXYC. XJAKS was characterized by compounds in the red frame, including (E)-3-hexenoic acid, 1-penten-3-ol, ethyl hexanoate, 2-methyl-benzaldehyde, 2-hydroxy-benzaldehyde, trans-4-undecenal, methyl laurate, 2-pentylfuran, and heptanal. e volatile markers of SXYC and XJAKS would be confirmed by the correlation analysis between E-nose and the relative quantitative results of GC-MS for the discriminating components in the following parts.  Table 3. E-nose results were not specific information on sample composition but rather a simple fingerprint through pattern recognition. Figures 3(b), 3(g), and 3(i) show that VOCs of SNXDL, SDZH, HBHH, SDYT, XJKEL, and SXLY responded almost the same tendency on sensors W5S, W1W, and W2W. Aroma of SXYC had the highest response value on sensors of W1C, W3C, and W5C (Figures 3(a), 3(c), and 3(e)). XJAKS had the highest response value on sensors W5S, W1S, and W2S inferred from Figures 3(b), 3(f ), and 3(h). PCA results (Figure 4) based on Figure 3 showed that SNXDL, SDZH, HBHH, SDYT, XJKEL, and SXLY relatively clustered. SXYC and XJAKS were far away from others. e classification results of aroma characteristics obtained by E-nose (Figure 4) showed the same trend as those obtained by GC-IMS ( Figure 2). e classification results of the two technologies strongly indicated that GC-IMS and E-nose had advantages in rapid classification and accuracy. Although targeted and qualitative analysis of winter jujubes was performed well by GC-IMS, the identification and quantitation of potential contributed markers in different winter jujubes were further combined with GC-MS analysis. green, yellow, and red frame in Figure 1(b) of winter jujubes). e results are shown in Table 4. Partial least squares discriminant analysis (PLS-DA) assesses the relationship between a descriptor matrix X and a response matrix Y in a supervised way [19]. e differential variations were amplified to illustrate the relationships between the groups in biplot. Identification of the most important variables for the prediction ability of the PLS-DA model is generally performed by analyzing the regression coefficients. VIP scores provide information about the importance of each  variable used in the grouping model of PLS-DA [30]. VIP scores greater than 1.0 are always treated as the discriminating factor. PLS-DA has been applied to identify primary and specialized metabolites that are responsible for the discrimination of fruits and vegetables [31]. PLS-DA and biplot (Figures 5(a) and 5(b)) got from Table 4 showed the same tendency as the PCA results got from GC-IMS and E-nose (Figures 2 and 4). SNXDL, SDZH, HBHH, SDYT, XJKEL, and SXLY relatively clustered, while SXYC and XJAKS were apart from others. As shown in Figure 6, 1-peten-3-ol, acetone, 2-methoxyphenol, methyl laurate, 3-pentanone, 2-formyltoluene, and ethyl hexanoate were the components with VIP scores >1, indicating that they were crucial components for the discrimination of  winter jujubes from eight regions. e relationship between the differential VOCs (Table 4) and sensors (Figure 3) was analyzed by heatmap (Figure 7). e potential markers for SXYC and XJAKS were analyzed according to the VIP scores and heatmap.

Potential Markers
For SXYC, the sensors with higher response values (W1C, W3C, and W5C) were in close relationship with heptanal, acetone, 6-methyl-5-hepten-2-one, 2-methoxyphenol, and 2-nitrophenetole ( Figure 7). However, only acetone and 2-methoxyphenol had VIP scores over 1. Research showed that 2-methoxyphenol was a product of pyrolysis of lignin with woody odor. Acetone existed naturally in plants with pungent odor [32]. e quantitative results in GC-MS showed that 2-methoxyphenol and acetone accounted for 31.67% of the differential VOCs in SXYC (Table 4). eir concentration in SXYC were several times higher than that in other jujubes (Table 4). Hence, 2methoxyphenol and acetone might be the potential markers of SXYC with woody and pungent aroma.

Conclusion
In conclusion, the difference and similarity in VOCs of winter jujube from eight regions of China were well analyzed by GC-IMS, E-nose, and GC-MS. All results showed that SNXDL, SDZH, HBHH, SDYT, XJKEL, and SXLY clustered together with the differentiation of SXYC and XJAKS. As is shown in Figure 9, benzyl alcohol, octanoic acid, 2-hexenal, linalool, 2-nonenal, and ethyl decanoate were the common volatiles of winter jujubes from eight regions of China. e combined analysis of VIP scores, heatmap, and aroma description indicated that 1-penten-3-ol, ethyl hexanoate, methyl laurate, and 2-formyltoluene were the potential markers for green and fruity aroma profile of SXYC, and XJAKS could be labeled by acetone and 2-methoxyphenol with wood-like and pungent aroma.

Data Availability
Data are contained within this article.

Conflicts of Interest
e authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.