Characterization of Aroma-Active Components and Antioxidant Activity Analysis of E-jiao (Colla Corii Asini) from Different Geographical Origins

Abstract E-jiao (Colla Corii Asini, CCA) has been widely used as a healthy food and Chinese medicine. Although authentic CCA is characterized by its typical sweet and neutral fragrance, its aroma components have been rarely investigated. This work investigated the aroma-active components and antioxidant activity of 19 CCAs from different geographical origins. CCA extracts obtained by simultaneous distillation and extraction were analyzed by gas chromatography–mass spectrometry (GC–MS), gas chromatography–olfactometry (GC–O) and sensory analysis. The antioxidant activity of CCAs was determined by ABTS and DPPH assays. A total of 65 volatile compounds were identified and quantified by GC–MS and 23 aroma-active compounds were identified by GC–O and aroma extract dilution analysis. The most powerful aroma-active compounds were identified based on the flavor dilution factor and their contents were compared among the 19 CCAs. Principal component analysis of the 23 aroma-active components showed 3 significant clusters. Canonical correlation analysis between antioxidant assays and the 23 aroma-active compounds indicates strong correlation (r = 0.9776, p = 0.0281). Analysis of aroma-active components shows potential for quality evaluation and discrimination of CCAs from different geographical origins. Graphical Abstract

1 Introduction E-jiao (Colla Corii Asini, CCA), made of donkey-hide gelatin, rice wine, rock candy and soya bean oil, has been widely used as a functional food and an ingredient in traditional Chinese medicine (TCM) [1][2][3]. Together with ginseng and pilose antler, CCA has been recognized as one of the "Three Treasures for Nourishing" since ancient times [4]. In TCM, CCA has been mainly used as a clinical hematic antanemic therapy and to treat a variety of conditions, including bleeding, dizziness, insomnia and a dry cough, for more than 2000 years [5]. CCA has demonstrated various biological activities in modern pharmacological investigations, including hemostasis, anti-aging activity, anti-tumor, immunomodulatory, bone repairing, anti-inflammatory, anti-fatigue, etc. [6].
The quality and chemical composition of CCA depend on many factors, such as geographical origins, raw materials, water sources, temperature and processing. Among various factors, geographical origin is highly correlated with other factors and is known to confer specific healthy and nutritional qualities to CCA. Traditionally, CCA is named and evaluated by its place of origin and quality difference exists among CCAs from different producing areas, e.g., for a long time Dong-E county of Shandong province has been recognized as the origin of authentic and top-quality CCA products [7]. Therefore, it is necessary to develop reliable and effective analytical methods for quality evaluation of CCAs from different geographical origins.
As a complex mixture, various components have been separated and identified from CCA, including collagen, amino acids, proteins, polysaccharides, volatile substances, trace elements, etc. [8,9]. Authentic CCA is characterized by its typical sweet and neutral fragrance, which is one of the most significant factors that shape the quality and affect consumer behaviors. However, most of the current investigations on CCA have been devoted to the analysis and bioactive effects of the non-volatile substances [10,11] and the investigations on volatile components of CCA have been rarely reported. A series of methods can be used to obtain the volatile flavour compounds of CCA like distillation [12], Headspace-solid phase microextraction (HS-SPME) [13], solid-phase extract (SPE) [14], simultaneous distillation and extraction (SDE) [15,16] and so on. Mao et al. [17] studied the volatile components of CCA from Shangdong by gas chromatography-mass spectrometry (GC-MS) analysis of distilled samples. They identified 23 volatile ingredients and 12 of them have characteristic smells. However, the contributions of the specific components to CCA aroma and the composition variations of volatiles in CCAs from different producing areas were not investigated. Although CCA contains many volatile components, only a small fraction of them contributes to its overall aroma. Therefore, besides qualitative and quantitative analysis of volatile components, some complementary sensory techniques, such as gas chromatographyolfactometry (GC-O), are required to evaluate the aroma contributions of volatile components. Aroma extract dilution analyses (AEDA) is a frequently used method combined with GC-O, which has been used to identify the key aroma compounds in foods [18], beverage like mushroom [19,20], durian wine [21] and so on.
Recently, CCA has drawn continuously increasing attention due to its potential anti-aging and antioxidant effects. In 2012, Wang et al. [22] investigated the potential anti-aging effect of CCA using D-galactose (gal) induced aged model mice. Their results indicated that CCA could improve the antioxidant activities of superoxide dismutase (SOD), catalase (CAT) and glutathione peroxidase (GSH-Px), scavenge free radicals such as malondialdehyde (MDA), as well as modulate the expression of aged-related gene. Such investigations have proved the anti-aging and antioxidant effects of CCA and cast new light on its antiaging mechanisms. The general antioxidant activity analysis will bring an important criterion for quality evaluation of different CCA samples. However, to our knowledge, little information about the general in vitro antioxidant activity of CCA has been reported to enable quality evaluation and comparison among CCAs from different geographical origins. Therefore, the aim of the present study was: (1) to identify, quantify and attribute the aroma contributions of the volatile compounds in CCA using GC-MS/GC-O-AEDA techniques and (2) to evaluate the quality of CCAs from different producing areas by analysis of antioxidant activity and key aroma profile using chemometrics.

Collection of CCA Samples and Chemicals
Nineteen CCAs from different geographical origins were collected from the market branch of each producer. All the samples were stored in a cool, dark area before analysis of aroma components and antioxidant activities. Information concerning the geographical origins and brands of the 19 CCA samples is listed in Table 1. 1,1-Diphenyl-2-picrylhydrazyl (DPPH) was supplied by Sigma-Aldrich (Steinheim, Germany). 2,2-azino-bis(3ethylbenzothiazoline-6-sulphonic acid) diammonium salt (ABTS) was purchased from the Beyotime Institute of biotechnology (Nantong, China). 1-Decanol and standard aroma compounds were obtained from Sigma-Aldrich (Shanghai, China) and Apple Flavor and Fragrance Co.
Ltd. (Shanghai, China). The other chemicals and solvents used in this study came from Sinopharm Chemical Reagent Co. Ltd. (Shanghai, China). All of them were analytical reagents.

Extraction of the Volatile Compounds
The simultaneous distillation extraction (SDE) technique was used to obtain aromatic extracts of CCA. Twenty grams of CCA and 100 mL pure water were put into a 500 mL distillation flask which was placed in a heater fixed at 80°C. Dichloromethane (60 mL) was pipetted into a 100 mL distilled flask, which was placed in a water bath at 62°C. SDE was performed for 3 h. After extraction, 100 μL of internal standard (1-decanol) was added to the distillate. The extract was dehydrated by anhydrous sodiumsulfate, and then condensed to 200 μL under a gentle stream of pure nitrogen.

GC-MS, GC-Flame Photometric Detector (FPD) and GC-O analysis
The GC-MS analysis was performed using a Hewlett-Packard 7890A GC with a 5975C mass selective detector Working conditions were same to the GC-MS. Postcolumn flow was split at a ratio of 1:1 to the FID and the ODP using two deactivated and uncoated fused silica capillaries (106 cm 9 0.15 mm I.D., 139 cm 9 0.1 mm I.D.) at the end of the capillary. Each sample was run in triplicate. Sulfur compounds were detected using a flame photometric detector (FPD) (Agilent Technologies, USA) installed on the Agilent 7890A GC. Separations were accomplished using two different capillary columns as described above in the GC-MS and GC-O analysis. The oven temperature program used was the same as that employed for the GC-O studies. Identification of sulfur compounds was based on matching RIs from authentic sulfur standards and RIs reported in the literature with those observed in the samples.
The quantification was performed using the internal standard method. A set of standard mixtures, previously prepared and containing known concentrations of the chemical standards (C analyte ) and the I.S. concentration (C I:S: ), were analyzed and their peak areas (A analyte and A I:S: ) recorded. For each chemical standard, a six-point calibration line of relative peak area (A analyte =A I:S: ) versus C analyte was drawn to confirm a linear detector response, from which the amount of the analyte could be determined. The unknown concentration of an analyte (C analyte;X ) was calculated according to the interpolation of the calibration line as: where DRF analyte was the Detector Response Factor (DRF) for an individual analyte. The internal standard (1-decanol) using a DRF of 1.0.

Aroma Extract Dilution Analysis (AEDA)
The original aroma extracts were also analyzed by GC-O to evaluate the contributions of the identified volatile components to CCA odor. For AEDA, the concentrated aromatic extract (200 μL) of CCA was 1:2 diluted stepwise using dichloromethane as the solvent to obtain serial dilutions (1:2, 1:4, 1:8 and up to 1:64) of the original extracts. Sniffing of dilutions was continued until no odorant could be detected by GC-MS and GC-O. Each odorant was thus assigned a flavour dilution (FD) factor representing the last dilution in which the odorant was still detectable. The larger the FD factor of an aroma component, the higher the degree of its flavour activity [23].

Sensory Assessment of CCA
Sensory analysis of CCA samples was performed by a group of 12 members, including six females and six males between 22 and 40 years old. A CCA sample (2.00 g) was dissolved in pure water (10 mL) and then placed in a tasting opaque glass coded for a specific number with three digits. The temperature of samples was kept at 50 ± 2°C during sensory analysis. This assessment consisted of 8 aroma descriptors (toasted, muttony, medicinal, fatty, milky, earthy, sour and waxy). The intensity from each panelist was given a score of 0-9 where 0 = none or not perceptible intensity, and 9 = extremely high intensity. The intensity of samples was normalized to reduce the response differences among the panelists. Results were expressed as the median intensity of the sensory perceptions of the 12 panelists. All of the samples were evaluated in triplicate.

ABTS Radical Cation Scavenging Activity
The determination of ABTS radical scavenging was carried out as reported by Luo et al. [24] with slight modifications. The stock solutions included ABTS solution and oxidant solution. The working solution was prepared by mixing the two stock solutions in equal quantities and allowing them to react for 16 h at room temperature in the dark. The solution was then diluted by mixing 1 mL working solution with 90 mL ethanol (80%) in order to obtain an absorbance of 0.7 ± 0.05 at 734 nm. A fresh ABTS solution was prepared for each assay. Samples (10 μL) with a concentration range of 2.00-20.00 mg/mL were mixed with 200 μL of fresh ABTS solution and the mixture was left at room temperature for 6 min. The absorbance was then measured at 734 nm.
The ABTS radical scavenging activity of an unknown sample was calculated as follows: where A 1 is the absorbance in the presence of the test sample; A 2 is the absorbance of contrast and A 3 is the absorbance of the control (ABTS solution without test sample). The values of the half maximal inhibitory concentration (IC 50 ) were determined as reported above. Tests were carried out in triplicate.

DPPH Radical Cation Scavenging Activity
DPPH radical scavenging activity was determined using the method as described by Brand-Williams and coworkers [25] with slight modifications. Samples (100 μL) with a concentration range of 2.00-50.00 mg/mL were added to 100 μL DPPH solution. The mixture was stirred vigorously and allowed to stand at room temperature in the dark for 30 min. The absorbance of the resulting solution was measured at 517 nm. A lower absorbance of the reaction mixture indicates higher free radical scavenging activity, and vice versa for higher absorbance. Antioxidant activities of test compounds were expressed as IC 50 , which is defined as the concentration of the test material required to cause a 50% decrease in initial DPPH concentration.
The DPPH radical scavenging activity of each sample was calculated as the percent inhibition according to the following equation: where A 1 is the absorbance in the presence of the test sample; A 2 is the absorbance of contrast and A 3 is the absorbance of the control (DPPH solution without test sample). Tests were carried out in triplicate.

GC-O Dilution Results
Volatile components with high concentrations do not always play important roles in odor contribution, because the odor level of a volatile component was also linked with its threshold. GC-O can easily differentiate which compounds have nothing to do with odor and which are important aroma-active components although they were present at trace amount [26,27]. The Dong-E CCA was used as the typical sample to define the common aroma-active volatiles in the 19 CCAs. A total of 23 aroma-active compounds were identified by GC-O and AEDA as shown in Table 2. Three compounds were found to be the most potent aroma-active compounds among the 23 identified components, including s-methyl thioacetate, 2,6-dimethylpyrazine and 2-ethyl-3,6-dimethylpyrazine. They were detected as having the highest FD factors. The odour characteristic of s-methyl thioacetate was reported to be sulfurous and burnt. It is a well-known powerful odorant that contributes to the characteristic aromas of various foods like onion, garlic and radish, due to its low odor threshold. Compounds 2,6-dimethylpyrazine and 2-ethyl-3,6-dimethylpyrazine exhibited "roasted nut", and "roast" notes, respectively. Their highest FD factor values suggested they are likely to make the main contributions to the toasted aroma of CCA. Pyrazines are mostly formed through the Maillard reaction between saccharides and amino residues [28,29]. A higher temperature would benefit the Maillard reaction, and produce more pyrazines. Similar to pyrazines, a higher temperature also facilitates furan formation through nonenzymic browning of sugars. Several furans were identified in this study, such as 2-pentyl-furan and 5-methyl furfural. The presence and the contents of various pyrazines and furans can reflect the operation parameters of heating process during the manufacturing of CCA.
With an FD factor of 32, 5 odor regions were detected, including dimethyl disulfide (cabbage/putrid), hexanal (fatty/apple/cut grass), 2,3-dimethylpyrazine (nut/peanut butter/cocoa), 2,3,5-trimethyl pyrazine (roast/potato) and 2,4-decadienal (fatty/deep fried), respectively. Odor threshold values of aldehydes are generally low thus they have important potential effect to the overall aroma of CCA. Hexanal was another influential aroma compound identified in CCA, which is widespread, as it has already been found in many other foods and fruits. 2,4-Decadienal is a powerful compound that contributes to the fatty note of CCA.
Among the 65 identified volatiles, aldehydes are the most abundant chemical group found in CCA. A total of 13 aldehydes were identified in all CCAs. The S1 (5836.6 μg/ kg), S15 (5554.6 μg/kg) and S19 (3676.3 μg/kg) CCAs were characterized by the highest aldehyde contents. Aldehydes are generally characterized by intense sensory description by panelists associated with green, cut grass, citrusy and sweet notes. Hexanal, nonanal and benzaldehyde were the major components among aldehydes. (E)-2hexenal has been demonstrated to be a predominant compound in guava [31]. These unsaturated aldehydes are generally considered as the oxidized products of unsaturated fatty acids.
Eight ketones were found in the CCAs, including 2-octanone, 1-octen-3-one, and 2,3-octanedione. They contribute to the milky, waxy, earthy and fatty aroma. The highest concentration of ketones was found in S5 CCA (2597.3 μg/kg) and the lowest amount of ketone compounds was found in S11 CCA (7.5 μg/kg). Significant differences were observed in terms of the contents of ketones compounds.
Volatile sulfur compounds are typically in the parts per trillion (ppt) range to low parts per billion (ppb) range [32], however, they are normal components of wine, beer, cheese or cooked meat aroma, and can even exert a decisive role on some characteristic sensory descriptors [33]. As shown in Table 3, a total of 7 sulfur volatiles were detected in the 19 CCAs. Some aroma-active sulfur volatiles, including dimethyl sulfide, dimethyl disulfide, s-methyl thioacetate, ethanethioic acid s-ethyl ester, 3-methyl-2-butene-1-thiol, 4-mercapto-4-methylpentan-2-one and 3-(methylthio) propionaldehyde, have very strong aroma notes, which were described as possessing "cooked onion, sulfuric", "cabbage-like, rotten egg", "sulfurous, burnt", "sulfurous", "amine, smoke", "black currant" and "cooked potato" aromas, respectively. The primary sources for the sulfur-containing volatiles are the degradation of the sulfur-containing amino acids, methionine and cysteine, which have been found in CCA [34]. The relative levels of these amino acids in different CCA samples are probably responsible for the differences in the sulfur volatiles. Among the sulfur volatiles, four compounds, namely dimethyl sulfided, dimethyl disulfide, s-methyl thioacetate and 3-(methylthio) propionaldehyde, were observed in all the 19 CCA samples. The highest amount of sulfur compounds was found in S5 CCA (3.0 μg/kg), followed by S6 (2.6 μg/kg) and S19 (2.6 μg/kg). Sulfur compounds had been previously reported in CCA and they are extremely potent and produce only weak sulfur FPD peaks. At the normal attenuation their peaks would generally be overlooked and no further examination would be performed without GC-O analysis indicating strong aroma signals at the corresponding retention times. Therefore, GC-O is a useful complementary tool for analysis of some extremely low-content and potent aroma components.

Principal Components Analysis (PCA) and Correlation Analysis
In order to study the potential application of aroma analysis to the quality evaluation and discrimination of different CCAs, the contents of the 23 aroma-active components in the 19 CCAs were submitted to PCA. Considering the fact that the odor level of each aroma-active component depends on both its content and threshold, the contents of aroma-active components in the 19 CCAs were mean centered and scaled to have a unit sample standard deviation. The first three principal components (PCs) jointly explained 71.80% of the total data variance observed. There were at least 5 very significant PCs, which accounted for 43. 35, 17.25, 11.20, 6.44 and 6.26% of the total  variance, respectively. Moreover, 7 (92.46%) and 9 PCs (96.97%) were required to explain 90 and 95% of the total data variances, respectively. The large number of significant PCs indicate the factors influencing the aroma of CCAs from different producing areas were multiple and complicated. Figure 1a represents the corresponding loadings plot that established the relative importance of each odour-active compound and the relations between odour-active compounds and samples. Main volatiles positively correlated to PC1 were 3-(methylthio) propionaldehyde (17), 2,3,5-trimethyl pyrazine (13), dimethyl sulfided (1) and dimethyl disulfide (3). According to the loading plot, hexanal (5), 2-pentyl-furan (7), 2-ethyl-3,6-dimethylpyrazine (14) were positively correlated to PC2. Combined with Fig. 1b, it is obvious that hexanal (5), 2-pentyl-furan (7), 2-ethyl-3,6dimethylpyrazine (14) were important to the aroma of Dong-E CCA and the results are similar to those by AEDA assay. Figure 1b shows the scores scatter plot on PC1 and PC2. The score plot indicates that the CCA samples are not arranged in a random way but could be generally divided into three groups (Fig. 1b). It is obvious that the levels of odour-active compounds in CCAs from different regions are distinctively different. In this study, we found that the chemical character of the samples from the HeNan (S11), BeiJing (S18) and GanSu (S19) provinces of China was significantly different from any of the three groups owing to the unique geographical factors.
The squares of correlation coefficients (r 2 ) among the 19 CCAs were computed to investigate their similarity based on the 23 aroma-active components. The r 2 of each pair among the 19 CCAs was less than 0.8 and the two highest r 2 values were 0.7132 (S16 and S7) and 0.6018 (S14 and S6). The low similarity in terms of correlation coefficients among the 19 CCAs indicates that the contents of the 23 aroma-active components are specific to the producing areas of CCAs. Therefore, the profiles of aroma volatiles demonstrate the feasibility and potential for discriminating the producing areas of CCAs. Figure 2 is a summary of the total aroma grouped into eight regions (toasted, muttony, medicinal, fatty, milky, earthy, sour and waxy) based on aroma similarities by sensory analysis. It can be seen from Fig. 2 that the toasted, muttony and medicinal note characters are the major contributors to CCA aroma. The fatty, waxy and earthy notes are midlevel aroma attributes. Finally, the remaining two aroma categories, sour and waxy contribute to background CCA aroma. According to sensory analysis, the relative aroma strengths of the 8 aroma categories varied somewhat between the 19 CCA samples. The strongest toasted aroma was found in S17 (6.25) and S2 (2.40) demonstrated the lowest toasted aroma. Sulfur compounds did play a key role in muttony note and S9 (7.84) had the highest value and S7 (3.93) had the lowest value. As to the medicinal note, S19 (5.40) possessed the strongest aroma and S14 (3.11) is the weakest. S7 (5.50), S8 (5.53) and S3 (5.85) exhibited highest in the earthy, fatty and waxy aroma categories, respectively. The r 2 among the 19 CCAs were also computed based on the sensory characters of CCAs. The r 2 of each pair among the 19 CCAs was less than 0.8 and the two highest r 2 values were 0.7132 (S16 and S7) and 0.6018 (S14 and Aroma-Active Components and Antioxidant Activity Analysis of E-jiao 79 S6). The conclusion from sensory analysis were generally consistent with that of aroma-active components analysis, indicating aroma is an important quality parameter of CCAs and its attributes largely depend on the producing areas of CCAs. Because sensory analysis is subjective and is less stable compared with GC-MS and GC-O, the latter two techniques will provide a useful alternative tool for quality evaluation and discrimination of geographical origins for CCA.

Antioxidant Activities and Canonical Correlation Analysis
In the present study, the ability of test samples to scavenge ABTS is assessed on the bases of their IC 50 values, defined as an effective concentration at which the ABTS radical was scavenged by 50%. IC 50 values of CCAs S1-S19 were given in Table 4. A low IC 50 value indicates strong antioxidant activity in a tested sample. In this assay, the most potent radical scavenger was S19 (2.094), followed by S6, S18, S2, S17, S14, S15, S3, S4, S16, S1, S13, S12, S5, S10, S11, S9, S8 and finally S7(4.778). The ability of test samples to scavenge DPPH is assessed on the bases of their IC 50 too [35]. The effect of antioxidants on DPPH radical scavenging is generally due to their hydrogen-donating ability. The reduction capacity of DPPH radicals was determined by the decrease in its absorbance at 517 nm [36]. The 19 CCA samples differed in their DPPH radical cation scavenging capacities (Table 4). S19 (18.827) showed higher antioxidant capacity than other samples in terms of DPPH. The results were similar to that of ABTS assays. In brief, the DPPH radical cation scavenging capacities of the nineteen samples exhibited the descending order of S19, S16, S18, S2, S6, S14, S3, S15, S4, S1, S13, S17, S12, S5, S10, S11, S 9, S8 and S7.
The correlation coefficient between DPPH and ABTS assays was 0.9117, indicating the extraction of CCAs were well controlled and antioxidant activities analysis was reliable. To further investigate the relationship between aroma-active components and the antioxidant activities, canonical correlation analysis was performed between the antioxidant activities (DPPH and ABTS assay values) and the contents of 23 aroma volatiles. Considering the number of aroma-volatile variables (23) is much larger than that of antioxidant activities (2), only the first 9 PCs of aroma volatiles (explaining over 95% of the total variance) instead of the 23 raw variables were used in canonical correlation analysis to avoid overfitting. The largest canonical correlation coefficients were 0.9776 (p = 0.0281). The results of canonical correlation analysis indicated that the antioxidant activities of CCAs were highly correlated with the aroma volatiles and the latter could reflect the quality of CCAs.

Conclusion
In the present study, 65 major volatile compounds were identified in CCA by SDE/GC-MS/FPD analysis, and 23 potent odorants have been identified by GC-O analysis. Through AEDA the main aroma-active compounds identified as being responsible for the typical aroma in CCA were s-methyl thioacetate, 2,6-dimethylpyrazine and 2-ethyl-3,6-dimethylpyrazine.
Moreover, PCA was performed to show the distribution of CCA samples and the relationship between aroma-active compounds and CCA samples. According to the loading plot, hexanal, 2-pentyl-furan and 2-ethyl-3,6-dimethylpyrazine were representative aroma compounds of Dong-E CCA. The result is similar to that obtained by aroma extract dilution analysis (AEDA).
Finally, the antioxidant activities of 19 kinds of CCA were evaluated by DPPH radical scavenging activity and ABTS radical cation scavenging activity. Canonical correlation analysis has proved the relationship between antioxidant activities and the 23 aroma-active components.
Analysis of aroma-active components shows potential for quality evaluation and discriminaton of CCAs from different geographical origins. Future investigations will be focused on the influencing factors of CCA antioxidant activities.