Comparison of Phenolic Compounds and the Antioxidant Activities of Fifteen Chrysanthemum morifolium Ramat cv. ‘Hangbaiju’ in China

This study investigated the phenolic compounds of 15 Chrysanthemum morifolium Ramat cv. ‘Hangbaiju’, including 6 ‘Duoju’ and 9 ‘Taiju’, using high performance liquid chromatography (HPLC). The antioxidant activities of these ‘Hangbaiju’ were estimated by DPPH, ABTS and FRAP assays. Results show that a total of 14 phenolic compounds were detected in these flowers, including 3 mono-caffeoylquinic acids, 3 di-caffeoylquinic acids, 1 phenolic acid and 7 flavonoids. ‘Duoju’ and ‘Taiju’ possess different concentrations of phenolic compounds, and ‘Taiju’ exhibits higher caffeoylquinic acids and stronger antioxidant activities than ‘Duoju’. Caffeoylquinic acids show a strong correlation with the antioxidant activities of the samples. Principal component analysis (PCA) reveals an obvious separation between ‘Duoju’ and ‘Taiju’, using phenolic compounds as variables. Apigenin-7-O-glucoside, 3,5-di-O-caffeoylquinic acid, luteolin and acacetin were found to be the key phenolic compounds to differentiate ‘Duoju’ from ‘Taiju’.


Sample Extraction
The 'Hangbaiju' samples were ground into powder by a Chinese medicine crusher (Shanghai Dianjiu Traditional Chinese Medicine Machinery Manufacturing Co. Ltd., Shanghai, China) and passed through a 60 mesh sieve. About 0.5 g sample weighed with the precision of 0.1 mg using an AB-135s scale (METTLER TOLEDO), which was mixed with 100 mL of methanol:water (70:30 v/v) in a glass tube, and sonicated at 50 • C for 40 min. Afterwards, the mixture was cooled to room temperature, and the final extract volume was brought up to 100 mL using the methanol:water solution [7]. The extract was filtered through a 0.22 µm PTFE membrane and stored at −20 • C for further analysis.

HPLC-DAD Analysis
A Waters E2695 HPLC coupled with a 2998 diode array detector (Waters, Milford, MA, USA) was used to analyze the phenolic compounds in the 'Hangbaiju' samples. A Phenomenex Luna C 18 column (250 mm × 4.6 mm, 5 µm, Torrance, CA, USA) was used to separate the phenolic compounds under a 0.8 mL/min flow rate. The mobile phase consisted of (A) acetonitrile and (B) 0.1% phosphoric acid in water (v/v). The sample injection volume was 10 µL, and the column was maintained at 35 • C during the elution program. The gradient was programed as follows, 0 to 11 min, 10% A to 18% A; 11 to 32 min, 18% A isocratic; 32 to 40 min, 18% A to 30% A; 40 to 48 min, 30% A to 35% A; 48 to 50 min, 35% A to 40% A; 50 to 55 min, 40% A isocratic; 55 to 60 min, 40% A to 70% A; and 60 to 70 min, 70% A to 10% A. Phenolic compounds were identified by comparing their retention time with corresponding external standards. The quantification was also conducted using the standards.

DPPH Assay
The antioxidant activity of the 'Hangbaiju' extracts and standard solutions using the DPPH assay followed the method of Turkoglu et al. [17] with minor modifications. Briefly, 20 mg of DPPH was dissolved in methanol in a 500 mL volumetric flask to a concentration of 0.101 M. The extract or standard solution (0.2 mL) was mixed with 3.8 mL of the DPPH solution. The mixture was well vortexed and kept at room temperature for 1 h. Afterwards, the absorbance of the mixture was recorded at 517 nm on a UV-5200PC spectrophotometer (Metash instrument, Shanghai, China). The reference was prepared by mixing 0.2 mL of 50% methanol solution with 3.8 mL DPPH solution. The DPPH radical inhibition rate of these samples was calculated using the equation below, where A ref and A sample were the absorbance of the reference and the sample, respectively. EC 50 represents the sample with its concentration that inhibits 50% of DPPH radicals. Each measurement was carried out in triplicate.

ABTS Assay
The ABTS assay was followed the method of Re et al. [18] with minor modifications. In brief, 176 µL of 140 mM potassium persulfate solution was mixed with 10 mL of 7 mM ABTS solution to yield ABTS working solution. The working solution was kept in the dark for 12 h. During the analysis, the ABTS working solution was diluted by ethanol to an absorbance value of 0.70 ± 0.02 at 734 nm on the UV-5200PC spectrophotometer. Afterwards, 0.1 mL extract or standard solution was mixed with 3.9 mL ABTS working solution and kept at room temperature for 6 min in the dark. The absorbance of the mixture was also recorded at 734 nm.
The 50% methanol solution (0.1 mL) mixed with 3.9 mL ABTS working solution was used as the reference. The ABTS radical inhibition rate of the sample was calculated using the equation below, where A ref and A sample were the absorbance of the reference and the sample, respectively. EC 50 was used to represent the inhibition activity of each sample against 50% of the ABTS radical. Each measurement was conducted in triplicate.

Ferric Reducing Antioxidant Power (FRAP) Assay
FRAP assay was carried out according to Benzie & Strain [19] with minor modifications. In brief, the FRAP working solution was prepared by mixing 0.1 M acetate solution (pH 3.6, acidic), 10 mM TPTZ solution and 20 mM ferric chloride solution at a 10:1:1 (v/v/v) ratio. Afterwards, 0.1 mL extract sample or standard solution was mixed with 3.9 mL FRAP working solution and vortexed. The resultant mixture was incubated at a 37 • C water bath for 10 min. After cooling to room temperature, the absorbance of the mixture was recorded at 593 nm. The reference was prepared by mixing 50% methanol with 3.9 mL FRAP working solution. Trolox was used as the external standard. The FRAP value of each sample was expressed as mg TEAC/g dry weight. Each analysis was performed in triplicate.

Statistical Analysis
All data were expressed as the mean ± standard deviation of triplicate tests. Analysis of variance (ANOVA) was carried out to investigate the significant difference among the means at a significant level of 0.05 using SPSS22.0 software (SPSS Inc, Chicago, IL, USA). Principal component analysis (PCA) (Metabo Analyst 4.0, http://www.metaboanalyst.ca) was carried out using phenolic compounds as variable to elucidate the similarity of 'Hangbaiju' samples.  The TMAC was found to be in the range of 2552.04 to 7402.83 µg/g DW in 15 'Hangbaiju' samples, in which the 'Duoju' samples had the content between 2552.04 and 4352.29 µg/g DW and the 'Taiju' samples between 2673.11 and 7402.83 µg/g DW. In comparison of the TMAC among 'Duoju', 'DJ6′ was higher than others (P < 0.05). The contents of 3-O-caffeoylquinic acid, 5-Ocaffeoylquinic acid and 4-O-caffeoylquinic acid in the 'DJ6′ were found to be 1167.59 µg/g DW, 2392.17 µg/g DW and 792.53 µg/g DW, respectively, which were also significantly higher than the other 'Duoju' (P < 0.05). It is worth noting that 5-O-caffeoylquinic acid was the dominant monocaffeoylquinic acid in the 'Duoju' samples. These results were consistent with a previous report [13].  The TMAC was found to be in the range of 2552.04 to 7402.83 µg/g DW in 15 'Hangbaiju' samples, in which the 'Duoju' samples had the content between 2552.04 and 4352.29 µg/g DW and the 'Taiju' samples between 2673.11 and 7402.83 µg/g DW. In comparison of the TMAC among 'Duoju', 'DJ6 was higher than others (P < 0.05). The contents of 3-O-caffeoylquinic acid, 5-O-caffeoylquinic acid and 4-O-caffeoylquinic acid in the 'DJ6 were found to be 1167.59 µg/g DW, 2392.17 µg/g DW and 792.53 µg/g DW, respectively, which were also significantly higher than the other 'Duoju' (P < 0.05). It is worth noting that 5-O-caffeoylquinic acid was the dominant mono-caffeoylquinic acid in the 'Duoju' samples. These results were consistent with a previous report [13]. Among the 'Taiju' samples, 'TJ3 had the highest TMAC (P < 0.05), with 2844.64 µg/g DW of 3-O-caffeoylquinic acid, 3139.04 µg/g DW of 5-O-caffeoylquinic acid and 1365.16 µg/g DW of 4-O-caffeoylquinic acid, respectively. 5-O-Caffeoylquinic acid was also the predominant individual mono-caffeoylquinic acid in the 'Taiju' samples. Different to 'Duoju', the 3-O-caffeoylquinic acid content in 'Taiju' was much higher than 4-O-caffeoylquinic acid (Tables 1 and 2).
It has been reported that 5-O-caffeoylquinic acid, 3,5-di-O-caffeoylquinic acid and luteoloside were the most crucial phenolic components in the chrysanthemum that influence the antioxidant and medicinal properties [7,13]. Therefore, it is important to compare the total content of these three key phenolic components. Among these 15 'Hangbaiju' samples, the TKPC was ranged from 7591.50 µg/g DW to 13,144.12 µg/g DW. In the 'Duoju' samples, 'DJ6 and 'DJ2 contained the highest TKPC. It was observed that 7 'Taiju' samples (TJ1, TJ3, TJ4, TJ5, TJ6, TJ8 and TJ9) had higher TKPC than any of the 'Duoju' samples. It should be noticed that 'TJ3 and 'TJ5 had similar TKPC values and were significantly higher than the other 'Taiju' samples (P < 0.05).

Antioxidant Activity
DPPH, ABTS and FRAP assays are often used simultaneously to evaluate the in vitro antioxidant activities of extracts or active substances [20][21][22]. Tables 3 and 4 show the antioxidant activity of these 'Hangbaiju' samples. In the DPPH assay, antioxidants work as hydrogen donors to react with stable DPPH free radical causing its discoloration [23,24]. The EC 50 of DPPH of these samples ranged from 1.69 mg/L to 3.04 mg/L. 'DJ2 exhibited the highest DPPH scavenging activity among the 'Duoju' (P < 0.05) samples, whereas 'TJ6 , 'TJ4 , 'TJ1 and 'TJ8 had the strongest DPPH scavenging capacity among 'Taiju' samples. In the ABTS assay, the stable-colored ABTS radicals are interact with antioxidants and this results in color loss [14,24]. In the present study, the 'Duoju' samples had the ABTS quenching EC 50 value of 2.13 mg/mL to 2.83 mg/mL, and between 1.82 mg/mL to 2.42 mg/mL for 'Taiju samples'. The highest ABTS quenching activity of 'Taiju' samples was found in 'TJ4 , whereas 'TJ1 , 'TJ5 , 'TJ6 and 'TJ8 possessed similar ABTS radicals scavenging capacity. FRAP assay is used to estimate the antioxidant activity of a compound through its capacity of reducing ferric into ferrous ions [14]. In this study, the 'Hangbaiju' samples had the FRAP value of 222.28 to 436.51 mg TEAC/g DW, and generally the 'Taiju' samples exhibited a higher reduction capacity than did the 'Duoju' samples. Among these 'Duoju' samples, 'DJ1 and 'DJ6 had the lowest and highest FRAP reduction capacity, respectively. The highest FRAP reduction ability among 'Taiju' was 'TJ8 (436.51 mg TEAC/g DW), and the 'TJ9 showed the lowest FRAP value.  It has been reported that more hydroxyl groups in the flavonoid molecular structure could enhance the antioxidant capacity [25]. Additionally, the acylation of caffeoyl group could improve the complexity of the caffeoylquinic acid, resulting in a stronger capacity of scavenging free radicals [26]. The antioxidant activities of individual phenolic components from these 'Hangbaiju' samples were also evaluated (Table 5). Hyperoside, luteolin, 4,5-di-O-caffeoylquinic acid, luteoloside, 3,4-di-O-caffeoylquinic acid, 3,5-di-O-caffeoylquinic acid, caffeic acid and 5-O-caffeoylquinic acid show similar DPPH radical scavenge capacity, but they are higher than that of 4-O-caffeoylquinic acid, 3-O-caffeoylquinic acid, apigenin, apigenin-7-O-glucoside and acacetin. However, the DPPH radical scavenging capacity of linarin was not detected.
It has been reported that phenolic compounds play a vital role to the antioxidant activity in C. morifolium [3]. A correlation study was conducted between the phenolic compounds and the antioxidant capacities of these 'Hangbaiju' samples (Table 6). It was observed that phenolic compounds in both 'Duoju' and 'Taiju' exhibit a good correlation with their antioxidant activities. For example, a positive correlation was established between 5-O-caffeoylquinic acid, TPAC and TKPC with the DPPH and ABTS quenching activity in the 'Duoju'. However, such a correlation was weak for 'Taiju'. In the FRAP assay system, TMAC and 5-O-caffeoylquinic acid showed a weak correlation with the FRAP reducing capacity. This indicates that phenolic acids including 5-O-caffeoylquinic acid, 3,5-di-O-caffeoylquinic acid and luteoloside might be the key compounds that contribute to the antioxidant activity of these 'Hangbaiju' samples. Data are the mean ± standard deviation of triplicate independent experiments. "-" represents 'not detected'. a, b, c, d, e, f, g, h, i, j, k Different letters in each raw indicate significant difference at a significant level of 0.05.

Principal Component Analysis (PCA)
PCA is commonly used to explain differentiation between samples and to obtain information on the variables that mainly influence the sample similarities and differences [27]. In order to differentiate these 'Hangbaiju', PCA was carried out using the detected phenolic compounds as variables. The first component (PC1) and second component (PC2) accounted for 72.8% and 15.3% of the total variance, respectively (Figure 2A). The 'Duoju' were aggregated together on the left downside of the score plot and segregated from the 'Taiju'. According to the loading plot ( Figure 2B), apigenin-7-O-glucoside and 3,5-di-O-caffeoylquinic acid had significant differences between the 'Duoju' and the 'Taiju' samples. Meanwhile, the content of luteolin in the 'Duoju' was about 3-4 times higher than that in the 'Taiju'. Therefore, luteolin also played a vital role in segregating 'Taiju' and 'Duoju' (Tables 1 and 2). It should be noted that acacetin was only present in some 'Duoju', but 'Taiju' contained a higher content of apigenin-7-O-glucoside than that of 'Duoju'. Therefore, acacetin and luteolin also affected the differentiation between 'Taiju' and 'Duoju' (Tables 1 and 2). the total variance, respectively (Figure 2A). The 'Duoju' were aggregated together on the left downside of the score plot and segregated from the 'Taiju'. According to the loading plot ( Figure 2B), apigenin-7-O-glucoside and 3,5-di-O-caffeoylquinic acid had significant differences between the 'Duoju' and the 'Taiju' samples. Meanwhile, the content of luteolin in the 'Duoju' was about 3-4 times higher than that in the 'Taiju'. Therefore, luteolin also played a vital role in segregating 'Taiju' and 'Duoju' ( Table 1 and Table 2). It should be noted that acacetin was only present in some 'Duoju', but 'Taiju' contained a higher content of apigenin-7-O-glucoside than that of 'Duoju'. Therefore, acacetin and luteolin also affected the differentiation between 'Taiju' and 'Duoju' ( Table 1 and Table 2).

Discussion
Phenolic acids and flavonoids are two groups of active substances in 'Hangbaiju', especially the caffeoylquinic acid. Although 5-O-caffeoylquinic acid and 3,5-di-O-caffeoylquinic acid play an absolute dominant role in mono-caffeoylquinic acids and di-caffeoylquinic acids, respectively. Monocaffeoylquinic acids can be converted to each other, as well as di-caffeoylquinic acid [28][29][30], therefore total mono-caffeoylquinic acid contents (TMAC) and total di-caffeoylquinic acid contents (TDAC) were used, rather than the content of single mono-caffeoylquinic acid or di-caffeoylquinic acid.
The flavonoid contents in C. morifolium rose during the flowering period and then declined; the general trend was a "peak shape" degree of flower openness [3]. The "peak shape" course may relate to the key enzyme in flavonoid biosynthesis -chalcone isomerase (CHI). During early florescence, CHI activity was enhanced, and the flavonoids gradually accumulated; at the end of florescence, CHI activity was inhibited, and the flavonoids gradually decreased [6,31]. 'Taiju' are the dried flower heads with opened ray florets but closed tubular florets, and CHI activity was enhanced, and the flavonoids gradually accumulated. But 'Duoju' represent the dried flower heads harvested with their ray florets and tubular florets being fully opened, so CHI activity was inhibited, and the flavonoids gradually decreased. Most contents of flavonoids in 'Taiju' are higher than in 'Duoju'.
It has been required by the Chinese Pharmacopoeia that the content of 5-O-caffeoylquinic acid, 3,5-di-O-caffeoylquinic acid and luteoloside in C. morifolium should be above 0.2 g/100 g DW, 0.7 g/100 g DW and 0.08 g/100 g DW to exhibit strong antioxidant activity and medicinal property [7]. In the present study, three 'Duoju' samples, namely 'DJ2′, 'DJ5′ and 'DJ6′, had more than 0.2 g/100 g DW

Discussion
Phenolic acids and flavonoids are two groups of active substances in 'Hangbaiju', especially the caffeoylquinic acid. Although 5-O-caffeoylquinic acid and 3,5-di-O-caffeoylquinic acid play an absolute dominant role in mono-caffeoylquinic acids and di-caffeoylquinic acids, respectively. Mono-caffeoylquinic acids can be converted to each other, as well as di-caffeoylquinic acid [28][29][30], therefore total mono-caffeoylquinic acid contents (TMAC) and total di-caffeoylquinic acid contents (TDAC) were used, rather than the content of single mono-caffeoylquinic acid or di-caffeoylquinic acid.
The flavonoid contents in C. morifolium rose during the flowering period and then declined; the general trend was a "peak shape" degree of flower openness [3]. The "peak shape" course may relate to the key enzyme in flavonoid biosynthesis-chalcone isomerase (CHI). During early florescence, CHI activity was enhanced, and the flavonoids gradually accumulated; at the end of florescence, CHI activity was inhibited, and the flavonoids gradually decreased [6,31]. 'Taiju' are the dried flower heads with opened ray florets but closed tubular florets, and CHI activity was enhanced, and the flavonoids gradually accumulated. But 'Duoju' represent the dried flower heads harvested with their ray florets and tubular florets being fully opened, so CHI activity was inhibited, and the flavonoids gradually decreased. Most contents of flavonoids in 'Taiju' are higher than in 'Duoju'.
Linarin (acacetin-7-O-rutinoside), luteoloside (luteolin-7-O-glucoside) and apigenin-7-O-glucoside are the glycoside of acacetin, luteolin and apigenin, respectively. Compared with apigenin, luteolin has a hydroxyl at the 3 position, and compared with linarin, the 4 hydroxyl of apigenin was methylated. Bobilya suggested that glycosidation and methylation of the hydroxyl group results in a decreased antioxidant activity of flavonoids [32]. Numerous studies have shown that the increase of hydroxyl groups in flavonoids can increase their antioxidant activity, especially the formation of o-dihydroxyl groups at the 3 and 4 positions [32][33][34]. Our study also echoes this claim. In the three antioxidant systems, luteolin showed higher antioxidant activity than apigenin, followed by acacetin, and luteoloside had higher antioxidant activity than apigenin-7-O-glucoside and linarin.

Conclusions
In conclusion, phenolic compounds and antioxidant capacity were analyzed and compared in 15 'Hangbaiju', including 6 'Duoju' and 9 'Taiju'. The compositions of phenolic compounds were significantly different between 'Duoju' and 'Taiju'. 'Taiju' contained a higher content of caffeoylquinic acids and higher antioxidant activities than 'Duoju', suggesting that the flowers harvested earlier had higher phenolic contents and antioxidant activity. A correlation study indicated that phenolic compounds, especially caffeoylquinic acids, played a vital role to the antioxidant capacity. Principal component analysis indicates that the difference on the phenolic composition could be used to differentiate 'Duoju' from 'Taiju' of C. morifolium.