Identification and Quality Assessment of Chrysanthemum Buds by CE Fingerprinting

A simple and efficient fingerprinting method for chrysanthemum buds was developed with the aim of establishing a quality control protocol based on biochemical makeup. Chrysanthemum bud samples were successively extracted by water and alcohol. The fingerprints of the chrysanthemum buds samples were obtained using capillary electrophoresis and electrochemical detection (CE-ED) employing copper and carbon working electrodes to capture all of the chemical information. 10 batches of chrysanthemum buds were collected from different regions and various factories to establish the baseline fingerprint. The experimental data of 10 batches electropherogram buds by CE were analyzed by correlation coefficient and the included angle cosine methods. A standard chrysanthemum bud fingerprint including 24 common peaks was established, 12 from each electrode, which was successfully applied to identify and distinguish between chrysanthemum buds from 2 other chrysanthemum species. These results demonstrate that fingerprint analysis can be used as an important criterion for chrysanthemum buds quality control.


Introduction
Chrysanthemums, colloquially known as mums, are herbaceous perennial flowering plants and have been cultivated for over 3 millennia. Chrysanthemums include more than 3000 varieties [1], including Ammobium alatum, perennial chamomile, Aster novi-belgii, and Calendula officinalis, which come from different regions, flower in different seasons, and may contain different active compounds. Chrysanthemum buds are one of the highest grades of chrysanthemum in use. Chrysanthemum buds are an important component in many traditional Chinese medicine (TCM) formulas [2] for its therapeutic effects, which include antioxidant, anti-inflammatory, antiviral (including HIV), antimutagenic, anticarcinogenic, antihepatotoxic, and antiaging activities [3]. Chrysanthemum buds are also a common health food/supplement used by many consumers [4] for "scattering cold, " "cleaning heat and toxin, " and "brightening eyes, " which are considered beneficial to human health.
Significant amounts of biologically active compounds have been found in chrysanthemum buds that play important roles in human body, mainly including flavonoids, carbohydrate, and essential oils [5]. Among these compounds, chlorogenic acid, luteolin, and glucoside have been confirmed to possess a variety of biological activities [6]. Traditionally, these active components were used to evaluate the quality of raw plant material. However, owing to the fact that there are hundreds of complex active components in chrysanthemum buds, it has been extremely difficult to identify all these substances and carry out quantitative analyses on them individually. What is more, the chemical composition of chrysanthemum buds can differ between different varieties. As a result, it became necessary to develop a new technology to capture the total chemical composition of chrysanthemum buds and to identify chrysanthemum varieties and verify their authenticity.
Fingerprinting is a method to capture total chemical information of herbs by chemical analytical techniques and is displayed as spectrograms, electropherograms, and other graphs. Fingerprint analyses produce a representative "fingerprint" that contains the greatest amount of information possible to accurately represent a sample and distinguish it from others. Fingerprint analysis of medicinal herbs has been the optimal measurement for identifying and assessing the variety and quality of the plants. Fingerprint analysis has been accepted as a strategy for the assessment of herbal 2 Journal of Analytical Methods in Chemistry medicines for the evaluation of medicinal products for herbal preparations by the U.S. Food and Drug Administration (FDA) [7] and the European Medicines Agency [8]. In China, the former State Drug Administration (SDA) also began to develop fingerprints of raw materials as a standard of quality control in 2000 [9].
Recently, several techniques have been developed which can characterize the nature and chemical composition of substances. HPLC [10] and GC [11], prime techniques used for fingerprint analysis, have high precision, sensitivity, and reproducibility. However, sample preparations, including preconcentration and derivatization, are often timeconsuming, complicated, and troublesome. Thin layer chromatography (TLC) [12] is a commonly used technique for screening of herbal liquid extracts. The ultra-performance liquid chromatography (UPLC) [13] approach has some advantages over HPLC, GC, and TLC, including a large decrease in analysis time and solvent consumption, the possibility of obtaining high efficiencies, and the ability to resolve coeluting compounds. However, its drawbacks include increased back-pressure and the availability of only few stable stationary phases. CE technology has been widely applied to the characterization of diverse samples due to its low cost, minimal sample volume requirement, short analysis time, and high separation efficiency [14][15][16]. Electrochemical detection (ED) is a commonly used chemical detection method because of the small size of both the detector and control instrumentation and low power demands [17]. CE coupled to electrochemical detection (CE-ED) is a useful technology offering high sensitivity and good selectivity for electroactive analytes. Based on the two main kinds of active compounds in chrysanthemum buds, flavonoids and polysaccharides including the hydroxyl (-OH) groups are electroactive at carbon and copper electrodes, respectively, which suggests that CE-ED is an appropriate method to investigate the chemical fingerprint of chrysanthemum buds.
The purpose of this study is to establish chromatographic fingerprints of chrysanthemum buds by CE-ED analysis. In this analysis, water and alcohol extraction methods will be successively employed to enhance extraction efficiency. Copper and carbon electrodes will be both used to guarantee that the fingerprints produced can encompass the main bioactive compounds. Two distinct chrysanthemum samples will be identified by the utility of the proposed fingerprint.  prepared with ethyl alcohol. The concentration of all stock solutions was 0.01 g mL −1 . All analytes were diluted to the desired concentration in running buffer for CE analysis.

Sample Collection and
Handling. Twelve batches of chrysanthemum bud samples were purchased from supermarkets in five main cultivation areas located in China ( Table 1). The chrysanthemum buds were dried at room temperature and finely ground using a blender (Joyoung Limited by Share Ltd., Shandong, China). The analytes in chrysanthemum buds were extracted as follows. First, the milled chrysanthemum buds (1 g) were suspended in 40 mL of deionized water and then ultrasonicated for 30 min to lyse the cells. Next, the mixture was heated at 90 ∘ C for 30 min to extract the water-soluble compounds. The suspension was cleared by centrifugation at 14800 rpm for 2 min using an Anke TGL-16C centrifuge (Shanghai Anting Instrument Factory, Shanghai, China), and the supernatant was filtered through a 0.22-m membrane to produce the polysaccharide fraction. To obtain the flavonoid fraction, the filtered residue was extracted with 50 mL 95% ethanol solution and ultrasonicated for 30 min. This suspension was centrifuged and stored as above. Before analysis, the samples were diluted with running buffer. All samples were prepared fresh every day.

Electrode Preparation.
In this study, all employed electrodes were made in our laboratory.
A scrap copper wire (25 cm long, 0.3 mm diameter) was sealed into a soft glass capillary (10 cm long) with glue water. The capillary was cut perpendicular to its length to expose the wire at both ends. A copper electrode was used as soon as the glue solidified.
A lead inside a graphite pencil (4 cm long, 0.3 mm diameter) was first burnished and carefully wound with a polished copper wire. Then, the lead was sealed into a soft glass capillary with glue water. Finally, the capillary was cut perpendicular to its length to expose the lead and wire at each end of the capillary. The carbon electrode was used as soon as the glue solidified.
At the start of each experiment, both ends of the copper or carbon electrode were polished with extra fine carborundum paper followed by the sonication in deionized water using KQ-100KDE ultrasonic generator purchased from Kunshan Ultrasonic Instruments Co., Ltd. (Kunshan, China) before being placed in the cell.

CE-ED Instrument.
CE analysis was performed on a laboratory-built CE-ED system [18]. A 30 kV high voltage power supply (Shanghai Institute of Nuclear Research, China) supplied the voltage between the ends of the capillary. The inlet end of the capillary was held at cathodic potential and the outlet end was maintained at ground. The inlet cell was filled with the separation running buffer, and the outlet end was placed in the detection cell filled with detection running buffer. A fused-silica capillary of 25 m (inner diameter) obtained from Hebei Yongnian Factory (Handan, China) was used for the separation. The samples were injected electrokinetically.
The design of the CE-ED system was based on the endcolumn approach. The working electrode (either copper or carbon) was placed at the outlet of the separation capillary, and detection was carried out in the reservoir containing the grounding electrode for the CE instrument. Before use, the surface of the working electrode was positioned carefully opposite to the capillary outlet using a micropositioner (Shanghai Lianyi Instrument Factory, China). A threeelectrode cell system composed of the working electrode, a platinum auxiliary electrode, and a saturated calomel electrode (SCE) was employed along with a BAS LC-3D amperometric detector (Biochemical System, West Lafayette, IN, USA). The electropherograms were processed with the HW-2000 software (Shanghai Qianpu Microsoftware, China).

CE Analysis.
As in previous CE-ED analyses [19], several key factors were investigated to find the optimal separation conditions. The running buffer was selected from Na 2 B 4 O 7 -H 3 BO 3 , phosphate salts, Na 2 B 4 O 7 -NaOH, and NaOH; pH and the concentration of the running buffer varied from 9 to 13 and from 10 to 50 mM, respectively; separation voltage ranged from 10 to 25 kV; the potential applied to copper working electrode ranged from 0.5 to 0.8 V, and the potential applied to carbon working electrode ranged from 0.8 to 1.1 V.
2.6. Data Analysis. The method was validated by identifying some key known compounds in chrysanthemum buds, such as chlorogenic acid, luteolin, glucose, and fructose. The relative standard deviations (RSDs), linearity, and detection limits of these compounds were calculated to determine the feasibility of this method.
Due to the novelty of fingerprinting analysis, only a few papers have been published on chemometrics [20]. In this study, data were analyzed with the professional software Computer-Aided Similarity Evaluation, which was developed based on chemometrics by the Research Center for the Modernization of Traditional Chinese Medicines (Central South University, Changsha, China). Ten batches of chrysanthemum buds were analyzed to establish the mean chromatograph as a representative standard fingerprint electropherogram. Data was analyzed using included angle cosine [21] and correlation coefficient [22] methods in order to compare their suitability for discriminating between chrysanthemum fingerprints.
The included angle cosine method considers the fingerprint spectrum data as a multidimensional space vector to convert the fingerprint spectrum similarity problem into the similarity between multidimensional vectors. The included angle cosine (cos ) is calculated by the following equation: while the correlation coefficient ( 1 ), which measures the relationship between the two properties, is calculated by the following equation: where and are the th elements in the two different electropherograms (namely, and , resp.) and is the number of the elements in the electropherograms. and are the mean values of the elements in electropherograms and , respectively.

Sample Identification.
Under the optimum analysis conditions, Chrysanthemum morifolium and Chrysanthemum indicum obtained from local supermarkets were analyzed by CE-ED. The electropherograms were compared with the standard fingerprint of chrysanthemum buds to distinguish between various chrysanthemums.

CE Analysis.
In order to achieve good separation of main components and quantify all of the bioactive chemical compounds in chrysanthemum buds, copper and carbon electrodes were utilized as the working electrode to analyze polysaccharides and flavonoids, respectively.

Optimization Condition of CE with Carbon Working
Electrode. The carbon electrode was used as the working electrode mainly to analyze flavonoid compounds in chrysanthemum bud samples. Running buffer selection was considered first because of its significant effect on separation. Na 2 B 4 O 7 -NaOH was chosen as the running buffer for its greater elution effect after comparing with the separation efficiency of Na 2 B 4 O 7 -H 3 BO 3 , phosphate salts, and Na 2 B 4 O 7 -NaOH.
The acidity and concentration of the running buffer also plays a key role in CE due to its effects on the zeta-potential ( ), the electroosmotic flow (EOF), and the overall charge of the analytes, all of which impact the separation and migration time of the analytes. When the pH of the same running buffer in separation cell and detection cell was lower than 9.89, two standard compounds (chlorogenic acid and luteolin) could not be separated and there were few peaks, demonstrating poor separation efficiency. On the other hand, when the pH of the running buffer was higher than 12, the migration time was over 1 h. So pH 11.25 Na 2 B 4 O 7 -NaOH (including 3.1 × 10 −3 g mL −1 boric acid ions) was selected as the optimum running buffer, balancing good separation with reasonable separation times.
The potential applied to the working electrode directly affected the sensitivity and detection limit of this method. Separation voltage affects the velocity of the electroosmotic flow and the migration time of the analytes. In the following analyses, the potential applied to the carbon electrode was maintained at 0.95 V, where the background current was not too high, while the signal-to-noise (S/N) ratio was the highest. Moreover, the working electrode demonstrated good stability and high reproducibility at this optimum potential.
The effect of the separation voltage on the migration time of the analytes was also studied. The results show that a higher separation voltage resulted in shorter migration times for all analytes but also resulted in increased baseline noise. In the following analyses, the separation voltage was maintained at 14 V.

Optimization of CE with Copper Working Electrode.
The copper electrode was used to analyze polysaccharide compounds in chrysanthemum buds. The optimal condition was selected with the same selection standards as for the carbon electrode. In order to obtain good separation and detection simultaneously [23], NaOH (pH 13.0) was selected as the optimal detection buffer because of the good response of the copper electrode in strong basic solution, and Na 2 B 4 O 7 (pH 9.24, 7.63 × 10 −3 g mL −1 ) was selected as the best separation buffer because it resulted in a good separation efficiency. The optimal potential to the copper working electrode was determined to be 0.67 V, and the optimal separation voltage was determined to be 20 kV.

Establishing the Fingerprint of Chrysanthemum Buds.
Because the similarity analysis determined that the 10 batches of chrysanthemum buds were highly similar, they were used to produce mean electropherograms for chrysanthemum buds using the copper working electrode (Figure 1) and the carbon working electrode (Figure 2). Together, these electropherograms comprise a comprehensive fingerprint of the bioactive compounds in chrysanthemum buds.

Identification of Markers and Method
Validation. Some compounds found in chrysanthemum buds, such as chlorogenic acid, luteolin, glucose, and fructose, were selected as marker compounds to validate this technology. Peaks 8 and 11 in Figure 1 represent glucose and fructose, respectively. Peaks 5 and 6 in Figure 2 represent chlorogenic acid and luteolin, respectively. The reproducibility of the peak current was evaluated by repeatedly injecting a standard solution under the optimum conditions. The RSDs of the migration time were 1.83%, 0.57%, 2.13%, and 1.41%, respectively. Additionally, a dilution series of standard solutions was also tested to measure the linearity of the current response for each of the four standard analytes. The linearity and detection limits are summarized in Table 2. As predicted, the observed Table 2: Results of the regression analysis on the calibration curves and the detection limits.

Compound
Regression equation 1 Correlation coefficient ( 2 ) Linear range (×10 −4 g mL −1 ) Detection 2 limits (10 −7 g mL −1 ) Glucose 1 y is the peak area; x is the concentration of the analytes (g mL −1 ). 2 The detection limits correspond to the concentrations of the signal-to-noise ratio of 3. reproducibility and detection limits of the four analytes were satisfactory.

Common Peaks Selection.
Peaks found in all samples were assigned as "common peaks, " standing for the main characteristic compounds and representing the chemical profile of sample. According to the selection criterion of common peaks in Chinese herbs, common peaks were selected as those with a migration time RSD lower than 5% and a minimum relative peak area of less than 0.5%. Ten chrysanthemum bud samples were analyzed by CE with both the copper and carbon electrodes. 12 common peaks were chosen from each of the two CE electropherograms as the common peaks for chrysanthemum buds for a total of 24.

Similarity
Analysis. The similarity of 10 batches of chrysanthemum bud samples from various locations was evaluated by correlation coefficient and the included angle cosine. The similarity between samples was calculated by the correlation coefficient method with the average of all samples as a standard, and the similarities between samples were calculated by the included angle cosine method with the average of all samples taken as a standard. If the value of angle cosine and correlation coefficient from samples are similar with > 0.90 and cos > 0.95, the samples are assigned to the same origin.
From the data analysis in Tables 3 and 4, ten chrysanthemum bud samples (numbers 1-10) had high similarity even though the concentration of active compounds among samples was not at the same level. From the results it can be concluded that the ten samples belong to the same species even though they are obtained from different place and different years.

Application of Standardized Fingerprint for Identification.
Fingerprinting analysis can be used to assess the quality of chrysanthemum buds that come from different sources. By examining the relative retention time and the relative peak area of the common peaks in a fingerprint, we can determine whether a raw herb is genuine. But the most important application of fingerprints is that they can be used to separate different chrysanthemum varieties from each other.
Under the optimal analysis conditions, two other chrysanthemum species (Chrysanthemum morifolium and Chrysanthemum indicum) were analyzed by this CE-ED method. By comparing their electropherograms with the standardized fingerprint of chrysanthemum buds (Tables  3 and 4), the distinctive features of each species have been identified. Some of the fingerprint common peaks are not found (carbon peaks 1, 2, 9, and 11; copper peaks 2, 5, 7, and 9) from both species. What is more, similarity analysis demonstrated that these species are significantly different