Evaluation of Medicinal Categorization of Atractylodes japonica Koidz. by Using Internal Transcribed Spacer Sequencing Analysis and HPLC Fingerprinting Combined with Statistical Tools

Atractylodes rhizomes have been used as the herbal medicine “Changchul” or “Baekchul,” according to their clinical purpose, in Korea, China, and Japan. Among the Atractylodes species, the medicinal use of Atractylodes japonica has been controversial, as it is categorized as both Changchul and Baekchul in those countries, and, moreover, parts of the rhizome have been differently used, depending on age of the plant, in Korea. Chromatographic fingerprinting by using HPLC combined with chemometric analyses and internal transcribed spacer (ITS) sequencing analysis were conducted to classify and identify 34 crude drugs derived from Atractylodes rhizomes. The identification of the samples, authenticated by their morphological features as A. japonica Koidz. (Changchul and Baekchul), A. chinensis Koidz., and A. macrocephala Koidz., was confirmed as A. japonica, A. chinensis, and A. macrocephala by ITS sequencing. The results from chemometric analyses showed that the chemical components of the crude drugs from A. japonica were significantly different from those from A. macrocephala but were similar to those from A. chinensis. The analyses also suggested that the categorization by age of A. japonica as Changchul or Baekchul is not recommended. The results indicate that A. japonica should be categorized as “Changchul” and should not be further categorized by age.


Introduction
The genus Atractylodes (Asteraceae) are perennial herbs distributed in Korea, China, and Japan. Their dried rhizomes have been classified into two kinds of herbal medicines according to their clinical purpose, "Baekchul" (Baizhu in Chinese, Byakujutsu in Japanese) and "Changchul" (Cangzhu in Chinese, Soujutsu in Japanese) [1]. In medicinal applications in Korea, Japan, and China, the rhizomes of A. lancea DC. and A. chinensis Koidz. have been classified as Changchul, while that of A. macrocephala Koidz. has been classified as Baekchul in the pharmacopeias of Korea, China, and Japan [2][3][4].
However, there has been disagreement between countries in classifying the rhizome of A. japonica Koidz.: Korean and Japanese pharmacopeias, as well as some studies from Korea and Japan, have classified the rhizome of A. japonica as Baekchul, whereas Chinese studies have classified it as "Gwan-Changchul" (Guan-Cangzhu in Chinese), a type of Changchul, which is not even listed in the Chinese pharmacopeia [5,6]. Moreover, A. japonica is recorded as a synonym of A. lancea in the Flora of China [7]. Confusion also occurs in local herbal markets in Korea, where the rhizomes of A. japonica have been used as Changchul when the fibrous substance has formed after it has grown over two years [8]. The dried rhizomes of Atractylodes species can be identified by their morphological features; however, it is difficult to discriminate them by macroscopic observation, a subjective method, due to their morphological similarity. Consequently, the misuse of Atractylodes rhizomes may occur 2 Evidence-Based Complementary and Alternative Medicine when identification is based only on their morphological features. Therefore, strict classification is required for the exact use of Atractylodes rhizomes for medicinal purposes.
Genetic identification achieved by hybridization, polymerase chain reaction (PCR), and sequencing techniques is considered more objective, precise, and reliable method for identifying and authenticating herbal species [9,10]. Internal transcribed spacer (ITS) regions, which are rapidly evolving regions of nuclear ribosomal DNA, have been widely used for the identification of plant materials, and phylogenetic analysis using ITS regions has been conducted to investigate the genetic variability of complicated herbal species [11][12][13]. It has been reported that the four species of Atractylodes (A. japonica, A. macrocephala, A. lancea, and A. chinensis) can be distinguished by their different genotypes on ITS sequences [14]. In contrast, one study conducted by PCR-restriction fragment length polymorphism (PCR-RFLP) analysis of the ITS region on nucleotide ribosome DNA (nrDNA) reported that A. japonica is not different from A. macrocephala [15]. Other studies have demonstrated that the geographical distributions, morphological features, and genetic differentiation between A. lancea and A. chinensis were not consistent, and A. japonica was most closely related to A. lancea, according to the results from ITS and trnL-F sequences [16,17]. The results of these studies demonstrate that the classification of Atractylodes rhizomes, particularly that of A. japonica, remains controversial.
Chemical fingerprinting by using HPLC is an effective and reliable method for the investigation of chemical components in herbal medicines, owing to its high separation efficiency and high detection sensitivity [18]. HPLC fingerprinting also enables a systematic and comprehensive approach to the identification and quantification of the components in herbal medicines [19]. HPLC fingerprinting, combined with chemometric statistical analysis, has been widely used for the quality control of herbal medicines and related herbal products [20,21]. Moreover, such techniques have also been used to discriminate interspecies differences in herbal medicines, by incorporating principal components analysis (PCA) [22][23][24]. In previous studies, A. lancea was chemically differentiated from A. chinensis by HPLC fingerprinting combined with PCA or orthogonal partial least squares-discriminant analysis [25,26]. However, the chemical discrimination of Atractylodes species, using HPLC fingerprinting and chemometric analysis, has not been conducted.
In the present study, chromatographic fingerprinting and chemometric statistical analyses, including PCA, hierarchical clustering analysis (HCA) analysis, and Pearson's correlation coefficient analysis were conducted to classify the crude drugs derived from Atractylodes rhizomes. The ITS sequences from nrDNA were also examined to identify the Atractylodes species.

Plant Materials and Reagents.
Thirty-four samples of crude drugs from Atractylodes rhizomes were collected or purchased from the wild, agricultural fields, or local markets in Korea and China. The samples were authenticated by their morphological features through identification criteria by authors. Fifteen rhizomes were identified as A. chinensis (coded as "AC"); seven samples of Baekchul and five samples of Changchul were "young" and "aged" dried rhizomes of A. japonica (coded as "AJB" and "AJC," resp.); and seven rhizomes were identified as A. macrocephala (coded as "AM") ( Table 1). The voucher specimens have been deposited in the herbarium of the College of Korean Medicine, Woosuk University (Jeonju, Jeonbuk, Korea).
Nine specimens (coded as Arabic number) of Atractylodes plants were collected as species reference samples for identification of Atractylodes rhizomes in Table 1 and those samples were deposited in the Korea Institute of Oriental Medicine as voucher specimens. The classification result of nine driedvoucher specimens is shown in Table 2.

Preparation of Genomic DNA.
The genomic DNA was extracted from the crude drugs of Atractylodes rhizomes according to the manuals of NucleoSpin5 Plant II kit (Macherey-Nagel, Dueren, Germany). For some samples, 10% cetyl trimethyl ammonium bromide (CTAB) and 0.7 M NaCl were used to remove the phenolic compounds and polysaccharides.

Determination of DNA Sequence of PCR Product.
PCR product separated from agarose gel was cloned using MG6 TOPcloner TA kit (Macrogen, Korea) and DNA sequence of cloned PCR product was determined through the interpretation performed by Macrogen (Korea).

Analysis of DNA Sequence and Preparation of Dendro-
gram. DNA sequence was analyzed using ClustalW multiple sequence alignment (Bioedit, v7.0.9; available at http://www .mbio.ncsu.edu/BioEdit/page2.html) and the phylogenetic tree was created by using DNADist (Bioedit). To study the relationship of Atractylodes, the nucleotide sequences of the genera Atractylis and Carlina deposited in NCBI GenBank were used. The genera Brachylaena, Cardopatium, Cirsium, Echinops, Phonus, and Tarchonanthus were also used as reference for phylogenetic relationship. Magnolia heptapeta, A. chinensis Koidz. Zanthoxylum rhoifolium, and Lilium cernuum were used as out-groups in the phylogenetic analyses, based on previous studies [28,29]. ITS sequences of these taxa were collected from GeneBank in NCBI (the accession numbers were shown in Table 3).

Preparation of Samples for HPLC Analysis.
The dried powder of the rhizomes (100 mg) was weighed and then extracted by sonication with 2 mL of ethanol (HPLC grade; Phillipsburg, NJ, USA) for 40 min. The extract was filtered through a 0.45 m membrane filter (Adventec, Tokyo, Japan) and concentrated in vacuum oven at 45 ∘ C. Concentrated extract was dissolved with methanol at the concentration of 5000 g/mL and filtered through a 0.45 m membrane filter, prior to HPLC injection.

HPLC Conditions for Chromatographic Fingerprinting.
An Agilent 1260 liquid chromatography system (Agilent Technologies, Palo Alto, CA, USA) equipped with an autosampler, degasser, quaternary solvent pump, and diode

ITS Genotype and Genetic Identification of Atractylodes Rhizomes.
The amplification of internal transcribed spacer (ITS) region produced overall 733 bp of nucleotide sequences from 34 samples listed in Table 1 and nine dried-voucher specimens ( Figure 1). The determined ITS nucleotide sequence of samples were confirmed by using DNA sequence registered in NCBI GenBank as well as previous paper [14] with comparison of the accession numbers: lancea (AB219407), A. chinensis (AB219408), and A. koreana (AB219409). As presented in Table 4, there was nucleotide substitutions observed on 37 sites on the ITS regions of Atractylodes samples. Type 1, ITS sequence of A. japonica, showed multiple sequences comparing to other species, while types 2 and 3 were the genotypes of A. macrocephala and A. lancea, respectively. Type 4, the genotype of A. chinensis, was identical to type 5, the genotype of A. koreana. All 7 samples labelled as "AM" were determined as A. macrocephala and the difference of DNA sequence between samples was not observed. Among 15 samples labelled as "AC," AC-1, AC-2, AC-4, AC-5, AC-6, and AC-11 were determined as A. chinensis, whereas the rest were A. japonica. The sequence of AC-1 was different from that of type 4 by 1 bp (A → G) at nucleotide position, 128 bp, which indicates intraspecific variation. All 7 samples labelled as "AJB" as well as 5 samples labelled as "AJC" were determined as A. japonica. AC-13 and AJB-5 were different form that of type 1 by 1 bp

Genetic
Relationship of the Atractylodes. Nine driedvoucher specimens were confirmed as A. japonica, A. lancea, and A. macrocephala. Thirty-four samples of Atractylodes rhizomes were identified as A. chinensis, A. japonica, and A. macrocephala. Phylogenetic classification based on ITS region was constructed and the inferred evolutionary relationships among Atractylodes rhizomes were represented on phylogenetic tree. The genus Atractylodes is well separated from other close genera and out-groups. The samples of A. japonica, A. lancea, and A. chinensis formed A. lancea complex, whereas those of A. macrocephala formed their own complex, namely, A. macrocephala complex. Five AC samples, AC-2, AC-4, AC-5, AC-6, and AC-11, were involved in A. chinensis group. AJB, AJC samples, and the rest AC samples were contained in A. japonica group (Figure 2).
Comparing the genotypes listed in Figures 1 and 2 and Table 4, the species of 34 samples in Table 1 of Atractylodes rhizomes were recoded as shown in Table 5.

HPLC Fingerprinting of Atractylodes Samples.
Through macroscopic observation of the results of HPLC fingerprinting, the chromatograms of A. japonica samples (AJBs + AJCs + AJs) and A. chinensis samples (ACs) showed peak patterns similar to each other, whereas the chromatograms of A. macrocephala samples (AMs) represented different peak patterns compared with those of A. japonica and A. chinensis samples. Moreover, chromatographic difference between Baekchul samples (AJBs) and Changchul samples (AJCs) was not apparently distinguished (Figures 3 and 4).

Chemometric Statistical Analysis of Atractylodes Samples.
The results of the chromatographic fingerprinting were further analyzed using principle component analysis (PCA), hierarchical clustering analysis (HCA), and Pearson's correlation analysis to evaluate the correlations between the samples whose code names were redetermined by genetic identification as listed in Table 5.
HCA results also showed similar result obtained from PCA. Below the height of 30000, while the samples of A. macrocephala formed their own clustering, those of A. japonica and A. chinensis were gathered together, where A. japonica samples and A. chinensis samples were mixed. Therefore, apparent discrimination between A. japonica samples and A. chinensis samples was not observed. Moreover, as seen in PCA score plot, AJB and AJC samples were not clearly distinguished in HCA dendrogram, as they were grouped undistinguishably within the same levels ( Figure 6 160  170  180  190  200  210  220  230  240  250  260  270  280  290  300   310  320  330  340  350  360  370  380  390  400  410  420  430

Discussion
Discrimination of herbal medicines according to their therapeutic effect is most important in using herbal medicines as therapeutic agents. To achieve such purpose, correct classification of herbal medicines should be preceded at their species levels. The rhizome of A. japonica has been at the center of controversy as it is differently classified in many countries; Korea and Japan have used that as "Baekchul," whereas China has used that as "Changchul." Furthermore, the rhizome of A. japonica is further divided into "Baekchul" and "Changchul" by the age; tuberous part grown for less than a year is used as Baekchul, while fibrous part grown for more than two years is used as Changchul. Hence, we classified Atractylodes samples which were genetically identified by ITS genotypes using analytical tools combined with chemometric statistics, in order to propose correct categorization of Atractylodes rhizomes, especially A. japonica rhizomes.  [14,17]. The samples of A. japonica, A. lancea, and A. chinensis formed A. lancea complex, in accordance with the previous research [17]. A. japonica is most closely related to A. lancea and next related to A. chinensis, as suggested in previous researches [17,30,31]. The samples identified as A. japonica were obviously separated from those identified as A. macrocephala by DNA sequencing analysis [14,16,32]. According to the results,  Genotype

Type 5 T A G T A C C A C C G T C C T T T G C A C G C T C C G C G C A C T C C A C *
Bold characters indicate nucleotide additives, M = A and C; R = A and G; W = A and T; Y = C and T. Type 1 genotype resulted from deposited ITS nucleotide sequences in NCBI Genbank of Atractylodes japonica.    there was difference in classification between morphological identification and genetic identification which might be caused by morphological similarity.

Classification of Atractylodes Samples Using Chemometric
Analysis. Principal components analysis (PCA) was performed for the clustering of the samples and for investigating the relationships among the samples using principal components on the PCA plot. Principal component 1 (PC1) explains most of the variance, and PC2, which is orthogonal to PC1, represents most of the variance not explained by PC1 [33]. HCA is another clustering method which classifies similar objects mathematically into the same group and the classifications were represented as tree diagram called dendrogram [34]. Pearson's correlation coefficient of each sample was calculated to evaluate the correlations between samples (−1 ≤ ≤ +1; the closer the is to 1, the more correlated the two fingerprints are) [35].
From the results of PCA, HCA, and Pearson's correlation analysis, the A. japonica samples were obviously distinct from A. macrocephala samples but distributed close to the A. chinensis samples without apparent separation of AJBs and AJCs samples. These results indicate that A. japonica is chemically different from A. macrocephala; however, the crude drugs derived from A. japonica and A. chinensis were not clearly distinguishable by their chemical components. These results indicate that the therapeutic effects of A. japonica and  AM1  AM3  AM5  AM7  AJB2  AJB4  AJB6  AJC1  AJC3  AJC5  AC2  AC3  AC4  AC5  AC6  AJ1  AJ2  AJ3  AJ4  AJ5  AJ6  AJ7  AJ8  AJ9   Average of Pearson's correlation coefficient   AM2  AM4  AM6  AJB1  AJB3  AJB5  AJB7  AJC2  AJC4  A. chinensis whose effects can be exerted by the chemical components are thought to be presumably analogous [36], although they are genetically different. Moreover, the separation of A. japonica rhizomes as "Baekchul" and "Changchul" is not recommended because they had the closest PC scores, equal levels of hierarchies, and similar Pearson's coefficients [37]; therefore, AJBs and AJCs can be assumed to be the same medicinal parts.

Assured Medicinal Categorization of Atractylodes japonica
Rhizomes. The categorization of A. japonica as Baekchul in Korea and Japan, therefore, should be removed, because its relationship to A. macrocephala, the only species regarded as Baekchul in China, is not close, which supports the results from previous studies [14,17,32,38]. Moreover, we do not recommend the segregation of A. japonica into Baekchul and Changchul, because the segregated crude drugs of A. japonica are not distinguishable by their chemical and genetic characteristics. This suggests that there is the lack of chemical or genetic reason to categorize the rhizome of A. japonica as Changchul and Baekchul based on plant age. Therefore, we propose that A. japonica should be separated from A. macrocephala and should be used as "Changchul" as the Chinese literatures define it, regardless of its age. Further biological and clinical experiments are necessary to confirm the results of the present study.
In conclusion, the 34 crude drugs derived from Atractylodes rhizomes were identified using ITS sequencing analysis and classified using HPLC fingerprinting and statistical tools such as PCA, HCA, and Pearson's correlation analysis. The macroscopically authenticated Atractylodes samples were genetically identified as A. japonica, A. chinensis, and A. macrocephala by ITS DNA sequencing. The hyphenated HPLC fingerprinting and statistical analyses showed that the chemical components of A. japonica were not related to those of A. macrocephala but related to those of A. chinensis. Moreover, the rhizome of A. japonica could not be segregated as Baekchul and Changchul by its age. The results from these chemical and genetic analyses demonstrate that A. japonica should be classified as "Changchul" and should not be classified by plant age.

Disclosure
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the paper.