Chemical Differentiation of Genetically Identified Atractylodes japonica, A. macrocephala, and A. chinensis Rhizomes Using High-Performance Liquid Chromatography with Chemometric Analysis

The rhizome of Atractylodes japonica, which is a herbal medicine used for gastrointestinal therapeutics, has been categorized with A. macrocephala rhizome or A. chinensis rhizome based on different therapeutic criteria in Korea, China, and Japan. In the present study, 61 A. japonica, A. macrocephala, and A. chinensis rhizomes were collected from Korea and China and were genetically identified by internal transcribed spacer sequencing analysis. Chromatographic profiles were obtained from high-performance liquid chromatography analysis of the methanol and hot-water extracts of Atractylodes rhizomes and chemical differentiation of the rhizomes was carried out using chemometric statistical analyses such as principal component analysis, hierarchical clustering analysis, and Pearson's correlation coefficient analysis. The results from chromatographic profiles and chemometric analyses demonstrate that A. japonica rhizomes showed apparent chemical differences from A. macrocephala and A. chinensis rhizomes in the methanol extracts. In contrast, no clear distinction was apparent for the hot-water extracts of Atractylodes rhizomes, especially A. chinensis rhizomes. These results indicate that there is a clear chemical difference between A. japonica and A. macrocephala rhizomes; however, the chemical diversity of A. chinensis rhizome shows different chemical relationships with A. japonica or A. macrocephala rhizome, dependent on the chemical features.


Introduction
The rhizomes of Atractylodes japonica, A. macrocephala, and A. chinensis (Asteraceae) have been therapeutically used to treat gastrointestinal disorders in Korea, China, and Japan. The medicinal classification of these rhizomes differs in the pharmacopeia of each country: the rhizomes of A. japonica and A. macrocephala are categorized together in Korea, while the rhizome of A. japonica is unofficial in the pharmacopeia but is categorized with that of A. chinensis in China [1][2][3]. This inconsistency, especially with respect to A. japonica rhizome, causes confusion in the use of the Atractylodes rhizomes between countries with different categorization of such medicinal herbs.
For the criteria of the use of Atractylodes rhizomes, we reported in our previous study that the rhizome of A. japonica showed a closer genetic relationship and higher chemical similarity with A. chinensis rhizome than with A. macrocephala rhizome, and more consistent therapeutic effects would be expected when similarly categorized. Therefore, we proposed that A. japonica rhizome is genetically and chemically distinct from A. macrocephala rhizome and that the former should be used under the same medicinal categorization as A. chinensis rhizome [4]. However, there were a number of limitations at study, such as insufficient sample numbers, a lack of detailed sample information, and single solvent extraction. Hence, genetic and chemical relationships In the present study, a total of 61 Atractylodes samples were collected and their original species were identified using internal transcribed spacer (ITS) sequences from nuclear ribosomal DNA (nrDNA), which is the identification method mostly used for genetic confirmation of plant species [5][6][7]. Genetically identified samples were extracted with either organic or aqueous solvents, and the extracts were analyzed using a high-performance liquid chromatographydiode array detector (HPLC-DAD) instrument to compare their chromatographic profiles [8]. The chemical relevance of the rhizomes of A. japonica, A. macrocephala, and A. chinensis was determined based on principal component analysis (PCA), hierarchical clustering analysis (HCA), and Pearson's correlation analysis.

Plant Materials and Reagents.
Methanol, water, and acetonitrile (HPLC grade) were purchased from J.T.Baker (Phillipsburg, NJ, USA). Trifluoroacetic acid (TFA) was purchased from Sigma-Aldrich (St Louis, MO, USA). Fortyone samples of A. japonica rhizomes, 11 samples of A. macrocephala rhizomes, and 9 samples of A. chinensis rhizomes were obtained from the wild, agricultural fields, or local herbal markets in Korea and China in 2016. The samples were authenticated by their morphological characteristics and by ITS sequencing analysis. The samples were coded as AJ for A. japonica rhizomes, AM for A. macrocephala rhizomes, and AC for A. chinensis rhizomes (Table 1). Voucher specimens have been deposited at the School of Korean Medicine, Pusan National University.

Preparation of Genomic DNA.
Genomic DNA was extracted from the crude drugs of Atractylodes rhizomes according to the modified manuals of NucleoSpin5 Plant II kit (Macherey-Nagel, Dueren, Germany). For some samples, 10% cetyltrimethyl ammonium bromide 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 by using a TOPcloner6 TA-Blunt kit (Enzynomics, Daejeon, Korea) and the DNA sequence of the cloned PCR product was determined after interpretation performed by Bioneer (Daejeon, Korea). Dendrogram. DNA sequences were analyzed using the ClustalW multiple sequence alignment (BioEdit, v7.0.9; available at http://www.mbio.ncsu.edu/BioEdit/page2.html) and the phylogenetic tree was created using DNADist (BioEdit). To study the relationships among Atractylodes rhizomes, 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 used as outgroups in the phylogenetic analyses, based on previous studies [10,11]. ITS sequences of these taxa were collected from NCBI GenBank. The ITS sequences of nine specimens (coded with Arabic numerals) of Atractylodes plants were collected as species reference samples for identification of Atractylodes rhizomes in previous research [4], and these were used as a reference for species.

Preparation of Samples for HPLC Analysis.
For the methanol extract, dried powder of the rhizomes (1.0 g) was extracted with 10 mL of methanol using an ultrasonic extractor (Power Sonic 520; Hwashin Tech, Daegu, Korea) for 30 min. The extract was centrifuged at 10,000 rpm for 5 min. The supernatant was transferred to a 1.5 mL microtube and evaporated using a nitrogen gas blowing concentrator (MGS-2200; Eyela, Miyagi, Japan). Residue was dissolved in methanol to a concentration of 10,000 g/mL and the extract solution was filtered through a 0.2 m syringe filter (BioFact; Daejeon, Korea) prior to HPLC injection.
For the hot-water extract, dried powder of the rhizome (0.5 g) was extracted in a 100 mL glass bottle with 20 mL of distilled water using a microwave oven (BP-111-RS; Microwave Research & Applications, Inc., IL, USA) at 90 ∘ C for 20 min. The water extract was transferred to a 15 mL conical tube and centrifuged at 3000 rpm for 10 min. The supernatant was lyophilized using a freeze-dryer (IlShin-BioBase; Dongducheon, Gyeonggi-do, Korea). The powder of dried extract (20 mg) was dissolved in 1 mL of HPLC grade water and then the solution was centrifuged at 10,000 rpm for 5 min. The supernatant was filtered through a 0.2 m syringe filter (BioFact) 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 array detector (DAD) was used for chromatographic fingerprinting. The data were processed with Chemstation software (Agilent Technologies). The separation of compounds was carried out on a Capcell Pak Mg II C 18 column (4.6 mm × 250 mm, 5 m; Shiseido, Tokyo, Japan) at 35 ∘ C. The flow rate was 1 mL/min and the injection volume was 10 L.

Chemometric Statistical Analysis.
The 61 samples that were genetically identified and recoded were used for PCA, HCA, and Pearson's correlation analysis. In total, 45 and 31 peaks were selected as reference peaks (> 1.0% of total peak area) for the methanol extract and water extract, respectively, at their optimal UV absorption, and their absolute areas were calculated by peak area integration for chromatographic fingerprinting. A matrix composed of the rows (Atractylodes sample) and columns (absolute area of each reference peak) was used to construct the PC plot and dendrogram and for Pearson's correlation analysis, using the open source software R (v. 3.4.3; The R Foundation for Statistical Computing).

ITS Genotype and Genetic Identification of Atractylodes
Rhizomes. Amplification of the ITS region produced overall 733 bp of nucleotide sequences from 61 samples listed in Table 1 and from nine dried voucher specimens [4]. The ITS sequences of samples were determined by comparing DNA sequences registered in NCBI GenBank as well as in [12], with the following accession numbers: for A. japonica AB219405, for A. macrocephala AB219406, for A. lancea AB219407, for A. chinensis AB219408, and for A. koreana AB219409. As shown in previous research [4], nucleotide substitutions were observed in 37 sites on the ITS regions of Atractylodes samples. Type 1, the ITS sequence of A. japonica, showed multiple sequences compared with the other species; 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 11 samples labeled AM, were determined as A. macrocephala; no difference of DNA sequence between samples was observed. Among the nine samples labeled AC, determined as A. chinensis, AC4, AC7, and AC8 showed differences in nucleotide sequence. The sequence of AC4, AC7, and AC8 differed from that of Type 4 by 1 bp (A → G) at nucleotide position 128 bp, which indicates intraspecific variation. All 41 samples labeled AJ were determined as A. japonica. Unlike in other species, some intraspecific variation was observed among the 41 AJ samples. Most of the differences were multisequences, which have already been shown to be Type 1 [4].

Genetic Relationship of the Atractylodes Rhizomes.
Sixtyone samples of Atractylodes rhizomes were identified as A. chinensis, A. japonica, and A. macrocephala. Phylogenetic classifications based on the ITS region were made and the inferred evolutionary relationships among Atractylodes rhizomes are represented as a phylogenetic tree. The genus Atractylodes is well separated from other close genera and outgroups. The samples of A. japonica, A. lancea, and A. chinensis formed A. lancea complex, whereas those of A. macrocephala formed their own A. macrocephala complex (Figure 1).

Chromatographic Profiling of the Methanol and Hot-Water
Extracts of Atractylodes Samples. Chromatographic profiles of representative samples (AJ1, AC1, and AM1) at all UV wavelengths were compared; a total of 45 peaks and 31 peaks for the methanol and water extracts, respectively, were selected for comparison by macroscopic observation and further chemometric analysis (Figures 2 and 3 and Table S1). There were distinct differences in chromatographic patterns of the methanol extracts of AJ, AC, and AM, and 28 out of 45 reference peaks were shared among the methanol extracts of Atractylodes samples (Figure 4). In contrast, chromatographic differences among the water extracts were less distinctive, with 18 out of 31 references peaks being shared among the chromatograms from the hot-water extracts ( Figure 5). In the principal component (PC) plot of the methanol extracts of the samples, three distinct clusters were observed: an AJ group (AJ1-41 including AC4), an AM group (AM1-10), and an AC group (AC1-9 except AC4). The distribution of PC1 score showed that the samples in the AJ group were plotted closer to those in the AM group than those in the AC group; in contrast, in the PC2 score, the samples in the AC group were located closer to those in the AJ group than those in the AM group ( Figure 6). However, the PC plot of the hot-water extracts of the samples differed from that of the methanol extracts. As the samples were distributed in a wide range of PC1 and PC2 scores, no distinct clusters were observed. AC4 and AC6 were plotted closer to the location of AJ samples, whereas AC5 and AC8 overlapped into the plots of AM samples (Figure 7).

Clustering Analysis of Atractylodes Samples
The dendrogram of methanol extracts from HCA showed similar groups as obtained from PCA; namely, AJ, AC, and AM samples formed their own groups (groups II, III, and IV) except for AC4 in the AJ group. However, several AJ samples were divided into a separate group below a height of 250000 and clustered in group I below a height of 100000  ( Figure 8). For the dendrogram of the methanol extracts, the water extracts of AJ samples were also divided into two distinct groups (groups I and II), whereas those of AC and AM samples were not clearly distinguished within their own groups because they were not separated in the subgroup of group II, below a height of 4000 ( Figure 9).

Evaluation of the Correlation between A. japonica, A. chinensis, and A. macrocephala Samples by Pearson's Correlation
Coefficient. In the methanol extracts, the mean value of the Pearson's correlation coefficient (r) of each AJ sample ranged from 0.01 to 0.31 (except for AJ41; r = -0.01) for the correlation of AM samples and from 0.01 to 0.1 for the correlation of AC samples (Figure 10(a)). The mean correlation coefficient of each AC sample ranged from -0.08 to 0.1 (except for AC4; r = 0.95) for the correlation of AJ samples and from -0.07 to 0.1 for the correlation of AM samples (Figure 10(b)). The mean correlation coefficient of each AM sample ranged from -0.05 to -0.01 for the correlation of AC samples and from -0.03 to 0.22 for the correlation of AJ samples (Figure 10(c)).
In the hot-water extracts, the mean correlation coefficient of each AJ sample ranged from 0.25 to 0.80 for the correlation of AM samples and from 0.30 to 0.81 for the correlation of AC samples (Figure 11(a)). The mean correlation coefficient of each AC sample ranged from 0.55 to 0.72 for the correlation of AJ samples and from 0.29 to 0.87 for the correlation of AM samples (Figure 11(b)). The mean correlation coefficient of each AM sample ranged from 0.49 to 0.79 for the correlation Evidence-Based Complementary and Alternative Medicine of AC samples and from 0.45 to 0.65 for the correlation of AJ samples (Figure 11(c)). The mean and median values of Pearson's correlation coefficient from Atractylodes samples are shown in Table 2.

Discussion
We previously proposed that the rhizome of A. japonica (AJ) should be considered for medicinal use as being in the same medicinal category as that of A. chinensis (AC) because of their genetic and chemical similarity to that of A. macrocephala (AM) [4]. However, an unresolved issue is whether the rhizome of A. japonica should be medicinally used in parallel with the rhizome of A. macrocephala or with the rhizome of A. chinensis. Therefore, building on our previous work, in this study, 61 Atractylodes rhizomes that were collected from Korea and China in 2016 were genetically identified by their original species and their chemical differentiation was carried out by chromatographic profiling and chemometric statistical analysis.
In the methanol extracts, three apparent clusters of Atractylodes samples in the PC plot (AJ samples + AC4, AC samples, and AM samples) showed different proximity among each group. PC1 and PC2 scores, the most and next most influential factors of clustering in the PC plot, provided the group of AJ samples with more proximity to the group of AM samples than that of AC samples which results in better chemical relevance, as the PC scores of the AJ samples were closer to those of AM samples [13,14]. The proximity among the groups of AJ, AC, and AM samples was also confirmed by the dendrogram from HCA, which produces clusters to classify samples by measuring the distance between samples [15,16]. The dendrogram also indicated apparent groups of AJ, AC, and AM samples, and samples in the AJ group (cluster IV) showed closer chemical similarity to those in the AM group (cluster III) than to those in the AC group (cluster II) below a height of 100000, as samples with a similar distance were involved in the same group [17]. Pearson's correlation coefficient (r) analysis among the samples also supports the clusters obtained from PCA and HCA. The overall correlation between different Atractylodes species was not strong and was even negative, indicating that the samples of AJ, AM, and AC groups represented a weak correlation and, hence, were separated from each other [18,19]. Stronger correlation was observed between samples from the AJ-AM group, with positive and higher values, than between samples from the AJ-AC group, with negative and lower values (close to 0) [20,21].
In the hot-water extracts, the clusters of the samples according to their original species were not distinct in the PC plot, in which the samples from the same species were mostly gathered together. The AJ samples exhibited higher chemical similarity to a number of AC samples from the most influential PC1 scores, whereas the PC2 scores of AJ samples indicated a closer relationship to AM samples. Rather, AC and AM samples showed higher proximity to each other. These relationships between samples were also observed in the dendrogram: the AJ samples formed separate groups, whereas the AC and AM samples were consolidated in subgroups of cluster II, which makes the apparent delineation between AC and AM samples more difficult, as evaluated by the higher values between two species. Moreover, AJ samples indicated a stronger correlation with AC samples than with AM samples overall. Unlike the distribution of the samples in the PC plot, about half of the AJ samples formed a separate AJ group (each cluster I in the dendrograms from the methanol and water extracts, respectively) in the dendrogram from HCA, showing least similarity with AC, AM, and the remaining AJ samples. In contrast, but similar to the results from PCA, each AJ sample was strongly correlated with other AJ samples from the methanol and hot-water extracts, with an value close to 1 [19] (Supplementary Figures S1 and S2). These results might be ascribed to the differences in mathematical and statistical methods between HCA and PCA [22][23][24].
In contrast to our previous study [4], the chemometric results of the present study demonstrate that the chemical correlation and relevance of AJ samples, particularly in the methanol extracts (organic extract), were weaker with AC samples than with AM samples, although AJ samples have a closer genetic relationship to AC samples than to AM samples. Regarding these results, an important difference from the previous work that should be considered, however, is that there was always a distance between AJ and AM samples in the previous and present study, which means that AJ samples were not chemically analogous to AM samples; that is, they had their own clusters of samples. Another point is that the location of AC samples changed from "inside" to "outside" the cluster of AJ samples, indicating that AC samples in the present work showed less chemical  homogeneity to AJ samples, as reported previously [25,26]. This might be explained by the fact that geographical diversity can lead to a wide range of variation in the content of chemical components from the rhizome of A. chinensis [27,28]. This is because environmental factors, such as climate, temperature, humidity, soil, and altitude, and the corresponding adaptation of plants are crucial for the biosynthesis and production of chemical components and secondary metabolites [29,30], and such environmental differences can also influence the chemical variation in the plants of genus Atractylodes [31][32][33][34]. Based on the results from chromatographic profiles and chemometric analyses, A. japonica rhizomes could apparently be differentiated from A. macrocephala and A. chinensis rhizomes by their methanol-soluble components, whereas A. chinensis rhizomes were not clearly distinguishable from A. japonica and A. macrocephala rhizomes by their hot-watersoluble components. These results indicate that the chemical difference between A. japonica and A. macrocephala was distinct in both extracts, as also reported previously [4]. However, the difference between A. japonica and A. chinensis was dependent on the chemical features that were identified despite their closer genetic relationship. Moreover, aqueous components from A. chinensis rhizome were chemically closer to those from A. macrocephala rhizome, although they are categorized as different therapeutic agents. Further pharmacological and clinical evidence is necessary to confirm the chemical correlation between Atractylodes rhizomes.

Conclusion
In this study, 61 Atractylodes rhizomes that were collected from Korea and China in 2016 were genetically identified by their original species by ITS DNA sequencing analysis: A. japonica, A. macrocephala, and A. chinensis. Chemical differentiation was carried out by chromatographic profiling and chemometric statistical analysis, namely, PCA, HCA, and Pearson's correlation analysis, using the methanol and hot-water extracts of Atractylodes samples. The results from chemical fingerprinting and statistical analyses demonstrated that A. japonica rhizomes were chemically distinct from A. macrocephala rhizomes. However, A. chinensis rhizomes represented diverse chemical variation showing a wide range of relationships to A. japonica and A. macrocephala rhizomes, presumably arising from their environmental differences.