Comprehensive study on genetic and chemical diversity of Asian medicinal plants, aimed at sustainable use and standardization of traditional crude drugs

The ITS sequences of specimens of the four Paeonia species as well as crude drug samples were of the same length, in which ITS1, 5.8S rRNA gene, and ITS2 regions were 267, 164, and 221 bp, respectively. Almost all the samples of WPR, RPR, and PR were identified as P. lactiflora, except for a few RPR samples from Sichuan Province that were identified as P. veitchii [8]. Significant intra-species polymorphism of the ITS sequences was detected within P. lactiflora. Clustering analysis on the basis of sequence similarity showed that P. lactiflora formed a large group distinct from P. veitchii, P. anomala, P. japonica, and P. suffruticosa Andrews (Fig. 1). Within the P. lactiflora group, there were two main subgroups: I and II. The P. lactiflora plants cultivated in southern China and Chinese WPR produced in Anhui, Zhejiang, and Sichuan provinces, as well as Japanese PR including Japanese medicinal cultivars “Bonten (S34)” and “Kitasaisho (S31),” belonged to subgroup I (WPR-type); whereas, P. lactiflora plants growing wild in northern China (P1, P2) and Mongolia (P3), and Chinese RPR (D12-D15) fell into subgroup II (RPR-type). Most horticultural varieties from Japan belonged to subgroup II.

The ITS sequences deposited in the International Nucleotide Sequence Database included two main types, differing by three nucleotides at positions 69, 458, and 523: type 1 showed three cytosines (C–C–C; U27682) [9], while type 2 had thymine (T), adenine (A), and T (T–A–T; JN572150) [10] at the three sites. According to the nucleotides at these three sites, crude drug samples and plants were divided into two main groups corresponding to the two subgroups (Fig. 1). One group showed T₆₉-A₄₅₈-T₅₂₃ at the three sites, which was the same as for type 2. The other group showed additive nucleotides (double nucleotides detected at the same site) of Y (T & C), M (A & C), and Y at the three sites (Y₆₉-M₄₅₈-Y₅₂₃), respectively. The two Japanese medicinal cultivars showed a sequence identical to a type of sequence from WPR (AB920144) of subgroup I. The five PR samples produced in Nara Prefecture (D7, D9, D10, D42, and D44), usually called “Yamato Shakuyaku,” were divided into two subgroups: I and II. Moreover, the PR from Niigata Prefecture (D6) was a mixture of individuals belonging to two subgroups, indicating Japanese PR was not only derived from medicinal cultivars of P. lactiflora.

The contents of eight components—paeoniflorin (P1), albiflorin (P13), 1,2,3,4,6-penta-O-galloyl-β-d-glucose (P19), (+)-catechin (P20), paeonol (P21), gallic acid (P22), methylgallate (P23), and benzoic acid (P24)—were quantitatively analyzed to clarify chemical properties of P. lactiflora, P. veitchii, P. anomala, and the crude drug samples including WPR, PR, and RPR (Fig. 2). Within the commercial samples derived from P. lactiflora, RPR samples produced in northern China had obviously higher contents of P1, P20, and P21 but lower content of P13 compared to WPR produced in southern China and most PR produced in Japan [8] (Fig. 3). Among 11 WPR samples, eight contained less than the 2.0% of paeoniflorin specified in the JP. Comparing high-performance liquid chromatography (HPLC) chromatograms of these samples with those of the PR collected from Japanese markets, commonly showed a conspicuous peak at retention time around 10.4 min. Further analysis using liquid chromatography (LC)/mass spectrometry (MS) clearly indicated that this notable peak was paeoniflorin sulfonate (P2), which could be produced by traditional processing with sulfur fumigation [11]. This indicated that the WPR available in Chinese markets was usually processed by sulfur fumigating, resulting in an extremely low P1 content; whereas the PR available in the Japanese market was not treated with this process. Apart from the WPR samples in which P2 was detected, quantitative data of the six constituents excluding P23 and P24 in Paeonia specimens and commercial samples was subjected to principal component analysis (PCA) and the PCA score plot showed four separate clusters of all samples. Besides the respective groups of P. veitchii and P. anomala, samples derived from P. lactiflora were clearly classified into two groups: one group included RPR (RPR group) and the other group was composed of WPR, PR produced in China, and most PR produced in Japan (WPR/PR group). The former was characterized by high P21, P20, and P1 contents, and the latter by a relatively high content of P13. Moreover, the P. veitchii group was far from the other groups and had high contents of P1, P19, and P22. The grouping in the PCA score plot was in accordance with clustering based on the similarity of ITS sequences. This result indicated that WPR and RPR were not only geographically isolated, but also genetically and chemically separated.

In addition, monoterpenoid profiling was performed using LC coupled with ion trap and time-of-flight MS (LC–IT-TOF-MS) to precisely characterize and quantify different types of peony root and the roots of related Paeonia species. The MS/MS fragmentation patterns of monoterpenoids with paeoniflorin (PF)-, albiflorin (AF)-, and sulfonated paeoniflorin (PFS)-type of skeletons were elucidated, which provided basic clues enabling subsequent identification of 35 monoterpenoids in LC–MS profiles of Paeonia species (Fig. 2). Mudanpioside C (P6) was the characteristic component of P. lactiflora, and 4-O-methyl-paeoniflorin (P3) was only detected in P. veitchii and P. anomala. Notably, six PFS-type monoterpenoids were detected in sulfur-fumigated WPR samples [12]. Quantification of 15 main monoterpenoids including 10 PF-type (P1, P3–P10, and P12), three AF-type (P13–P15), and two other types, a new compound paeoniflorol (P16) and lactiflorin (P18), in 56 samples revealed that five monoterpenoids [P1, benzoylpaeoniflorin (P5), galloylpaeoniflorin (P7), oxypaoniflorin (P9), and P13] predominated in all samples, but with quite different relative contents. Of the samples derived from P. lactiflora, RPR samples showed higher contents of five PF-type monoterpenoids [salicylpaeoniflorin (P4), P6, mudanpioside J (P8), P9, and benzoyloxypaeoniflorin (P10)] as well as P1 compared to WPR/PR, but comparatively lower contents of AF-type monoterpenoids [a new compound, 4-epi-albiflorin (P14), paeonivayin (P15), and P13] (Fig. 4). For the two other types of monoterpenoids, P16 had a high content in RPR, whereas P18 had a high content in WPR and PR. The above differences in monoterpenoid profiles between WPR and RPR might be some of the reasons for their different traditional uses and different bioactivities including anti-allergic effects. The samples derived from P. veitchii had the highest P4 and P7 contents, very low P9 and P10 contents, and no P6, and clearly differed from those derived from other sources [12].

The same analytical methods as peony root were applied to two medicinal and 61 horticultural cultivars of P. lactiflora, which were genetically of RPR- and WPR-types. There were high similarities between RPR- and WPR-types of cultivars in regard to the eight main constituents (Fig. S1) [13]. Moreover, monoterpenoid composition of the respective P. lactiflora cultivars showed high similarity regardless of RPR- or WPR-type (Fig, 4) [12], differing from the situation between commercial WPR and RPR.

Because several studies have reported anti-allergic activity of peony root [14], a bioactivity screening experiment concerning the inhibitory effect against immunoglobulin E (IgE)-mediated mast cell degranulation was performed to compare the activity of RPR from the Inner Mongolia Autonomous Region of China and PR from Japan and to determine a characteristic cultivar with anti-allergic activity among 17 horticultural cultivars. Among them, a RPR sample and two cultivars, “Edulis Superba (ES)” and “Harunoyosooi (HY),” significantly inhibited β-hexosaminidase release stimulated by 2,4-dinitrophenylated bovine serum albumin on IgE-sensitized rat basophil leukemia-2H3 cells at a concentration of 1.0 mg/mL. In order to elucidate the active compounds, bioassay-guided fractionations were subsequently conducted on the RPR sample and cultivar ES, and anti-allergic activities of isolated compounds were, respectively, examined. From the 60% and 80% aqueous methanol subfractions of methanol extract of RPR sample that showed activity, 29 compounds were isolated and identified as three new monoterpenoids [P14, P16, and 4′-hydroxypaeoniflorigenone (P17)], 14 known monoterpenoids, five flavonoids, and seven other types of compounds [15]. In the same way, 26 compounds were isolated from the ethyl acetate- and n-butanol-soluble fractions of methanol extract of cultivar ES, including one new norneolignan, paeonibenzofuran (26), five monoterpenoids, 10 flavonoids, and 10 other types of compounds [16]. Among compounds isolated from the RPR sample, nine monoterpenoids (P1, P4, P5, P7, P8, P10, P11, P16, and P18), as well as P19 and P23, showed moderate anti-allergic activities (Table S4). Among monoterpenoids, a new compound P16 exhibited the greatest effectiveness (IC₅₀ 41.17 μM), followed by P4 and P7. However, AF-type monoterpenoids had no activity [15]. Moreover, mudanpioside E (P12), quercetin (P27), and quercetin-3-O-β-d-glucopyranoside (P28) isolated from cultivar ES, showed potent inhibitory activity (IC₅₀ 40.34, 25.05, and 42.55 μM, respectively), followed by paeonolide (P25), (2R)-(-)-naringenin-7-O-β-d-glucopyranoside (P29), and P26 [16]. Based on these experimental results, two horticultural cultivars with moderate anti-allergic activity were selected as promising candidates with potential as medicinal resources of RPR, as well as the active components being clarified.

A medicinal cultivar “Bonten (BT)” belonging to WPR-type has been widely cultivated in Toyama Prefecture. To find new resources of peony root of RPR-type from P. lactiflora cultivars, with both horticultural and medicinal utilities, PCA was performed on quantitative data of six constituents (P1, P13, and P19–P22) from 63 P. lactiflora cultivars added to those from commercial samples except P2-containing samples, and Paeonia specimens. In the PCA score plot, samples and specimens derived from P. lactiflora were clearly classified into two groups: RPR and WPR/PR [13]. Three horticultural cultivars were included in the RPR group; the two cultivars ES and HY among them showed anti-allergic activity as described in the previous section, and were nominated as candidates for brand peony roots.

Next, with the aim to determine the influence of different post-harvest processing methods of boiling, peeling, drying, and storing on chemical composition and morphological features of the produced peony root and to establish an appropriate and practicable method for production of brand peony root with superior quality, 15 kinds of processing methods were applied to the 15 groups of fresh roots of cultivar BT after 4 years of cultivation. The roots produced were analyzed using HPLC and the contents of eight components (P1, P13, and P19–P24) and internal root color were compared (Fig. 5). The results showed that low temperature (4 °C) storage of fresh roots for approximately 1 month after harvest resulted in stable and high content of P1, possibly due to suppression of enzymatic degradation including enzymes involved in paeoniflorin hydrolysis [17]. This storage also prevented discoloration of roots and facilitated production of desirable bright color in roots, traditionally believed to be of high quality. In addition, quantitative ¹H nuclear magnetic resonance (qHNMR) analysis demonstrated that sucrose content increased significantly in root after low-temperature storage for 1 month, resulting in increased ethanol extract yield [18]. Boiling treatment triggered decomposition of polygalloylglucoses and led to significantly increased contents of P19, P22, and P23, shown by monitoring gallotannin changes using LC-IT-TOF–MS. Peeling treatment resulted in decreased P13 content; additionally in the group without boiling, the P20 content also decreased. For roots washed and boiled after low-temperature storage, P19 content was slightly higher for no peeling compared to peeling treatment, and for drying at 30 °C using a drying machine compared with indoor drying. As a result, the optimized processing method to produce high contents of main active components in the root was low-temperature storage for approximately 1 month after harvest, followed by washing, boiling, and drying at 30 °C with a drying machine [17]. Similar experiments to those described above were applied to two P. lactiflora cultivars: ES and HY. Following application of the optimized processing method, the main constituents were quantified and compared among three cultivars, with the order for levels of P1 and P13 of ES > HY > BT, and for P20 of HY > ES > BT. Moreover, P23 was found in cultivars ES and BT. Although P21 was not detected following the optimized processing roots of the three cultivars, ES and HY contained P21 in the dried roots treated without boiling. Using the above results, we selected suitable cultivars and processing methods. Recognizing the potential of these discoveries, a collaborative project between academia and local government is currently in progress in Toyama Prefecture to produce brand peony root.

For glycyrrhiza, one of the crude drugs with export restricted by the Chinese Government, to reveal chemical characteristics of G. uralensis growing in Mongolia, eight major bioactive constituents in the underground parts were quantitatively analyzed and compared with glycyrrhiza produced in China [20]. Most of the 15 specimens from eastern, southern, and western Mongolia contained 26.95–58.55 mg/g of glycyrrhizin, exceeding the criterion assigned in the JP [6], and was highest in the sample from Tamsagiyn Hooloy, Dornod Province, eastern Mongolia. The total content of three flavanones (liquiritin apioside, liquiritin, and liquiritigenin) was in the range of 3.00–26.35 mg/g and of three chalcones (isoliquiritin apioside, isoliquiritin, and isoliquiritigenin) was 1.13–10.50 mg/g. The content of glycyrrhizin and subtotal contents of flavanones and chalcones from Mongolian G. uralensis were obviously lower than those of the crude drugs available in the Japanese market derived from Chinese wild specimens, but higher than those derived from cultivated specimens in the Chinese market. Glycycoumarin, a constituent specific to G. uralensis, was detected in all Mongolian specimens and its content was higher in samples from Sergelen and Tamsagiyn Hooloy, which were comparable to that of Tohoku-kanzo in Japan derived from wild Chinese G. uralensis. Thus, Mongolian G. uralensis could be a glycyrrhiza resource and was generally of JP grade. However, the producing area should be taken into consideration for ensuring high quality.

After a field survey of Ephedra plants in Mongolia, molecular and chemical assessments on plant specimens were conducted to clarify whether they could be an alternative resource of ephedra herb used in Japan [21]. The distribution of E. sinica, E. equisetina, E. monosperma Gmelin ex C. A. Meyer, E. przewalskii Stapf, E. glauca Regel, E. regeliana Florin, E. lomatolepis Schrenk, and unknown Ephedra sp. (assumed to be E. dahurica Turz.) was confirmed. Among them, E. sinica and E. equisetina are prescribed as the botanical origin of ephedra herb in JP [6]. On the basis of nucleotide sequences of nuclear 18S rRNA gene and chloroplast trnK gene, E. sinica, E. equisetina, and E. monosperma presented completely identical sequences to the corresponding species from China. Since the population in southwestern Mongolia showed a high likelihood of hybrid origin, further sequence analysis of the partial nuclear ITS1 region was conducted to obtain detailed evidence of hybridization status as well as to elucidate the marker sequences of pure lines of each species [22]. As a result, the ITS1 sequences from all eight Ephedra species were roughly divided into five types: I–V. E. equisetina and E. monosperma had similar sequences (Type V), differing from other species. Although the remaining five species possessed similar sequences, they were divided into four types based on the nucleotides at four informative sites. Among them, type II sequences had additive nucleotides at four sites observed in E. sinica, Ephedra sp., E. glauca, and E. regeliana, which provided useful information for tracing hybrid origin. Morphological, genetic and distribution evidences suggested that hybridization of Ephedra species occurred widely in southwestern Mongolia, and several Ephedra species including E. przewalkskii were involved in these events. Integrated with trnK-, 18S-, and ITS-sequence types, pure lines of each species were proposed. The pure line (type I [AB600683]) of E. sinica was only found in eastern areas, while E. equisetina, which consisted of pure line, was found in the mountainous region of central–western areas. Quantitative analysis of five ephedrine alkaloids [ephedrine (E1), pseudoephedrine (E2), norephedrine, norpseudoephedrine, and methylephedrine] revealed that almost all Mongolian Ephedra plants, except E. przewalskii and E. lomatolepis, contained high amounts of total ephedrine alkaloids (TAs), with range 1.86–4.90% of dry weight [21]. The E. sinica (types I and II) contained 1.95–4.16% TAs and showed a higher percentage of E1 in TAs than other species. Within E. sinica, types I and II specimens collected from eastern grassland areas showed a high proportion of E1 in TAs (mean 51.4% and 54.0%, respectively); whereas, type II specimens from central areas showed a high proportion of E2 in TAs (mean 55.5%). The proportions of E1 and E2 might be affected more by growing environment than genetic factors. Ephedra equisetina had the highest content of TAs (3.98–4.90%), with more than 90% of TAs being E2. Both E. sinica and E. equisetina satisfied the JP criterion [6] of containing no less than 0.7% of summed content of E1 and E2, with 1.43–3.68% and 3.81–4.59%, respectively. Therefore, these two species growing in Mongolia were a suitable new resource of ephedra herb used in Japan. Given that the pure line of E. sinica is limited to eastern areas, and E. equisetina has a limited distribution, promoting the cultivation of selected specimens in suitable regions is crucial for ensuring a stable supply of ephedra herb and supporting environmental preservation.

Saposhnikovia root and rhizome (SR) derived from Saposhnikovia divaricata (Turcz.) Schischk. have been used as a diaphoretic, febrifuge, analgesic, and anti-phlogistic in China and Japan, including frequently as an ingredient in Kampo formulas [5]. In recent years, natural SR resources have been depleted because of increasing demand, therefore, a large amount of SR derived from cultivated plants has become available in the Chinese and Japanese markets. However, some of these do not meet the CP and JP requirements due to lower amounts of prim-O-glucosylcimifugin (S1) and 4′-O-β-d-glucosyl-5-O-methylvisamminol (S3) and the higher yield of dilute-ethanol-soluble extract [6, 7]. For sustainable utilization of SR resources through high-performance cultivation by selecting suitable plant resources and cultivation areas, Mongolia was chosen because of the large population of wild growing S. divaricata, especially in the eastern part including Khentii and Dornod Provinces, and field investigation and metabolomic analysis were performed on collected plant specimens.

To evaluate the quality of Mongolian S. divaricata, metabolomic profiling of all root parts of 43 specimens from Mongolia, as well as eight SR samples and two plant specimens from China, were conducted using LC-IT-TOF–MS. The LC–MS profiles of the 70% methanol extracts of the specimens and SR samples showed uniformity, and 13 chromones and 17 coumarins were tentatively identified [23]. Among them, a new compound, 3′-O-(6″-O-malonyl)-glucosylhamaudol (S10) [24] and 17 known compounds were isolated and unambiguously identified. Orthogonal partial least squares-discriminant analysis (OPLS-DA) based on LC–MS data revealed that Mongolian specimens were clearly distinguished from Chinese SR and characterized by an abundance of S1. Moreover, Mongolian specimens could be discriminated by their growing regions based on the content of eight compounds (S1–S3, S6, S7, and S9–S11) (Fig. 6A). Specimens from Khalkhgol in far eastern Mongolia contained higher amounts of dihydrofurochromones S1, cimifugin (S2), and S3, while those from Holonbuir contained relatively higher amounts of hamaudol (S6), 3′-O-acetylhamaudol (S7), 3′-O-angeloylhamaudol (S9), and praeruptorin B (S11) [23].

Subsequently, to accurately determine the contents of metabolites in S. divaricata roots from eight different regions of Mongolia and to investigate their geographical variation, nine chromones (S1–S9) and four coumarins were quantified using HPLC-diode-array detection. All 44 Mongolian specimens contained S1 (3.98–20.79 mg/g) and S3 (1.06–6.68 mg/g), with their total content (5.04–25.06 mg/g) exceeding the criterion (2.4 mg/g) assigned in the CP [24] (Fig. 6B). The contents of S1, S7, ledebouriellol (S8), and S9 were significantly higher in the Mongolian specimens than in Chinese SR samples. Moreover, specimens from Norovlin in northeastern Mongolia had the highest level of total dihydrofurochromones (S1–S4; 12.20–26.80 mg/g) including S1 (9.18–16.22 mg/g) and S3 (2.60–6.68 mg/g), while the level of dihydrofurochromones was more variable in specimens from Khalkhgol (5.59–23.41 mg/g) and Holonbuir (3.37–26.12 mg/g) (Fig. 6D). The level of total dihydropyranochromones (S5–S9) tended to be higher in specimens from Bayan-Uul (1.86–5.04 mg/g) in northeastern Mongolia. The OPLS-DA based on HPLC data revealed that the Mongolian specimens tended to be separated into three groups based on growing regions, in which several chromones contributed to each group (Fig. 6C). Furthermore, for characterizing metabolites such as polyacetylenes and sugars, qHNMR analysis showed that Mongolian specimens had less sucrose and a substantial amount of polyacetylenes including panaxynol (S12) [24]. Thus, the chemical characteristics of Mongolian S. divaricata specimens were clarified and showed that specimens from northeast Mongolia, including Norovlin, had superior properties due to higher amounts of major chromones. Although Norovlin was proposed as a prospective region for S. divaricata cultivation, the amount of the compounds varied widely due to growing conditions. Therefore, cultivation method should be studied as a next step.

We investigated a new method for discriminating Curcuma species and for standardizing crude drugs, especially Ezhu/Gajutsu, using molecular analysis based on the intron length polymorphisms (ILPs) in genes encoding diketide-CoA synthase and curcumin synthase, because trnK sequence comparison could not differentiate C. kwangsiensis (gl type) from C. wenyujin with the same K(gl)Wtk type of sequence. Curcuminoids are the most important components in Curcuma drugs. The curcuminoid biosynthesis route of C. longa has been determined, in which two type III polyketide synthases, diketide-CoA synthase (DCS) and curcumin synthase (CURS1) are involved. The DCS catalyzes formation of feruloyldiketide-CoA by condensing feruloyl-CoA and malonyl-CoA; and CURS1 mainly catalyzes formation of curcumin from feruloyl-CoA and the feruloyldiketide-CoA produced by the action of DCS [31]. Moreover, CURS2 and CURS3, isozymes of CURS1, are also responsible for curcuminoid synthesis and have been discovered in C. longa. The CURS2 prefers feruloyl-CoA as a starter substrate and CURS3 prefers both feruloyl-CoA and p-coumaroyl-CoA [32]. Similarly, DCS2, another isozyme of DCS (DCS1), has been identified. Two isozymes of DCS similar to DCS1 and DCS2 and three isozymes of CURS similar to CURS1, CURS2, and CURS3 of C. longa in amino acid sequence were obtained from several Curcuma species irrespective of whether they contain curcuminoids in their rhizomes. The nucleotide sequences of DCS1 and DCS2 of C. longa, which encode DCS1 and DCS2, respectively, contain two introns (DCS intron I and II), which divide the coding region into three exons; and those of CURS1, CURS2, and CURS3, which encode CURS1, CURS2, and CURS3, respectively, each contain two exons and one intron (CURS intron). Although the sequences of exons were conserved, the sequences of introns were quite variable, even in size, which offer potential for diversity analysis. The ILPs are useful genetic markers because they represent variation in the transcribed portion of the genome. The ILP markers developed in various gene regions of many plant species are used for gene tagging, diversity analysis, and comparative studies [33, 34]. Therefore, we developed a molecular method using ILP markers in DCS1, DCS2, and CURS1–CURS3 to discriminate Curcuma species and identify their related crude drug samples [35].

First, two intron regions I and II in DCS1 and DCS2 and one intron region in CURS1–CURS3 were amplified separately via PCR using each of the three pairs of primers. One primer of each pair was labeled with different fluorescent dyes, enabling the respective amplicons to be detected and discriminated. The PCR product of each intron region was mixed with size standard and analyzed using capillary electrophoresis. The length of the amplified fragments was then determined by comparison with size standard markers that included 36 single-stranded labeled fragments. As the result, specimens of C. phaeocaulis, C. zedoaria (Jp), C. wenyujin, and C. aromatica (Jp) showed a species-specific fragment profile, whereas each specimen of C. kwangsiensis (gl type) had an individual fragment profile, showing intraspecies variation. However, all C. kwangsiensis (gl type) specimens fell into a single group in the dendrograms constructed by the fragment patterns in the three intron regions of each specimen and commercial sample [35]. Thus, C. kwangsiensis (gl type) specimens were distinguishable from C. wenyujin specimens using the ILP profile. Combined with variable chemical constituents in C. kwangsiensis specimens, we speculated that there was cross-hybridization between C. kwangsiensis (gl type) and other Curcuma species, supported by the fact that seed setting sometimes occurs in this species. The molecular method we developed has potential for global classification of Curcuma.

To elucidate specific molecular markers of medicinally used Curcuma species in Asia, and to resolve the confusion on the reported botanical origin of crude drugs, molecular analysis based on ILP in DCS and CURS genes and trnK gene sequences was performed using 59 plant specimens and 42 crude drug samples from 13 Curcuma species, obtained from Asian countries including China, Japan, Thailand, Indonesia, India, Nepal, Malaysia, Myanmar, and Sri Lanka [28]. The plant specimens, mostly introduced from the former seven countries, were collected from several medicinal plant gardens in Japan. The botanical origins of crude drug samples were inferred from their local names using relevant literature as a reference. Among these, C. comosa Roxb. was identified as the likely botanical origin of the Thai crude drug “Wan chak modluk.” In the ILP analysis, the length of the amplified DNA fragments ranged within 213–276 bp in DCS intron I region, 274–308 bp in DCS intron II region, and 194–256 bp in CURS intron region. The ILP patterns of the respective species revealed high intraspecies consistency in C. aromatica from Japan and China; C. zedoaria from Japan, Indonesia, and India; and C. phaeocaulis, C. aeruginosa Roxb., C. wenyujin, and C. zanthorrhiza Roxb.; but showed intraspecies polymorphism in C. longa, C. kwangsiensis, C. amada Roxb., C. mangga Valeton et Zijp, and C. comosa. The similarity of ILP patterns in the specimens and samples of the respective species led to clear clustering in the neighbor-joining tree, with 11 main groups corresponding to the respective species: groups L (C. longa), JA (C. aromatica), Ze (C. zedoaria), Ae (C. aeruginosa), P (C. phaeocaulis), W (C. wenyujin), K (C. kwangsiensis), Za (C. zanthorrhiza), A/M (C. amada or C. mangga), Pe (C. petiolata), and C (C. comosa) (Fig. 7). Groups Pe and C formed a clade, separated from the large clade, which was further divided into two subclades. Group L formed one subclade and was further divided into three subgroups and this grouping was highly consistent with the geographical origins of the included samples; they were tentatively assigned as China–Japan (L1), Thailand (L2), and India–Indonesia (L3) groups. Another subclade comprising the other species was further divided into two branches: one composed of groups JA, Ze, Ae, P, W, and K; and the other composed of Za and A/M. Intraspecies polymorphism in group L is believed to arise from a combination of geographic variation and the presence of numerous cultivars in C. longa. For groups A/M and C, the polymorphism is attributed to taxonomic confusion and instances of crossbreeding in C. amada, C. mangga, and C. comosa. For the trnK sequence, in addition to five typical sequences [Ltk, Ptk, Atk, K(pl)Ztk, and K(gl)Wtk], two new types were found with high similarities to the K(pl)Ztk type, but differing in poly-A and poly-T numbers at positions 205 and 502, respectively. The K(pl)Ztk type of sequence detected in C. zedoaria from Japan had six As and 14 Ts at these two sites [renamed K(pl)Ztk(6A14T)], whereas new types had seven As and 15 or 13 Ts—named K(pl)Ztk(7A15T)[LC636648] or K(pl)Ztk(7A13T) [[LC636649], respectively. Curcuma aeruginosa, C. amada, and C. mangga possessed the Ptk type of sequence; C. petiolata, the Ltk type; and C. zanthorrhiza, the K(pl)Ztk(6A14T) type. The five crude drug samples from Thailand, four “Wan chak modluk” and one “Wan maha mek,” which belonged to group C in ILP pattern (C. comosa) were of K(pl)Ztk(7A15T) type. The three Thai samples, one “Wan chak modluk” of group C and two “Kamin oi” of group L2 in ILP pattern, possessed the K(pl)Ztk(7A13T) type of sequence. Considering that group L2 was of C. longa, the botanical origins of these “Khamin oi” were determined to be a hybrid between C. longa (paternal line) and another species (maternal line) with a K(pl)Ztk(7A13T) type of trnK sequence. This species might be C. comosa, even though “Khamin oi” was inferred to be C. zedoaria based on Thai literature. In a similar situation, although the crude drug “Kasturi manjal” from India and “Wan narn kum” from Thailand were considered to be C. aromatica in both countries, the botanical origins of two and one samples obtained were demonstrated as C. zanthorrhiza belonging to group Za in ILP pattern and with mostly K(pl)Ztk(6A14T) type of trnK sequence. Therefore, these crude drugs should be used with caution. Moreover, molecular information showed that C. aromatica and C. zedoaria cultivated in Japan were closely related to C. aromatica from China and Thailand, and C. zedoaria from Indonesia and India, respectively. Thus, ILP markers in DCS and CURS genes combined with trnK gene sequences were demonstrated as useful for taxonomic arrangement of Asian Curcuma species and standardization of Asian Curcuma drugs. For more concise results regarding these difficult questions, however, further study including morphological comparison with the specimens from type locality and molecular investigation on variability of ILP pattern in hybrid plants is needed.

To elucidate the sequence differences in intron regions of the DCS and CURS genes and to search for specific sequences suitable for identification of Curcuma species, six plant specimens from five Curcuma species (C. longa, C. zedoaria, C. phaeocaulis, C. aromatica, and C. zanthorrhiza) that showed distinct ILP patterns were subjected to subcloning coupled with sequencing analysis for the DCS intron I and CURS intron regions [36]. More than 30 sequences of each region from each specimen were grouped into genes DCS1, DCS2, or CURS1–CURS3 and subsequently the sequences of the same genes were compared. Sequences belonging to the same gene showed inter-species similarity, and thus these intron sequences were less informative within each single gene region. The determined sequences from each specimen showed 3–5 kinds of sequence lengths in DCS intron I region, and 5–7 kinds of sequence lengths in the CURS intron region. The fragment numbers and lengths matched those of the corresponding ILP patterns, effectively clarifying the origin of the ILP pattern in Curcuma species.

The EOs comprising various bioactive compounds have been widely detected in Curcuma species. Due to the widespread distribution and misidentification of Curcuma species and differences in processing methods, inconsistent reports on major compounds in rhizomes of the same species from different geographical regions are not uncommon. This inconsistency leads to confusion and inaccuracy in compound detection of each species and also hinders comparative study based on EO composition. Then, to characterize EO compositions of Curcuma species, as well as to determine the compositional variation among different species, and between plant specimens and their related crude drug samples, 47 plant specimens of 11 Curcuma species and 20 crude drug samples identified genetically were analyzed using HS-SPME–GC–MS [37]. Plant specimens of the same species showed similar EO patterns, regardless of geographical source. Based on EO patterns, all specimens and samples were separated into eight main groups: C. longa (L); C. phaeocaulis, C. aeruginosa, and C. zedoaria (P-Ae-Ze); C. zanthorrhiza (Za); C. aromatica and C. wenyujin (JA-W); C. kwangsiensis (K); C. amada and C. mangga (Am-M); C. petiolata (Pe) group; and C. comosa (C) groups (Thai crude drug samples were used). From EOs of all specimens and samples, 54 compounds (C1–C54) were identified (Table 1). The eight groups contained characteristic sesquiterpenes as well as monoterpenes belonging to the bisabolane type (L); the elemane- and germacrane-types (P-Ae-Ze); the bisabolane-type (Za); the germacrane-type besides eucalyptol (C5) (JA-W); the elemane- and germacrane-types besides camphor (C10) (K); the caryophyllane-type besides β-pinene (C2) and β-myrcene (C3) (Am-M); the caryophyllane- and germacrane-types besides C2 and C5 (Pe); and the humulane-type besides C5 (C), recognizing two subtypes whether the santalene- and bisabolane types were simultaneously present (type I) or not (type II) (Table 2). Most of the major compounds detected in the plant specimens were also observed in crude drug samples, although a few compounds such as β-curcumene (C26), turmerone (C40), and 4,5-epoxygermacrone (C52) degraded due to processing procedures or over time. We supposed C26 may convert into ar-curcumene (C29), and C52 into curcumenol (C51). Two “Khamin oi” samples from Thailand determined genetically to be hybrids between C. longa and C. comosa with a K(pl)Ztk(7A13T) type of trnK sequence, contained major compounds similar to C. longa and C. comosa type II. Identification of the marker compounds to discriminate each group or each species was achieved after re-discrimination study using the GC–MS data of 47 plant specimens using OPLS-DA. For example, specimens belonging to group P-Ae-Ze were used for OPLS-DA, and C. phaeocaulis, C. aeruginosa, and C. zedoaria were separated from each other by curzerenone (C41) and curzerene (C32); zingiberene (C23), ar-curcumene (C29), and germacrone (C43); and C5 and C10, respectively [37]. By relying on several marker compounds of each species and considering the likelihood of degraded compounds, the identification of Asian Curcuma drugs can be achieved to a reasonable degree.

In summary, molecular analyses of the ILP markers in the DCS and CURS genes and trnK gene sequences combined with EO composition analysis were demonstrated to be useful for taxonomic resolution of Asian Curcuma species and the standardization of Curcuma crude drugs.

The medicinal properties of C. longa are mainly attributed to its abundant content of curcuminoids including curcumin, demethoxycurcumin, and bisdemethoxycurcumin, reportedly possessing anti-inflammatory, anti-oxidant, and anti-cancer activities [38], as well as anti-metastatic activity of cancer [39]. From Supplement II to JP15 [40], Turmeric, “Ukon” in Japanese (the Latin name, Curcumae Rhizoma was changed to Curcumae Longae Rhizoma in Supplement I to JP17 [41]) is prescribed to contain not less than 1.0% and not more than 5.0% of total curcuminoids using the LC assay described in the JP with the standard solution of curcumin. Moreover, one identification test must be done using the sample solution as directed in the assay to confirm that the peak area of curcumin is larger than that of demethoxycurcumin, and is larger than 0.69 times that of bisdemethoxycurcumin, which means the content of curcumin is higher than those of both other curcuminoids [6]. Before such a standard was determined, we obtained the following HPLC results to quantify each curcuminoid: turmeric samples from Japanese, Chinese, Indian, and Thailand markets contained 0.24–3.64%, 0.56–4.47%, 2.02–4.08%, and 3.28–4.92% of total curcuminoids, respectively, and curcumin occupied 54–72% of total curcuminoids in all samples. Among other Curcuma crude drugs, “Khamin oi” samples from Thailand derived from a hybrid between C. longa and C. comosa with a K(pl)Ztk(7A13T) type of trnK sequence contained 1.2–2.0% curcuminoids, in which demethoxycurcumin content exceeded that of curcumin [30]. Therefore, this identification test, as well as quantifying curcuminoids, is essential to avoid curcuminoid-containing crude drugs other than turmeric.

For the other Curcuma crude drug, “Gajutsu” in Japanese, although C. zedoaria has been prescribed as the botanical origin until Supplement I to JP17 [41], C. phaeocaulis and C. kwangsiensis were newly added, and the English and Latin names were changed to Curcuma Rhizome and Curcumae Rhizoma from Zedoary and Zedoariae Rhizoma, respectively. As mentioned above, Gajutsu samples derived from C. kwangsiensis cultivated in Guangxi Zhuangzu Autonomous Region lacked genetic and chemical stability due to their hybrid origin, whereas those from C. phaeocaulis in Sichuan Province and those from C. zedoaria in Japan were stable. Since Gajutsu has been used to treat “Oketsu” in Japanese (the syndrome of blood stasis, insufficient blood circulation, sometimes caused by inflammation) [5], the anti-Oketsu effects were examined by two pharmacological studies using genetically identified crude drugs: effects on vasomotion in rat aortic rings, especially on endothelial-dependent relaxation effect of the blood vessel [42]; and anti-inflammatory activity using an arthritis mice model induced by Complete Freund’s Adjuvant injection [43]. In the former study, all methanol extracts of five Curcuma drugs exhibited intense nitric oxide (NO)-independent relaxation and all water extracts showed relaxation effects as a sum of the methanol-soluble compounds-induced relaxation and polysaccharides-induced contraction. Only the water extract of C. zedoaria showed NO-dependent relaxation and NO-independent relaxation, suggesting the drug derived from C. zedoaria has the potential to cure Oketsu through various acting points [42]. In the latter study, the methanol extract of C. phaeocaulis significantly inhibited paw swelling in the right hind footpad of mice and serum haptoglobin concentration, whereas that of C. kwangsiensis significantly inhibited serum haptoglobin concentration when orally administrated 1 day before the injection. Moreover, for in vitro assay, cyclooxygenase (COX)-2 activity was significantly inhibited by the methanol extract of C. phaeocaulis. Therefore, C. phaeocaulis could be a useful drug among Curcuma species for treating acute inflammation [43]. Subsequently, furanodienone and C51 were identified as the major active anti-inflammatory constituents of C. phaeocaulis, through a new approach for investigation of bioactive constituents in crude drugs using partial least squares regression model and detailed analysis of the regression vector, followed by isolation of these compounds and their COX-2 inhibitory assays. The COX-2 inhibitory activities of furanodienone and C51 were weaker (IC₅₀ 4.6 and 34.5 μM, respectively) than those of the positive control indomethacin; however, selectivity indices (IC₅₀ of COX-1/IC₅₀ of COX-2) of both compounds (11.4 and 4.6) were higher than that of indomethacin (0.05) [44]. In previous section, franodienone was not mentioned, because this compound was heat-sensitive and could be completely transformed into C41 in our GC–MS condition [45]. According to our quantitative analysis using the HPLC method, the methanol extract of C. phaeocaulis rhizome contained franodienone and C41 at an average ratio of 2.5:1. Referring to the aforementioned results and considering the distribution amounts, it is evident that C. phaeocaulis, C. kwangsiensis, and C. zedoaria are the likely botanical origins of Gajutsu, Curcuma Rhizome as recognized in the JP [6].