Chemical and genetic diversity of Astragalus mongholicus grown in different eco-climatic regions

Astragalus mongholicus Bunge (Fabaceae) is an important plant source of the herbal drug known as Radix Astragali, which is used worldwide as a medicinal ingredient and a component of food supplement. Russian Federation, Mongolia, Kazakhstan, and China are the main natural distribution areas of A. mongholicus in the world. However, the quality of medicinal plant varies among different locations. As for A. mongholicus, limited literature focused on its biodiversity mechanism. Here, we combined the chemometric analysis of chemical components with genetic variation, as well as climatic and edaphic traits, to reveal the biodiversity mechanism of A. mongholicus. Results showed that the detected chemical, genetic and climatic traits comprehensively contributed to the quality diversity of A. mongholicus. The eight main chemical components, as well as the inorganic elements of P, B and Na were all significant chemical factors. The precipitation and sunshine duration were the main distinguishing climatic factors. The inorganic elements As, Mn, P, Se and Pb were the distinguishing edaphic factors. The systematic method was firstly established for this medicinal plant in order to illustrate the formation of diversity in terms of quality, and provide scientific evidence for geographic indications and climatic adaptation in production and in the clinical application of herbal medicinal plants.


Introduction
Astragalus mongholicus Bunge (Fabaceae) is the main plant source of the herbal drug Radix Astragali, which is recorded by European Pharmacopoeia (EP, 8 Japanese, and Korean systems of medicine but also widely cultivated and used outside of Asia, such as Germany and the United States [1] for its immunostimulant, anti-perspirant, antidiarrheal, anti-diabetic and tonic properties [2]. According to previous studies, the content of astragalosides, calycosin-7-glucoside and formononetin content varied among different habitats, which indicated that different habitats possessed medicinal materials of different composition, quality and strength [3,4]. This phenomenon was related to biodiversity, local adaptation or 'geoherbalism' which involves the use of geo-herbs for higher clinical efficacy [5]. There is a relationship between geographical indications and geoherbalism [6]. In China, geoherbalism can be protected as traditional Chinese medical knowledge under the geographical indication regulatory framework [7]. Generally, geo-herbs have higher quality and are distributed in specific geographic origins with characteristic natural conditions and ecological environments. Furthermore, geo-herbs involve particular techniques for cultivation, harvesting and processing. Therefore, the quality and clinical effects surpass those of the material coming from the same botanical origin but produced in other regions [8]. The content of chemical components and therapeutic effects are the essential differences between the same medicinal materials from geo-producing areas and other locations. The total sugar, reducing sugar and soluble polysaccharide content in Asian ginseng (Panax ginseng C.A. Mey., Araliaceae) root samples from fifteen different areas showed variations related to geographic location [9]. The stem of Dendrobium candidum Wall. ex Lindl. (Orchidaceae) obtained from six different areas showed significantly different content of mannose and polysaccharide [10]. Several relevant studies have focused on ecological and geographical differences using combined technologies and statistical analyses to determine the biodiversity and geoherbalism mechanism in herbal medicines [11][12][13][14][15]. For A. mongholicus, studies have tried to explain the differences in the molecular variation and soil properties. However, limited literature accounted for the mechanisms of biodiversity by systematical method and reciprocal correlation based on chemical, genetic, climatic and edaphic traits [16][17][18][19]. In the present study, A. mongholicus was used as a representative medicinal plant to interpret the mechanism of biodiversity and local adaptation that could be attributed to internal (chemical components and genetic variation) and external (climatic and edaphic traits) causes based on chemometric and correlation analysis, which is essential for the geographic indications and further scientific development of herbal medicinal plants.

Materials information
The A. mongholicus samples were collected from three locations, namely, the provinces of Inner Mongolia, Shanxi and Gansu, which are the main producing areas in the People's Republic of China (S1 Table, Fig 1). The samples were collected on 28 October, 2012 after twoyears of cultivation in unified size and quality grade. The raw medicinal materials were cleaned by water, removing the fibrous root and root apex, and then hung to dry in the sun naturally and uniformly. The dried samples were crushed into powder separately for further analysis of components. The samples were handled identically and met the same basic inclusion criteria. All corresponding voucher specimens were deposited to the Herbarium of the Institute of Medicinal Plant Development (IMPLAD) at the Chinese Academy of Medical Sciences in Beijing, China.
The content of Astragalus polysaccharide was determined by the phenol-sulfuric acid method [20]. The inorganic elements in the herb and soil were determined by inductively coupled plasma atomic emission spectrometry (ICP-AES) [21]. The method mainly referred to the general principles of ICP-AES.

DNA extraction, PCR amplification and sequencing
The material specimens were dried by natural methods, and 30 mg of dried plant material was used for DNA extraction. Samples were rubbed for 2 min at a frequency of 30 r/s in a FastPrep bead mill (Retsch MM400, Germany). The total genomic DNA was isolated from the crushed material according to the manufacturer's instructions (Plant Genomic DNA Kit, Tiangen Biotech Co., China). We made the following modifications to the protocol: chloroform was diluted with isoamyl alcohol (24:1), and the buffer solution GP2 was diluted with isopropanol (same volume). The powdered sample, 700 μL of 65˚C GP1 and 1 μL of β-mercaptoethanol were mixed for 10-20 s before the mixture was incubated for 60 min at 65˚C. Subsequently, 700 μL of a chloroform: isoamyl alcohol mixture was added, and the solution was centrifuged for 5 min at 12, 000 rpm (~13400×g). The supernatant was removed and transferred into a new tube before adding 700 μL isopropanol and mixing for 15-20 min. The mixture was centrifuged in CB3 spin columns for 40 s at 12,000 rpm. The filtrate was discarded and 500 μL GD (adding quantitative anhydrous ethanol before use) was added before centrifugation at 12,000 rpm for 40 s. The filtrate was discarded, and 700 μL PW buffer (quantitative anhydrous ethanol was added before use) was used to wash the membrane before centrifugation for 40 s at 12,000 rpm. This step was repeated with 500 μL PW, followed by a final centrifugation step for 2 min at 12,000 rpm to remove the residual wash buffer. The spin column was dried at room temperature for 3-5 min and then centrifuged for 2 min at 12,000 rpm to obtain the total DNA.
General PCR reaction conditions and universal DNA barcode primers were used for the ITS, ITS2, and psbA-trnH barcodes, as presented in S2 Table. PCR amplification was performed on 25 μL reaction mixtures containing 2 μL of the DNA template (20-100 ng), 8.5 μL of ddH 2 O, 12.5 μL of 2×Taq PCR Master Mix (Beijing TransGen Biotech Co., China), and 1/1 μL of the forward/reverse (F/R) primers (2.5 μM). The reaction mixtures were amplified in a 9700 GeneAmp PCR system (Applied Biosystems, USA). Amplicons were visualized by electrophoresis on 1% agarose gels. The purified PCR products were sequenced in both directions by an ABI 3730XL sequencer (Applied Biosystems, USA).

Climatic data collection
The climatic data were collected from the China meteorological data sharing service system (Website: http://cdc.cma.gov.cn/home.do). The data of 20 climatic factors from 1951 to 2012 were included in our analysis.

Data analysis
The PCA, one-way ANOVA, and correlation analyses for the chemical, climatic and molecular factors as well as the inorganic elements in the soil were performed with the SPSS for Windows (release version 22.0; SPSS Institute, Cary, NC, USA).
The DNA sequencing peak diagrams were obtained and proofread before the contigs were assembled with CodonCode Aligner 5.0.1 (CodonCode Co., USA). The complete ITS2 sequences were obtained via the HMMer annotation method based on the Hidden Markov model [22]. All of the sequences were aligned using ClustalW. The ML trees were constructed based on the Tamura-Nei model, and bootstrap tests were conducted with 1,000 repeats to assess the confidence of the phylogenetic relationships by MEGA 6.0 software [23].

Chemical analysis
The content of the main chemical components in the root of A. mongholicus obtained from different locations were determined and analyzed (Fig 2). The main chemical organic constituents and inorganic elements were respectively shown in Fig 2(a) and 2(b). The content of eight main constituents was distinct among the different locations, and the samples from Shanxi Province had the highest total content compared with other two origins. For the content of inorganic elements, samples from Gansu Province possessed slightly higher average content than those from the other two locations. Furthermore, the content of toxic metal elements, including Pb, Cd, Cr and As were in the range of the safety values (The Pharmacopoeia of the People's Republic of China requires that Radix Astragali should contain not-more-than (NMT) 5 ppm of Pb, 0.3 ppm of Cd, 2 ppm of As, 0.2 ppm of Hg, and NMT 20 ppm of Cu). From the loading diagram and PCA analysis, three origins were significantly divided into three parts (Fig 2(c) and 2(d)).

Genetic variation analysis
Three DNA barcodes of the ITS, ITS2, and psbA-trnH intergenic region were obtained and used for the analysis of the genetic variation among A. mongholicus from different origins [24]. Maximum likelihood (ML) trees were constructed based on the Tamura-Nei model (Fig 3). The barcodes could distinguish A. mongholicus from three different locations. The psbA-trnH intergenic region performed better than ITS and ITS2 regions. The results showed that A. mongholicus from different origins possessed genetic variation.

Climatic and edaphic traits analysis
For this study, 20 climatic factors and 19 edaphic factors collected from different locations were analyzed. The loading diagram and PCA were employed (Fig 4), which demonstrated that different origins possessed discriminating climatic and edaphic conditions. The analysis for climatic factors showed that the mean temperature, extreme maximum temperature, maximum wind speed and the maximum daily precipitation distributed around the coordinate axis were the principal component traits that influenced the biodiversity of A. mongholicus from different origins (Fig 4(a) and 4(b)). The inorganic elements As, Mg, Ca and K were the principal component traits that influenced the biodiversity of different origins in terms of edaphic factors (Fig 4(c) and 4(d)). According to the PCA, Shanxi and Inner Mongolia provinces were   more closely clustered than Gansu province, and this result was consistent with the distribution of A. mongholicus (Fig 1).

ANOVA analysis
One-way analysis of variance (ANOVA) was performed for the chemical, molecular, climatic and edaphic factors (Table 1). We aimed to analyze each factor and to find the distinguishing factors for the subsequent correlation analysis. As shown in Table 1, compared to ITS region, ITS2 was more significant in molecular factor. The eight main chemical components, as well as the inorganic elements of P, B and Na were all significant chemical factors. The precipitation, mean air speed, sunshine duration, evaporation capacity, extreme wind speed, mean relative humidity, percentage of sunshine, mean minimum air temperature, the number of days of daily precipitation (!0.1 mm) and the extreme minimum air temperature were the distinguishing climatic factors. The inorganic elements As, Mn, P, Se and Pb were the distinguishing edaphic factors. The results provided a reference for the mutual authentication of the correlation analysis of the four influence factors.

Correlation analysis and diversity evaluation
Based on the loading diagram, PCA and ANOVA analysis for four factors, the correlation between four factors were conducted (Fig 5). For the chemical and genetic traits, the genetic distance of ITS2 was negatively correlated with the chemical components, except for the inorganic elements of P and Se in the roots, which were positively correlated with ITS2. The genetic distance of ITS was negatively correlated with Ca, Fe, Ba, Na and Cr, but was positively correlated with P. The psbA-trnH intergenic region showed no significant differences in terms of genetic distance. Thus, we excluded this region in subsequent analysis. For the chemical and climatic factors, the main chemical components calycosin-7-glucoside, polysaccharides, kaempferol, quercetin, and ononin were strongly positively correlated with the climatic traits of mean air speed, sunshine duration, evaporation capacity, extreme wind speed and percentage of sunshine, but these components were strongly negatively correlated with the precipitation, mean relative humidity and mean minimum air temperature. Meanwhile, formononetin and calycosin were strongly negatively correlated with the mean air speed, sunshine duration, evaporation capacity, extreme wind speed and percentage of sunshine, but were strongly positively correlated with the precipitation, mean relative humidity and mean minimum air temperature. Astragaloside was weakly positively correlated with the mean air speed, sunshine duration, evaporation capacity, extreme wind speed and percentage of sunshine but was weakly negatively correlated with the precipitation, mean relative humidity and mean minimum air temperature. For the chemical and edaphic factors, polysaccharide, kaempferol and ononin were strongly positively correlated with the inorganic elements Pb, As and Mn, but were weakly negatively correlated with Mg. Formononetin and calycosin were strongly positively correlated with Pn, As and Mn but weakly negatively correlated with Mg. Calycosin-7-glucoside and quercetin had no correlation with Pb, As, Mn and Mg. The inorganic element of B in the roots was positively correlated with Pb, As and Mn, but negatively correlated with Mg. Conversely, P in the roots was negatively correlated with Mg, but positively correlated with Pb, As and Mn. The inorganic elements in the roots were not correlated with the climatic factors.

Discussion
Different geographic locations, climatic conditions and edaphic traits may contribute to the different quality levels of medicinal plants. In this study, the A. mongholicus samples were collected from three main producing areas, namely, Shanxi, Inner Mongolia and Gansu provinces in China. The samples were analyzed in terms of their chemical, genetic, climatic and edaphic traits, which were the main factors of influence corresponding to the quality of herbal material. The eight main active constituents included six flavonoids, astragaloside and Astragalus polysaccharide [25]. The content of the main active constituents showed that  A. mongholicus produced in Shanxi province contained the highest total content. By contrast, the content of the inorganic elements in the roots were not significantly different. Remarkably, the content of Astragaloside was higher in the samples of Inner Mongolia than those from the other two locations. As an isolated active constituent, Astragaloside mainly confers a protective effect against ischemic-heart disease and brain injury. The content of flavonoid and Astragalus polysaccharide were higher in the samples from Shanxi province. Flavonoid and polysaccharide components of A. mongholicus are mainly used for anti-neoplastic properties and inhibition of atherosclerosis formation. The present results were consistent with those of previous studies [26,27].
To explore the genetic variation between different locations, we employed three commonly used DNA barcodes in medicinal plants, ITS, ITS2 and psbA-trnH intergenic region, to analyze A. mongholicus from three different locations. The ML tree showed that the samples from different locations were divided into different clusters. The results were consistent to a certain degree with those of previous studies [28,29]. According to the report by Liu et al., the ITS sequence for A. mongholicus from different regions were highly conservative at an intra-specific level. In our study, we confirmed the performance of ITS and its partial sequence of ITS2 in A. mongholicus. The ML tree of ITS and ITS2 were in substantial agreement, because ITS2 was a part of the ITS sequence. Furthermore, we combined the chloroplast barcodes, psbA-trnH intergenic region, which performed well compared with ITS and ITS2 sequence in our previous studies [30][31][32]. The psbA-trnH intergenic region performed better than the other two barcodes because it divided the samples into three observable clusters. The results indicated that A. mongholicus from different locations possessed stable genetic variation in the psbA-trnH intergenic region, which could be used as a molecular marker to distinguish samples from different locations.
In addition, the analysis of climatic and edaphic traits demonstrated that A. mongholicus from different locations had different climatic and edaphic conditions. The PCA results showed that the three locations were obviously divided into three clusters. Nonetheless, the quality of Radix Astragali is the result of comprehensive influences. The ecological environment, which comprises the climatic and edaphic traits, was the main external cause, whereas the chemical component and genetic variation were the main internal causes that influenced the quality of Radix Astragali. Therefore, we analyzed the correlation between the chemical components and the genetic variation as well as the climatic and edaphic traits.
The ITS2 barcode was strongly negatively correlated with the main chemical components of Astragalus polysaccharide, astragaloside, kaempferol, and ononin. By contrast, ITS was not correlated with the main chemical components. The results indicated that the variation of genetic distance was negatively correlated with the chemical components on a certain level. In terms of the edaphic inorganic element and chemical components, we found that the harmful elements of Pb and As were strongly negatively correlated with the content of Astragalus polysaccharide, astragaloside, kaempferol, and ononin, but were strongly positively correlated with the content of formononetin and calycosin. These results were interesting because several heavy metals, such as Cu and Zn, are essential trace minerals of plants. Meanwhile, some heavy metals, such as Pb, Cd and Hg, are trace minerals that are not necessary for the growth of plants, and a very small amount can cause toxic contamination of the plants [33]. The inorganic element of Mn, which is a transition metal, was strongly negatively correlated with the Astragalus polysaccharide, kaempferol and ononin content, but strongly positively correlated with formononetin and calycosin. The inorganic element Mg was negatively correlated with formononetin and calycosin, but was positively correlated with Astragalus polysaccharide, astragaloside, kaempferol, and ononin. Among the climatic factors and chemical components, the mean air speed, sunshine duration, evaporation capacity, percentage of sunshine and extreme wind speed were strongly positively correlated with the chemical content of calycosin-7-glucoside, polysaccharide, kaempferol, quercetin, ononin, and astragaloside, thereby indicating that the above mentioned climatic traits affected the quality of A. mongholicus.

Conclusion
This work proposes a systematic and comprehensive method for estimating the quality of A. mongholicus from different locations based on internal (chemical component and genetic variation) and external (climatic and edaphic traits) causes. The correlation analysis proved that the genetic, climatic and edaphic traits were similarly closely correlated with the content of the chemical components. This proposed method presents useful data for estimating the quality of A. mongholicus. Thus, the findings reported here firstly establish a scientific method to clarify the mechanisms of the biodiversity and local adaptation for medicinal plants. Further study should be focused on genome-wide studies, and developing suitable methods for quantitative traits analysis, which is conducive to better understand the biodiversity and local adaptation in plant population, especially for medicinal plants.
Supporting information S1