Introduction

The genus Stephania Lour. (Menispermaceae) includes approximately 60 species of herbaceous or woody vines with tropical and subtropical distribution in Asia and Africa. China is the diversity center of Stephania, with 40 species distributed mainly to the south of the Yangtze River. This genus has been classified into three subgenera (subgen. Stephania, subgen. Tuberiphania, and subgen. Botryodiscia) [1, 2]. Many species of the genus Stephania are used as traditional medicines for the treatment of diseases such as dysentery, fever, tuberculosis, rheumatism, and cancer [2, 3]. Stephania plants contain a variety of alkaloids that possess various pharmacological activities, e.g., antimicrobial, analgesic, anticancer, anti-viral, anti-inflammatory, and antipsychotic activities [2,3,4]. Apart from their diverse medicinal uses, several Stephania plants exert toxic effects [3]. Thus, clinical misuse of Stephania plants may be ineffective or even detrimental to patients.

In recent decades, overexploitation of medicinal Stephania plants has resulted in 22 Stephania species being listed on the China Biodiversity Red List [5]. However, indiscriminate harvesting of Stephania plants is still prevalent, mainly due to misidentification of species with similar morphological characteristics. Despite many studies describing the taxonomic delimitation, bioactive components, and genetic diversity of Stephania plants [1,2,3,4, 6,7,8], relatively few articles have addressed identification at species level [1, 9]. Due to the morphological similarities and variations, it is very difficult to identify many Stephania species with certainty using the conventional morphological methods [1]. A UPLC-QTOF-MS/MS method developed for profiling the alkaloids of Stephania plants can differentiate between the three species being investigated [9]. However, the three species, belonging to the three different subgenera, have relatively pronounced morphological features for differentiation, and may be relatively easy to be distinguished by chemicals. The efficiency of phytochemical analysis for distinguishing between Stephania plants, especially closely related species belonging to the same subgenus, needs further study. Thus, it is difficult to efficiently identify Stephania species using the currently reported methods.

DNA barcoding has become an effective tool for species identification of medicinal plants [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30]. The psbA-trnH intergenic spacer [14, 19,20,21] and nuclear ribosomal ITS or ITS2 [22, 23] have been successfully applied to species identification. Barcode combinations such as ITS+matK [25] and ITS/ITS2+psbA-trnH [13, 15, 24, 26, 29] have achieved high identification rates at the species level. Although our group has studied the phylogeny and infragenetic classification of Stephania using ITS and trnL-F regions [7], species-level differentiation remains problematic because of insufficient information for accurate species identification and limited sample size of closely related medicinal species.

In this study, six barcoding loci, including ITS, ITS2, psbA-trnH, matK, rbcL, and trnL-F, were evaluated either as single region or in combination for species discrimination in the genus Stephania, and their species discrimination efficiencies were compared by different analytical methods. In addition, some insights into the phylogenetic relationship between Stephania species from a DNA barcode perspective were provided.

Materials and methods

Plant materials

Sixty-four accessions from 23 species were sampled (Table S1), of which 12 species were listed on the China Biodiversity Red List. Two additional samples of Cyclea hypoglauca (Schauer) Diels were used as out-groups. All species were identified by Dr. Yun Kang, Fudan University. The fresh leaf and root samples were desiccated by silica gel immediately after collection. The voucher specimens were deposited in the herbarium of the School of Pharmacy, Fudan University (SHMU).

DNA extraction, polymerase chain reaction (PCR) amplification, and sequencing

Total genomic DNA was extracted from samples using the Plant Genomic DNA Kit (Axygen, Biheng Biotechnology Co., Shanghai, China) following the manufacturer’s protocol. Six DNA barcode sequences were amplified using the primers and conditions described in Table 1. The PCR amplification was performed in a 25 µL reaction mixture with 30 ng template genomic DNA, 2.5 µL of 10 × PCR buffer (Mg2+ Plus) (TaKaRa, Biheng Biotechnology Co., Shanghai, China), 2 µL dNTP (2.5 mM each), 0.2 µL Taq DNA Polymerase (5 U/µL), and 0.5 µL of 10 µM forward and reverse primers (Sangon Biotech (Shanghai) Co., Ltd., Shanghai, China). The PCR products were detected by 1.0% agarose gel electrophoresis. Finally, the PCR products were purified using PCR purification kit (AxyPrepTM Gel Extraction Kit, Axygen Biosciences) and sequenced on an ABI 3730XL automated sequencer (Thermo Fisher Scientific Co., Suzhou, China).

Table 1 Primers and PCR conditions
Full size table

Sequence alignment and data analysis

Bidirectional sequences were assembled and edited by Contig and the edited sequences were aligned using Clustal X2 [31]. The assembled sequences were submitted to the National Center for Biotechnology Information (NCBI). Intraspecific and interspecific genetic distances were calculated using the Kimura two-parameter (K2-P) model [32] by MEGA 6.0 [33]. Further, the barcoding gap histograms were estimated. Jmodeltest 2.1.10 [34] was used to calculate the most suitable evolution model for the DNA barcode sequences of Stephania species. Phylogenetic trees were constructed by MEGA 6.0 [33] and Mrbayes 3.1.2 [35]. The discriminatory power of the DNA barcodes was evaluated by TaxonDNA [36].

Results

Sequence analysis

In this study, the amplification and sequencing reactions were performed with high success (100%) for all loci of the 64 samples from 23 Stephania species. Among six DNA barcodes, ITS provided the largest number of variable and informative sites (182/179), followed by psbA-trnH (173/162), trnL-F (103/98), matK (91/82), ITS2 (76/74), and rbcL (30/28). All barcodes except rbcL contained insertions and deletions (indels). The largest length variation was detected in psbA-trnH, consisting of 18 indels. The GC content of psbA-trnH ranged from 24.8 to 27.5% and was lower than the other five barcodes. Intraspecific distances in the six barcodes ranged from 0.0000 to 0.0278, while interspecific distances ranged from 0.0000 to 0.2832 (Table 2).

Table 2 Statistics of candidate DNA barcodes
Full size table

DNA barcoding gap assessment

A PWG-distance method based on K2-P distance was adopted to estimate the barcoding gap among the six barcodes. When interspecific and intraspecific distances of the six barcodes were compared, it was found that interspecific distances were greater than intraspecific distances in Stephania species (Tables 2, S2–S5). ITS exhibited the shortest mean intraspecific distance (0.0005), and ITS2 showed the longest mean interspecific distance (0.1019). For an ideal barcoding gap, the minimum interspecific divergence should be greater than the maximum intraspecific variation. However, the overlap of genetic distances increases as the number of closely related species increases [14, 27]. Yet, despite this overlap, histograms of the barcoding gap showed that there existed a defined range between the intraspecific and interspecific divergence of the six barcodes (Fig. S1). Interspecific variation was greater than intraspecific variation. The results were confirmed by Wilcoxon two sample tests and a median test (Tables S4, S5).

Species discrimination

TaxonDNA was employed to analyze all sequences generated in this study. Both the “best match” and “best close match” methods achieved similar species discrimination success (Table 3). PsbA-trnH and trnL-F obtained the highest species discrimination for single regions (both 83.33%), followed by rbcL (50.00%), matK (48.48%), ITS (42.42%), and ITS2 (36.36%). With respect to the two-region combinations, ITS+psbA-trnH and ITS+trnL-F exhibited the highest discriminatory power (both 93.93%), followed by ITS2+psbA-trnH (92.92%), psbA-trnH+matK (83.83%), psbA-trnH+rbcL (83.83%), and psbA-trnH+trnL-F (83.83%). Combinations of three and four regions did not increase the discrimination rate any further (Table 3).

Table 3 Species identification success based on best match and best close match
Full size table

Tree-based analyses (Figs. 1, S2–S6) showed that the species resolution rates for combination barcodes were higher than those of single-region barcodes. Trees using one barcode could not provide an accurate resolution for Stephania species. The neighbor-joining (NJ) trees displayed similar clustering patterns to those of the Bayesian inference (BI) trees. The highest identification rate (100%) was achieved by the combination of nuclear ITS and chloroplast psbA-trnH using either NJ or BI method (Fig. 1, S2).

Fig. 1
figure 1

Phylogenetic tree of ITS+psbA-trnH barcode using neighbor-joining method

Full size image

Discussion

The results of this study showed that the barcode candidates exhibited different ability of species discrimination in the genus Stephania and combination barcodes could achieve higher discrimination rates than single-region barcodes. The Consortium for the Barcode of Life (CBOL) Plant-Working Group proposed ITS/ITS2 as a core barcode in seed plants [23]. The psbA-trnH intergenic spacer is the most viable region of the chloroplast genome in angiosperms, and its species identification ability is generally higher than that of other chloroplast genes due to its fast evolution rate [14, 20, 21]. In this study, based on the best and best close match (Table 3), all four chloroplast DNA barcodes had higher rates of accurate identification than the two nuclear barcodes. The chloroplast psbA-trnH and trnL-F achieved the highest success rate (83.33%), while the nuclear ITS and ITS2 obtained a low success rate (about 40%) with a substantial amount of ambiguous (about 50%) and incorrect identification (about 10%). Tree-based analysis also demonstrated that the best barcode was able to authenticate most, but not all, Stephania species. Consequently, combinations of the single-region barcodes were analyzed for higher identification efficiency.

CBOL proposed matK+rbcL as the best DNA barcode to identify plants at a genetic level [28]. However, the number of matK+rbcL variable sites was insufficient to differentiate between closely related species in our study (Table 2). Further, the species discrimination rate of matK+rbcL was only 63.63% based on best match and best close match methods (Table 3). The phylogenetic trees based on either matK+rbcL or all four chloroplast genes combined (Figs. S5, S6) were also unable to authenticate all species. Previous studies showed that combining chloroplast with nuclear genes could greatly improve species identification [15, 18, 23, 25, 29], which is consistent with the satisfactory discrimination achieved by ITS+psbA-trnH and ITS+trnL-F combinations in our study. Based on the best match and best close match methods, ITS+trnL-F and ITS+psbA-trnH showed the same discrimination success rate (both 93.93%; Table 3); however, the phylogenetic tree (NJ, BI) of ITS+psbA-trnH indicated a higher discrimination rate (100%) (Figs. 1, S2). Thus, ITS+psbA-trnH was chosen as the best DNA barcode to identify Stephania species.

Phylogenetic trees reveal interrelations between different species, which can be used for species identification. Based on flower, inflorescence, leaf, and tuber traits, species of the genus Stephania in China have been classified into three subgenera, namely S. subgenus Botryodiscia, S. subgenus Stephania, and S. subgenus Tuberiphania [2]. In this study, the phylogenetic trees (NJ, BI) base on ITS+psbA-trnH contained three major clades. Those clades corresponded to the three subgenera of the genus Stephania. S. hainanensis and S. succifera belong to the same subgenus (Tuberiphania), and both are distributed in Hainan province. The samples of these two species clustered as two independent clades in the phylogenetic tree constructed based on the ITS+psbA-trnH combination. The two clades were well supported as sister groups with high bootstrap values (Figs. 1, S2), indicating that these two closely related species could be distinguished at the genetic level. The three sympatric species, S. kwangensis, S. micrantha, and S. kuinanensis—all of which are the source plants of the traditional Chinese medicine Radix Stephaniae—have similar morphology. Thus, they are usually harvested indiscriminately even though S. kuinanensis is an endangered species. The samples of the three species also formed different clades in the phylogenetic trees constructed using ITS+psbA-trnH (NJ, BI). These results show that tree-based phylogeny is a feasible method to identify these closely related species in the genus Stephania.

Further, S. kwangensis samples, collected from different populations, formed two clades in the phylogenetic trees established by ITS+psbA-trnH (NJ, BI). There might be two explanations for this pattern of S. kwangensis. One is that different evolutionary lineages of S. kwangensis exhibit similar morphology because of similar habitats, introgression, and incomplete lineage sorting. Another possible explanation is that there might be a cryptic species exists within S. kwangensis. Thus, this DNA barcoding method could provide valuable information regarding the formation of variants, hybrids and cryptic species at the genetic level.

The proposed DNA barcoding method is applicable to not only leaf, but also to root samples for species identification, which improves its practicability because roots are the medicinal parts for most Stephania plants. However, this technique has an inherent limitation which is related to the quality of genomic DNA. In this study, high-quality total DNA could be extracted from samples dried immediately with silica-gel and stored in dry conditions even for seven years. Genomic DNA from air-dried or sun-dried samples was seriously degraded, especially for moldy samples. Then the following amplification and sequencing were failed. In general, DNA may be degraded partially or even completely by various factors including extensive heat treatment, irradiation, and microbiological corrosion [30]. Consequently, this method may be not feasible for some processed crude drugs. Thus, other methods such as chemical analysis and microscopic observation are needed to be developed for identifying these samples. Herein, we recommend DNA barcoding combined with other techniques to further resolve the identification problems in Stephania plants.

Conclusions

In this study, six barcodes were analyzed either as single region or in combination for their potential to discriminate 23 Stephania species, and a combination of ITS+psbA-trnH was recommended as the best DNA barcode for the species discrimination based on tree-based, distance-based, and sequence similarity-based methods. Both ITS and psbA-trnH are relatively short regions (< 600 bp) and easy to be amplified (100%), and this barcode combination exhibits high discrimination power. The results demonstrate that DNA barcoding is an efficient and practical method for the species identification of Stephania. The application of this method will be of great help to the correct usage of Stephania plants and the protection of endangered resources.