Identification of medical plants of 24 Ardisia species from China using the matK genetic marker.

BACKGROUND: Ardisia is a group of famous herbs in China, which has been used as medical plants for more than 900 years. However, the species from the genus are so analogous that it is difficult to discriminate them just by morphological characteristics. DNA barcoding is a new technique that uses a short and standard fragment of DNA sequences to identify species.
OBJECTIVE: Choose a suitable DNA marker to authenticate Ardisia species.
MATERIALS AND METHODS: Four markers (psbA-trnH, internal transcribed spacer 2 [ITS2], rbcL, matK) were tested on 54 samples of 24 species from genus Ardisia. The success rates of polymerase chain reaction amplification and sequencing, differential intra- and inter-specific divergences, DNA barcoding gap and identification efficiency were used to evaluate the discrimination ability.
RESULTS: The results indicate that matK has the highest interspecific divergence and significant differences between inter- and intra-specific divergences, whereas psbA-trnH, ITS2 and rbcL have much lower divergence values. Matk possessed the highest species identification efficiency at 98.1% by basic local alignment search tool 1 [BLAST1], method and 91.7% by the nearest distance method.
CONCLUSION: The matK region is a promising DNA barcode for the genus Ardisia.

INTRODUCTION

Ardisia genus is a group of flowering plants belonging to Myrsinaceae family, native to tropical American, Austronesia, India Peninsula, East and South Asian, minority spread over Oceania. The genus includes about 300 species in the world and 68 species in China, which is widely and commonly cultivated in south area of Yangtze River.[1] Most species of Ardisia are medicinal plants and a few of them are ornamental plants in China. Some of them are famous on medicinal value. For example, Ardisia japonica (Hornst.) Blume is commonly used for treating chronic bronchitis; Ardisia crenata Sims var. crenata is used as oxytocics and anti-pregnancy drugs. Ardisia pusilla A. de Candolle. is used to treat traumatic injuries.[2] But the species from the genus are so analogous that it is very difficult to discriminate them just by morphological characteristics and it is often taken place that many species of the genus are confused and used by other different species. So, it is very important to accurately identify these medical plants from Ardisia.

DNA barcoding, which was first proposed by Hebert et al.[3] is a new technique that uses a short and standardized fragment of DNA sequences to identify species, and recently it has become a hotspot of biodiversity research.[4] In subsequent research,[567] Hebert et al. found that the CO1 gene is a standard DNA barcode for animals. But the studies on plant barcodes are much more complicated than that of animals, because of the hybridization and reticulate evolutionary histories.[89] Recently, a number of single loci and combined loci have been suggested as candidate barcode sequences for plant identification,[101112] but there was no consensus on universal DNA barcode for all plant species. For every concrete group of species, especially those which contain many closely related species, applicable loci have to be studied and choose. Some scholars have done DNA barcoding researches in related species and genera, but no one has evaluated feasibility of the method in plants of Ardisia.

In this context, we choose four regions intensively recommended (psbA-trnH, matK, rbcL, internal transcribed spacer 2 [ITS2]) to test and evaluate the feasibility of these regions as candidate DNA barcodes to discriminate medicinal species in China from Ardisia and try to find a new a digital identification method for medicinal plants of Ardisia.

MATERIALS AND METHODS

Plant materials

The experimental samples were collected from (1) South China Botanical Garden, Guangdong Research Institute of Traditional Chinese Medicine, Guangdong province, and authenticated by Prof. Yuewen Cai of the Institute; (2) Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Yunnan province, and authenticated by senior Engineer Chunfen Xiao of the Garden; (3) Wuhan Botanical Garden, Chinese Academy of Sciences, Hubei province, and authenticated by Engineer Shouzhong Zhang of the Garden. All voucher images and specimens were deposited in the herbarium of Hubei University of Chinese Medicine. The information of 54 samples belonged to 27 species are given in Table 1.

Table 1

The collection sites and GenBank accession of 54 samples of the Ardisia genus

DNA extraction, amplification, and sequencing

First, leaf tissues were dried in silica gel. A total of 10 mg of each of the dried tissues was rubbed for 1 min at a frequency of 30 times/s in a FastPrep bead mill (Retsch MM400, Germany). Total DNA was extracted using the Plant Genomic DNA Kit (Tiangen Biotech Co., China). The polymerase chain reaction (PCR) reaction mixture consisted of 2 μL (˜60 ng) DNA, 4 μL of 25 mM MgCl2, 5 μL of 10 × PCR buffer, 2U of Taq DNA polymerase, 4 μL of 2.5 mM deoxy-ribonucleoside triphosphates [dNTPs] mix (Biocolor BioScience & Technology Co., China), 2.0 μL 2.5 μM of primers (Synthesized by Sangon Co., China), the final volume was 50 μL. The sequences of the universal primers for the DNA barcode to be tested and general PCR reaction conditions were obtained from previous studies by Chen et al.[13] PCR products were first examined with 1.5% agarose gel electrophoresis and purified using the Gel Band Purification Kit (Tiangen Biotech Co., China) and then sequenced in both directions with the primers used for PCR amplification on a 3730XL sequencer (Applied Biosystems, USA). The sequences were submitted to GenBank [Table 1].

Data analyses

The original forward and reverse sequences were assembled and edited using CodonCode Aligner 3.0 (CodonCode Co., USA) to estimate the quality of the generated sequence traces. Sequences alignment and checking were conducted by Clustal W. The ITS2 sequences were retrieved according to Keller et al.[14] and other sequences were retrieved using CodonCode Aligner. All the experimental materials were used to investigate the amplification efficiency of each sequence. The inter/intra-specific variation of the samples was calculated according to Luo et al.[15] and Zhu et al.[16] and Wilcoxon signed rank tests[17] were used to check the result. DNA barcoding gap was produced using Taxon DNA.[18] After the data from GenBank database were brought into, basic local alignment search tool 1 [BLAST1] and the nearest distance method were performed as described previously[19] to assess the identification efficiency of each candidate sequence.

RESULTS

PCR amplification efficiency and the success rate of sequencing

The efficiency of PCR amplification and the success rate of sequencing of the four candidates were compared. The result showed that the efficiency of PCR amplication of rbcL, psbA-trnH, ITS2 and matK region were 100%, 100%, 100% and 88.9%. And they were all successfully sequenced by 100% [Table 2]. The sequence length, Guanine and Cytosine [GC] content of the four regions based on the results of the CodonCode Aligner and Clustal W alignment were presented [Table 2].

Table 2

Success rate of sequencing, length range, GC content

The analysis of intra-specific variations and inter-specific divergences

An ideal barcode should show the low intra-specific variations and high inter-specific divergences in order to distinguish different species. Here, six parameters were used to characterize inter-specific versus intra-specific variation [Table 3]. Through comparison of interspecific genetic distances among congeneric species for four candidate barcode, ITS2 region exhibited the highest interspecific divergence with all four metrics, followed by psbA-trnH and matK, while rbcL provided the lowest [Table 3]. We also found that rbcL showed the lowest level of intraspecific variation with all four metrics, followed by psbA-trnH and matK, while ITS2 provided the highest [Table 3].

Table 3

Analysis of inter-specific divergence between congeneric species and intra-specific variation for the whole sample

Validation of the different sequences’ inter/intra-specific variation

The results of Wilcoxon signed rank tests confirmed that matK provided much higher inter-specific divergence among congeneric species [Table 4] and the higher variation between conspecific individuals [Table 5].

Table 4

Wilcoxon signed rank test for interspecific variations

Table 5

Wilcoxon signed rank test for intraspecific variations

Assessment of barcoding gap

Barcodes should exhibit a “barcoding gap” between interspecific and intraspecific distances.[1720] Although the histogram did not show a clear gap between intraspecific variation and interspecific divergence in the distributions of the four tested loci (matK, rbcL, ITS2, psbA-trnH intergenic spacer) [Figure 1], the results of Wilcoxon two-sample tests showed that the distribution of inter-specific divergences for the four barcodes were higher than that of intra-specific variations [Table 6]. All the four candidate sequences showed significant difference (P < 0.05).

Figure 1

The barcoding gap between interspecific and intraspecific divergences for four candidate barcodes. Histograms showing the relative distribution of pairwise (y-axis) intraspecific (blue bar) and interspecific (red bar) divergence distance estimates (x-axis) for internal transcribed spacer 2(ITS2), psbA-trnH, rbcL and matK intergenic spacers, respectively. The divergences were calculated using the Kimura 2-parameter model. Barcoding gaps were assessed by the Wilcoxon two-sample tests, and all were significant (P < 0.05)

Table 6

Wilcoxon two-sample tests for distribution of intra- versus inter-specific divergences

Evaluation of identifying ability of barcodes

Two methods of species identification, including BLAST1 and the nearest distance method were used to test the applicability of using different regions for unique species identification. In the BLAST1 method, the results showed that the matK region identified correctly 98.1% of the samples at the species level. In contrast to matK, the identification efficiency of psbA-trnH, ITS2 and rbcL were much lower at the species level. The results confirmed that matK had the highest success rate at the species level identification with both two methods [Table 7].

Table 7

Comparison of identification efficiency for potential DNA barcodes loci using different methods of species identification

DISCUSSION

The screening of DNA barcode for the Ardisia genus

Optimal DNA barcode should meet following requirements: (1) Significant inter-species variance; (2) Sufficient small intra-species variance; (3) It should be amplified by single primer and have quality sequence by dual sequencing.[21] In this research, we tested four DNA regions (psbA-trnH, ITS2, rbcL and matK) using 55 plant samples belonging to 27 closely related species from the Ardisia genus.

PsbA-trnH fragment has one of the biggest evolution rate among chloroplast compartment and flanked with approximate 75 bp conservative sequences at two ends, which can be used for designing universal primer.[81122] Yao et al. found it universal with high success rates of amplification, which is highly recommended in barcode research[2324] In our study, we found psbA-trnH sequence has a successful identification rate of 70.4%. Although there is a significant difference at intra- and inter-species levels, it has low identification efficiency. Therefore, it is not suitable as the Ardisia barcode sequence.

Many researchers have proposed the ITS2 region as a suitable marker for taxonomic classification.[1325] However in our study, the identification efficiency with ITS2 is only 51.9%. So ITS2 is also not suitable as a barcode for the identification of Ardisia species.

RbcL and matK are recommended as plant barcode sequence in the latest Consortium for the Barcode of Life [CBOL] Research.[26] There are large amount of data for rbcL in Genbank, which is universal, being easily amplified and compared, but its variance mainly exist in intra-species rather than inter-species.[1727] As described before, rbcL fragment was chosen as plant barcode candidate by Kress et al.[8] However, there is no significant difference between the intra-species and inter-species in the research, moreover, the efficiencies of identification by BLAST1 and nearest distance are only 29.1% respectively, therefore, rbcL is not proper as the Ardisia DNA barcode sequence.

The matK fragment is emerging as a gene with potential contribution to plant molecular systematic and evolution.[28293031]

The fragment has a quicker evolution compared with other fragments. In the research, the matK region had the highest identification success rate at the species level; meanwhile, it exhibited well in PCR amplification and sequencing efficiency, differential intra- and inter-specific divergences and DNA barcoding gap. Therefore, we suggested matK region as the DNA barcode for the genus Ardisia.

Discussion on samples with unsuccessful identification

In our study, the matK sequence was chosen as a DNA barcode to identify the species of Ardisia genus. Among the 48 samples, which were successfully sequenced, there was one sample (A. japonica) that could not be distinguished from A. pusilla. These two species are sister species both attached to the group of Sect. Bladhia. They show little differences in morphology and closely relationship between them and that may be the reason that they were difficult to differentiate from each other.

The present research found that of the four candidate loci (psbA-trnH, ITS2, matK, rbcL), matK produced the highest rate of successful identification in 91.7% at the species level and it can correctly discriminate 22 Chinese medicinal species from Ardisia according to the nearest distance method. Therefore, it is proposed that the matK region can be used as a DNA barcode to identify these medicinal plants from Ardisia. Collection of more samples and deep researches for those species of ambiguous identification are necessary to provide more effective information about phyletic evolution and more reliable method for the identification of genus Ardisia.

Measuring the success rates of identification methods

CBOL recommended rbcL and matK together as plant barcode sequence, but needs enough matching data from the experiment, which could increase cost, therefore, we just focus on the probability of single sequence, BLAST1/and the nearest distance methods are employed. Meanwhile, identification efficiency was measured in order to display the ability for all sequences. BLAST method compares sample's DNA sequence with total sequence in terms of base, which was ranked by base difference; the advantage of this method is high velocity and accuracy. Nearest distance method compares sample's DNA with the “Kimura 2-parameter (K2P)” distance of total sequence, which is based on overall comparison.[19] It can quantitate difference in single sequence with low velocity, meanwhile lost locus and variable locus are processed equally, which easily leads to the slight difference between data and facts, which is the reason why these two verifications are not uniform.

We will measure how the total data change when each sequence exist or not during the process of efficiency identification. When the data are abnormal, we blast the suspicious sequence with GenBank in order to exclude “false positive” data. The same as other authors, we define “inter-species variance” as the variance among different species under a genus without breaking the genus. It might get smaller results than real fact by using “inter-species” in above extent. Layhaye et al.[17] also got the same conclusion as we have, therefore, we will use some new method of identification, e.g., probability of correct identification [PCI], in order to exclude man-made disturbance.

DNA barcode technique has already been used in animal research and increasing used in plant research, which will assist non-systematic scholars to quickly and accurately identify different species. DNA barcode cannot replace traditional taxonomy, but it is accurate, abundant and unique with high repeats as digital DNA sequence, leading to a useful tool for taxonomists.[262732] This research explored the application of DNA barcode technique and provided a new method and insight for molecular identification and relationship. As limits of sampling condition in this research, some species had no duplicates; some nearest sibling species were included under a genus. There should be more effective information and reliable method when more samples are included in further research in the future.