HPLC Fingerprint and Chemical Pattern Recognition of Wild and Cultivated Millettia speciosa Champ.

Jianguang ZHANG, Xuemin SHI, Ming CHEN, Xiaoping ZHOU, Xiangle MENG

1. Qinzhou Health School, Qinzhou 535000, China; 2. Beibu Gulf University, Qinzhou 535011, China

Abstract [Objectives] To establish HPLC fingerprints of wild and cultivated Millettia speciosa Champ., and identify medicinal materials combined with chemical pattern recognition methods, and provide a reference system for the identification and quality control of M. speciosa from different sources. [Methods] 20 batches of M. speciosa from different sources were determined by HPLC method, and the similarity analysis and evaluation were performed using the Similarity Evaluation System of Traditional Chinese Medicine Chromatographic Fingerprints (2012 edition). Principal component analysis (PCA) and least partial squares method-discrimination analysis (PLS-DA) were used to conduct chemical pattern recognition research on wild and cultivated M. speciosa. [Results] The HPLC fingerprints of wild and cultivated M. speciosa were established, 10 common peaks were calibrated, and the similarity of 20 batches of samples was greater than 0.9; PCA can better classify M. speciosa from different sources into 2 categories, and PLS-DA can completely distinguish between wild and cultivated M. speciosa. [Conclusions] The established M. speciosa fingerprint, combined with chemical pattern recognition methods, can effectively distinguish between wild and cultivated M. speciosa, so it can provide a reference for quality control and evaluation of M. speciosa.

Key words Millettia speciosa Champ., Wild, Cultivated, Fingerprint, Quality control

1 Introduction

MillettiaspeciosaChamp. takes the dried root as the medicine, also known as Dalishu and Shanlianou,etc., and is mainly distributed in Guangxi, Guangdong, Hainan and of China, and is famous medicinal and edible plant in areas south of the Five Ridges. Chemical component studies have shown that the effective active components ofM.speciosaare mainly flavonoids, sugars, saponins and alkaloids[1-2]. Modern pharmacological studies have indicated thatM.speciosahas obvious anti-inflammatory and antioxidant functions[3-4]. At present, more and more people takeM.speciosaas health foods and develop related health products, such as health wine, health tea, health beverages,etc., resulting inM.speciosabeing in short supply in the market, and wildM.speciosaresources are almost exhausted and unable to meet market demand. In Guangxi, the area of artificially cultivatedM.speciosais expanding, and the related researches onM.speciosaare mainly concentrated on chemical components and quantitative analysis methods[5-7], but there is still no report about the fingerprints of wild and cultivated products ofM.speciosa. As it is known to all, differences in ecological environment, planting of medicinal materials, genetic factors,etc. will cause the difference between the secondary metabolites of wild and cultivated medicinal materials. To explore the reasons for the differences in secondary metabolites and their main factors is to distinguish between wild and cultivated medicinal materials, and thus making further quality evaluation and control is one of the main contents of the quality research of traditional Chinese medicine.

The fingerprint of traditional Chinese medicine can comprehensively reflect the quality of medicinal materials[8-10]. It is an effective way to control the quality of medicinal materials, and provide an effective method for quality evaluation and control. The fingerprint of traditional Chinese medicine is integral and vague, and it is quite complicated, making it almost impossible to process chromatographic data. In recent years, the rapid development of chemometrics has carried out in-depth mining of fingerprint data, which provides an effective method for people to calculate complex system data. At present, fingerprints combined with chemical pattern recognition methods have been widely applied in the quality control of Chinese herbal medicines[11-15], and good analysis results have been obtained.

In this experiment, we used the RP-HPLC method to study the fingerprints of wild and cultivatedM.speciosa. Through similarity evaluation, using chemical pattern recognition technologies such as PCA and PLS-DA, we analyzed the chemical components of wild and cultivatedM.speciosaand made classification, and screened out the main chemical components that cause differences between different samples, so as to provide a reference for the quality control ofM.speciosaand the development of new drugs.

2 Materials

Agilent 1220 high performance liquid chromatography (HPLC) system (American Agilent Technologies Co., Ltd.); METTLER AE240 Electronic Analytical Balance (Mettler-Toledo Instruments Co., Ltd); KH5200B ultrasonic cleaner (Kunshan Ultrasonic Instruments Co., Ltd.); methanol adopted chromatographically pure, and water adopted Milli Q ultrapure water (Millipore, Bedford, USA); hypaphorine reference substance (batch No.19012210, Chengdu Pufei De Biotech Co., Ltd.). The sources ofM.speciosawild and cultivated products are shown in Table 1. The material was identified by us as a leguminous plant of the genusMilletia, and the specimens were kept in the Pharmaceutical Laboratory of Qinzhou Health School.

Table 1 Information of Millettia speciosa Champ. samples

3 Methods and results

3.1 Chromatographic conditionsChromatographic column: Agilent TC-C18(250 mm×4.6 mm, 5 μL); mobile phase: acetonitrile (A)-0.1% acetic acid aqueous solution (B); gradient conditions: 0-5 min, 3%A; 5-10 min, 3%-18%A; 10-20 min, 18%-25%; 20-40 min, 25%-45%A; 40-50 min, 45%-70%A; 50-60 min, 70%-90%A; 60-68 min, 90%A; detection wavelength: 290 nm; flow rate: 0.8 mL/min; column temperature: 30 ℃; injection volume: 20 μL. Under such chromatographic condition, the chromatographic peak separation effect ofM.speciosawild and cultivated samples is good. The chromatographic peaks are shown in Fig.1-2.

Fig.1 Chromatogram of wild Millettia speciosa Champ. product

Fig.2 Chromatogram of cultivated Millettia speciosa Champ. product

3.2 Preparation of reference solutionPrecisely weighed 5.20 mg of hypaphorine reference substance dried to constant weight, placed to a 10-mL volumetric flask, added methanol to dilute to the desired scale, shook up, obtained the stock solution of reference substance.

3.3 Preparation of test sample solutionTook 1 g of powder ofM.speciosa(screened with No.2 sieve), precisely weighed, placed in a 100-mL conical flask, precisely added 25 mL of 50% ethanol, weighed, conducted ultrasonic treatment (power 200 W at 40 KHz) for 60 min, cooled down, then weighed again, made up for the weight loss with 50% ethanol, shook up, filtered with 0.45 μm microporous membrane, took the filtrate, and obtained the test sample solution.

3.4 Methodology examination

3.4.1Precision test. Weighed one piece ofM.speciosamedicinal powder (S16), prepared the test solution in accordance with the method in Section3.3, and continuously injected 6 samples under the chromatographic conditions in Section3.1, with the chromatographic peak of hypaphorine No.3 as the reference peak (S), calculated theRSDof relative retention time of each common peak (<1.24%), and theRSDof relative peak area was <2.51%. Test results indicate that the instrument had excellent precision.

3.4.2Stability test. Weighed one piece ofM.speciosamedicinal powder (S16), prepared the test solution in accordance with the method in Section3.3, and injected samples at 0, 2, 4, 8, 10, and 12 h under the chromatographic conditions in Section3.1, with the chromatographic peak of hypaphorine No.3 as the reference peak (S), calculated theRSDof relative retention time of each common peak (<2.28%), and theRSDof relative peak was <2.42%. The test results show that the test solution ofM.speciosawas stable within 12 h at room temperature.

3.4.3Reproducibility test. Weighed one piece ofM.speciosamedicinal powder (S16), prepared the test solution in accordance with the method in Section3.3, and injected 6 samples under the chromatographic conditions in Section3.1, with the chromatographic peak of hypaphorine No.3 as the reference peak (S), calculated the relative retention time of each common peak, theRSDwas <2.20%, and the relative peak areaRSDwas <1.85%. Test results indicate that method had high reproducibility.

3.5 Pattern recognition analysisUsing theSimilarityEvaluationSystemofTraditionalChineseMedicineChromatographicFingerprints(2012 edition), we analyzed the fingerprints ofM.speciosa, determined the common peaks, generated the control fingerprint, and performed the similarity calculation. Using the common peak area as a variable, importing SIMCA-P14.1 software, we performed PCA, PLS-DA, and variable projection importance (VIP) analysis, studied the difference between wild and cultivatedM.speciosa, and screened the chemical components that cause differences between samples.

3.6 Establishment of fingerprints and determination of common peaksTook 20 batches ofM.speciosasamples from different sources in Guangxi, prepared the test solution in accordance with the method in Section3.3, and performed the determination under the chromatographic conditions in Section3.1, then obtained chromatograms of 20 batches ofM.speciosa. Imported the chromatograms of 20 batches ofM.speciosainto theSimilarityEvaluationSystemofTraditionalChineseMedicineChromatographicFingerprints(2012 edition) to make analysis, took S16 as the reference chromatogram, performed multi-point calibration with the time window width of 0.5, performed automatic matching, established a common mode, calculated the fingerprint similarity of the reference substance, marked 10 common peaks (Fig.1-2), and obtained a superimposed chromatogram, as shown in Fig.3.

Fig.3 Fingerprint overlay of 20 batches of Millettia speciosa Champ.

A total of 10 common peaks were obtained from 20 batches of different sources ofM.speciosain Guangxi. Through comparing with the chromatogram of the hypaphorine reference substance (Fig.4), it was confirmed that the peak No.3 was hypaphorine. Because this peak existed in all samples, the resolution was good, and the content was high, so we selected hypaphorine (peak 3) as the reference peak (S). Taking the peak area of the reference peak as 1, we calculated theRSDvalue of each chromatographic peak and the reference peak of retention time and relative peak area. The results indicate that theRSDof the equivalent retention time of the common peaks the chromatograms of the sample was in the range of 0.15%-0.28%. These tell us that the relative retention times of the common peaks were stable, indicating that the HPLC fingerprint system was stable; and theRSDof relative peak area of different samples was in the range of 31.96%-140.02%, the difference was large, indicating that there are certain differences in the content of different batches ofM.speciosasamples.

Fig.4 Chromatogram of hypaphorine reference product

3.7 Similarity analysisImported the chromatograms of 20 batches ofM.speciosainto theSimilarityEvaluationSystemofTraditionalChineseMedicineChromatographicFingerprints(2012 edition) to make analysis, took S16 as the reference chromatogram, make multi-point correction using the average correlation coefficient method, and calculate the similarity evaluation. The results are shown in Table 2. Combined with Fig.1-2, it can be seen that except for the similarity of sample S17 which was 0.909, the similarity of the other samples was greater than 0.95, indicating that the wild and cultivatedM.speciosafrom different sources in Guangxi had high similarity, the overall chemical components were similar, and the overall quality was relatively stable, but there were certain differences in chemical component content. These may be related to factors such as cultivation methods, growth years, and growth environment.

Table 2 Similarity of HPLC fingerprint of Millettia speciosa Champ.

3.8 Chemical pattern recognition

3.8.1Unsupervised pattern recognition. Taking the peak areas of 10 common peaks in 20 batches ofM.speciosasamples as variables, we obtained a 20×10 order data matrix. With the aid of SIMCA-14.1 statistical software, we performed PCA analysis on the collected 20 batches of wild and cultivatedM.speciosasamples.

According to the PCA results of 10 common peaks of 20 batches ofM.speciosamedicinal materials, we extracted the first two main components to plot a score chart, as shown in Fig.5. According to Fig.5, the wild and cultivatedM.speciosahave a good separation trend along the t[1] axis. Wild products are located on the left side of the score chart, and cultivated products are located on the right side of the score chart, but wild products S1, S5, and S7 samples and the cultivars S12, S12, S14, and S19 samples are not classified according to their respective sources. In order to better examine whether the chemical pattern recognition method can completely distinguish wild and cultivatedM.speciosaproduced in Guangxi, we used PLS-DA method to analyze the samples.

Fig.5 PCA score chart of Millettia speciosa Champ. samples

Fig.6 PLS-DA score chart of Millettia speciosa Champ. samples

We extracted the VIP value chart of the 10 variables in the PLS-DA model (Fig.7), arranged the VIP values of the 10 common peaks and peak areas from large to small, and selected the common peaks with a VIP value greater than 1, and the result shows that No.10 peak (VIP value of 1.988 4), No.6 peak (VIP value of 1.648 3), No.3 peak (VIP value of 1.074 9) are all greater than 1, indicating that these three chemical components have a significant impact on the classification of wild and cultivatedM.speciosasamples in Guangxi, which are main symbolic components that cause the difference in components between wild and cultivatedM.speciosa. The No.3 peak is designated as hypaphorine through the reference substance. The VIP value of the remaining chromatographic peaks is less than 1, so their corresponding components have little influence on distinguishing samples.

Fig.7 VIP of the differential markers of Millettia speciosa Champ. samples

4 Discussions

4.1 Extraction method testIn this experiment, we tested different extraction solvents (water, absolute ethanol, 20% ethanol, 50% ethanol, 80% ethanol, methanol), extraction time (30, 60, and 90 min), and solid-to-liquid ratio (10, 25, 40 times). According to the results, we determined that 50% ethanol as the solvent, 1∶25 as the solid-to-liquid ratio, and 60 min as the ultrasonic extraction time, under such conditions, the extraction effect is optimal.

4.2 Chromatographic condition testIn this experiment, we tested the effects of different mobile phases (acetonitrile-0.1% acetic acid, methanol-water, acetonitrile-water, methanol-0.1% formic acid), different column temperatures (20, 25, 30, 40 ℃), and different absorption wavelengths (240, 260, 280, 290, 300, 320 nm) on the fingerprints. After overall consideration, we selected acetonitrile-0.1% acetic acid aqueous solution as the mobile phase gradient elution, column temperature of 30 ℃, Agilent TC-C18column (250 mm×4.6 mm, 5 μm), the flow rate of 0.8 mL/min, and the detection wavelength of 290 nm. Under these conditions, the baseline of the HPLC fingerprint is stable, the chromatographic peaks are abundant, the peak shape is good, and the resolution is good.

4.3 Establishment of fingerprint and evaluation of similarity

Through studying the HPLC fingerprints of wild and cultivatedM.speciosa, we established a method in this experiment, and it is simple, stable and reliable. From 20 batches ofM.speciosa, we obtained a total of 10 common characteristic peaks, which can comprehensively reflect the chemical characteristics of the medicinal material. The similarity of the 20 batches of samples is 0.909-0.999. It can be seen that the overall similarity of the chemical components of wild and cultivatedM.speciosais relatively high, and the quality of the medicinal materials is relatively stable, but theRSDvalues of relative peak area are different, indicating a certain difference in the content of chemical components.

4.4 Chemical pattern recognitionChemical pattern recognition method is widely applied in the quality control of traditional Chinese medicines. Combined with traditional Chinese medicine fingerprints, near-infrared fingerprints,etc. for evaluation, it has obtained a series of research results, which can better distinguish Polygalae Radix[12], Taibai rice[13]and Crataegi Folium[14]wild and cultivated varieties. Combined with the chemical pattern recognition method, and taking the peak area of the common peak as a variable, we established the PCA and PLS-DA classification models of wild and cultivatedM.speciosa. The comprehensive evaluation results show that PCA can distinguish wildM.speciosaproducts from cultivated products, and PLS-DA can completely distinguish wildM.speciosaproducts from cultivated products, indicating that artificial introduction and domestication, cultivation management, geographical environment and other factors have caused the changes in content of certain chemical components ofM.speciosa, which leads to differences between wildM.speciosaproducts and cultivated products in Guangxi. Further screening of markers shows that three chemical components including hypaphorine are the main components that distinguish wildM.speciosaproducts from cultivated products.

The fingerprint ofM.speciosamedicinal materials established in this experiment is stable, reliable, fast and simple, and can comprehensively reflect the chemical component information ofM.speciosa. The similarity evaluation of the fingerprints of traditional Chinese medicine showed that the wild and cultivatedM.speciosahave good similarity and the quality is relatively stable; the established PCA and PLS-DA chemical pattern recognition models can clearly distinguish wildM.speciosafrom cultivated products. Combining fingerprint and chemical pattern recognition is a method to evaluate biotechnology ofM.speciosafrom different sources. It is an effective means forM.speciosaquality control and is favorable for standardized planting and clinical application ofM.speciosa.