J S S
ISSN 1615-9306 · JSSCCJ 38 (11) 1813–2006 (2015) · Vol. 38 · No. 11 · June 2015 · D 10609
JOURNAL OF
SEPARATION SCIENCE
Methods Chromatography · Electroseparation Applications Biomedicine · Foods · Environment
11 15
www.jss-journal.com
1850 Qiong Lou1,2 Xiaolan Ye1 Yingyi Zhou2 Hua Li2 ∗ Fenyun Song1 ∗ 1 Department
of Pharmaceutical Analysis, School of Pharmacy, Guangdong Pharmaceutical University, Guangzhou, P.R. China 2 Department of Traditional Chinese Medicine, Guangdong Institute for Food and Drug Control, Guangzhou, P.R. China Received January 12, 2015 Revised March 2, 2015 Accepted March 10, 2015
J. Sep. Sci. 2015, 38, 1850–1857
Research Article
Chemical fingerprint of Ganmaoling granule by double-wavelength ultra high performance liquid chromatography and ultra high performance liquid chromatography with quadrupole time-of-flight mass spectrometry A method incorporating double-wavelength ultra high performance liquid chromatography with quadrupole time-of-flight mass spectrometry was developed for the investigation of the chemical fingerprint of Ganmaoling granule. The chromatographic separations were performed on an ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m) at 30⬚C using gradient elution with water/formic acid (1%) and acetonitrile at a flow rate of 0.4 mL/min. A total of 11 chemical constituents of Ganmaoling granule were identified from their molecular weight, UV spectra, tandem mass spectrometry data, and retention behavior by comparing the results with those of the reference standards or literature. And 25 peaks were selected as the common peaks for fingerprint analysis to evaluate the similarities among 25 batches of Ganmaoling granule. The results of principal component analysis and orthogonal projection to latent structures discriminant analysis showed that the important chemical markers that could distinguish the different batches were revealed as 4,5-di-Ocaffeoylquinic acid, 3,5-di-O-caffeoylquinic acid, and 4-O-caffeoylquinic acid. This is the first report of the ultra high performance liquid chromatography chemical fingerprint and component identification of Ganmaoling granule, which could lay a foundation for further studies of Ganmaoling granule. Keywords: Chemical fingerprint / Ganmaoling granule / Quadrupole time-of-flight mass spectrometry / Ultra high performance liquid chromatography DOI 10.1002/jssc.201500037
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction Traditional Chinese medicine (TCM; raw herbs, extractions, or preparations) has been widely used for healthcare in China since ancient times and has become increasingly popular around the world. Unlike modern pharmacology that often looks at the single chemical entity aimed at a specific single target, TCM often views complex matrices and one herb usually contains a myriad of components. Their overall efficacy is based on the synergic effect of multicompound Correspondence: Dr. Fenyun Song, Department of Pharmaceutical Analysis, School of Pharmacy, Guangdong Pharmaceutical University, Guangzhou 510006, P.R. China E-mail: [email protected] Fax: +86-020-89567081
Abbreviations: DAD, diode array detector; GML, Ganmaoling; OPLS-DA, orthogonal projection to latent structures discriminant analysis; PCA, principal component analysis; TCM, traditional Chinese medicine C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
and multiingredient preparations. Therefore, to evaluate the quality and authenticity of the complex preparations only by selecting a few markers or pharmacologically active components is troublesome [1]. Better analytical methods to assure their efficacy, safety, and stability are in great demand. Currently, the chromatographic fingerprint strategies play a significant role in the QC of TCM, which can not only determine the characteristic patterns of each plant type but also reveal the inherent relationships between multiple compounds [2]. Moreover, both the China Food and Drug Administration (CFDA) and European Medicines Agency (EMEA) have clearly accepted fingerprint chromatograms for the assessment of herbal medicines [3]. Additionally, it is a common practice to combine fingerprinting strategies with ∗ Additional correspondence: Dr. Hua Li, Department of traditional Chinese medicine, Guangdong Institute for Food and Drug Control, Guangzhou 510180, P.R. China E-mail:[email protected] Fax: +86-020-81886161
www.jss-journal.com
J. Sep. Sci. 2015, 38, 1850–1857
Liquid Chromatography
1851
acid (MUST-14043010), 4-O-caffeoylquinic acid (MUST14041410), 3,5-di-O-caffeoylquinic acid (MUST-14022612), 3,4-O-dicaffeoylquinic acid (MUST-14041411), and 4,5-di-Ocaffeoylquinic acid (MUST-14041414) were purchased from Beijing century audiocodes biological technology (Beijing, China). The purity of each standard was over 98%, which was suitable for UHPLC determination. Acetonitrile of HPLC grade was purchased from Honeywell Burdick & Jackson Honeywell Burdick & Jackson (Morristown, New Jersey, USA). Formic acid and methanol of HPLC grade were obtained from Merck (Germany). Ultrapure water was generated by means of a Milli-Q water purification system (Millipore, Milford, MA, USA). Other reagents were used as received. Twenty-five batches of GML granule were obtained from Huarun san-jiu pharmaceutical. These are 1301039H, 1212019H, 1303012H, 1301042H, 1306004H, 1303030H, 1303026H, 1304020H, 1304010H, 1210022H, 1212026H, 1304020H, 13020490H made by Huizhou manufacturer and 13010310, 1303033H, 1306006H, 1304023H, 1305035D, 1302045D, 1302003D, 1302370, 1305029D, 1308150, 13020260, 13010320 made by Zaozhaung manufacturer.
chemometrics analysis, such as principal component analysis (PCA), orthogonal projection to latent structures discriminant analysis (OPLS-DA), and similarity analysis to call attention to the chemical distinctions between the samples and identify their individual chemical markers [4, 5]. Recently, several methods including TLC, HPLC, GC, and CE are usually employed to develop the fingerprint, and HPLC fingerprint analysis is the first choice due to its advantages and popularization [6–9]. The latest UHPLC technique is designed to undertake separations under the high pressure that results from using nonporous silica [10, 11]. Ganmaoling (GML) granule is a popular compound preparation of traditional Chinese herbs, which consists of four strong antiviral herbs: Ye Ju Hua (F. Chrysanthemi Indici), Jin Zhan Yin Pan (B. biternata Merr. et Sherff), Gang Mei (Radix Ilex Asprella), and San Cha Ku (Radix Evodia Lepta), with therapeutic actions of reliving cold symptom, preventing and treating virus infection. It has been widely used in clinical practice for treating headache, fever, stuffy nose, sore throat, and so on, since 1970s as an over-the-counter (OTC) drug in China as well as North America [12]. Some chemicals such as 3-O-caffeoylquinic acid, 5-O-caffeoylquinic acid, 4-Ocaffeoylquinic acid, linarin, 3,4-di-O-caffeoylquinic acid, luteolin 7-O-glucoside, acetaminophen, caffeine are considered to be the main active ingredients in this formula. Those compounds have physiological activities such as fever-reducing, antibacterial, antiviral, and antipyretic activity, and other protective effects [13–15]. However, few studies have carried out the QC of GML granule effectively and systematically. This is due to unidentified components and difficult-to-obtain separation because of the presence of several isomeric quinic acids with similar structures. At present, an advanced technology, UHPLC coupled with Q-TOF-MS has rapidly been applied to identify unknown components of TCM, which combines advantages of ultrahigh resolution, short duration of analyses, and high sensitivity [16–21]. In this study, a characteristic fingerprint of GML granule using double-wavelength UHPLC had been developed, and the main components were identified by Q-TOF-MS initially. Significantly, PCA and orthogonal projection to latent structures discriminant analysis (OPLS-DA) were applied to evaluate the fingerprints of GML granule from 25 batches. It was regarded as the foundation of QC of GML granule completely.
The granule was ground into a powder using a mortar. Each sample powder (4.0 g) was accurately weighed and extracted using ultrasonic bath with 25 mL of methanol for 30 min. The extract was then filtered through a No. 40 filter paper, and 10 mL of the final extract was collected and transferred into an evaporating dish and evaporated to dryness on a 100⬚C water bath. The residue was dissolved in 50% methanol, adjusted to 5 mL, and filtered through 0.45 m membrane to obtain the test solution. Standard stock solutions of 11 compounds (3-Ocaffeoylquinic acid, 5-O-caffeoylquinic acid, 4-O-caffeoylquinic acid, 3,5-di-O-caffeoylquinic acid, 3,4-di-O-caffeoylquinic acid, 4,5-di-O-caffeoylquinic acid, linarin, luteoloside, caffeine, chlorphenamine maleate, acetaminophen) were prepared by dissolving appropriate amounts of standards into 50% methanol to achieve final concentrations of 70, 70, 70, 70, 70, 70, 450, 150, 15, 15 g/mL and 4 mg/mL, respectively.
2 Materials and methods
2.3 Double-wavelength UHPLC analysis
2.1 Chemicals and material
UHPLC was conducted on a Waters ACQUITY system (Waters, Milford, MA, USA) equipped with a binary pump with an online degasser, autosampler, column oven, diode array detector (DAD), and Empower Data Manager. The UHPLC fingerprint was carried out on an ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m) at 30⬚C with sample injection volume of 2 L at a flow rate of 0.15 mL/min. The mobile phase consisted of acetonitrile (B) and 0.1% formic acid in water (A) with a gradient elution program—that is 0–6 min,
Standards of 3-O-caffeoylquinic acid (MUST-14043010), buddlejoside (111528-200606), acetaminophen (100018-200408), caffeine (171215-201211), luteolin 7-O-glucoside (111720201307), and chlorphenamine maleate (100047-200606) were purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China; purity >98%). Standards of 5-O-caffeoylquinic C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2.2 Preparation of sample solution and standard solution
www.jss-journal.com
1852
Q. Lou et al.
6% B; 6–10 min, 6–15% B; 10–24 min, 15–30% B; 24–28 min, 30–6% B; and 2 min post run, 6% B. During optimization of the UHPLC analytical procedure, it was observed that most of the components in samples had their largest chromatographic responses at 310 nm and that many of the remaining analytes had their most intense responses at 264 nm. Thus, the fingerprints in this study were collected at detection wavelengths of 310 and 264 nm. 2.4 UHPLC–Q-TOF-MS analysis The Agilent 1290 UHPLC–DAD–Q-TOF-MS (CA, USA) system consisted of G4204A Quaternary Pump, G4226A automatic sample injector, G1316C thermostatted column compartment, G4212A DAD and 6540 UHD Accurate-Mass Q-TOF was used. The UHPLC fingerprint was carried out on an ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m) at 30⬚C with sample injection volume of 2 L. The mobile phase consisted of 0.1% formic acid in water (A) and acetonitrile (B) using gradient program at a flow rate of 0.15mL/min. Analysis began with an isocratic step of 6% B for 6 min, then 15–30% (B) in 10–24 min, 30–6% (B) in 24–28 min, and reequilibrated for 2 min. The DAD was set at 310 and 264 nm. Each sample was detected with an ESI source operating in positive and negative ion modes. Q-TOF parameters were as follows: scan range m/z, 100–1700; dry gas, 8 L/min; dry gas temperature, 350⬚C; nebulizer pressure, 30 psi; capillary voltage, 4 kV; fragmentor and skimmer voltages, 175 and 65 V.
2.5 Data analysis UHPLC data were analyzed by the professional software Similarity Evaluation System for Chromatographic Fingerprint of TCM (Version 2009), which was used to align the peak areas with retention times. The data of aligned peak areas were exported. PCA and OPLS-DA were used to describe variations in the data with minimum latent variables that were demonstrated with SIMCA-P+ (CA,USA). This facilitated grouping or classification of the fingerprints and detected outliers [22, 23].
3 Results and discussion 3.1 Optimization of UHPLC condition Several UHPLC analytical parameters including separation column, mobile phase, its elution mode and flow rate, detection wavelength (310, 327, 350, and 264 nm) of DAD, and column temperature were optimized to provide sufficient information about the analyzed compounds, and to achieve good resolution and reasonable analysis time. Different types of columns were tested to reach an optimum separation on the ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m). After several trials using different mobile phases including acetonitrile and methanol, 0.1% formic acid water solution and C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2015, 38, 1850–1857
methanol, 0.1% formic acid water solution and acetonitrile, 0.1% TFA water solution and acetonitrile, and 0.08% formic acid water solution and acetonitrile, 0.1% formic acid water solution and acetonitrile was determined as the most appropriate eluent with a step linear gradient due to the satisfactory resolution and acceptable peak parameters it provided. The 30⬚C column temperature and flow rate of 0.15 mL/min were found to be optimal parameters. These conditions were also suitable for UHPLC–Q-TOFMS analysis. The optimal condition for MS (collision energy, fragmentation voltage, and capillary voltage) was selected, and ion modes (positive and negative) were applied to identify the corresponding signals. The optimal UHPLC and MS conditions are detailed in Sections 2.3 and 2.4. 3.2 Optimization of the extraction method To extract the investigated compounds from the GML granule, the extraction solvent (methanol, ethanol, and 50% methanol), the extraction method (ultrasonic extraction and refluxing), and the extraction time (15, 30, and 40 min) were carried out for the single factor test. The optimal extraction method of the GML granule was 4.0 g of the dry powder with 25 mL of 50% methanol heated in an ultrasonic water bath for 30 min. To acquire more accurate results, the sample was concentrated twice. 3.3 Validation for UHPLC fingerprint The method validation of fingerprint analysis was performed based on the relative retention time (the ratio of peak retention time of sample constituents to the reference standard) and the relative peak area (the ratio of peak area of sample constituents to the reference standard). Among these active components, buddleoside indicates a proper retention time and higher content; therefore, we chose it as the reference substance. All common peaks’ relative retention time and relative peak area were obtained on the basis of this substance. The precision of the proposed method, on the basis of analyzing six replicate samples, were below 0.78 and 4.82% for RSD of relative retention time and relative peak area of all peaks, respectively. The stability test was performed with sample solutions for 24 h, which resulted in RSDs of relative retention times and peak areas that were less than 0.91 and 4.70%, respectively. The reproducibility test was performed with six sample solutions extracted from one batch of preparation, which yielded RSDs less than 0.90 and 4.70%. These results indicated that double-wavelength UHPLC fingerprint method developed for this study was adequate, valid, and applicable. The typical fingerprint chromatograms with 25 “common peaks” in the double wavelength are shown in Fig. 1. 3.4 Identification of the ingredients Based on the standard substances and mass spectral information of UHPLC–Q-TOF-MS, 11 compounds were identified. www.jss-journal.com
J. Sep. Sci. 2015, 38, 1850–1857
Liquid Chromatography
1853
Figure 1. Typical chromatograms of test solution of GML granule (a1, b1) and mixed reference solution (a2, b2). 1, chromatogram at 310 nm; 2, chromatogram at 264 nm.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.jss-journal.com
1854
Q. Lou et al.
J. Sep. Sci. 2015, 38, 1850–1857
Figure 2. The total ion current (TIC) chromatogram of mixed reference solution (d1, d2) and test solution of GML granule (e1, e2). 1, TIC in positive mode; 2, TIC in negative mode.
The total ion current (TIC) chromatograms of 11 mixed reference solution (d1, e1) and test solution of GML granule (d2, e2) are shown in Fig. 2. According to the accurate masses, the molecular formula of each constituent was deduced, and the fragmentations and accurate mass of each fragmental ion were obtained. Compound 7, 9, 10, 14, 15, 16, 17, 20, 23, 24, and 25 were unambiguously identified by comparing with reference standards. The results of UHPLC–Q-TOF-MS analysis and chemical structure are summarized in Table 1 and Fig. 3, where six peaks of [M–H]− at m/z 515 and 353 occurred with a higher presence of quinic acid derivatives [2, 14, 24, 25]. The molecular weight of compound 7 is 354.3087 and its chemical formula is C16 H18 O9 . MS2 spectra of this compound showed [M–H]– at m/z 353.1, [M–H– 162]− ([M–H–C6 H10 O5 ]− ) at m/z 191.0, indicating a loss of caffeoyl acid anion from the mother ion. A characteristic C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
fragment ion of the basic peak at m/z 135.0 with chemical formula C8 H7 O2 – resulted from the loss of 1,3, 4-trihydroxycyclohexane carboxylic acid anion and CO2. Therefore, compound 7 was deduced to be neochlorogenic (5-O-caffeoylquinic acid), which was confirmed by comparing its spiking with the reference compound. Compounds 7, 9, and 10 are a group of isomers that have identical molecular ions and similar fragment pathways. Compound 10 showed the characteristic fragments of 191.1 and 127.0 [M–H–162–18–18–28]– ([M–H– C6 H10 O5 –H2 O–H2 O–CO]− ), and therefore was deduced to be 4-O-caffeoylquinic acid. Additional peak was observed at m/z 191.0 in MS2 data of compound 9, which was produced by loss of a caffeoyl acid anion, suggesting compound 9 is likely chlorogenic acid (3-O-caffeoylquinic acid). www.jss-journal.com
Liquid Chromatography
J. Sep. Sci. 2015, 38, 1850–1857
1855
Table 1. UHPLC–Q-TOF-MS identification of the major chemical compounds in GML granule
Peaks no
Rt /min
Identification
Molecular formula
[M+X]+ [M–X]−
Expected m/z
Fragments
Experimental m/z
7 9 10 14
3.62 5.98 6.78 13.72
5-O-caffeoylquinic acid 3-O-caffeoylquinic acid 4-O-caffeoylquinic acid Luteolin 7-O-glucoside
C16 H18 O9 C16 H18 O9 C16 H18 O9 C21 H20 O11
15 16 17 20 23
14.93 15.33 16.66 20.61 3.01
3,4-Di-O-caffeoylquinic acid 3,5-Di-O-caffeoylquinic acid 4,5-Di-O-caffeoylquinic acid Buddleoside 4-Acetamidophenol
C25 H24 O12 C25 H24 O12 C25 H24 O12 C28 H32 O14 C8 H9 NO2
353.1 353.1 353.1 447.1 449.1 515.1 515.1 515.1 593.2 152.0
6.10 14.75
Caffeine Chlorpheniramine maleate
C8 H10 N4 O2 C16 H19 ClN2 .C4 H4 O4
353.3 353.3 353.3 447.1 449.1 515.5 515.5 515.5 593.6 152.2 150.2 195.2 274.9
191.0, 135.0 191.0 191.1, 127.0 285.1 287.0 353.1, 173.0 353.1, 191.0, 135.0 353.1, 173.0, 135.0 285.0 110.0
24 25
[M–H]− [M–H]− [M–H]− [M–H]− [M+H]+ [M–H]− [M–H]− [M–H]− [M+H]+ [M+H]+ [M–H]− [M+H]+ [M+H–C4 H4 O4 ]+
138.0, 110.0 230.0
195.1 275.1
Figure 3. Chemical structures of 11 compounds identified by UHPLC–Q-TOF-MS in GML granule.
The molecular formula of compound 15 was established as C25 H24 O12 by UHPLC–Q-TOF-MS analysis (m/z 515.1 [M– H]− ). A series of distinctive product ions at m/z 353.1 [(M– H)–162]− and 173.0 [(M–H)–162–162–18]– were observed from the loss of successive caffeoyl acid anions and H2 O. Compound 15 was elucidated to be isochlorogenic acid B (3,4-di-O-caffeoylquinic acid) based on the results of DAD detection. Compounds 15, 16, and 17 also are a set of isomers. MS2 spectra of compound 16 showed the fragment peaks at m/z 353.1, 191.0, 135.0 were forced by eliminating two C6 H10 O5 C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
molecules from the precursor ion at m/z 515.1, suggesting that compound 16 was likely isochlorogenic acid A (3,5-di-Ocaffeoylquinic acid). Compound 17 showed the characteristic fragments of 353.1, 173.0, and 135.0, and based on the results of DAD detection, which was deduced to be isochlorogenic acid C (4,5-di-O-caffeoylquinic acid). The molecular formula of compound 14 was identified as C21 H20 O11 by UHPLC–Q-TOF-MS analysis (as m/z 447.1 [M– H]− , m/z 449.1 [M+H]+ ). The pseudo molecular ion at m/z 447.1 [M–H]– yielded a fragmentation ion at m/z 285.1 [M–H-162]– and ion at m/z 449.1 [M+H]+ produced a www.jss-journal.com
1856
J. Sep. Sci. 2015, 38, 1850–1857
Q. Lou et al.
Figure 4. OPLS-DA score plot (A) and loading plot (B) for all samples of GML granule.
fragment ion at m/z 287.0 [M+H–162]+ in the MS/MS spectrum by loss of glucose. Compound 14 was elucidated to be luteolin 7-O-glucoside. The molecular formula of compound 20 was established as C28 H32 O14 by UHPLC–Q-TOF-MS analysis (m/z 593.2 [M+H]+ ). A distinctive product ion at m/z 285.0 [M+H– 308]+ was observed resulting from loss of C12 H20 O9 + . Compound 14 was elucidated to be buddleoside, based on the results of DAD detection. The molecular weight of compound 23 is 151.2, and its chemical formula is C8 H9 NO2 . MS2 spectra of this compound showed [M+H–42]+ at m/z 110.0 indicating the loss of C2 H2 O+ . The molecular formula of compound 20 was established as C16 H19 ClN2 .C4 H4 O4 by UHPLC–Q-TOF-MS analysis (as m/z 275.1 [M+H–116]+ ) by the loss of C4 H4 O4 + . The representative product ion at m/z 230.0 [M+H–116–45]+ was observed by loss of NC2 H7 + . Compound 25 was elucidated to be chlorpheniramine maleate, based on the results of DAD detection.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
3.5 PCA and OPLS-DA PCA is an advanced technique which was widely applied to summarize multivariate variation in a few principal components remaining maximum reasonable variability. To assess variations in quality and to differentiate between samples, PCA was adopted based on the contents of 25 compounds in 25 samples by analyzing the final matrix of double-wavelength fingerprints. The contents of 25 fingerprint peaks were applied to evaluate variations in the GML granule. As shown in the score plot of the first two principal components of doublewavelength fingerprint, Fig. 4A, all of the samples could be classified into two groups. Additionally, sample 6 and sample 8 were obviously different from the other samples. To make sure that the sample was an outlier, PCA analysis within the batches was carried out. The points for sample 6 and sample 8 were graphed far from the other points. In Hostelling’s T2 range plot, the point above the green horizontal line revealed the probability that the sample was less than 5% similar to the other samples [4]. Therefore, the samwww.jss-journal.com
J. Sep. Sci. 2015, 38, 1850–1857
ple was removed from the analysis. As shown in Fig. 4B, the scores also show some dispersion among the batches; this may due to the samples’ places of origin and times of collection. To confirm the two-group data model studied above, OPLS-DA was performed. This enabled us to calculate discriminant functions as linear combinations of the selected descriptors, which maximized the ratio between-class sum of squares and the within-class sum of squares [26, 27]. As shown in Fig. 4A, each point represented a particular sample and was automatically color-coded according to its class. The OPLS-DA model in this experiment resulted in a (1 + 1) component with a Q2 of 64.5%, R2(X) of 48.3%, and R2(Y) of 89.1% of the variables, which showed the good quality of the model. By examining the loadings plot, the important chemical markers separating these samples were revealed. From the loadings plot, the variables with larger covariance and correlation values were considered to be the discriminatory markers, while the variables with larger variable importance in the projection (VIP) values were more relevant for sample classification. As shown in Fig. 4B, the loadings plot of the OPLS-DA results demonstrated that peaks 17, 18, 16, and 10 most influenced the discrimination among different batches of GML granule.
4 Concluding remarks In conclusion, the double-wavelength UHPLC fingerprint and UHPLC–Q-TOF-MS techniques were used in this study that allowed the identification and comparison of GML granule and can be utilized to assess the quality of the GML granule from different batches. This method had been validated for precision, reproducibility and stability. Meanwhile, the mathematical statistical method PCA and OPLS-DA in our study was used for quantitative comparison of GML granule. It can be concluded that the main classification pattern was caused by their native geographical distribution. And the result demonstrates that the established method is a powerful and meaningful tool to comprehensively conduct the QC of GML granule. The authors acknowledge support for this work by the National Evaluation Project of Guangdong Institute for Food and Drug Control, which was also a task of pharmacopeia, cooperated with Huarun san-jiu pharmaceutical. The authors have declared no conflict of interest.
Liquid Chromatography
1857
[3] Yao, H., Shi, P. Y., Shao, Q., Fan, X. H., Chin. Med. 2011, 6, 1–9. [4] Qi, M., Xiong, A., Geng, F., Yang, L., Wang, Z., J. Sep. Sci. 2012, 35, 1470–1478. [5] Zeng, Y., Araujo, P., Grung, B., Zhang, L., Analyst 2011, 136, 1507–1514. [6] Liang, X. M., Jin, Y., Wang, Y. P., Jin, G. W., Fu, Q., Xiao, Y. S., J. Chromatogr. A 2009, 1216, 2033–2044. [7] Lazarowych, N. J., Pekos, P., Drug Inf. J. 1998, 32, 497– 512. [8] Lu, G.H., Chan, K., Liang, Y.Z., Leung, K., Chan, C. L., Jiang, Z. H., Zhao, Z. Z., J. Chromatogr. A 2005, 1073, 383–392. [9] Li, F. M., Yuan, B., Xiong, Z. L., Lu, X. M., Qin, F., Chen, H. S., Liu. Z. G., Biomed. Chromatogr. 2006, 20, 634– 641. [10] Yan, X., Liu, W., Yu, Z. T., Han, C. X., Zeller, F. J., Hasm, S. L. K., Yan, Y. M., Aust. J. Crop. Sci. 2014, 8(1), 140– 147. [11] Wren, S. A. C., Tchelitcheff, P., J. Chromatogr. A 2006, 1119, 140–146. [12] Ge, S., He, T. T., Hu, H., Chin. Med. 2014, 9, 22. [13] Wang, Y. J., Wen, J., Zheng, W. H., Zhao, L. S., Fu, X. H., W, Z. Z., Xiong, Z. Z., Li, F. M, X, W., Biomed. Chromatogr. 2014, doi 10.1002/bmc.3241. [14] Ye, Q., Deng, C. H., J. Chromatogr. Sci. 2009, 47, 287–290. [15] Sukumaran, P., Nair, A. G., Chinmayee, D. M., Mini, I., Sukumaran, S. T., Appl. Biochem. Biotechnol. 2012, 167, 1795–1801. [16] Yang, M., Sun, J. H., Lu, Z. Q., Chen, G. T., Guan, S. H., Liu, X., Jiang, B. H., Ye, M., Guo, D. A., J. Chromatogr. A. 2009, 1216, 2045–2062. [17] Chen, X. F., Wu, H. T., Tan, G. G., Zhu, Z. Y., Chai, Y. F., J. Pharm. Anal. 2011, 1, 235–245. [18] Zhou, J. L., Qi, L.W., Li, P., J. Chromatogr. A. 2009, 1216, 7582–7594. [19] Zhou, Y., Xu, G., Choi, F. F. K., Ding, L. S., Han, Q. B., Song, J. Z., Qiao, C. F., Zhao, Q. S., Xu, H. X., J. Chromatogr. A 2009, 1216, 4847–4858. [20] Wu, F. F., Sun, H., Wei, W. F., Han, Y., Wang, P., Dong, T. W., Yan, G. Y., Wang, X. J., J. Sep. Sci. 2011, 34, 3194– 3199. [21] Pang, X., Kang, L. P., Yu, H. S., Zhao, Y., Xiong, C. Q., Zhang, J., Shan, J. J., Ma, B. P., J. Sep. Sci. 2012, 35, 1538–1550. [22] Dunn III, W., Scott, D., Glen, W., Tetrahedron. Comput. Methodol. 1989, 2, 349–376. [23] Chen, Z., Tao, H., Liao, L., Zhang, Z., Wang, Z., J. Sep. Sci. 2014, 37, 2253–2259. [24] Wang, Z., Clifford, M., Acta Pharm. Sin. 2008, 43, 185.
5 References [1] Xie, P. S., Chen, S. B., Liang, Y. Z., Wang, X. H., Tian, R.T., Upton, R., J. Chromatogr. A 2006, 1112, 171–180. [2] Chen, Z. Y., Liao, L. P., Yang, Y. Y., Zhang, Z. J., Wang, Z. T., J. Sep. Sci. 2014, 37, 1762–1772.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[25] Jaiswal, R., Sovdat, T., Vivan, F., Kuhnert, N., J. Agric. Food Chem. 2010, 58, 5471–5484. [26] Chen, Y., Zhu, S.-B., Xie, M.-Y., Nie, S.-P., Liu, W., Li, C., Gong, X.-F., Wang, Y.-X., Anal. Chim. Acta 2008, 623, 146– 156. ´ ´ [27] Jurado, J., Alcazar, A., Pablos, F., Mart´ın, M., Gonzalez, A., Talanta 2005, 66, 1350–1354.
www.jss-journal.com
ISSN 1615-9306 · JSSCCJ 38 (11) 1813–2006 (2015) · Vol. 38 · No. 11 · June 2015 · D 10609
JOURNAL OF
SEPARATION SCIENCE
Methods Chromatography · Electroseparation Applications Biomedicine · Foods · Environment
11 15
www.jss-journal.com
1850 Qiong Lou1,2 Xiaolan Ye1 Yingyi Zhou2 Hua Li2 ∗ Fenyun Song1 ∗ 1 Department
of Pharmaceutical Analysis, School of Pharmacy, Guangdong Pharmaceutical University, Guangzhou, P.R. China 2 Department of Traditional Chinese Medicine, Guangdong Institute for Food and Drug Control, Guangzhou, P.R. China Received January 12, 2015 Revised March 2, 2015 Accepted March 10, 2015
J. Sep. Sci. 2015, 38, 1850–1857
Research Article
Chemical fingerprint of Ganmaoling granule by double-wavelength ultra high performance liquid chromatography and ultra high performance liquid chromatography with quadrupole time-of-flight mass spectrometry A method incorporating double-wavelength ultra high performance liquid chromatography with quadrupole time-of-flight mass spectrometry was developed for the investigation of the chemical fingerprint of Ganmaoling granule. The chromatographic separations were performed on an ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m) at 30⬚C using gradient elution with water/formic acid (1%) and acetonitrile at a flow rate of 0.4 mL/min. A total of 11 chemical constituents of Ganmaoling granule were identified from their molecular weight, UV spectra, tandem mass spectrometry data, and retention behavior by comparing the results with those of the reference standards or literature. And 25 peaks were selected as the common peaks for fingerprint analysis to evaluate the similarities among 25 batches of Ganmaoling granule. The results of principal component analysis and orthogonal projection to latent structures discriminant analysis showed that the important chemical markers that could distinguish the different batches were revealed as 4,5-di-Ocaffeoylquinic acid, 3,5-di-O-caffeoylquinic acid, and 4-O-caffeoylquinic acid. This is the first report of the ultra high performance liquid chromatography chemical fingerprint and component identification of Ganmaoling granule, which could lay a foundation for further studies of Ganmaoling granule. Keywords: Chemical fingerprint / Ganmaoling granule / Quadrupole time-of-flight mass spectrometry / Ultra high performance liquid chromatography DOI 10.1002/jssc.201500037
Additional supporting information may be found in the online version of this article at the publisher’s web-site
1 Introduction Traditional Chinese medicine (TCM; raw herbs, extractions, or preparations) has been widely used for healthcare in China since ancient times and has become increasingly popular around the world. Unlike modern pharmacology that often looks at the single chemical entity aimed at a specific single target, TCM often views complex matrices and one herb usually contains a myriad of components. Their overall efficacy is based on the synergic effect of multicompound Correspondence: Dr. Fenyun Song, Department of Pharmaceutical Analysis, School of Pharmacy, Guangdong Pharmaceutical University, Guangzhou 510006, P.R. China E-mail: [email protected] Fax: +86-020-89567081
Abbreviations: DAD, diode array detector; GML, Ganmaoling; OPLS-DA, orthogonal projection to latent structures discriminant analysis; PCA, principal component analysis; TCM, traditional Chinese medicine C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
and multiingredient preparations. Therefore, to evaluate the quality and authenticity of the complex preparations only by selecting a few markers or pharmacologically active components is troublesome [1]. Better analytical methods to assure their efficacy, safety, and stability are in great demand. Currently, the chromatographic fingerprint strategies play a significant role in the QC of TCM, which can not only determine the characteristic patterns of each plant type but also reveal the inherent relationships between multiple compounds [2]. Moreover, both the China Food and Drug Administration (CFDA) and European Medicines Agency (EMEA) have clearly accepted fingerprint chromatograms for the assessment of herbal medicines [3]. Additionally, it is a common practice to combine fingerprinting strategies with ∗ Additional correspondence: Dr. Hua Li, Department of traditional Chinese medicine, Guangdong Institute for Food and Drug Control, Guangzhou 510180, P.R. China E-mail:[email protected] Fax: +86-020-81886161
www.jss-journal.com
J. Sep. Sci. 2015, 38, 1850–1857
Liquid Chromatography
1851
acid (MUST-14043010), 4-O-caffeoylquinic acid (MUST14041410), 3,5-di-O-caffeoylquinic acid (MUST-14022612), 3,4-O-dicaffeoylquinic acid (MUST-14041411), and 4,5-di-Ocaffeoylquinic acid (MUST-14041414) were purchased from Beijing century audiocodes biological technology (Beijing, China). The purity of each standard was over 98%, which was suitable for UHPLC determination. Acetonitrile of HPLC grade was purchased from Honeywell Burdick & Jackson Honeywell Burdick & Jackson (Morristown, New Jersey, USA). Formic acid and methanol of HPLC grade were obtained from Merck (Germany). Ultrapure water was generated by means of a Milli-Q water purification system (Millipore, Milford, MA, USA). Other reagents were used as received. Twenty-five batches of GML granule were obtained from Huarun san-jiu pharmaceutical. These are 1301039H, 1212019H, 1303012H, 1301042H, 1306004H, 1303030H, 1303026H, 1304020H, 1304010H, 1210022H, 1212026H, 1304020H, 13020490H made by Huizhou manufacturer and 13010310, 1303033H, 1306006H, 1304023H, 1305035D, 1302045D, 1302003D, 1302370, 1305029D, 1308150, 13020260, 13010320 made by Zaozhaung manufacturer.
chemometrics analysis, such as principal component analysis (PCA), orthogonal projection to latent structures discriminant analysis (OPLS-DA), and similarity analysis to call attention to the chemical distinctions between the samples and identify their individual chemical markers [4, 5]. Recently, several methods including TLC, HPLC, GC, and CE are usually employed to develop the fingerprint, and HPLC fingerprint analysis is the first choice due to its advantages and popularization [6–9]. The latest UHPLC technique is designed to undertake separations under the high pressure that results from using nonporous silica [10, 11]. Ganmaoling (GML) granule is a popular compound preparation of traditional Chinese herbs, which consists of four strong antiviral herbs: Ye Ju Hua (F. Chrysanthemi Indici), Jin Zhan Yin Pan (B. biternata Merr. et Sherff), Gang Mei (Radix Ilex Asprella), and San Cha Ku (Radix Evodia Lepta), with therapeutic actions of reliving cold symptom, preventing and treating virus infection. It has been widely used in clinical practice for treating headache, fever, stuffy nose, sore throat, and so on, since 1970s as an over-the-counter (OTC) drug in China as well as North America [12]. Some chemicals such as 3-O-caffeoylquinic acid, 5-O-caffeoylquinic acid, 4-Ocaffeoylquinic acid, linarin, 3,4-di-O-caffeoylquinic acid, luteolin 7-O-glucoside, acetaminophen, caffeine are considered to be the main active ingredients in this formula. Those compounds have physiological activities such as fever-reducing, antibacterial, antiviral, and antipyretic activity, and other protective effects [13–15]. However, few studies have carried out the QC of GML granule effectively and systematically. This is due to unidentified components and difficult-to-obtain separation because of the presence of several isomeric quinic acids with similar structures. At present, an advanced technology, UHPLC coupled with Q-TOF-MS has rapidly been applied to identify unknown components of TCM, which combines advantages of ultrahigh resolution, short duration of analyses, and high sensitivity [16–21]. In this study, a characteristic fingerprint of GML granule using double-wavelength UHPLC had been developed, and the main components were identified by Q-TOF-MS initially. Significantly, PCA and orthogonal projection to latent structures discriminant analysis (OPLS-DA) were applied to evaluate the fingerprints of GML granule from 25 batches. It was regarded as the foundation of QC of GML granule completely.
The granule was ground into a powder using a mortar. Each sample powder (4.0 g) was accurately weighed and extracted using ultrasonic bath with 25 mL of methanol for 30 min. The extract was then filtered through a No. 40 filter paper, and 10 mL of the final extract was collected and transferred into an evaporating dish and evaporated to dryness on a 100⬚C water bath. The residue was dissolved in 50% methanol, adjusted to 5 mL, and filtered through 0.45 m membrane to obtain the test solution. Standard stock solutions of 11 compounds (3-Ocaffeoylquinic acid, 5-O-caffeoylquinic acid, 4-O-caffeoylquinic acid, 3,5-di-O-caffeoylquinic acid, 3,4-di-O-caffeoylquinic acid, 4,5-di-O-caffeoylquinic acid, linarin, luteoloside, caffeine, chlorphenamine maleate, acetaminophen) were prepared by dissolving appropriate amounts of standards into 50% methanol to achieve final concentrations of 70, 70, 70, 70, 70, 70, 450, 150, 15, 15 g/mL and 4 mg/mL, respectively.
2 Materials and methods
2.3 Double-wavelength UHPLC analysis
2.1 Chemicals and material
UHPLC was conducted on a Waters ACQUITY system (Waters, Milford, MA, USA) equipped with a binary pump with an online degasser, autosampler, column oven, diode array detector (DAD), and Empower Data Manager. The UHPLC fingerprint was carried out on an ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m) at 30⬚C with sample injection volume of 2 L at a flow rate of 0.15 mL/min. The mobile phase consisted of acetonitrile (B) and 0.1% formic acid in water (A) with a gradient elution program—that is 0–6 min,
Standards of 3-O-caffeoylquinic acid (MUST-14043010), buddlejoside (111528-200606), acetaminophen (100018-200408), caffeine (171215-201211), luteolin 7-O-glucoside (111720201307), and chlorphenamine maleate (100047-200606) were purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China; purity >98%). Standards of 5-O-caffeoylquinic C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2.2 Preparation of sample solution and standard solution
www.jss-journal.com
1852
Q. Lou et al.
6% B; 6–10 min, 6–15% B; 10–24 min, 15–30% B; 24–28 min, 30–6% B; and 2 min post run, 6% B. During optimization of the UHPLC analytical procedure, it was observed that most of the components in samples had their largest chromatographic responses at 310 nm and that many of the remaining analytes had their most intense responses at 264 nm. Thus, the fingerprints in this study were collected at detection wavelengths of 310 and 264 nm. 2.4 UHPLC–Q-TOF-MS analysis The Agilent 1290 UHPLC–DAD–Q-TOF-MS (CA, USA) system consisted of G4204A Quaternary Pump, G4226A automatic sample injector, G1316C thermostatted column compartment, G4212A DAD and 6540 UHD Accurate-Mass Q-TOF was used. The UHPLC fingerprint was carried out on an ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m) at 30⬚C with sample injection volume of 2 L. The mobile phase consisted of 0.1% formic acid in water (A) and acetonitrile (B) using gradient program at a flow rate of 0.15mL/min. Analysis began with an isocratic step of 6% B for 6 min, then 15–30% (B) in 10–24 min, 30–6% (B) in 24–28 min, and reequilibrated for 2 min. The DAD was set at 310 and 264 nm. Each sample was detected with an ESI source operating in positive and negative ion modes. Q-TOF parameters were as follows: scan range m/z, 100–1700; dry gas, 8 L/min; dry gas temperature, 350⬚C; nebulizer pressure, 30 psi; capillary voltage, 4 kV; fragmentor and skimmer voltages, 175 and 65 V.
2.5 Data analysis UHPLC data were analyzed by the professional software Similarity Evaluation System for Chromatographic Fingerprint of TCM (Version 2009), which was used to align the peak areas with retention times. The data of aligned peak areas were exported. PCA and OPLS-DA were used to describe variations in the data with minimum latent variables that were demonstrated with SIMCA-P+ (CA,USA). This facilitated grouping or classification of the fingerprints and detected outliers [22, 23].
3 Results and discussion 3.1 Optimization of UHPLC condition Several UHPLC analytical parameters including separation column, mobile phase, its elution mode and flow rate, detection wavelength (310, 327, 350, and 264 nm) of DAD, and column temperature were optimized to provide sufficient information about the analyzed compounds, and to achieve good resolution and reasonable analysis time. Different types of columns were tested to reach an optimum separation on the ACQUITY UPLC HSS C18 column (2.1 × 50 mm, 1.8 m). After several trials using different mobile phases including acetonitrile and methanol, 0.1% formic acid water solution and C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
J. Sep. Sci. 2015, 38, 1850–1857
methanol, 0.1% formic acid water solution and acetonitrile, 0.1% TFA water solution and acetonitrile, and 0.08% formic acid water solution and acetonitrile, 0.1% formic acid water solution and acetonitrile was determined as the most appropriate eluent with a step linear gradient due to the satisfactory resolution and acceptable peak parameters it provided. The 30⬚C column temperature and flow rate of 0.15 mL/min were found to be optimal parameters. These conditions were also suitable for UHPLC–Q-TOFMS analysis. The optimal condition for MS (collision energy, fragmentation voltage, and capillary voltage) was selected, and ion modes (positive and negative) were applied to identify the corresponding signals. The optimal UHPLC and MS conditions are detailed in Sections 2.3 and 2.4. 3.2 Optimization of the extraction method To extract the investigated compounds from the GML granule, the extraction solvent (methanol, ethanol, and 50% methanol), the extraction method (ultrasonic extraction and refluxing), and the extraction time (15, 30, and 40 min) were carried out for the single factor test. The optimal extraction method of the GML granule was 4.0 g of the dry powder with 25 mL of 50% methanol heated in an ultrasonic water bath for 30 min. To acquire more accurate results, the sample was concentrated twice. 3.3 Validation for UHPLC fingerprint The method validation of fingerprint analysis was performed based on the relative retention time (the ratio of peak retention time of sample constituents to the reference standard) and the relative peak area (the ratio of peak area of sample constituents to the reference standard). Among these active components, buddleoside indicates a proper retention time and higher content; therefore, we chose it as the reference substance. All common peaks’ relative retention time and relative peak area were obtained on the basis of this substance. The precision of the proposed method, on the basis of analyzing six replicate samples, were below 0.78 and 4.82% for RSD of relative retention time and relative peak area of all peaks, respectively. The stability test was performed with sample solutions for 24 h, which resulted in RSDs of relative retention times and peak areas that were less than 0.91 and 4.70%, respectively. The reproducibility test was performed with six sample solutions extracted from one batch of preparation, which yielded RSDs less than 0.90 and 4.70%. These results indicated that double-wavelength UHPLC fingerprint method developed for this study was adequate, valid, and applicable. The typical fingerprint chromatograms with 25 “common peaks” in the double wavelength are shown in Fig. 1. 3.4 Identification of the ingredients Based on the standard substances and mass spectral information of UHPLC–Q-TOF-MS, 11 compounds were identified. www.jss-journal.com
J. Sep. Sci. 2015, 38, 1850–1857
Liquid Chromatography
1853
Figure 1. Typical chromatograms of test solution of GML granule (a1, b1) and mixed reference solution (a2, b2). 1, chromatogram at 310 nm; 2, chromatogram at 264 nm.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.jss-journal.com
1854
Q. Lou et al.
J. Sep. Sci. 2015, 38, 1850–1857
Figure 2. The total ion current (TIC) chromatogram of mixed reference solution (d1, d2) and test solution of GML granule (e1, e2). 1, TIC in positive mode; 2, TIC in negative mode.
The total ion current (TIC) chromatograms of 11 mixed reference solution (d1, e1) and test solution of GML granule (d2, e2) are shown in Fig. 2. According to the accurate masses, the molecular formula of each constituent was deduced, and the fragmentations and accurate mass of each fragmental ion were obtained. Compound 7, 9, 10, 14, 15, 16, 17, 20, 23, 24, and 25 were unambiguously identified by comparing with reference standards. The results of UHPLC–Q-TOF-MS analysis and chemical structure are summarized in Table 1 and Fig. 3, where six peaks of [M–H]− at m/z 515 and 353 occurred with a higher presence of quinic acid derivatives [2, 14, 24, 25]. The molecular weight of compound 7 is 354.3087 and its chemical formula is C16 H18 O9 . MS2 spectra of this compound showed [M–H]– at m/z 353.1, [M–H– 162]− ([M–H–C6 H10 O5 ]− ) at m/z 191.0, indicating a loss of caffeoyl acid anion from the mother ion. A characteristic C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
fragment ion of the basic peak at m/z 135.0 with chemical formula C8 H7 O2 – resulted from the loss of 1,3, 4-trihydroxycyclohexane carboxylic acid anion and CO2. Therefore, compound 7 was deduced to be neochlorogenic (5-O-caffeoylquinic acid), which was confirmed by comparing its spiking with the reference compound. Compounds 7, 9, and 10 are a group of isomers that have identical molecular ions and similar fragment pathways. Compound 10 showed the characteristic fragments of 191.1 and 127.0 [M–H–162–18–18–28]– ([M–H– C6 H10 O5 –H2 O–H2 O–CO]− ), and therefore was deduced to be 4-O-caffeoylquinic acid. Additional peak was observed at m/z 191.0 in MS2 data of compound 9, which was produced by loss of a caffeoyl acid anion, suggesting compound 9 is likely chlorogenic acid (3-O-caffeoylquinic acid). www.jss-journal.com
Liquid Chromatography
J. Sep. Sci. 2015, 38, 1850–1857
1855
Table 1. UHPLC–Q-TOF-MS identification of the major chemical compounds in GML granule
Peaks no
Rt /min
Identification
Molecular formula
[M+X]+ [M–X]−
Expected m/z
Fragments
Experimental m/z
7 9 10 14
3.62 5.98 6.78 13.72
5-O-caffeoylquinic acid 3-O-caffeoylquinic acid 4-O-caffeoylquinic acid Luteolin 7-O-glucoside
C16 H18 O9 C16 H18 O9 C16 H18 O9 C21 H20 O11
15 16 17 20 23
14.93 15.33 16.66 20.61 3.01
3,4-Di-O-caffeoylquinic acid 3,5-Di-O-caffeoylquinic acid 4,5-Di-O-caffeoylquinic acid Buddleoside 4-Acetamidophenol
C25 H24 O12 C25 H24 O12 C25 H24 O12 C28 H32 O14 C8 H9 NO2
353.1 353.1 353.1 447.1 449.1 515.1 515.1 515.1 593.2 152.0
6.10 14.75
Caffeine Chlorpheniramine maleate
C8 H10 N4 O2 C16 H19 ClN2 .C4 H4 O4
353.3 353.3 353.3 447.1 449.1 515.5 515.5 515.5 593.6 152.2 150.2 195.2 274.9
191.0, 135.0 191.0 191.1, 127.0 285.1 287.0 353.1, 173.0 353.1, 191.0, 135.0 353.1, 173.0, 135.0 285.0 110.0
24 25
[M–H]− [M–H]− [M–H]− [M–H]− [M+H]+ [M–H]− [M–H]− [M–H]− [M+H]+ [M+H]+ [M–H]− [M+H]+ [M+H–C4 H4 O4 ]+
138.0, 110.0 230.0
195.1 275.1
Figure 3. Chemical structures of 11 compounds identified by UHPLC–Q-TOF-MS in GML granule.
The molecular formula of compound 15 was established as C25 H24 O12 by UHPLC–Q-TOF-MS analysis (m/z 515.1 [M– H]− ). A series of distinctive product ions at m/z 353.1 [(M– H)–162]− and 173.0 [(M–H)–162–162–18]– were observed from the loss of successive caffeoyl acid anions and H2 O. Compound 15 was elucidated to be isochlorogenic acid B (3,4-di-O-caffeoylquinic acid) based on the results of DAD detection. Compounds 15, 16, and 17 also are a set of isomers. MS2 spectra of compound 16 showed the fragment peaks at m/z 353.1, 191.0, 135.0 were forced by eliminating two C6 H10 O5 C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
molecules from the precursor ion at m/z 515.1, suggesting that compound 16 was likely isochlorogenic acid A (3,5-di-Ocaffeoylquinic acid). Compound 17 showed the characteristic fragments of 353.1, 173.0, and 135.0, and based on the results of DAD detection, which was deduced to be isochlorogenic acid C (4,5-di-O-caffeoylquinic acid). The molecular formula of compound 14 was identified as C21 H20 O11 by UHPLC–Q-TOF-MS analysis (as m/z 447.1 [M– H]− , m/z 449.1 [M+H]+ ). The pseudo molecular ion at m/z 447.1 [M–H]– yielded a fragmentation ion at m/z 285.1 [M–H-162]– and ion at m/z 449.1 [M+H]+ produced a www.jss-journal.com
1856
J. Sep. Sci. 2015, 38, 1850–1857
Q. Lou et al.
Figure 4. OPLS-DA score plot (A) and loading plot (B) for all samples of GML granule.
fragment ion at m/z 287.0 [M+H–162]+ in the MS/MS spectrum by loss of glucose. Compound 14 was elucidated to be luteolin 7-O-glucoside. The molecular formula of compound 20 was established as C28 H32 O14 by UHPLC–Q-TOF-MS analysis (m/z 593.2 [M+H]+ ). A distinctive product ion at m/z 285.0 [M+H– 308]+ was observed resulting from loss of C12 H20 O9 + . Compound 14 was elucidated to be buddleoside, based on the results of DAD detection. The molecular weight of compound 23 is 151.2, and its chemical formula is C8 H9 NO2 . MS2 spectra of this compound showed [M+H–42]+ at m/z 110.0 indicating the loss of C2 H2 O+ . The molecular formula of compound 20 was established as C16 H19 ClN2 .C4 H4 O4 by UHPLC–Q-TOF-MS analysis (as m/z 275.1 [M+H–116]+ ) by the loss of C4 H4 O4 + . The representative product ion at m/z 230.0 [M+H–116–45]+ was observed by loss of NC2 H7 + . Compound 25 was elucidated to be chlorpheniramine maleate, based on the results of DAD detection.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
3.5 PCA and OPLS-DA PCA is an advanced technique which was widely applied to summarize multivariate variation in a few principal components remaining maximum reasonable variability. To assess variations in quality and to differentiate between samples, PCA was adopted based on the contents of 25 compounds in 25 samples by analyzing the final matrix of double-wavelength fingerprints. The contents of 25 fingerprint peaks were applied to evaluate variations in the GML granule. As shown in the score plot of the first two principal components of doublewavelength fingerprint, Fig. 4A, all of the samples could be classified into two groups. Additionally, sample 6 and sample 8 were obviously different from the other samples. To make sure that the sample was an outlier, PCA analysis within the batches was carried out. The points for sample 6 and sample 8 were graphed far from the other points. In Hostelling’s T2 range plot, the point above the green horizontal line revealed the probability that the sample was less than 5% similar to the other samples [4]. Therefore, the samwww.jss-journal.com
J. Sep. Sci. 2015, 38, 1850–1857
ple was removed from the analysis. As shown in Fig. 4B, the scores also show some dispersion among the batches; this may due to the samples’ places of origin and times of collection. To confirm the two-group data model studied above, OPLS-DA was performed. This enabled us to calculate discriminant functions as linear combinations of the selected descriptors, which maximized the ratio between-class sum of squares and the within-class sum of squares [26, 27]. As shown in Fig. 4A, each point represented a particular sample and was automatically color-coded according to its class. The OPLS-DA model in this experiment resulted in a (1 + 1) component with a Q2 of 64.5%, R2(X) of 48.3%, and R2(Y) of 89.1% of the variables, which showed the good quality of the model. By examining the loadings plot, the important chemical markers separating these samples were revealed. From the loadings plot, the variables with larger covariance and correlation values were considered to be the discriminatory markers, while the variables with larger variable importance in the projection (VIP) values were more relevant for sample classification. As shown in Fig. 4B, the loadings plot of the OPLS-DA results demonstrated that peaks 17, 18, 16, and 10 most influenced the discrimination among different batches of GML granule.
4 Concluding remarks In conclusion, the double-wavelength UHPLC fingerprint and UHPLC–Q-TOF-MS techniques were used in this study that allowed the identification and comparison of GML granule and can be utilized to assess the quality of the GML granule from different batches. This method had been validated for precision, reproducibility and stability. Meanwhile, the mathematical statistical method PCA and OPLS-DA in our study was used for quantitative comparison of GML granule. It can be concluded that the main classification pattern was caused by their native geographical distribution. And the result demonstrates that the established method is a powerful and meaningful tool to comprehensively conduct the QC of GML granule. The authors acknowledge support for this work by the National Evaluation Project of Guangdong Institute for Food and Drug Control, which was also a task of pharmacopeia, cooperated with Huarun san-jiu pharmaceutical. The authors have declared no conflict of interest.
Liquid Chromatography
1857
[3] Yao, H., Shi, P. Y., Shao, Q., Fan, X. H., Chin. Med. 2011, 6, 1–9. [4] Qi, M., Xiong, A., Geng, F., Yang, L., Wang, Z., J. Sep. Sci. 2012, 35, 1470–1478. [5] Zeng, Y., Araujo, P., Grung, B., Zhang, L., Analyst 2011, 136, 1507–1514. [6] Liang, X. M., Jin, Y., Wang, Y. P., Jin, G. W., Fu, Q., Xiao, Y. S., J. Chromatogr. A 2009, 1216, 2033–2044. [7] Lazarowych, N. J., Pekos, P., Drug Inf. J. 1998, 32, 497– 512. [8] Lu, G.H., Chan, K., Liang, Y.Z., Leung, K., Chan, C. L., Jiang, Z. H., Zhao, Z. Z., J. Chromatogr. A 2005, 1073, 383–392. [9] Li, F. M., Yuan, B., Xiong, Z. L., Lu, X. M., Qin, F., Chen, H. S., Liu. Z. G., Biomed. Chromatogr. 2006, 20, 634– 641. [10] Yan, X., Liu, W., Yu, Z. T., Han, C. X., Zeller, F. J., Hasm, S. L. K., Yan, Y. M., Aust. J. Crop. Sci. 2014, 8(1), 140– 147. [11] Wren, S. A. C., Tchelitcheff, P., J. Chromatogr. A 2006, 1119, 140–146. [12] Ge, S., He, T. T., Hu, H., Chin. Med. 2014, 9, 22. [13] Wang, Y. J., Wen, J., Zheng, W. H., Zhao, L. S., Fu, X. H., W, Z. Z., Xiong, Z. Z., Li, F. M, X, W., Biomed. Chromatogr. 2014, doi 10.1002/bmc.3241. [14] Ye, Q., Deng, C. H., J. Chromatogr. Sci. 2009, 47, 287–290. [15] Sukumaran, P., Nair, A. G., Chinmayee, D. M., Mini, I., Sukumaran, S. T., Appl. Biochem. Biotechnol. 2012, 167, 1795–1801. [16] Yang, M., Sun, J. H., Lu, Z. Q., Chen, G. T., Guan, S. H., Liu, X., Jiang, B. H., Ye, M., Guo, D. A., J. Chromatogr. A. 2009, 1216, 2045–2062. [17] Chen, X. F., Wu, H. T., Tan, G. G., Zhu, Z. Y., Chai, Y. F., J. Pharm. Anal. 2011, 1, 235–245. [18] Zhou, J. L., Qi, L.W., Li, P., J. Chromatogr. A. 2009, 1216, 7582–7594. [19] Zhou, Y., Xu, G., Choi, F. F. K., Ding, L. S., Han, Q. B., Song, J. Z., Qiao, C. F., Zhao, Q. S., Xu, H. X., J. Chromatogr. A 2009, 1216, 4847–4858. [20] Wu, F. F., Sun, H., Wei, W. F., Han, Y., Wang, P., Dong, T. W., Yan, G. Y., Wang, X. J., J. Sep. Sci. 2011, 34, 3194– 3199. [21] Pang, X., Kang, L. P., Yu, H. S., Zhao, Y., Xiong, C. Q., Zhang, J., Shan, J. J., Ma, B. P., J. Sep. Sci. 2012, 35, 1538–1550. [22] Dunn III, W., Scott, D., Glen, W., Tetrahedron. Comput. Methodol. 1989, 2, 349–376. [23] Chen, Z., Tao, H., Liao, L., Zhang, Z., Wang, Z., J. Sep. Sci. 2014, 37, 2253–2259. [24] Wang, Z., Clifford, M., Acta Pharm. Sin. 2008, 43, 185.
5 References [1] Xie, P. S., Chen, S. B., Liang, Y. Z., Wang, X. H., Tian, R.T., Upton, R., J. Chromatogr. A 2006, 1112, 171–180. [2] Chen, Z. Y., Liao, L. P., Yang, Y. Y., Zhang, Z. J., Wang, Z. T., J. Sep. Sci. 2014, 37, 1762–1772.
C 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[25] Jaiswal, R., Sovdat, T., Vivan, F., Kuhnert, N., J. Agric. Food Chem. 2010, 58, 5471–5484. [26] Chen, Y., Zhu, S.-B., Xie, M.-Y., Nie, S.-P., Liu, W., Li, C., Gong, X.-F., Wang, Y.-X., Anal. Chim. Acta 2008, 623, 146– 156. ´ ´ [27] Jurado, J., Alcazar, A., Pablos, F., Mart´ın, M., Gonzalez, A., Talanta 2005, 66, 1350–1354.
www.jss-journal.com
