Chemical fingerprinting and quantitative constituent analysis of Siwu decoction categorized formulae by UPLC-QTOF/MS/MS and HPLC-DAD

Background Siwu decoction categorized formulae (SWDCF) are widely used for treating gynecological diseases. This study aims to elucidate the differences of bioactive constituents in SWDCF by ultra-high performance liquid chromatography coupled with time-of-flight mass spectrometry (UPLC - QTOF - MS /MS) and HPLC-DAD. Methods An efficient method based on UPLC - QTOF - MS /MS was developed for identifying the chemical profiles of SWDCF. HPLC-DAD method was used for quantifying seven chemical markers in SWDCF. Results Eighty four components were identified or characterized, including ten organic acids, thirty glycosides (monoterpene or iridoid or phenylpropanoids glycosides), fourteen lactones, eighteen flavonoids, and eleven alkaloids in the complex system. The datasets of tR-m/z pairs, ion intensities and sample codes were processed with supervised orthogonal partial least squared discriminant analysis to compare these decoction samples. After a clear classification was established, OPLS-DA was performed and 16 common components with relative quantity in SWDCF samples were determined. Gallic acid, protocatechuic acid, vanillic acid, caffeic acid, paeoniflorin, ferulic acid, and senkyunolide I were selected as the chemical markers to identify SWDCF by HPLC-DAD. Conclusion The chemical profiles with 84 components in SWDCF, including monoterpene glycosides, acetophenones, galloyl glucoses, even some isomers in the complex system were characterized by UPLC–QTOF–MS/MS.


Conclusion:
The chemical profiles with 84 components in SWDCF, including monoterpene glycosides, acetophenones, galloyl glucoses, even some isomers in the complex system were characterized by UPLC-QTOF-MS/MS.

Background
Chinese medicine (CM) is holistically formulated for treating complicated CM syndromes ("ZHENG" in Chinese) [1]. The researches on categorized formulae (CF) based on a basic formula composition may help understand the rules of formulation and Fang-Zheng consistency in CM [2].
Siwu decoction (SWD) is a classical prescription that is widely used for the treatment of women's diseases in CM, such as relief of emmeniopathy, climacteric syndrome, dysmenorrhea and other estrogen-related diseases [3]. A recent study demonstrated that the SWD can be synergistically used with Western medicine [4]. SWD consists of four herbs, i.e., Angelicae sinensis Radix, Chuanxiong Rhizoma, Paeoniae Radix Alba, and Rehmanniae Radix [5]. A series of SWD-based formulae used to treat women's diseases especially for primary dysmenorrheal (PD) are considered as Siwu decoction categorized formulae (SWDCF), including Taohong Siwu decoction (THSWD), Xiangfu Siwu decoction (XFSWD), Shaofu Zhuyu decoction (SFZYD), and Qinlian Siwu decoction (QLSWD). The compositions and applications of SWDCF were described in Table 1. These formulae are usually adopted to treat different symptoms of different kinds of PD.
The investigations of the components of SWDCF are important to reveal their effects and action mechanisms. However, the complicated chemical profiles of SWDCF components demand a rapid and efficient method for chemical profiling of SWDCF.
This study aims to profile the constituents of SWDCF by UPLC-QTOF-MS/MS with MarkerLynx analysis and identify SWDCF with several chemical markers.

Chemicals, reagents and materials
Gallic acid, protocatechuic acid, vanillic acid, caffeic acid, paeoniflorin, ferulic acid, and senkyunolide I ( Figure 1) were purchased from National Institute for the Control of Pharmaceutical and Biological products (China).
Acetonitrile was HPLC-grade from Merck (Darmstadt, Germany) and deionized water was purified by a Millipore water purification system (Millipore, Milford, MA, USA) and filtered with 0.22 μm membranes. Other reagent Table 1 The composition and application of SWDCF

Apparatus and chromatographic conditions UPLC-QTOF-MS/MS qualitative analysis
Chromatography was performed on an AcQuity ™ UPLC system with a conditioned autosampler (Waters Corp., Milford, MA, USA) at 4°C. The separation was carried out on an AcQuity UPLCTM BEH C 18 column (100 mm × 2.1 mm i.d., 1.7 μm; Waters Corp., Milford, MA, USA) maintained at 35°C. The mobile phase consisted of 0.1% formic acid (HCOOH) in water as solvent A and acetonitrile (ACN) as solvent B. The gradient conditions of the mobile phase were: 0 min 95% A, 9.0 min 56% A, 12.0 min 26% A, 20.0 min 10% A, 22.0 min 10% A, 25.0 min 95% A. The flow rate was 0.40 mL/min. The sample injection volume was 5 μL.
Mass spectrometric detection was carried out on an AcQuity Synapt Mass Spectrometer equipped with an electrospray ionization (ESI) interface (Waters, Milford, MA, USA). High purity nitrogen was used as the nebulizer and auxiliary gas; argon was utilized as the collision gas.

HPLC-PDA quantificative analysis
The quantification analysis was performed on a Waters-2695 Alliance HPLC (Waters Corporation, Milford, MA, USA) equipped with an on-line degasser, an auto-sampler and a 2996 photodiode array detector. UV detection was achieved at 210-400 nm. A Waters Sun Fire ™ C 18 column (4.6 × 250 mm, 5 μm, serial no. 186002560 Waters Corporation, USA) was used. A linear gradient elution of A CH 3 OH and B (CH 3 COOH: H 2 O = 0.1: 100) was used. The gradient program is 85% B in 0-8 min, 85-80% B in 8-10 min, 80-74% B in 10-30 min, 74-66% B in 30-40 min, 66-34% B in 40-60 min, 34-10% B in 60-85 min. The solvent flow rate was 1 mL/min and the column temperature was set at 30°C. Re-equilibration duration was 15 min between individual runs. A Waters 2996 photo diode array was connected to the liquid chromatography for detection of the raw data.

Preparation of sample solutions
The mixtures of SWDCF (Table 1) were crushed into small pieces and refluxed with 10 times water for 2 h twice. The filtrates from each decoction were combined and concentrated to 1.0 mg/mL at 70°C. The filtrates were added 95% ethanol until the concentration of ethanol was adjusted to 50%. After centrifugation at 3000 × g for 10 min, the supernatant was stored at 4°C and filtered through a 0.22 μm membrane filter before the UPLC-QTOF-MS analysis and the HPLC-DAD analysis.

Validation of the HPLC method Calibration curves, limits of detection and quantification
The working standard solutions with at least six different concentrations (1, 148-1480.0 μg/mL; 2, 38.40-384.00 μg/mL; 3, 32.0-320.0 μg/mL; 4, 35.6-356.0 μg/mL; 5, 476.0-4760.0 μg/mL; 6, 136.0-1360.0 μg/mL; 7, 46.0-460.0 μg/mL) were analyzed, and the calibration curves were calculated by linear regression of the double logarithmic plots of the peak area versus the concentration of the reference solution injected. The limits of detection and quantification (LODs and LOQs) under the present chromatographic conditions in this study were determined by diluting the standard solution when the signal-to-noise ratios (S/N) of analytes were about 3 and 10, respectively. The S/N was calculated as the peak height divided by the background noise value.

Precision, repeatability and accuracy
The intra-day and inter-day variations, which were chosen to determine the precision of the developed method, were investigated by determining the seven analytes in six replicates during a single day and by duplicating the experiments on three consecutive days. Variations of the peak area were taken as the measures of precision and expressed as relative standard deviations (R.S.D.).
Repeatability was confirmed with six independent analytical sample solutions prepared according to the methods describing before and expressed by R.S.D. These SWTCF sample solutions was stored at 4°C, and injected into the HPLC apparatus at 0, 2, 4, 8, and 12 hr, respectively, to evaluate the stability of the solution.
Recovery test was performed by adding accurate amounts of the seven standards into a certain amounts separately to evaluate the accuracy of this method. The spiked samples were then extracted, processed, and quantified in accordance with the methods mentioned above. Six replicates were performed for the test. The average recovery percentage was calculated by the formula: recovery (%) = (observed amount − original amount)/spiked amount × 100%.

MarkerLynx ™ analysis
The data obtained from positive and negative ion mode of all determined samples were analyzed by MarkerLynx v4.1 software (Waters, Manchester, UK) with the help of multivariate statistical analysis, to reveal any potential changed components in SWDCF. The original data were processed using the following parameters: initial retention time of 0 min, final retention time of 25 min, and mass in the range 100-1000 Da, with a mass tolerance of 0.02 Da, mass window of 0.02 Da, retention time window of 0.1 min, noise elimination level 6. For peak integration, peak width at 5% of the height was 1 s, peak-to -peak baseline noise was automatically calculated, and peak intensity threshold was 10. No specific mass or adduct was excluded. Isotopic peaks were excluded for analysis. For data analysis, a list of the intensities of the peaks detected was generated using tR and mass data (m/z) pairs as the identifier of each peak. An ID was assigned to each of these tR-m/z pairs in the order of their UPLC elution for data alignment [15]. The process was repeated for each run. After completion, the correct peak intensity data for each tR-m/z pair of the entire batch of samples were aligned in the final data table.
The ions that showed the same tR (with a tolerance of 0.1 min) and m/z value (with a tolerance of 0.05 Da) in different samples were considered as the same ion. For those peaks hard to be detected in the sample, the ion intensities were documented as zero in the final data table. Before submitted for multivariate analyses, the ion intensities for each detected peak were normalized against the sum of the peak intensities. The resulting threedimensional data comprising of peak number (tR-m/z pair), sample name and ion intensity were analyzed by OPLS-DA with the MarkerLynx software.

Identification of SWDCF constituents by UPLC-QTOF-MS/ MS
The multiple chemical components in the SWDCF samples were identified and characterized by both negative and positive ESI modes. The total current chromatograms at the two modes were shown in Additional file 1: Figures S1-1 and 1-2. Diagnostic fragmentations were then checked by MS/MS to confirm the results. Eighty-four constituents were identified by comparing the tR, UV λmax , and MS fragments characteristics of the compounds. The analyzed and identified compounds were listed in Table 2. Among these 84 compounds, there were ten organic acids, thirty glycosides (monoterpene or iridoid glycosides or phenylpropanoids), fourteen lactones, eighteen flavonoids, and eleven alkaloids.

Identification of alkaloids
The alkaloids compounds derived mainly from Chuanxiong Rhizoma, Corydalis Rhizoma, and Coptidis Rhizoma. At the positive ion mode, 11 alkaloids constituents were analyzed and identified from SWDCF by comparing with reference compounds and literature data [35]. UPLC-Q-TOF-MS method was employed to identify the constituents from SWDCF. In ESI-TOF-MS experiment, accurate molecular mass of the components can be obtained. As ESI was a soft ionization technique, the interface produces little fragmentation of analytes and generally forms protonated molecular ions [M + H] + for positive ionization mode or [M-H]for negative ionization mode. Comparing the mass spectra of the compounds with the standards and those in the literature, the common and different components were unequivocally identified from every formula of SWDCF. These data would be provided the bioactive components for activities of different formulae.

The markers obtained by MarkerLynx ™
The principal components analysis (PCA) was done by the Waters MarkerLynx ™ software. In this study, SWD, THSWD, XFSWD, SFZYD and QLSWD were injected six times each in two ESI modes. Unsupervised PCA was performed to globally evaluate the chemical consistency among these five SWDCF decoctions. The data from both positive and negative ion modes were displayed as scores plots (Additional file 1: Figures S2-1 and 2-2). The scores plots demonstrated a clear classification trend among SWDCF samples, with all the observations falling within the Hotelling T2 (0.95) ellipse, which confirmed the fact that chemical difference exists among SWDCF. The results showed that the SFZYD and QLSWD were significantly different to SWD, THSWD, and XFSWD, while SWD, THSWD, and XFSWD were close to each other. These data indicated that the chemical composition and quantity of components changed after combining SWD with different herbs.  Chemical markers were analyzed to find out the changed common components contributing most to the SWDCF.

Optimization of the HPLC conditions
A small amount of acid was added into the mobile phase which could inhibit the ionization of these components to improve the peak shape and restrain the peak tailing due to the existence of acidic ingredients in SWDCF samples. 0%, 0.1% and 0.2% aqueous formic acid and acetic acid solutions were compared. The results showed that 7 compounds could be baseline separated when 0.1% aqueous formic acid solution was selected.
DAD detection was set at the wavelength range of 190-400 nm. For the satisfactory sensitivity, resolution and lower noise, four wavelengths at 230 nm, 260 nm, 277 nm, and 320 nm were selected for determining the different compounds in SWDCF. According to absorption curve of the tested analytes, the paeoniflorin had optimal sensitivity for detection at 230 nm, 320 nm for ferulic acid, 260 nm for gallic acid, vanillic acid, and caffeic acid, and 277 nm for senkyunolide I. Thus, a switching UV wavelength method was established by a variable-wavelength spectrophotometric detector. Under the optimized HPLC-UV conditions, the investigated analytes were well separated and detected in 85 min ( Figure 3).
Prior to sample analysis the optimal process of extraction had to be investigated. According to the applied form of decoction, the SWDCF were extracted by refluxing with water for twice. Ethanol (95%) was added to the filtrates until the concentration of ethanol was adjusted to 50% and seven compounds were almost completely extracted.
As shown in Table 4, the intra-and inter-day precisions, repeatability and stability of the seven analytes were less than 3%. The overall recoveries lay between 92.20% and 104.60% with RSD less than 3.54% for seven components in all samples. These results indicated that the HPLC fingerprint chromatograms had a good repeatability, precision, accuracy, and recovery (Tables 4 and 5) and the developed HPLC-DAD method was a reliable and useful method for assessment of SWDCF.

Sample analysis
The HPLC-DAD method was then subsequently applied to simultaneously determine the chemical markers including gallic acid, protocatechuic acid, vanillic acid, caffeic acid, paeoniflorin, ferulic acid, senkyunolide I in SWDCF samples. The results (Table 6) showed there were remarkable differences among the contents of the chemical markers analyzed in different samples. Paeoniflorin (5) was found to be a predominant constituent in both of QLSWD and SFZYD, while the lowest contents in SWD except XFSWD, suggesting that the active compounds of paeoniflorin was dissolved increasedly after the SWD combined with other herbs. The contents of senkyunolide I in SWDCF were decreased except SFZYD varing from 31.30 to 84.70 mg/g. The phenolic acids including gallic acid, caffeic acid, ferulic acid, vanillic acid, and protocatechuic acid were increased significantly except in XFSWD.

Conclusion
The chemical profiles with 84 components in SWDCF, including monoterpene glycosides, acetophenones, galloyl glucoses, even some isomers in the complex system were characterized by UPLC-QTOF-MS/MS.

Additional file
Additional file 1: Figure S1.