Chemical Profile Analysis and Comparison of Two Versions of the Classic TCM Formula Danggui Buxue Tang by HPLC-DAD-ESI-IT-TOF-MSn

Danggui Buxue Tang (DBT) is a Traditional Chinese Medicine (TCM) formula primarily used to treat symptoms associated with menopause in women. Usually, DBT is composed of one portion of Radix Angelicae Sinensis (RAS) and five portions of Radix Astragali (RA). Clinically, Radix Hedysari (RH) is sometimes used by TCM physicians to replace RA in DBT. In order to verity whether the chemical constituents of the DBT1 (RA:RAS = 5:1, w/w) and DBT2 (RH:RAS = 5:1, w/w) share similarities the chemical profiles of the two DBTs crude extracts and urine samples were analyzed and compared with the aid of HPLC-DAD-ESI-IT-TOF-MSn, which determines the total ion chromatogram (TIC) and multi-stage mass spectra (MSn). Then, the DBT1 and DBT2 were identified and compared on the basis of the TIC and the MSn. In the first experiment (with crude extracts), 69 compounds (C1–C69) were identified from the DBT1; 46 compounds (c1–c46) were identified from the DBT2. In the second experiment(with urine samples), 44 compounds (M1–M44) were identified from the urine samples of rats that had been administered DBT1, and 34 compounds (m1–m34) were identified from the urine samples of rats that had been administered DBT2. Identification and comparison of the chemical compositions were carried out between the DBT1 and DBT2 of the crude extracts and urine samples respectively. Our results showed that the two crude extracts of the DBTs have quite different chemical profiles. The reasons for their differences were that the special astragalosides in DBT1 and the isoflavonoid glycosides formed the malonic acid esters undergo single esterification and acetyl esters undergo acetylation in DBT1. In contrast, the urine from DBT1-treated rats strongly resembled that of DBT2-treated rats. These metabolites originate mainly from formononetin, calycosin and their related glycosides, and they were formed mainly by the metabolic process of reduction, deglycosylation, demethylation, hydrogenation and sulfation. The HPLC-DAD-ESI-IT-TOF-MSn method was successfully applied for the rapid chemical profiles evaluation of two DBTs and their related urine samples.

The plants RA and RH belong to the same botanical family but different genus, and have long been widely used as the same crude herb in DBT [1]. This is always a question of whether RH can replace RA in the DBT decoction. Chemically, RA-containing DBT showed higher amounts of calycosin-7-Oβ-D-glucoside, ferulic acid, ononin, calycosin, astragaloside IV, astragaloside III, and Z-ligustilide. Only formononetin was higher in RH-containing DBT. In parallel, the estrogenic, osteogenic and erythropoetic effects of RA-containing DBT1 showed better activities than that of RA-containing DBT2 [1]. So far, the chemical differences between DBT1 and DBT2 has not been investigated. Therefore, we designed a systematic comparison of the chemical ingredients of DBT1 and DBT2.
Two experiments were designed, including thorough elucidation of the chemical profiles of DBT1 and DBT2 crude extracts and illumination of the metabolites of DBT1 and DBT2 after being administrated to rats. The chemical profiles of the two DBTs were compared by determining the total ion chromatogram(TIC) and the multistage mass spectra (MS n ) from HPLC-DAD-ESI-IT-TOF-MS n . Subsequently, DBT1 and DBT2 were identified and compared on the basis of the TIC and the MS n [11]. The results will be provide a solid evidence to understand the chemical profiles of the two different versions of DBT.

Optimization the Conditions of HPLC and Mass Spectrometry
In order to obtain desirable HPLC and mass spectrometry chromatograms, the procedures for preparation of the urine samples and crude extracted samples of the two DBTs were optimized in terms of the extraction solvents and extraction times. Methanol and acetonitrile were initially selected as the extraction solvents, but methanol is less poisonous and produced almost the same chromatograms as acetonitrile, so it was applied as the final extraction solvent. For comparison, different columns (Phenomenex RP C 18 , Agilent RP C 18 ) were tested for sample separation, and Phenomenex RP C 18 gave the best chromatographic resolution. The column was eluted with a gradient mobile phase that consisted of water-formic acid (100:0.1, v/v) (A), acetonitrile (B) and at a flow rate of 1.0000 mL/min, in addition, 0.1% (v/v) formic acid was added to improve the mass spectrometry ionization efficiency and enable symmetric peak shapes [12]. Both the positive ion (PI) and negative ion(NI) modes were tested for the experiment. Since MS and MS n fragmentations gave more information about the isoflavones in PI mode but about saponins in NI mode, the analysis was simultaneously conducted in both PI and NI mode.

The Identification and Analysis of 19 Reference Compounds
Nineteen reference compounds which might represent the major structural types of the DBTs were analyzed. The characteristic fragment ions of 19 references are very useful for determining the structural skeleton and the substitution patterns of those related compounds in two DBTs. Their high resolution mass spectra (HRMS) data are summarized (see Table 1 and Supplementary: pages S2-S8). The base peak chromatograms (BPCs) detected in NI and PI mode were recorded (see Figure 1a). Furthermore, the structure of 19 reference compounds were shown ( Figure 1b) [11,13].
According to our preliminary research, there three main types of isoflavonoids are found in Radix Astragali which were named as a: isoflavone (T1, T2, T5, T6, and T8), b: isoflavan (T4, T7, and T10) and c: pterocarpan (T3, and T9) [11,[13][14][15]. To facilitate the structural identification of the isoflavonoids in the DBTs, the fragmentation behaviors of the three types of isoflavonoids were analyzed, which might represent the major structural types. We first studied the MS n fragmentation behaviors in PI and NI mode, and found that the fragmentation behaviors in PI mode could give more information about the structure than in NI modes. Then we elucidated the structure of the three types of isoflavonoids mostly from the PI mass spectra.
For example, we identified the characteristic malonate-glucose-, acetyl-glucose-and glucose-binding ingredients with a neutral loss of 248 Da, 204 Da, and 162 Da, otherwise, glucuronide metabolites with a neutral loss of (−176 Da) and sulfated metabolites with a neutral loss of (−80 Da) from the molecular ion peaka in the MS 2 spectra [13,16,17].   The HRMS data of these identified compounds are summarized (see Table 2 and Supplementary: pages S9-S35). The BPCs detected in PI (1BPC) and NI (4BPC) mode were also recorded ( Figure 3a) [17][18][19].
Using the [M+H] + , [M-H] − or [M+Na] + data of C7, C8, C10, C12, C13, C14, C21, C25, C26, and C28, we predicted their formulas, which indicates that they are the ingredients of the samples. However, at this point, their exact structures could not be identified.     The HRMS data of these identified compounds are summarized (see Table 2 and Supplementary: pages S36-S51). The BPCs detected in NI and PI modes were recorded (see Figure 3b).
From the analysis based on the comparison of TIC and MS n : 69 compounds (C1-C69) were identified from the crude extracts of DBT1, 46 compounds (c1-c46) were identified from the crude extracts of DBT2. The isoflavonoids glycosides had experienced acetylation (seven compounds, C18, C32, C34, C40, C44, C46, and C47), formed the malonate acid esters (five compounds, C31, C35, C42, C43, and C51) and with special astragalosides (six compounds, C49, C52, C53, C54, C60, and C64) in DBT1. Thus, the number of identified components in DBT1 was significantly more than in DBT2 (the chemical structural diversity of isoflavonoids which were detected in DBT1 more than in DBT2 are shown in Figure 4). Among these, the 24 common chemical constituents accounted for approximately 27% to the total 91 identified compounds. However, and the proportion of the total isoflavonoids and saponins to the total identified ingredients accounted for nearly 62% (see Table 3).

Profiling and Identifying Chemical Profiles of the Urine Samples after Administration of the DBT1 and DBT2 Performed by HPLC-DAD-ESI-IT-TOF-MS n
In the study of the existing literature, ingredients such as isoflavones, saponins in the two DBTs had no obvious differences in chemical profiles between the serum and bile samples collected from enterohepatic circulation. In addition, they had a lower concentration in serum samples, even when giving at dosages of 60-120 g/kg (w/w) several times to rats within 24 h [13,17,20]. Thus, this approach is not conducive to tracing these minor components. This study chose the normal usage of 10 g/kg (w/w) by comparing the urine samples of rats that were administrated two different DBTs, so as to improve the detection through enrichment of the treatments.

Identification of the Chemical Profiles of Urine Sample after Administration of the DBT1 by HPLC-DAD-ESI-IT-TOF-MS n
The HRMS data of these identified metabolites are summarized (see Table 4 and Supplementary: pages S52-S66). The BPCs detected in NI mode were recorded (see Figure 3c). In addition, the main proposed structures of these metabolites identified from the urine samples of rats that had been administrated DBT1 were showed (see Figure 5).
Using MS data with the reference compounds, M15 was identified as calycosin, and M23 was identified as formononetin [11,13].
When predicting their formulas, M39, M41, and M42 were identified as related metabolites of saponins [13,17].    The HRMS data of these identified metabolites are summarized (see Table 4 and Supplementary: pages S67-S77). The BPCs detected in NI model were recorded (see in Figure 3d). The main proposed structures of metabolites identified from the urine samples of rats that had been administrated DBT2 were showed (see Figure 6).
The MS data of m13, and m19 show that they are the reference compounds. In addition, m13 was identified as calycosin, and m19 was identified as formononetin [11,13].
In this part of the experiment, decoctions of DBT1 and DBT2 were administered to rats, and an analysis was conducted on the rats' urine for metabolites: 44 compounds (M1-M44) were identified from the urine samples after DBT1 was administrated, 34 compounds (m1-m34) were identified from the urine samples after DBT2 was administrated. The number of the chemical components in the urine samples from DBT1 was administrated to rats are slightly more than these of DBT2. The proportion of the 19 common constituents to the total 59 identified compounds accounted for approximately 33%. The proportion of 48 total isoflavonoids and saponins to the total identified compounds, however, reached approximately 82% (See Table 3).
The phase Ⅱ metabolites from the urine samples of DBT1 and DBT2 were dominantly sulfated products, with rare or no glucuronide metabolites. This result still requires further research as the literature reports the chief presence of mainly glucuronide metabolites [13][14][15][16]. Metabolites that originated from RAS are relatively rare or not detected. This is likely due to the low proportion of RAS in DBT and even the low dosage that was given to rats in this study, those chemical constituents of RAS being easy to volatilize, or a loss when sampling was conducted by concentrated processes [22,23]. At this normal dosage of DBT and due to poor absorption, the content levels of astragalosides are much lower. In addition, saponins and their metabolites that originated from astragalosides are rarely detected [24][25][26][27].
The HPLC-DAD-ESI-IT-TOF-MS n method adopted in our research was confirmed to be a powerful method to evaluate the chemical profiles of the crude extracts and the related urine samples. As we know, the chemical composition found in a Chinese herbal decoction is rather complicated. In this study, the chemical profile analysis of DBT1 and DBT2 was conducted, which provided a comprehensive understanding of that those isoflavonoids that play an important role in the main common chemical basis when they are used in clinical practice. Some identified metabolites are known to have many bioactivities, such as calycosin, formononetin, daidzein and equol (which are well-known phytoestrogens), and most of them displayed many beneficial effects to humans [28][29][30].
Through the comparison of chemical profiles of two DBTs at our used normal dosage, the similarity of urine samples is higher than that of the crude extract samples. This leads us to believe that the main chemical basis of the chemical constituents is almost the same. Whether the DBT2 of RH:RAS can replace the DBT1 of RA:RAS, we need to further investigate different ratios of RH:RAS with the RA: RAS (5:1) when using equivalent pharmacological research.

Sample Preparation
In clinical use, DBT is typically boiled with water twice, then the two decoctions are combined and applied [1]. Therefore, in our study, DBT1, consisting of RA 100 g and RAS 20 g was boiled in 1,000 mL of water (w:v) for 45 min, and then the decoction was filtered. The residue was again boiled in 700 mL of water (w:v) for 30 min. The two decoctions were evaporated to dryness under reduced pressure at 50 °C to 100 mL volume. Samples of DBT2, consisting of RH 100 g and RAS 20 g, were prepared in the same way.

Data Analysis
All data were recorded and processed by Shimadzu software LCMS solution version 3.60, Formula Predictor version 1.2 and Accurate Mass Calculator (Shimadzu).

Conclusions
A comparison was conducted on the similarities and differences of crude extracts and urine samples of DBT1 and DBT2. The chemical profiles of the crude extracts comprised a total of 115 proposed chemical components. There were 24 common ingredients, which was accounted for 27% in the total 91 identified components. There were a total of 56 isoflavonoids and saponins identified, which accounted for nearly 62% in the total identified components. Since isoflavonoid glycosides had acetylation (C18, C32, C34, C40, C44, C46, and C47), the formation of malonate acid esters (C31, C35, C43, C42, and C51) and special astragalosides (C49, C52, C53, C54, C60, and C64) in DBT1, the identified compounds from DBT1 were significantly greater than DBT2. Of these, C18, C34, C35, C42, C47, and C51 were identified from DBT for the first time.
In total, 78 proposed chemical components in the urine samples of rats that had been administrated DBT1 and DBT2, respectively, were found. These included 19 common ingredients, which accounted for approximately 33% in the total identified constituents. In addition, 48 of total isoflavonoids and saponins were found, which accounted for nearly 82% of the total 59 identified components. The differences between those metabolites in the urine samples were revealed to be less than the crude extracts. These identified metabolites are mainly originated from formononetin, calycosin and their related glycosides, which are formed mainly through the metabolic processes of reduction, deglycosylation, demethylation, hydrogenation and sulfation. Through the comparison of chemical profiles of two DBTs at our used doses, the similarity of urine samples is higher than that of the crude extract samples, one can think the main chemical constituents are almost the same as when administrated to rats.
The HPLC-DAD-ESI-IT-TOF-MS n method was successfully applied for the chemical profile comparison of two different DBTs and its related medicinal materials. The proposed assay provides an important reference and can be a suitable method for the rapid and accurate chemical basis evaluation of TCM or their related prescriptions.