Two Standards‐Aided LC‐MS Technologies Uncover 30 “Unreported” Compounds in Natural Bear Bile Powder to Add a New Pharmacopoeia Quality‐Marker

ABSTRACT

Natural bear bile powder is a precious traditional Chinese medicine that has a long history of use in clinical practice. The current study analyzed four batches of natural bear bile powder samples (B‐1∼B‐4) using standards‐aided ultra‐high‐performance liquid chromatography‐quadrupole‐Orbitrap‐mass spectrometry and ultra‐performance liquid chromatography coupled with electrospray‐ionization quadrupole time‐of‐flight mass spectrometry. A total of 37 compounds (1‐37) were accurately identified, while 30 “unreported” compounds were found in natural bear bile powder for the first time. In particular, two “unreported” isomers (2'‐hydroxydaidzein and genistein) were strictly distinguished by means of ΔR.T. matching in chromatographic stick‐peak and MS/MS peak matching. The quantification work indicated that taurochenodeoxycholic acid (TCDCA) and its epimer tauroursodeoxycholic acid (TUDCA) had the highest contents in natural bear bile powder. However, TCDCA showed different binding energy towards pregnane X receptor from TUDCA in the molecular docking simulation. Based on the analytical chemistry, molecular docking simulation, and documentary evidence, TCDCA was recommended as a potential Q‐marker for pharmacopoeia. The new Q‐marker was validated to be feasible by analysis of 11 batches of natural bear bile powder samples (B‐5∼B‐15), and the specific ratio of TCDCA/TUDCA (0.58∼0.98) can distinguish natural BBP from artificial BBP and other animal biles. All these findings not only lay a foundation for manufacturing technique improvement of artificial natural bear bile powder, but also promote pharmacopoeia to update the quality‐control method.

Abbreviations

BBP

bear bile powder

HPLC‐CAD

high performance liquid chromatography—charged aerosol detector

HPLC‐ELSD

high performance liquid chromatography—evaporative light scattering detector

LC‐MS

liquid chromatography—mass spectrometry

MS

mass spectra

PXR

pregnane X receptor

Q‐marker

quality‐marker

R.T.

retention time

SIM

selected ion monitoring

Standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS

liquid chromatography coupled with electrospray‐ionization quadrupole time‐of‐flight mass spectrometry

Standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS

standards‐aided ultra‐high‐performance liquid chromatography quadrupole‐Orbitrap‐mass spectrometry

TCDCA

taurochenodeoxycholic acid

TCM

traditional Chinese medicine

TIC

total ion current

TUDCA

tauroursodeoxycholic acid

1. Introduction

Bear bile powder (BBP) is a precious traditional Chinese medicine (TCM) and has been used in TCM clinical practice for over 13 centuries [1]. The powder is prepared from bile of Selenaretos thibetanus Cuvier through surgical drainage; conventionally, this kind of product is termed natural BBP or drained BBP. Surgical drainage, however, is considered to violate animal welfare, and thus other biomimetic materials have also been developed to prepare artificial BBP [1, 2]. However, artificial BBP is still challenged in pharmacology and even chemistry at present.

The pharmacological exploration showed that natural BBP possessed various traditional functions or beneficial effects in enterohepatic system [3], including hepatoprotection [4], anti‐hepatocellular carcinoma [5, 6], and anti‐inflammation [7]. These traditional functions or pharmacological effects, of course, are associated with some specific signaling pathways, for example, pregnane X receptor (PXR, Nr1i2). From the perspective of chemistry, natural BBP mainly comprises cholic acid (bile acid) derivatives. Particularly, tauroursodeoxycholic acid (TUDCA) and taurochenodeoxycholic acid (TCDCA) are regarded as two main bioactive compounds [6, 8]. However, pharmacopoeia excludes TCDCA and only defines TUDCA as the Q‐marker of natural BBP. This implies that its pharmacopoeia Q‐marker can be further optimized based on the in‐depth chemical analysis.

Therefore, the current study tries to employ a state‐of‐the‐art LC‐MS technology, that is, standards‐aided ultra‐high‐performance liquid chromatography quadrupole‐Orbitrap‐mass spectrometry (standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS), to systematically investigate the chemistry of natural BBP. The standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS technology has successfully been used for TCM analysis and could provide reliable results [9]. This is because the UHPLC‐Q‐Orbitrap‐MS/MS apparatus itself possesses the merits of high resolution and high efficiency. With the aid of authentic standards, a set of parameters can be obtained, including definite molecular ion peak, m/z value of MS/MS fragment, and retention time (R.T.) value. Through matching these parameters with standards, some “unreported” compounds would also be found in the current study based on this method [10]. The so‐called “unreported” compound means one known compound that has not been reported to occur in natural BBP previously, after literature review was performed in Google Scholar, Web of Science, PubMed, and the China National Knowledge Infrastructure (CNKI).

On this basis, the current study further adopts standards‐aided ultra‐performance liquid chromatography coupled with electrospray‐ionization quadrupole time‐of‐flight mass spectrometry (standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS) technology [11] to fulfill the quantitative analysis and validation experiments based on different batches of BBP samples. All of these would not only provide chemical insight into BBP, but also facilitate the manufacturing technique improvement of artificial BBP. Of course, it will also help pharmacopoeia to update its old Q‐marker.

2. Materials and Methods

2.1. BBP Samples

Fifteen batches of natural BBP samples (B‐1∼B‐15) were collected for the study and are now stored in our laboratory. All natural BBP and artificial BBP samples were purchased from four different pharmaceutical companies that were legally registered and approved by Chinese government. The samples were stored in a dry, sealed condition at 4°C in our laboratory. The main information and intention were detailed in Supporting Information S1 Table S1.

2.2. Authentic Standards and Chemicals

Trigonelline (Cas. 535‐83‐1, C₇H₇NO₂, M.W. 137.138, 98%), L‐carnitine (Cas. 541‐15‐1, C₇H₁₅NO₃, M.W. 161.201, 98%), nicotinamide (Cas. 98‐92‐0, C₆H₆N₂O, M.W. 122.127, 98%), hypaphorine (Cas. 487‐58‐1, C₁₄H₁₈N₂O₂, M.W. 246.305, 98%), apigenin 7‐O‐glucuronide (Cas. 29741‐09‐1, C₂₁H₁₈O₁₁, M.W. 446.364, 98%), paprazine (Cas. 36417‐86‐4, C₁₇H₁₇NO₃, M.W. 283.322, 98%), parthenolide (Cas. 20554‐84‐1, C₁₅H₂₀O₃, M.W. 248.322, 98%), pterosin B (Cas. 34175‐96‐7, C₁₄H₁₈O₂, M.W. 218.296,98%), isoalantolactone (Cas. 470‐17‐7, C₁₅H₂₀O₂, M.W. 232.323,98%), N‐trans‐feruloyltyramine (Cas. 66648‐43‐9, C₁₈H₁₉NO₄, M.W. 313.348, 98%), 5‐hydroxymethylfurfural (Cas. 67‐47‐0, C₆H₆O₃, M.W. 126.111, 97%), indole‐3‐acetic acid (Cas. 87‐51‐4, C₁₀H₉NO₂, M.W. 175.187, 97%), TUDCA (Cas. 14605‐22‐2, C₂₆H₄₅NO₆S, M.W. 499.707, 97%), 3‐epi‐bufalin (Cas. 465‐20‐3, C₂₄H₃₄O₄, M.W.386.532, 97%), ursodeoxycholic acid (Cas. 128‐13‐2, C₂₄H₄₀O₄, M.W. 392.572, 97%), cholic acid (Cas. 81‐25‐4, C₂₄H₄₀O₅, M.W. 408.579, 97%), hyodeoxycholic acid (Cas. 83‐49‐8, C₂₄H₄₀O₄, M.W. 392.572, 97%), echinocystic acid (Cas. 510‐30‐5, C₃₀H₄₈O₄, M.W. 472.716, 97%), pedunculoside (Cas. 42719‐32‐4, C₃₆H₅₈O₁₀, M.W. 650.840, 97%), and betulin (Cas. 473‐98‐3, C₃₀H₅₀O₂, M.W. 442.728, 97%) were obtained from Herbest Biotech Co. Ltd (Baoji, China). 5,7‐Dihydroxychromone (Cas. 31721‐94‐5, C₉H₆O₄, M.W. 178.141, 97%) was obtained from BioBioPha Co. Ltd. (Kunming, China). 4‐Hydroxycinnamic acid (Cas. 501‐98‐4, C₉H₈O₃, M.W. 164.158, 97%), 7‐hydroxycoumarin (Cas. 93‐35‐6, C₉H₆O₃, M.W. 162.142, 97%), daidzin (Cas. 552‐66‐9, C₂₁H₂₀O₉, M.W. 416.378, 97%), 6‐gingerol (Cas. 23513‐14‐6, C₁₇H₂₆O₄, M.W.294.386, 97%), and 4‐isopropyltoluene (Cas. 99‐87‐6, C₁₀H₁₄, M.W. 134.218, 97%) were obtained from Alfa Biotechnology Co. Ltd. (Chengdu, China). Eicosapentaenoic acid (Cas. 10417‐94‐4, C₂₀H₃₀O₂, M.W. 302.451, 97%) and docosapentaenoic acid (Cas. 24880‐45‐3, C₂₂H₃₄O₂, M.W. 330.504, 97%) were obtained from Sopo Biological Technology Co. LTD (Guangzhou, China). Proline (Cas. 147‐85‐3, C₅H₉NO₂, M.W. 115.132, 97%), sucrose (Cas. 57‐50‐1, C₁₂H₂₂O₁₁, M.W. 342.297, 98%), and phenylalanine (Cas. 63‐91‐2, C₉H₁₁NO₂, M.W. 165.189, 97%) were purchased from J&K Scientific Co. Ltd. (Beijing, China). Protocatechuic acid (Cas. 99‐50‐3, C₇H₆O₄, M.W. 154.120, 97%) was purchased from Sichuan Weikeqi Biological Technology Co. Ltd. (Chengdu, China). Riboflavin (Cas. 83‐88‐5, C₁₇H₂₀N₄O₆, M.W. 376.364, 97%) and 3‐oxo‐4‐cholestene (Cas. 601‐57‐0, C₂₇H₄₄O, M.W. 384.638, 97%) were purchased from Sigma‐Aldrich (Shanghai, China). Genistein (Cas. 446‐72‐0, C₁₅H₁₀O₅, M.W. 270.237, 97%) was purchased from TCI Chemical Co. (Shanghai, China). TCDCA (Cas. 516‐35‐8, C₂₆H₄₅NO₆S, M.W. 499.707, 97%) was obtained from Acmec Biochemical Technology Co. Ltd (Shanghai, China). Methanol and water were of MS purity grade; Formic acid and ammonium acetate were purchased as analytical grade (A.R.) from the Guangzhou Chemical Reagent Factory (Guangzhou, China).

2.3. Preparation of Authentic Standard Solution and Sample Solution

2.3.1. Preparation of Authentic Standard Solution

All authentic standards listed in Section 2.2 were dissolved in methanol at 40 µg/mL concentration, respectively. The solutions were individually filtered through 0.22 µm nylon membrane. The filtrates were kept at 2°C–6°C for further analysis.

2.3.2. Preparation of Sample Solution

Natural BBP sample B‐1 (40 mg) was accurately weighed and then dissolved using 800 µL of 80% methanol (v/v) under the condition of ultrasonic heating (37°C, 70 kHz) for 20 min using HNY‐300L ultrasonic cleaning instrument (Huanan Ultrasonic Equipment Co. Ltd, Guangzhou, China). The solution was then filtered through a 0.22 µm nylon membrane to prepare sample solutions at different concentrations (0.5, 2.5, 10, and 50 mg/mL). The above preparation protocols were repeated using B‐2∼B‐4 sample solutions at different concentrations (0.5, 2.5, 10, and 50 mg/mL) and B‐5∼B‐15 sample solution at 2.5 mg/mL concentration.

2.4. Standards‐Aided UHPLC‐Q‐Orbitrap‐MS/MS Qualitative Analysis of B‐1

The analysis of BBP sample (B‐1) was carried out using standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS technology. In brief, a sample solution (50 mg/mL) was separated on an Accucore RP‐MS LC C₁₈ column (100 mm × 2.1 mm, 2.6 µm, Thermo Fisher) employing a binary mobile phase. The binary mobile phase consisted of A and B. Phase A contained water with 0.1% formic acid (v/v), and Phase B was 100% methanol. The binary phase run at a flow rate of 0.4 mL/min and was programmed as follows: from 0.0 to 5.0 min, 10% B; from 5.0 to 14.5 min, a linear increase from 10% B to 100% B; from 14.5 to 16.0 min, maintained at 100% B; and from 16.0 to 20.0 min, returning to 10% B. The column chamber temperature was at 40°C, while the sample tray was kept at 4°C. The MS parameters were optimized as follows: spray voltage set to 4.5 kV in negative mode; sheath gas (N₂) flow rate adjusted to 40 arbitrary units; auxiliary gas (N₂) flow rate set to 10 arbitrary units; capillary temperature was at 450°C. Resolution for the full MS scan was established at 70 000 full width at half maximum (FWHM), and the MS/MS scan was at 17 500 FWHM. AGC target was configured at 2 × 10⁵; stepped normalized collision energies was set at values of 20, 50, and 90. The trapping range of mass spectrum varied from m/z 100 to 1500.

The injection, data acquisition, and analysis were conducted using the TraceFinder 4.1 software package, which was integrated into the UHPLC‐Q‐Orbitrap‐MS system. Prior to data acquisition, background signals were subtracted by a blank solvent. The acquired data were then exported to Xcalibur 4.1 (one of the software in TraceFinder 4.1) for m/z extraction, facilitating the generation of corresponding mass spectra. The processing parameters were defined as follows: mass range set between 100 and 1500 Da; mass tolerance of 5 ppm; signal‐to‐noise (S/N) threshold established at 5:1; and isotopic pattern fit threshold set at 90%.

The qualitative analysis (i.e., accurate identification) of compound was achieved through a manual comparison with authentic standards based on four key parameters: R.T., molecular ion peak, MS/MS profile, and characteristic fragments, according to our previous method [9].

2.5. Standards‐Aided UHPLC‐ESI‐Q‐TOF‐MS/MS Quantitative Analysis of all 15 Batches (B‐1∼B‐15)

The standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS analysis was based on our previous method [12]. In brief, the samples (B‐1∼B‐15) were dissolved in methanol at an appropriate concentration and then filtered using a 0.22 µm filter. The filtrates were injected into UHPLC‐ESI‐Q‐TOF‐MS/MS system equipped with a C₁₈ column (2.1 mm inner diameter × 100 mm, 1.6 µm, Phenomenex, Torrance, CA, USA). The column was eluted at a flow rate of 0.2 mL/min with the following gradient elution program: 0–2 min, maintained at 30% B; 2–10 min, 30%–0% B; and 10–12 min, 0%–30% B. The mobile phase consisted of acetonitrile (Phase A) and 0.1% formic acid in water (Phase B). The injection volume was 3 µL.

The Q‐TOF‐MS/MS was performed on a Triple TOF 5600 ^plus mass spectrometer (AB SCIEX, Framingham, MA, U.S.A.) equipped with an ESI source. The scan range was m/z 100∼2000. The following parameter settings were used: ion spray voltage, +4500 V; ion source heater, 550°C; curtain gas (N₂), 30 psi; nebulizing gas (GS1, air), 50 psi; and TIS gas (GS₂, air), 50 psi. The declustering potential (DP) was +100 V, and the collision energy (CE) was +40 V with a collision energy spread (CES) of 20 V.

Prior to data acquisition, underground signals were subtracted by utilizing a blank solvent. To validate the quantitative method, seven parameters were analyzed before quantitative analysis, including linear regression equation, correlation coefficient (R value), linearity range, and lower limit of quantification (LLOQ), intra‐day accuracy, inter‐day accuracy, and recovery rate. The calculations of intra‐ accuracy and inter‐day accuracy were performed using analyzing B‐1 sample solution within the same day and over two consecutive days, respectively. The recovery test was also performed in the study based on our spiking test [13], to characterize the matrix effect. Four batches of samples (B‐1∼B‐4) were quantitatively analyzed for all 37 compounds (1‐37), while another 11 batches of samples (B‐5∼B‐15) were quantitatively analyzed only for TCDCA (20) and TUDCA (22).

2.6. Molecular Docking Simulation

The molecular docking simulation was based on the previous method [14]. The chemical structures of TCDCA and TUDCA were obtained from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/) and further optimized using Gaussian 16 until there was no imaginary frequency in Linux system. The basis set was (U)B3LYP‐D3(BJ)/6–31 + G(d,p) [15]. The optimized structures were viewed and exported via Gaussian View 6.1.1. The crystal structure of Nr1i2 protein (PXR) was obtained from the RCSB Protein Database (https://rcsb.org/) and processed using PyMol software. Molecular docking was simulated using CB‐DOCK2 (https://cadd.labshare.cn/cb‐dock2/php/index.php). Finally, visualization of the docking results and binding modes was performed in PLIP (https://plip‐tool.biotec.tu‐dresden.de/plip‐web/plip/index) and PyMOL software.

2.7. Statistical Analysis

The data of contents were presented as the mean value of triplicate measurements of same samples (n = 3). The content evaluation was based on a linear regression obtained with the corresponding authentic standards. The linear regression was analyzed using Origin 6.0 Professional software (Origin‐Lab Corp., Northampton, MA, USA).

3. Results and Discussion

3.1. Standards‐Aided UHPLC‐Q‐Orbitrap‐MS/MS Qualitative Analysis of B‐1 Uncovers 30 “Unreported” Compounds

The standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS analysis was capable of accurate identification of various natural compounds, as described in our previous study [16]. The accurate identification relied on the matching of MS/MS fragmention between authentic standard and sample compound. Usually, LC‐MS analysis without authentic standard may cause inaccurate identification (or tentative identification), due to the lack of comparable evidence.

The accurate identification has also distinguished a pair of C₁₅H₁₀O₅ isomers, 2'‐hydroxydaidzein (16) and genistein (21), in the study. The distinction relied on ΔR.T. value matching and MS/MS peak matching, according to our method [15]. Two standard isomers (16 and 21) were preliminarily injected into the apparatus for LC‐MS analysis, to obtain their ΔR.T. value and MS/MS peaks. The ΔR.T. value of two standard isomers was calculated as −0.59 min (7.80–8.39 min, Supporting Information S2 Figures S2.16 and S2.21). Thereafter, the tested‐sample solution (i.e., extract solution) was also injected into the apparatus under the same conditions. The ΔR.T. value was calculated to be −0.55 min (7.82–8.37 min, Supporting Information S2 Figures S2.16 and S2.21). The ΔR.T. value matching has preliminarily indicated that the chromatographic stick‐peak at 7.82 min was 2'‐hydroxydaidzein (16) and that at 8.37 min was genistein (21) (Figure 1A).

FIGURE 1.

The distinction of two isomers, 2'‐hydroxydaidzein (16) and genistein (21). (A) The chromatographic stick‐peaks of two isomers using m/z 269.0450 extraction under negative mode (7.5 µg BBP sample). (B) MS/MS fragmention pathways of 2'‐hydroxydaidzein (16). (C) MS/MS fragmention pathways of genistein (21). Their original data were detailed in Supporting Information S2 Figures S2.16 and S2.21.

The reason why the study adopted stick‐peak rather than the ordinary chromatography peak was that the stick‐peak could sharply distinguish the R.T. values. The reason why the study introduced ΔR.T. value rather than R.T. value, could be attributed to the fact that the R.T. values was affected by the concentration. This is because the concentration of tested‐sample solution was usually much higher than that of standard solution. For example, in the study, the tested‐sample solution was at 2.5∼50 mg/mL (Section 2.3.2), while the standard compound concentration was 40 µg/mL (Section 2.3.1). Thus, 2'‐hydroxydaidzein (16) ran more slowly in tested‐sample solution for the crow in chromatographic column, compared with the standard solution. However, the ΔR.T. value could not affected by the concentration factor.

Of course, further identification depended on MS/MS peak matching. As seen in Supporting Information S2 Figures S2.16 and S2.21, the sample compound 16 showed identical MS/MS peaks with the corresponding standard, while the sample compound 21 presented the similar MS/MS peaks to the corresponding standard. However, there were great differences between the two (16 and 21). The former gave rise to three specific peaks (m/z 241, 225, and 181, Figure 1B) while the latter did not (Figure 1C). This has strictly distinguished two C₁₅H₁₀O₅ isomers (16 and 21).

Coincidentally, two isomeric compounds have not been found in the natural BBP previously. Accordingly, they were described as “unreported” compounds in the current study. As shown in Figures 2 and 3, a total of 37 compounds (1‐37) have been detected from natural BBP (B‐1); while 30 “unreported” compounds were found in natural BBP for the first time. The identification confidence of all these compounds was classified into Level 1, according to Sumner's et al., Zhong's et al., and Fock's et al. definitions [17, 18, 19]. This is because that (1) the deviation of molecular ion between experimental m/z value and theoretical m/z value were at low level (δ = 0.0–4.6, Supporting Information S2 Table S2); (2) the diagnostic MS/MS peak and MS/MS profile of sample were highly similar to those of standard (Supporting Information S2 Figures S2.1–S2.37); (3) The ΔR.T. values between sample and the corresponding standard were also low (0∼0.08 min). For example, the ΔR.T. value of carnitine was 0.01 min (0.60–0.59 min), while that of sucrose was also 0.01 min (0.63–0.62 min) (Supporting Information S2 Figures S2.1 and S2.2). For the convenience of discussion, the main results and documentary evidence of 37 compounds (1‐37) were detailed in Supporting Information S2 Table S2; especially, the identification of TUDCA (22) and its C‐7 epimer TCDCA (20) was also shown in Figure 4.

FIGURE 2.

The structures of 37 compounds (1‐37) in natural BBP products (B‐1∼B‐15). Superscript U means “unreported” compound that has not been reported to occur in natural BBP previously. The purple knot connects two isomers.

FIGURE 3.

The collection of total ion chromatograms (TICs) of natural BBP samples  (B‐1 ~ B‐4, 150 µg). (A) Under negative mode; (B) under positive mode. The red font indicates “unreported” compound.

FIGURE 4.

The identification of TCDCA (20) and TUDCA (22) in B‐1. (A) The chromatographic peaks of TCDCA (20) and TUDCA (22) using m/z 498.2894 extraction under negative mode (7.5 µg BBP sample). (B) MS/MS fragmention pathways of TCDCA (20) and TUDCA (22). The original data for 20 and 22 were detailed in Supporting Information S2.

3.2. Standards‐Aided UHPLC‐ESI‐Q‐TOF‐MS/MS Quantitative Analysis of B‐1∼B‐4

As shown in Section 2.2, the current study collected 37 authentic standards (1‐37). All of them were then analyzed by UHPLC‐ESI‐Q‐TOF‐MS/MS, to establish their calibration curves between concentration and peak area. To optimize the peak area, some peaks were determined by manual integration. As seen in Supporting Information S3 Table S3, the established calibration curves showed excellent linear relationship (the correlation coefficient R = 0.9912∼0.9999).

Herein the study used B‐1 to validate the method. As in Supporting Information S3 Table S3, the relative standard deviation (RSD) values of contents of 37 compounds (1‐37) were distributed in the range of 0.26%∼4.52 %. The intra‐day accuracy and inter‐day accuracy were calculated as 91.40%∼118.64% and 98.36%∼104.41%, respectively. LLOQ values varied from 0.3 to 12.3 ng/mL. Under the SIM model, these sample compounds were observed to exhibit separated peaks and thus to show good selectivity (Supporting Information S3 Table S3). All these parameters have validated that our quantification method were scientifically sound and the content results were reliable.

As seen in Table 1, the contents of 37 compounds (1‐37) were in a wide range, from 12.0 ± 0.37 ppm to 44.60 ± 0.20%. Regardless of different sources, different manufacturers, and even different pharmaceutical companies, the distribution of 37 compounds (1‐37) was generally similar to each other in 4 batches of natural BBP (B‐1∼B‐4). In particular, two cholic acid epimers (20 and 22) always exhibited the highest contents in four natural BBPs (B‐1∼B‐4). In the current study, the contents of TCDCA (20) were calculated as 25.02 ± 0.41%∼36.57 ± 0.28% in B‐1∼B‐4. The results were generally consistent with the Qiao's et al. results (25.44%∼47.19%) [20], Shi's et al. results (16.6%∼48.1%) [21], Yuan's et al. results (31%∼41%) [22], Lei's results (12.99%∼43.50%) [6], and Yin's results (20.42%∼32.11%) [23]. The contents of TUDCA (22) in B‐1∼B‐4 however varied from 32.77 ± 5.55 to 44.60 ± 2.04%. These results were roughly paralleled by Qiao's et al. results (25.44%∼47.19%) [20], Shi's et al. results (18.8%∼44.8%) [21], Yin's et al. results (30.13 %∼38.37 %) [23], and Lei's et al. results (12.99%∼43.50%) [6].

TABLE 1.

The contents of 37 compounds (1–37) in 4 batches of bear bile powder (B‐1∼B‐4) determined by standards‐aided UHPLC‐Q‐TOF‐MS/MS (µg/g).

No	Name	B‐1	B‐2	B‐3	B‐4
1	Carnitine	14.2 ± 0.07	5.1 ± 0.06	35.0 ± 0.2	4.5 ± 0.08
2	Sucrose a	876.3 ± 12.1	61.2 ± 0.5	33 ± 0.3	880.2 ± 9.3
3	Trigonelline	29.8 ± 0.3	21.2 ± 0.9	24.5 ± 0.1	29.2 ± 1.3
4	Proline	3.3 ± 0.07	3.4 ± 0.02	3.0 ± 0.03	1.5 ± 0.05
5	Nicotinamide	5.0 ± 0.1	4.8 ± 0.02	6.1 ± 0.04	1.8 ± 0.05
6	5‐Hydroxymethylfurfural	6.2 ± 0.02	6.4 ± 0.03	4.4 ± 0.04	1.3 ± 0.02
7	Phenylalanine	34.3 ± 0.8	17.4 ± 0.04	4.4 ± 0.03	8.7 ± 0.2
8	Protocatechuic acid	12.3 ± 0.3	37.2 ± 1.7	7.4 ± 0.1	6.9 ± 0.08
9	Hypaphorine	2.2 ± 0.006	0.6 ± 0.002	0.6 ± 0.009	0.1 ± 0.003
10	Riboflavin	169.8 ± 2.5	202.7 ± 0.9	650.5 ± 4.7	86.2 ± 1.6
11	Daidzin	165.4 ± 2.2	288.4 ± 2.6	91.4 ± 2.5	65.2 ± 0.9
12	7‐Hydroxycoumarin	0.3 ± 0.01	0.3 ± 0.01	1.3 ± 0.06	0.2 ± 0.003
13	4‐Hydroxycinnamic acid	7.0 ± 0.07	3.4 ± 0.03	2.3 ± 0.1	8.1 ± 0.1
14	Indole‐3‐acetic acid	63.6 ± 1.6	68.4 ± 0.6	18.1 ± 0.2	13.3 ± 0.6
15	5,7‐Dihydroxychromone	1.2 ± 0.006	1.7 ± 0.03	0.2 ± 0.006	1.4 ± 0.02
16	2'‐Hydroxydaidzein	606.6 ± 16.1	729.2 ± 5.4	278.6 ± 1.9	60.1 ± 1.9
17	Apigenin 7‐O‐glucuronide	348.5 ± 5.3	70.2 ± 1.7	808.4 ± 3.6	170.2 ± 1.6
18	Paprazine	2.0 ± 0.06	1.3 ± 0.05	0.4 ± 0.003	0.6 ± 0.03
19	N‐trans‐Feruloyltyramine	191.5 ± 2.3	132.3 ± 1.5	72.1 ± 0.9	4.1 ± 0.1
20	TCDCA	3.6 × 105 ± 2.8 × 103	3.3 × 105 ± 2.9 × 103	2.5 × 105 ± 4.1 × 103	2.8 × 105 ± 1.9 × 103
21	Genistein	910.3 ± 4.3	1093.8 ± 8.2	350.3 ± 3.4	130.1 ± 2.7
22	TUDCA	4.5 × 105 ± 2.0 × 103	4.5 × 105 ± 3.5 × 103	3.3 × 105 ± 8.7 × 103	3.6 × 105 ± 4.3 × 103
23	Parthenolide	170.2 ± 0.9	22.6 ± 0.6	85.4 ± 1.3	1.2 ± 0.009
24	Pterosin B	9.3 ± 0.2	1.6 ± 0.02	6.5 ± 0.2	0.4 ± 0.01
25	6‐Gingerol	8.4 ± 0.1	15.2 ± 0.1	60.7 ± 0.5	0.6±0.02
26	3‐Epi‐bufalin	21.2 ± 0.9	4.3 ± 0.03	1.7 ± 0.03	55.0 ± 1.5
27	Isoalantolactone	12.3 ± 0.06	21.8 ± 0.4	87.1 ± 1.6	23.2 ± 0.3
28	Ursodeoxycholic acid	1894.1 ± 65.2	1039.1 ± 11.2	2029.9 ± 25.1	2100.1 ± 14.1
29	Hyodeoxycholic acid	5234.3 ± 22.3	4232.7 ± 79.3	8912.3 ± 330.2	7900.3 ± 78.2
30	Cholic acid	142.4 ± 0.6	118.2 ± 0.2	106.2 ± 2.5	109.1 ± 0.6
31	4‐Isopropyltoluene	0.3 ± 0.001	0.3 ± 0.01	0.1 ± 0.003	0.8 ± 0.04
32	Pedunculoside	0.7 ± 0.02	0.6 ± 0.02	7.1 ± 0.04	0.7 ± 0.005
33	Echinocystic acid	70.3 ± 2.3	78.5 ± 1.6	450.4 ± 13.3	74.5 ± 2.1
34	Eicosapentaenoic acid	—	—	—	1.2 ± 0.05
35	Betulin	673.6 ± 14.3	—	—	660.1 ± 15.1
36	Docosapentaenoic acid (C22:5n‐3)	—	—	—	94.2 ± 2.3
37	3‐Oxo‐4‐cholestene	6.3 ± 0.03	3.7 ± 0.04	48.3 ± 0.3	11.8 ± 0.2
	TCDCA/TUDCA ratio	0.80	0.73	0.76	0.78

Note: The sign “—” means that the compound was absent in bear bile powder. The content value calculated based on the peak area and linear regression equation and expressed as mean ± SD (n = 3). The quantification information was detailed in Supporting Information S3.

^a Sucrose, auxiliary material for improvement of taste.

The quantitative analysis of B‐1∼B‐4 has suggested that our results were basically acceptable and two cholic acid epimers (20 and 22) were indeed the most abundant compounds in natural BBP. Of course, there were still subtle differences in content among 4 batches natural BBP (B‐1∼B‐4). This can be attributed to different sources or processing methods. Finally, these quantification results across different sources and pharmaceutical companies have also provided a basis for new Q‐marker screening (see below).

Meanwhile, these standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS quantification results have brought at least three significances. (1) The quantification experiment itself could be considered as the cross‐validation of identification of all 37 compounds in standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS analysis. This further indicated that the identification of these compounds is highly reliable, despite the fact that there may be systematic errors between different LC‐MS technologies, and the discovery of 30 “unreported” compounds was highly robust and convincing. (2) As seen in Table 1, the sum of contents of 37 compounds of B‐1∼B‐4 varied from 59.23% to 82.33%, implying that the majority of natural compounds have been included in our analysis. Therefore, our analysis results, especially those of 30 “unreported” compounds, will guide scientists to optimize their manufacturing technique to produce highly simulated artificial PPB. (3) Quantification of all 37 compounds (1‐37) has also offered a premise to screen new Q‐markers (see below).

3.3. New Q‐Marker Screening: Relevant Principles, Validation Experiment, and Analysis Protocol Recommendation

3.3.1. Testability and Traceability Principles for New Q‐Marker Screening

New Q‐marker screening must adhere to some principles, such as testability, which was proposed by Liu Changxiao [24]. In the above quantitative analysis, 90 µg BBP samples yielded peak areas of 9567∼10537 million for TCDCA (20) and TUDCA (22) (Supporting Information S4 Figure S4); and even 7.5 µg BBP sample could produce∼4 million peak area of 20 and 22 (Supporting Information S3 Table S3). These peak area values far exceeded the basic LC‐MS requirement (peak area≈10000). Correspondingly, two cholic acid epimers (20 and 22) always exhibited the highest contents (Table 1). Their high content was also supported by previous studies from other teams, as mentioned above. Based on our experimental evidence and previous literature, TCDCA (20) was preliminarily screened as a new Q‐marker for its high testability. In other words, new Q‐marker (20), along with the old Q‐marker (22), adhered to Liu's testability principle [24].

As shown in Figure 5, TUDCA (22) is the C‐7 epimer of TCDCA (20). The 7‐OH group, however, can be oxidized into 7‐C═O group bearing an sp² ‐C atom to show planar configuration, under bacterial 7β‐hydroxysteroid dehydrogenase (7β‐HSDH) catalysis [25, 26]. The planar configuration is well‐known to share equal attack chances from two sides. Thereafter, the 7‐C═O would be equally distinguished into 7S‐OH (conventionally called “7β‐OH”) and 7R‐OH (conventionally called “7α‐OH”), after enzymatic redox reaction with the aid of co‐factors NAD⁺ or NADP⁺ (Figure 5). This can explain why TCDCA (20) showed generally equal content to TUDCA (22), and both of them could be simultaneously detected in different samples, by different teams, and using different analytic technologies [1, 6, 8, 23, 27, 28]. In a word, owing to the specific biogenetic pathway, TCDCA (20) shows high content, which is generally equal to that of TUDCA (22). Therefore, TCDCA (20) can always be detected in natural BBP to display good testability, similar to TUDCA (22).

FIGURE 5.

The interconversion TCDCA and TUDCA via enzymatic redox reaction.

However, good testability does not mean good traceability, because some high‐content Q‐markers may be immediately transformed via external conditions (such as heating, oxidation, digestion, and metabolism). Two epimers 20 and 22, however, could be easily determined in rat plasma for pharmacokinetic study [29]. This implies that the two remain stable and traceable after natural BBP is processed by drainage, collection, dryness, storage, and even metabolism. Therefore, it can be summarized that, similar to TUDCA (22), TCDCA (20) possesses not only good testability but also good traceability.

3.3.2. Pharmacological Relevance Principle for New Q‐Marker Screening

Furthermore, pharmacological relevance principle was also considered as a requirement of Q‐marker screening [24]. From the perspective of pharmacology and physiology, cholic acids play a critical role in human enterohepatic system and can alter gene expression in the liver and small intestine via activating Nr1i2 receptor [3]. Therefore, the current study compared their binding energies towards Nr1i2 receptor, by means of molecular docking simulation. As seen in Figure 6A, TCDCA (20) and TUDCA (22) could bind to the same site of Nr1i2. However, TCDCA (20, −12.2 kcal/mol) showed lower binding energy than TUDCA (22, −9.8 kcal/mol). Our results regarding Nr1i2 binding could partially explain why TCDCA had potential for treating liver fibrosis [30] and played a beneficial role in anti‐gastric cancer [31]. Other molecular docking information was shown in Supporting Information S5.

FIGURE 6.

The main results of molecular docking simulation. (A) Construction of pharmacophore models using TCDCA (or TUDCA) binding to the Nr1i2 receptor and complex structure. The binding energy is the primary indicator for evaluating the binding strength between ligands and receptors, where a lower value indicates stronger ligand–receptor affinity. (B) Visualized results of the interactions of TCDCA (20) with the amino acid residues of Nr1i2. (C) Visualized results of the interactions of TUDCA (22) with the amino acid residues of Nr1i2. The 7‐OH TCDCA (or TUDCA) is circled in red in B and C. Other molecular docking information was shown in Supporting Information S5.

However, the Nr1i2 binding energies were different between TCDCA (20) and TUDCA (22, Figure 6). The Nr1i2 binding difference was attributed only to the stereoconfiguration of 7‐OH group, because this is the sole difference between two epimers 20 and 22 (Figure 5). The visualized results (Figure 6B,C) revealed that the 7‐OH group in TCDCA (20) could form a hydrogen bonding with LEU‐209 residue, while that in TUDCA (22) could not. This indicates that 7R‐stereo‐configuration has advantage over 7S‐stereo‐configuration in binding Nr1i2 receptor. On the other hand, the Nr1i2 receptor has been suggested to not only play a role in enterohepatic system [3], but also be involved in inflammatory bowel disease [32] and cell carcinoma [33]. The versatile role of Nr1i2 receptor can explain why two epimers (20 and 22) sometimes play different roles in some pharmacological effects [5, 34, 35, 36]. Especially, Sugata's team pointed out that TCDCA (rather than TUDCA) played an important role in the development of hepatic fibrosis in the early phase of cholestasis without endotoxaemia [37].

Although molecular docking provides only a theoretical prediction of ligand–receptor interactions; however, the accumulated documents have clearly indicated that TCDCA played an indispensable role in the natural BBP and the role of TUDCA is sometimes different from that of TUDCA. Therefore, TCDCA could be considered a candidate complementary Q‐marker for natural BBP for further validation and integration with existing pharmacopoeia standards.

3.3.3. Specificity Principle for New Q‐Marker Screening

Recently, it has been reported that biocatalyzed chicken bile could also generate TUDCA (22), the old Q‐marker [33]. Thus, the old and single marker (22) lacked specificity to characterize the quality of various animal bile. Accordingly, the combination of TCDCA (20) and TUDCA (22) was expected to improve the specificity of Q‐marker. In a word, the above discussion regarding testability, traceability, pharmacological relevance, and specificity principles clearly shows that TCDCA (20) may be used as a potential Q‐marker for natural BBP, similar to TUDCA (22).

3.3.4. Validation Experiment: Analysis of New Q‐Marker (20) in 11 Batches of BBP Samples (B‐5∼B‐15)

To further validate the feasibility of new Q‐marker (20), other 11 batches of BBP samples (B‐5∼B‐15) were also analyzed by UHPLC‐ESI‐Q‐TOF‐MS/MS strategy under the SIM model (i.e., molecular formula = C₂₆H₄₄NO₆S⁻ or m/z = 498.2894). These samples exhibited high peak sign (5.6 × 10⁹∼9.9 × 10⁹ for 20 and 5.6 × 10⁹∼9.8 × 10⁹ for 22, Supporting Information S4 Figure S4) and high content (25.02 ± 0.41%∼36.57 ± 0.28% for 20 and 32.77 ± 0.87%∼44.60 ± 0.20 % for 22, Table 1 and Supporting Information S3 Table S3). This has validated that new Q‐marker (20) could be easily detected in natural BBP samples. Besides our team, six other teams have also successfully detected Q‐marker TCDCA (20) in natural BBP [1, 6, 8, 20, 21, 27]. All these have validated that detection of the new Q‐marker (20) would be feasible in various natural BBP samples, by various research teams, and through various analytic technologies (including HPLC‐CAD [22], HPLC‐ELSD [38], and LC‐MS [29]).

Furthermore, artificial bear bile sample was also introduced in the study for the UHPLC‐ESI‐Q‐TOF‐MS/MS quantitative analysis. The quantification results showed that the artificial BBP contained about 0.25% TCDCA and the TCDCA/TUDCA ratio was 0.20 (Supporting Information S4 Figure S4M). A review of the literature revealed that the content of TCDCA in chicken, bovine, or porcine bile was approximately 64% (25), 2% (39), and 4% (39), respectively. However, TUDCA was not detected in these samples [8, 38, 40] and hence the ratio of TCDCA/TUDCA was undefined in these animal samples. In the study, however, the ratio of natural BBP fell within the characteristic range of 0.58∼0.98 (Table 1 and Supporting Information S4 Figure S4), suggesting that the TCDCA/TUDCA ratio of natural bear bile was different from other animal biles or artificial bear bile.

The TCDCA/TUDCA ratio was subsequently investigated for the methodological parameters, including precision and robustness. For robustness, the sample solutions of B‐1 and B‐3 were stored at 4°C (refrigerator storage temperature) and 25°C (room temperature), and analyzed freshly prepared (0 h) and after 9 months of storage to evaluate the effects of different temperatures and long‐term storage time on the determination of TCDCA/TUDCA. The methodological validation experiment showed that the RSD values for all analyses were less than 3%. The matrix effects of TCDCA and TUDCA (recovery rates 99.55% and 102.30%) did not virtually interfere with the quantification of the TCDCA/TUDCA ratio. These results confirmed that the TCDCA/TUDCA ratio method was accurate and reliable for the quality control of natural BBP (Supporting Information S6 Table S6). All these results demonstrate that the TCDCA/TUDCA ratio (0.58∼0.98) can be used to recognize natural bear bile not only from other animal biles, but also from artificial bear bile.

3.4. Recommended Analysis Protocol for New Q‐marker (20)

Unlike to ordinary Chinese herbal medicines, natural BBP is a powder and thus can be directly dissolved in organic solvents (e.g., methanol) without an extraction process. Considering the high content of two epimers (20 and 22), the injection amount should be strictly controlled. Otherwise, their chromatographic peaks would be too broad (Supporting Information S7 and Figure S7). Therefore, the analysis protocol for new Q‐markers could be recommended as Figure 7.

FIGURE 7.

Recommended analysis protocol for new Q‐marker TCDCA (20), along with current TUDCA (22). The red asterisk “*” means new Q‐marker.

Natural BBP sample is dissolved in methanol; this solution is then filtered through 0.22 µm membrane to prepare a tested‐sample solution (≈2.5 mg/mL). Then, an aliquot of filtrate (3 µL) is injected into LC‐MS apparatus for analysis. Under negative mode, the new Q‐marker (20), together with old Q‐marker (22), can be simultaneously detected using the SIM method (molecular formula = C₂₆H₄₄NO₆S ⁻ or m/z = 498.2894). This LC‐MS analysis is highly specific and feasible, due to the absence of interference peaks (Figure 4). If there is no LC‐MS apparatus, HPLC analysis may also be available; however, there are always some interference peaks, regardless of whether an evaporative light scattering detector (ELSD) or ultraviolet (UV) detector is used [38, 39, 40]. Thus, HPLC analysis can only be used as an alternative strategy for Q‐marker analysis.

Nevertheless, chiral chromatographic column is unnecessary for such separation; this is because TCDCA (20) and TUDCA (22) are a pair of epimers, a kind of diastereoisomers with evident differences in physical or chemical properties. Therefore, the two can be easily separated by an ordinary chromatographic column. As seen in Supporting Information S3 Table S3, the LLOQ values of TCDCA and TUDCA were 9.2 and 8.6 ng/mL. Two values were available for the common LC‐MS method. All these results suggest that the detection of new Q‐marker could be conducted by conventional laboratories or quality inspection institutes in the future.

Finally, it should be noted that:

(1) Before standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS quantification, all identified compounds have actually been qualitatively identified by this tool. 3‐Oxo‐4‐cholestene (37) was a good instance. As seen in Supporting Information S2 Figure S2.37, the MS/MS profile and characteristic MS/MS peaks between the sample compound and authentic standard were similar in the UHPLC‐ESI‐Q‐TOF‐MS/MS analysis. This has also verified that the aforementioned UHPLC‐Q‐Orbitrap‐MS/MS identification of 3‐oxo‐4‐cholestene (37) was scientifically sound. Therefore, the standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS quantification can be considered as a double‐check of identification of all 37 compounds (1‐37) in Section 3.1, to ensure reliability of identification of 30 “unreported” compounds in standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS analysis. A single LC‐MS technology is well‐known to have systematic errors from apparatus. The combination of two LC‐MS technologies, that is, standards‐aided UHPLC‐Q‐Orbitrap‐MS/MS and standards‐aided UHPLC‐ESI‐Q‐TOF‐MS/MS, can effectively remove the possible systematic errors. Notably, two LC‐MS technologies have their own advantages and disadvantages [41]. This is the reason why the study introduced two LC‐MS technologies for the analysis.

(2) Although the current study has found 30 “unreported” compounds using two LC‐MS technologies; however, none of them belongs to cholic acid derivatives (Figure 2). In the quantification experiment, these non‐cholic acid derivatives showed low content and were even at trace levels. The maximum value was 0.11 ± 0.00082% (21 genistein in B‐2). On the other hand, genistein possessed a wide range of pharmacological effects (e.g., anti‐cancer) [40]; Therefore, ignorance of their existence in natural BBP is unwise, because sometimes a compound at low content can exert important action. Nevertheless, these “unreported” non‐cholic acid derivatives are unhelpful for BBP's quality control, due to their poor testability and specificity. These non‐cholic acid derivatives are known to be widely distributed in dietary plants. For example, genistein and daidzin co‐exist in soy products, and thus soybean may act as the feed of bears [42].

(3) As seen in Table 1, the sum of contents of 37 compounds of B‐1∼B‐4 varied from 59.21 ± 13.19% to 82.33 ± 5.76% in natural BBP. This high proportion implies that the majori of natural compounds have been included in our current analysis. Therefore, our analysis results (especially those of 30 “unreported” compounds), will guide scientists to optimize their manufacturing technique to produce highly simulated products.

(4) As seen in Supporting Information S3 Table S3, in the quantification experiment, we conducted a recovery test of only 19 compounds (1‐3, 7, 8, 10, 11, 14, 15, 16, 19–23, 26, and 28–30) to validate the method. This is because others showed very low peak signals and thus were not adequate for spike testing. However, this did not hinder the achievement of conclusion, because the current study was a Q‐marker screening work rather than a typical quantitative analysis work. The Q‐marker screening only needed the content of main compounds for ranking (Section 3.3.1).

4. Conclusion

Through two Standards‐aided LC‐MS technologies, a total of 37 compounds have been found in natural BBP, while 30 “unreported” compounds have been discovered for the first time in natural BBP. The total content of 37 compounds accounts for high proportion of natural BBP. Nevertheless, TUDCA and its epimer TCDCA play a dominating role in the content of natural BBP. TCDCA shows different binding action towards Nr1i2 receptor from TUDCA. Therefore, TCDCA can be considered a candidate complementary Q‐marker for natural BBP, and the TCDCA/TUDCA ratio (0.58∼0.98) can be used to recognize natural BBP from artificial BBP and other animal bile. All these findings lay a foundation for manufacturing technique improvement of artificial BBP and will promote pharmacopoeia to update the quality‐control method.

Author Contributions

Ziqing Li: data curation, formal analysis, methodology, writing – original draft. Haoyu Huang: data curation, formal analysis, methodology. Heping Zheng: conceptualization. Gongchang Zheng: conceptualization. Xi Zhao: visualization. Yongbai Liang: formal analysis. Rongxin Cai: data curation, methodology. Hanxiao Chai: data curation, formal analysis, methodology. Hongwei Song: writing – original draft. Xican Li: writing – original draft, writing – review and editing. All authors read and approved the final manuscript.

Funding

This work was supported by the National Nature Science Foundation of China (Grant Number 82374485).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: jssc70513‐sup‐0001‐SuppMat.zip.

Data Availability Statement

The authors confirm that the data supporting the findings of this study are available in the Supporting Information of this article.