Analysis on molecular interaction mechanism of four hapten flavonoids in Shuang-huang-lian powder injection with bovine serum albumin  
Haixiu Jiang, Yanqing Guan, Xiaotian Zhang, Ya Han, Ningning Guo, Hong Wang, Shizhong Chen* 
Department of Natural Medicines, School of Pharmaceutical Sciences, Peking University Health Science Center, Beijing 100191, China      

Abstract: Baicalein, baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide are the major flavonoids of the Shuang-huang-lian powder injection. These flavonoids are thought to be haptens that can induce allergic reactions. The interaction mechanism of these haptens with bovine serum albumin (BSA) was investigated by surface plasmon resonance (SPR) and molecular modeling method. The SPR study indicated that these compounds could specifically bind to the BSA with one binding site and equilibrium dissociation constant (KD) values were determined. Molecular modeling explored the mechanism of interaction under simulated physiological condition. The result of molecular modeling indicated that flavonoids could bind with BSA in the hydrophobic pocket of sub-domain II with hydrogen bonding as the main acting force.        
Keywords: Hapten flavonoids; SPR;Molecular modeling; BSA; Interaction
CLC number: R284                Document code: A                 Article ID: 10031057(2017)536606
1. Introduction
As a new formation of traditional Chinese medicine (TCM), traditional Chinese medicine injection (TCMI) is widely applied in China. However, allergic reactions due to the presence of hapten components in TCMIs have become a big concern over past few decades, which consequently impeded the extensive applications of TCMIs[1]. Among these TCMIs, Shuang-huang-lian powderinjection (SHLPI) has a bad fame because of its allergicreactions (side effects). SHLPI, composed of the extract of three herbs: Radix Scutellariae, Flos Lonicerae and Fructus Forsythiae, is an essential TCMI in Chinese Pharmacopoeia[2]. Flavonoid constituents, including baicalein, baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide, in TCMI have been investigated as haptensin our previous studies[3]. However, little has been knownabout the hapten mechanism with protein. Some methods have been developed for identifying haptens, including commonly used technologies, such as gold immunity chromatographic assay, enzyme linked immunosorbent assay (ELISA), immunosensor[4], immuno chromato test[5] and immuno-chip[6]. These methods are only applicable for purified compounds, and have obvious drawbacks because they are time-consuming and easy to lose minor active compounds during the purification process.
Bovine serum albumin (BSA) polypeptide is composedof 583 amino acids, and it is one of the two most commonly used carrier proteins to study the immune responses of hapten-carrier conjugates[7]. BSA has three homologous domains (IIII), of which two sub-domains IIA and IIIA are the primary binding sites of compounds/haptens[8]. Moreover, its similarity of spatial structure to human serum albumin[7] highlights the potential applications of BSA as a tool for investigating the mechanism between proteins and haptens.
In the present study, the BSA was applied to assess the interaction mechanism with baicalein, baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide by surfaceplasmon resonance (SPR) and molecular modeling. Thus, the combined data will help us better understand the structure-activity relationships between these haptenflavonoids and their binding sites on the BSA molecule. 
2. Materials and methods
2.1. Materials
Amine coupling reagents, N-hydroxysuccinimide (NHS),1-ethyl-3-(3-dimethylaminopropyl)-carbodiimidehydrochloride(EDC), acetate (pH 4.5, GE Healthcare Bioscience AB, 20278961311), PBS-P+ buffer, 10× (pH 7.5, GE Healthcare Bioscience AB, 210801), were purchased from commercial sources. The flavonoids (baicalein, baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide) shown in Figure 1 were obtained from Chengdu Must Bio-Technology Co., Ltd. (Chengdu, China). The purity of each standard was above 98% as determined by high performance liquid chromatography (HPLC). 

Figure 1.
Structures of baicalein, baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide.
2.2. Instrumentations and methods
SPR analysis was conducted at 25 °C using Biocore T200 (GE Healthcare, Piscataway, NJ) equippedwith Sensor Chip CM5 (GE Healthcare Bioscience AB, 10239303).
2.2.1. Immobilization of BSA
BSA in acetate (pH 4.5) was injected over a CM5 sensor chip at 10 µg/mL and captured on the carboxymethyl dextran matrix via an amine coupling reaction[9]. The surface was activated by injecting a solution containing 200 mM EDC and 50 mM NHS for 5 min. The BSAwas injected, and the surface was then blocked by injecting 1 M ethanolamine at pH 8.5 for 7 min.This reaction immobilized 16 335 resonance units (RU) of BSA.
2.2.2. Binding to BSA
Binding analysis with baicalein, scutellarin and Chrysin-7-O-β-D-glucuronide (10 different concentrations)was performed in PBS buffer, and baicalin (10 different concentrations) was performed in PBS buffer with 2% DMSO at a flow rate of 30 μL/min at 25 °C. Associationtime was set up at 60 s, and dissociation time was set up at 100 s.
2.2.3. Determination of KD values for BSA
All acquired sensorgrams were processed using a double referencing procedure. To determine KD values for the hapten/BSA interaction, the equilibrium response data were fit to an independent-binding-sites model:
Where Rmax is the maximal response, C is the concentration of binding compounds, and KDI is the KD. For a single-site interaction, I = 1, and for multiple binding sites, I = 2[10].
The KD for evaluation of the BSA-hapten binding affinity was determined by the steady-state affinity fittinganalysis of the Biocore T200 Evaluation Software (GE Healthcare).
2.2.4. Docking simulation
The docking algorithm Glide of Maestro 10.2 (Schrodinger LLC) was used for docking studies of active compounds. The complete crystal structure (PDB ID: 4OR0) of BSA reported recently was used due to its high resolution of 2.58 Å[11]. The docking procedures in the user’s guide were followed. For protein preparation, hydrogen atoms were added, disulfide bonds were created, and all unwanted waters beyond 5.0 Å from the het groups were deleted from the crystalstructure. The protein was then subjected to a series of  restrained minimization using the optimized potential for liquid simulations-all atom (OPLS_3) force field, and processed using “Protein Preparation module”. After protein generation, the iron was used as the center for grid generation (length ≤20 Å) using “Receptor Grid Generation”. The chemical structures of compounds were drawn by ChemSketch and prepared by “LigPrep module” with Epik to explore protonation and tautomericstates at pH 7.0±2.0. Moreover, the docking simulation was performed using standard precision mode. The electrostatic interactions of the docking were treated with distance-dependent dielectric solvation. Default parameters wereused as described in the software. Interactions betweenBSA and compounds were analyzed, and the docking scores of BSA-ligand were used for comparison.
3. Results and discussion
3.1. Interaction analysis using SPR
3.1.1. Interaction analysis between four flavonoids and BSA
Figure 2 shows the response data collected for various concentrations of baicalein, baicalin, scutellarin and 7-O-β-D-glucuronide binding to immobilized BSA.

Figure 2. A plot of response (RU) obtained from SPR analysis betweenfour flavonoid constituents and BSA. Various concentrations of baicalein (0.97, 1.95, 3.89, 7.78, 15.60, 31.1 and 62.3 μM), baicalin (0.97, 1.95, 3.89, 7.78, 15.60, 31.1 and 62.3 μM), scutellarin (1.95, 3.91, 7.81, 15.60, 31.1, 62.5, 125, 250, 500 and 1000 μM) and Chrysin-7-O-β-D-glucuronide (52, 78, 117, 176, 263, 395, 593, 889, 1330 and 2000 μM) were injected over the immobilized BSA on the sensor chip, and then the corresponding response was recorded.
Baicalein at a concentration ranging from 0.97 to 62.3 µM was injected over BSA surface. We foundthat the SPR response was increased upon baicalein injection over the immobilized BSA in a dose-dependentmanner. A global fitting using Biocore T200 Evaluation Software revealed that the KD between baicalein and BSA was 3.036×10−5 M. These results suggested that baicalein associated with BSA.
The SPR response was found to increase upon baicalein injection over the immobilized BSA in a dose-dependent manner. The KD between baicalein and BSA was 4.691×10−4 M. These results suggested that baicalin associated with BSA.  
In addition, we found that the SPR response was increased upon scutellarin injection over the immobilized BSA in a dose-dependent manner. The KD between scutellarin and BSA was 8.815×10−4 M. These results suggested that scutellarin associated with BSA.
Lastly, we found that the SPR response was increased upon Chrysin-7-O-β-D-glucuronide injection over the immobilized BSA in a dose-dependent manner. The KD between Chrysin-7-O-β-D-glucuronide and BSA was 8.0815 × 10−4 M. These results suggested that Chrysin-7-O-β-D-glucuronide associated with BSA.
3.1.5. Globally fitting equilibrium data comparison
In order to determine KD values for the different compounds, the data were fit globally to a single-binding-site model, where the I = 1 in equitation, constrained with one maximum response and allowing KDto be a local parameter[12]. From Table 1, the affinity of flavonoid (baicalein) was stronger than that of flavonoid glycosides (baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide). For baicalin and its aglycone baicalein, the steric hindrance of baicalin may weaken the BSA-binding affinity due to its C7 glycosylation. In addition, the activity of scutellarin with an additional C4-OH was smaller thanthat of baicalin, which may be due to the increased polarity and size of scutellarin, causing possible reduced capacity to penetrate into the tryptophan rich hydrophobic regions of BSA. The result was consistent with our previous flavonoids hapten studies.

Table 1. Hapten/BSA KD values determined by SPR.

3.2. Molecular docking between flavonoids and BSA
To gain better understandings of the interactions between baicalein (A), baicalin (B), scutellarin (C) and Chrysin-7-O-β-D-glucuronide (D) with BSA protein, we carried out docking simulation of baicalein (A), baicalin (B), scutellarin (C) and Chrysin-7-O-β-D-glucuronide (D) to BSA binding site (PDB ID: 4OR0). Figure 3 shows the binding poses of these compounds.

Figure 3. Docking simulation results showing interactions of baicalein (A), baicalin (B), scutellarin (C) and Chrysin-7-O-β-D-glucuronide (D) with BSA protein. The residues of BSA are represented using blue ribbons, and the ligand structure is represented using purple tube. The hydrogen bond between the ligand and BSA is represented using a yellow dashed line. 

Figure 3 shows that hydrogen bond could play an important role in ligand binding between the four natural products and BSA. Hydrophobic bonds are mainly located in the flavones nucleus. Therefore, it was speculated that hydrophobic interactions were more possibly involved in the binding process betweenthe nucleus and the BSA than hydrogen bonds. The detailed information about the hydrogen bonds in the four natural products are explained as follows: baicaleinformed two H-bonds (ARG-198 and ARG-256)and one Pi-Pi stacking (Arg-198) with BSA. Baicalin formed five H-bonds (ARG-198, ARG-217, GLU-291, GLU-152 and SER-191), one electrostatic interaction (ARG-194) and two Pi-Pi stackings (Arg-198, Arg-256) with BSA. Scutellarin forms two H-bonds (ARG-217 and LYS-221), two Pi-Pi stackings (TRP-213 and ARG-194) and two electrostatic interactions (ARG-217and TRP-213) with BSA. Chrysin formed three H-bonds (ALA-290, GLU-291 and TYR-149) and one Pi-Pi stacking (PHE-222) with BSA. The docking results suggested that more hydrogen bonds were formedbetween baicalin and BSA than baicalein and BSA.We conjectured that more hydroxyl groups existedin the glucose units of baicalin molecule, whereas less hydroxyl groups were found in baicalein molecule. Therefore, more hydrogen bonds were formed in the former than the latter. Hydrogen bonds are important non-covalent interactions in ligand binding forbaicalin, whereas the interactions between baicalein and HSA may belong to some other types, such asPi-Pi stacking, electrostatic interactions, as well as hydrophobic forces.
In all, the molecular docking showed that interactions existed between the four flavonoids and the BSA. Among these, hydrogen bonds were mainly concerning to baicalin, and the power of hydrogen bonds greatly depended on the number of the hydroxyl group in these natural products.
4. Conclusions
In our study, we investigated the interactions of four hapten flavonoids, including baicalein, baicalin, scutellarin and Chrysin-7-O-β-D-glucuronide, with BSA by SPR and molecular modeling. Interaction analysis using SPR confirmed the binding of hapten flavonoids to BSA with a dissociation constant of 10−4 M10−5 M, and the interactions between flavonoids and BSA were specific. Among the four hapten flavonoids tested in this study, the affinity between baicalein and BSA was the strongest.
Furthermore, molecular docking assay between flavonoids and BSA showed that hydrogen bonding was the major force. In addition, the way of electrostaticinteraction and Pi-Pi stacking interaction was also observed. The differences of the binding modes might be the reason of the glucose units in molecule.
In the previous studies, though several screening methods have been established to study haptens in TCMI, research showed that there is more than one compound involved in allergic reaction. Compared with previous studies on the sensitization mechanism, this study explained how haptens generated allergic reactions by binding to certain areas of BSA in a more essential way. We believed that our data not only provided a better understanding of the molecular mechanism of allergic action, which is essential in studying the binding of more flavonoid compounds with BSA, but also offered valuable scientific view of generating superior derivatives with reduced side effects by drug design.
This work was supported by Nature Science Foundation of Beijing, China (Grant No. 7142088). We are grateful for Professor Ge Fu of State Key Laboratory of Natural and Biomimetic Drugs, Peking University for assisting the Biocore experiment.  
[1] Xu, Y.B.; Dou, D. Chin. J. Chin. Mater. Mad.2015, 40, 2765–2773.
[2] Chen, X.; Howard, O.; Yang, X. Life Sci.2002, 70, 2897–2913.
[3] Sun, H.; Liu, M.; Lin, Z.; Jiang, H.; Niu, Y.; Wang, H.; Chen, S. J. Pharm. Biomed. Anal.2015, 115, 86–106.
[4] Situ, C.; Crooks, S.; Baxter, A.; Ferguson, J.; Elliott, C. Anal. Chim. Acta. 2002, 473, 143–149.
[5] He, F.; Zheng, K.; Zeng, J.; Dai, R.; Zan, Z.; Liu, W.; Shi, J.Chin. J. Chin. Mater. Med.2012, 37, 2836–2841.
[6] Qu, H.; Zhao, Y.; Wang, Q. J. Beijing Univ. Tradit. Chin. Med.2008, 31, 23.
[7] Carter, D.; Ho, J. Adv. Protein Chem.1994, 45, 153–203.
[8] Sun, Y.; Wei, Y.; Wu, L. Sci. Technol. Food Ind. 2010, 31, 129–131.
[9] Takami, M.; Takakusagi, Y.; Kuramochi, K. Mol.2011, 16, 4278–4294.
[10] Rich, R.; Day, Y.; Morton, T. Anal. Biochem.2001, 296, 197–207.
[11] Wang, M.; Lai, T.; Wang, L. Chem. Commun. (Camb). 2015, 51, 7867–7870.
[12] Homola, J. Chem. Rev.2008, 108, 462–493.

江海秀, 管颜青, 张笑天, 韩雅, 郭宁宁, 王弘, 陈世忠*
北京大学医学部药学院天然药物学系, 北京 100191 
摘要: 黄芩素、黄芩苷、野黄芩苷及白杨素-7-O-葡萄糖醛酸苷是双黄连注射剂中的主要黄酮类成分, 也是其中的半抗原成分。本研究分别采用表面等离子体传感共振技术(SPR)和分子模拟分别对四种化合物与牛血清白蛋白的相互作用机制进行了研究。表面等离子共振结果表明四种黄酮类半抗原化合物可以特异性地与牛血清白蛋白(BSA)1:1的方式相结合, 并得到药物与蛋白平衡解离常数KD值。分子模拟结果表明, 黄芩素、黄芩苷、野黄芩苷及白杨素-7-O-葡萄糖醛酸苷与牛血清白蛋白在亚结构域II的疏水腔内结合, 主要作用力为氢键。 
关键词: 黄酮类半抗原; 表面等离子体传感共振技术; 分子模拟; 牛血清白蛋白; 相互作用
Received: 2017-01-12, Revised: 2017-02-21, Accepted: 2017-03-16.
Foundation item: Nature Science Foundation of Beijing, China (Grant No. 7142088).
*Corresponding author. Tel.: +86-010-82802723, E-mail:   
本作品采用知识共享署名-非商业性使用 4.0 国际许可协议进行许可。