The roots of A. dahurica were purchased from Bozhou Materia Medica Market (Bozhou, China) in 2012 and identified at that institution. The voucher specimen (M20120907) was deposited in The Firs People's Hospital of Jingmen City (Jingmen, China).

Water was prepared from distilled water by a Milli-Q system (EMD Millipore, Billerica, MA, USA). Anti-dinitrophenyl (DNP)-IgE (D8406) and DNP-human serum albumin (D-5059-10) were purchased from Sigma-Aldrich (Merck KGaA, Darmstadt, Germany). The media and reagents for the cell culture were supplied by Gibco (Thermo Fisher Scientific, Inc., Waltham, MA, USA). ELISA kits for TNF-α (H052), IL-1β (H002) and IL-4 (H005) were purchased from the Nanjing Jiancheng Bioengineering Institute (Nanjing, China). The luciferase assay system including NF-κB luciferase reporter plasmid pGL4.32 and the Renilla luciferase reporter vector plasmid pRL-TK was supplied by Promega Corporation (Madison, WI, USA). The Lipofectamine^® 2000 transfection reagent was purchased from Invitrogen (Thermo Fisher Scientific, Inc.). The other solvents employed in the present study were of analytical purity grade. Sephadex LH-20 was purchased from GE Healthcare Life Sciences (Chalfont, UK), Toyopearl HW-40C from Tosoh Corp. (Tokyo, Japan), silica gel from Qingdao Ocean Chemical Co., Ltd. (Qingdao, China) and octadecylsilane (ODS) was supplied by Daiso Chemical Co., Ltd. (Osaka, Japan). Porous resin D₁₀₁ was obtained from Tianjin Blosorb Biotechnology Co., Ltd. (Tianjin, China).

The dried roots of A. dahurica (10.0 kg) were ground and extracted with 95% ethanol (20 l each time) under reflux for 5 h three times. Following evaporation of the solvent under reduced pressure, the residue was obtained and suspended in water (5.0 l) followed by successive partition with petroleum ether (PE; 5.0 l each time), ethyl acetate (EA; 5.0 l each time) and n-butanol (n-BuOH; 5.0 l each time) for three times. Subsequently, the solvents were removed and three parts were obtained.

The PE part (110.0 g) was subjected to column chromatography (CC) on a silica gel eluted with gradient PE/EA (from 100:0 to 20:80, v/v) and 6 fractions were obtained according to the thin-layer chromatography (TLC) assay results. Briefly, the samples in EA were loaded on the TLC plates (Qingdao Jiyida Silica Reagent Co., Ltd., Qingdao, China) and eluted with PE/EA (70:30, v/v). The spots were visualized under ultraviolet light. Fraction 3 was further separated by silica gel CC to produce compound 14 (25.0 mg) as colorless needle-like crystals. Fraction 5 was also isolated on a silica gel CC to obtain compound 13 (10.5 mg), and fraction 6 was subjected to HW-40C and eluted with isocratic dichloromethane methanol (DCM/MeOH; 3:1, v/v) to produce compound 1 (33.0 mg).

The EA extract (260.0 g) was subjected to silica gel CC and eluted by gradient DCM/MeOH (from 100:0 to 0:100, v/v). Furthermore, the eluate was combined into 12 fractions based on the TLC assay. Fraction 2 was subjected to the HW-40C CC eluted with isocratic DCM/MeOH (2:1, v/v) and then recrystallized to produce compounds 4 (33.7 mg) and 6 (15.5 mg). Fractions 3 and 4 were combined and separated on the silica gel CC with gradient DCM/MeOH (from 90:0 to 50:50, v/v) to produce four subfractions. Subfractions 2 and 4 were further purified with Sephadex LH-20 eluted with DCM/MeOH (50:50, v/v) and compounds 1 (81.0 mg) and 2 (63.0 mg) were obtained. Fraction 6 was repeatedly subjected to silica gel CC with DCM/MeOH as the eluent, and the eluates were subsequently crystalized to produce compounds 7 (21.0 mg), 5 (18.0 mg) and 11 (11.0 mg). Fraction 8 was disposed by the silica gel CC with gradient DCM/MeOH and subsequently Sephadex LH-20 CC eluted with isocratic DCM/MeOH (1:1, v/v) to produce compounds 9 (9.5 mg) and 10 (13.0 mg). Fraction 9 was purified by the Sephadex LH-20 CC directly and then preparative TLC to obtain compound 12 (8.5 mg). Finally, fraction 11 was separated on an ODS CC eluted with MeOH/H₂O (50:50 to 95:5, v/v) and purified by Sephadex LH-20 CC eluted with MeOH to gain compounds 3 (12.5 mg) and 8 (13.0 mg).

The n-BuOH extract (130.0 g) was subjected to porous resin D₁₀₁ CC with gradient ethanol/H₂O (10:90, 30:60, 90:10 and 100:0, v/v) to produce 4 fractions. Fraction 1 was separated on ODS CC eluted with MeOH/H₂O (from 30:70 to 95:5, v/v) to produce compounds 1 (16.5 mg), 3 (10.5 mg) and 4 (13.0 mg). Furthermore, fraction 2 was also subjected to ODS CC to produce compound 15 (26.0 mg).

The compounds isolated were dissolved in DMSO and the nuclear magnetic resonance (NMR) spectra were recorded on a Bruker DRX400 NMR spectrometer (Bruker, Billerica, MA, USA) to elucidate their chemical structures. Mass spectrometry was performed on Agilent LC-MS with 1260 LC system and 6400 Series Triple Quadrupole mass spectrometer equipped with electrospray ionization (Agilent Technologies, Santa Clara, CA, USA). The structures were determined through comparing the data obtained from NMR and MS with those in previously published literature (see Results).

RBL-2H3 cells were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 µg/ml streptomycin at 37°C in a humidified 5% CO₂ atmosphere.

As the levels of histamine represented the degree of mast cell degranulation, the histamine contents in culture media were measured. The o-phthaldialdehyde spectrofluorometric procedure was employed as previously described (26). For the IgE-mediated histamine release, RBL-2H3 cells (1×10⁵ cells/well in a 96-well plate) were sensitized with anti-DNP IgE in PBS (10 µg/ml) and incubated at 37°C overnight. Next, the cells were pretreated with compounds obtained from the plant (20 µM in PBS containing 0.1% DMSO) or PBS containing 0.1% DMSO at 37°C for 1 h prior to the challenge with DNP-human serum albumin (HSA; 500 ng/ml). The cells were then separated from the media by centrifugation at 10,000 × g for 5 min at 4°C. Furthermore, the fluorescent intensity was recorded by a fluorescent plate reader (Molecular Devices, LLC, Sunnyvale, CA, USA) at an excitation wavelength of 360 nm and an emission wavelength of 440 nm.

Molecular docking is able to provide the visual detail of the interaction between a ligand and a biomacromolecule. In the present study the Surflex-Dock Sybyl v2.0 program (Tripos Inc., St. Louis, MO, USA) was used. 3D structures of the compounds obtained were prepared by Sybyl sketch and optimized by Tripos force field. The crystal structure of the histamine H₁ receptor was retrieved from the RSCB Protein Data Bank (www.rcsb.org/pdb/home/home.do) (PDB code: 3RZE), and hydrogen atoms and charges were added. After removing the water, the docking process was performed (27).

The levels of inflammatory cytokines in the media from RBL-2H3 cells challenged with DNP-HSA were determined by ELISA. ELISA was performed on a 96-well plate using ELISA kits according to the manufacturer's instructions. After terminating the reaction of the substrate through the addition of stop buffer (2N H₂SO₄), the absorbance was recorded on a spectrophotometer at a wavelength of 450 nm.

RBL-2H3 cells (1×10⁶ cells/well in a 24-well plate) were transfected with both the NF-κB luciferase reporter plasmid pGL4.32 and the Renilla luciferase reporter vector plasmid pRL-TK at 100 and 9.6 ng per well, respectively. Following transfection using lipofectamine^® 2000 at 37°C for 24 h, the medium was replaced with fresh serum-free medium. Cells were treated with compounds obtained from the plant (20 µM in PBS containing 0.1% DMSO) or PBS containing 0.1% DMSO before stimulation. Cells were subsequently washed with PBS twice and lysed with PLB for 20 min according to the manufacturer's instructions for the luciferase reporter assay system. The Promega protocol was run on the GloMax-Multi JR detection system (Promega Corporation). Briefly, 50 µl luciferase assay reagent was mixed with 10 µl lysate. The fluorescence intensity of firefly luciferase was determined. Then 50 µl Stop & Glo^® reagent was added into the mixture and the fluorescence intensity of Renilla luciferase was measured. Furthermore, the relative luciferase activity was measured by normalizing the firefly luciferase activity against the internal control (Renilla luciferase).

Lipophilicity/hydrophilicity has a strong influence on the absorption, distribution, metabolism and excretion properties of pharmacological agents in vivo. Hydrophobic agents tend to bind to hydrophobic sites whereas hydrophilic ones favor hydrophilic positions (28). Therefore, the lipophilicity/hydrophilicity expressed as logp was theoretically calculated through the physicochemical module of the Sybyl program while inputting the chemical structures to elucidate the physicochemical properties of these compounds. Generally, the values of logp for the drugs are between 2 and 5, which suggests that the drugs may be easily delivered to binding sites.

The polar surface area (PSA) derived from the Sybyl program as above is defined as the surface sum over all polar atoms, primarily oxygen and nitrogen atoms, also including their attached hydrogen atoms. It is typically used to optimize a drug's ability to permeate cells. Furthermore, molecules with PSA values of <140 Ų tend to be poor at permeating cell membranes.

All values are expressed as the mean ± standard deviation, and GraphPad Prism 5.0 (GraphPad Software, Inc., La Jolla, CA, USA) was used for statistical analysis. Significant differences of experimental data from different groups were compared by one way analysis of variance followed by Dunnet's test for multiple comparisons and Student's t-test for single comparisons. P<0.05 was considered to indicate a statistically significant difference.

Phytochemical investigation on the roots of A. dahurica has led to the isolation of 15 compounds. As shown in Fig. 1, their structures were identified as isoimperatorin (1) (29), imperatorin (2) (29), oxypeucedanin (3) (30), oxypeucedanin hydrate (4) (29), bergapten (5) (29), byakangelicin (6) (29), phellopterin (7) (29), byakangelicol (8) (31), isopimpinellin (9) (32), xanthotoxol (10) (29), xanthotoxin (11) (29), pimpinellin (12) (31), scopoletin (13) (31), β-sitosterol (14) (29) and daucosterol (15) (29) on the basis of spectra analysis including ¹H-nuclear magnetic resonance (NMR), ¹³C-NMR and mass spectrometry in combination with the data comparison from previously published studies (29–32).

The contents of released histamine in the media for RBL-2H3 cells indicated mast cell degranulation. As shown in Fig. 2, anti-DNP IgE-sensitized RBL-2H3 cells released a significantly increased level of histamine when DNP-HSA was added, compared with control cells. Compounds 1–13 significantly reduced the content of histamine in the media, compared with DNP-HSA cells, compound 5 induced the greatest decrease of the three.

Molecular docking revealed the interaction between the histamine H₁ receptor and isolated coumarins. Out of all the coumarins, compound 3 exhibited a more potent affinity to the histamine H₁ receptor than the others. Its total score (a measure of its affinity recorded by the Sybyl program) was 8.46, which was higher than for doxepin (a ligand in the crystal structure of the histamine H₁ receptor) which had a total score of 7.57. In addition, compounds 1, 2, 4 and 6 exhibited scores of 6.11, 6.36, 6.60 and 6.60, respectively, whereas the scores for compounds 7 and 8 were 5.81 and 5.77, respectively. As shown in Fig. 3, compound 3 is able to fully enter the binding pocket of histamine H₁ receptor and present a similar pose with doxepin. Hydrophobic and electrostatic interactions are important for the formation of the ligand-receptor complex (Fig. 4), although there are no evident hydrogen bonds between small molecules and biomacromolecules.

To further evaluate the compounds from A. dahurica on chronic allergic inflammation, the levels of TNF-α, IL-1β and IL-4 in the media for RBL-2H3 cells were determined (Fig. 5). Following challenge with DNP-HSA, the levels of TNF-α, IL-1β and IL-4 in the media significantly increased compared with control cells. When pretreated with the obtained compounds, the levels of TNF-α, IL-1β and IL-4 for compounds 1–12 were significantly decreased compared with DNP-HSA cells, to different extents. Compounds 1, 2, 5 and 7 induced the greatest decrease in levels of these inflammatory cytokines. However, compounds 13–15 exhibited no significant difference on TNF-α, IL-1β and IL-4.

In order to further confirm the effects of these compounds on the expression of inflammatory cytokine genes, the activation of NF-κB was investigated following treatment with these compounds. As a result, the activation of NF-κB was significantly increased in RBL-2H3 cells stimulated by DNP-HSA and significantly ameliorated when compounds 1–12 were administered (Fig. 6). Of all the compounds, compounds 5 and 7 exhibited the greatest potency, followed by compounds 1 and 2.

As shown in Table I, the values of logp for the identified compounds 1–3, 5 and 7–11 were between 2 and 5. Conversely, the values of compounds 4, 6 and 12–15 were not within that range. In addition, the PSA values of these compounds were <140 besides compounds 6 and 15. However, the PSA of compound 15 was markedly higher than that of compound 6.