Reagents and equipment
Anhydrous ethanol (analytical grade ≥ 99.7%), methanol (HPLC grade) and dimethyl sulfoxide (DMSO, ≥ 99.9%) were purchased from Merck. Lipopolysaccharide (LPS, L2880) was from Sigma. α-Naphthyl isothiocyanate (ANIT, E1909091) was from Aladdin. White sugar (20,180,404), Yili whole milk powder (20,180,409), Jinluo edible lard (20,180,317) and Yujiangyuan virgin olive oil (20,191,008) were bought from the local supermarket. Adrenalin hydrochloride (Adr, 1 mg/mL for livestock, 20,190,307) was bought from Shanghai Quanyu Biotechnology Animal Pharmaceutical Co., Ltd. Drugs included Cisapride Tablet (20,180,518, Anglikang), Compound Danshen Dripping Pill (170,912, Tasly), Ursodeoxycholic Acid Capsule (L19001A, Kangzhe) and 28 chemical references of rhubarb (purity ≥ 98%, Nanjing Jin Yibai Biological Technology Co., Ltd., details listed in Additional file 1: Table S1). Antibodies used for western blotting were the same as our preceding research [31] and provided in the Supplementary Materials. Mouse enzyme-linked immunosorbent assay (ELISA) kits covering MTL, SS, VIP, AchE, TG, Na+-K+-ATPase, TNF-α, IL-1β, IL-6, HSP-70, (T-)SOD, NO, TXB2, 6-keto-PGF1α, PGE2, ET-1, Mg2+, Ca2+, MDA, GSH, ALT, AST, ALP, GST, GGT, TBIL, DBIL and TBA were all obtained from Shanghai Enzyme-linked Biotechnology Co., Ltd. Kits of coagulation four indices of TT, PT, APTT and FIB were from Steellex. Serum and tissue Fe3+ test kits (colorimetry, A039-1-1, A039-2-1) were from Nanjing Jiancheng Bioengineering Institute. The experiments were equipped with micro-plate reader (EnSpire, Molecular Devices), protein electrophoresis apparatus (PowerPac, BioRad), gel imager (#1,708,195, BioRad), platelet aggregation/coagulation analyzer (LG-PABER-1, Steellex), and UHPLC Acquity™ system coupled to Synapt™ Q-TOF mass spectrometer (Waters Corp.).
Preparation of rhubarb liquids
The rhubarb was collected from Gannan Prefecture, Gansu Province, China, and was processed into decoction pieces by Gannan Baicao Biotechnology Development Co., Ltd. They were identified by Professor Hui Yan (Department of Pharmacognosy, Nanjing University of Chinese Medicine) as Rheum tanguticum Maxim. ex Balf. (No. NJUTCM-20,171,015) [32]. According to the data mining of multifunctional preparation and usage of rhubarb before [33], it was found that the most commonly used extraction methods were water and wine decoction repeated twice for 15–200 min. Based on the preliminary detection of component changes among different extraction methods (Additional file 1: Figure S1), we designed the rhubarb groups from two aspects of 9 different concentrations of ethanol-water (EW) solvents and 2 extraction time. Short time (S) was 10 min for the first decoction and 5 min for the second while long time (L) was 120 min for the first and 80 min for the second. A total of 18 rhubarb groups were set up including water-S, water-L, 10% EW-S, 10% EW-L, 20% EW-S, 20% EW-L, 35% EW-S, 35% EW-L, 50% EW-S, 50% EW-L, 65% EW-S, 65% EW-L, 80% EW-S, 80% EW-L, 90% EW-S, 90% EW-L, ethanol-S, ethanol-L. The highest dose of Chinese Pharmacopoeia was selected to be converted to 1.95 g/kg for mouse dosage. The process was to precisely weigh rhubarb powder passing through No.4 Pharmacopoeia sieve, fully soak with 2 times the amount of extraction solvent and then add 8 times the amount of corresponding solvent to boil out by reflux twice. After merging two filtrates when they were hot, the extracting solution was concentrated till no alcohol taste using rotatory evaporation (50 °C), and finally diluted with water as the suspension of 0.195 g/mL for use.
Pretreatment of rhubarb samples and chemical references
Rhubarb samples: Before vacuum concentration, 1 mL of each liquid was pipetted and accurately added into the homologous solvent to a concentration of 0.03 g/mL. Then 100 µL of each sample was mixed as the quality control (QC) sample. Reference solution: The 28 chemical references were critically weighed 1 mg each adding with a trace amount of DMSO for hydrotropy and methanol till 1 mL, followed by ultrasound to become the transparent solution. 100 µL of each mother liquor was added to 1 mL of methanol for dilution, and then 100 µL of each dilution was taken as the mixing standard sample. All above samples were placed in HPLC glass vials through microfiltration membrane (0.22 μm) and stored at 4 °C for testing.
UPLC-Q-TOF/MSE profiling
To make the relative content determination of liquid ingredients more accurate, the acquisition time of mass spectrometry scanning was changed from 0.3 s to 0.1 s, and the wavelength range of ultraviolet (UV) detection was set at 190–420 nm. The phase A was water containing 0.1% formic acid and B was acetonitrile. We optimized the gradient elution program: 0 ~ 2 min, 3 ~ 10% B; 2 ~ 5 min, 10 ~ 15% B; 5 ~ 8 min, 15% B; 8 ~ 11 min, 15 ~ 23% B; 11 ~ 16 min, 23 ~ 30% B; 16 ~ 20 min, 30 ~ 60% B; 20 ~ 24 min, 60 ~ 80% B; 24 ~ 25 min, 80 ~ 3% B. Remaining device parameters were the same as we described previously [34]. At the beginning of the analytical batch, QC sample was continuously injected 5 times to balance the system. Then 18 rhubarb samples were injected successively, and each sample was parallel for 3 times. Regular injection of corresponding blank solvent and QC samples can be used to evaluate the reliability and repeatability of instrument status throughout the workflow sequence. Under the same conditions, 28 rhubarb standard samples and the mixing standard sample were detected in turn.
Fuzzy chemical identification
By the fuzzy chemical identification method [23], TCM components can be identified quickly, in which the core is to classify uncertain ones because compounds with the same mother nuclei possess similar action properties and interaction rules. It not only avoids some defects such as complex identification procedures of a great many components or scarcity of chemical references, but also reflects the vague and holistic view of TCM. It needs to be implemented in conjunction with the full-spectrum information database of nearly 300 chemical compounds of rhubarb (Additional file 1: Table S2) we established through querying numerous relevant literature reports and network platforms such as TCMSP (tcmspw.com), SymMap (www.symmap.org) and PubChem (pubchem.ncbi.nlm.nih.gov). Under the premise of selecting different types of known components with high contents in rhubarb as the reference, we summarized the peak order, fragmentation information and pathways of different categories, and then established corresponding compound group networks to recognize unknown components rather than to confirm all components necessarily.
Mouse models and determination of pharmacodynamic indexes
BALB/c mice, weighing 22 ± 2 g, were purchased from Shanghai Sippr-BK Laboratory Animal Co., Ltd. (license number: SCXK (Hu) 2018-0006). Adaptive feeding environment and all animal related procedures were strictly in accordance with the criteria of the Animal Ethics Committee of Nanjing University of Chinese Medicine. These mice were free to eat conventional food and drink water. Before modeling, they were stratified by weight and randomly divided into 21 groups including the control, model, positive and 18 rhubarb-treated groups.
Constipation model for evaluation of E1 and E2
The constipation model with gastrointestinal accumulated heat induced by dyspepsia was based on our established method [31]. The positive drug was Cisapride Tablet for the treatment of gastrointestinal dynamic diseases, and its maximum dosage was 3.9 mg/kg converted for mice. We divided 168 male mice (No. 20,180,006,004,662) into groups of 8 on average. Besides the control group, 0.8 mL of self-made high-calorie feed mixed by sugar, milk powder and lard [31] was additionally given to each mouse by intragastric administration 3 times a day at an interval of 6 h. The modeling lasted seven days. Positive and rhubarb-treated groups were administered with corresponding drug liquids (0.01 mL/g) in the morning of the 6th to 8th day while the control and model groups were intragastrically given isodose water at the same time. After 30 min of administration on day 8, all mice were intragastrically given Indian ink (0.01 mL/g) and immediately placed in the metabolic cages alone to observe defecation characteristics of first black stool time, the number of black stools and fecal weights within 12 h. On the 9th day, we weighed these mice, collected their serum, and then dissected 0.5 cm of duodena behind the pylorus, stomachs, colons and colonic contents. Biochemical index levels of MTL, SS, VIP, TG in mouse serum and AchE, Na+-K+-ATPase, TNF-α, IL-1β in duodenal tissues were detected by ELISA, and expressions of common inflammatory proteins in the colon covering p-NF-κB p65, NF-κB p65, p-p38, p38, p-ERK, ERK, p-JNK, JNK and TLR4 were determined by western blotting as depicted previously [31].
Blood stasis syndrome for evaluation of E3 and E4
The blood stasis syndrome induced by noxious heat can be established by LPS-induced inflammation combined with Adr-induced stagnation of the circulation of vital energy [35]. Existing literature reports basically used LPS to simulate the process of exogenous heat toxin, and its modeling dose for mice was 3 mg/kg prepared by 0.9% normal saline (NS). Adr injection for livestock will assist to achieve the effect of blood stasis in a short time, and its combined dosage was determined as 5 mg/kg through our continuous exploration in order to greatly reduce the mortality of mice. In this model, Compound Danshen Dripping Pill was chosen as the positive drug (105.3 mg/kg for the maximum dose of mice), which was often used clinically to regulate Qi and activate blood circulation. We separated 252 female mice (No. 20,180,006,007,816) into 12 mice in each group. On the first day of modeling, except injecting isodose 0.9% NS for the control group, all mice were intraperitoneally injected with LPS at 0.01 mL/g for the first time and their anal temperatures should be measured punctually 4 h later. If the temperatures were abnormal, the model was successful, and then corresponding drug liquids were administered once every night (0.01 mL/g, isodose water given to the control and model groups) for 6 consecutive days. On the 5th day, mice were subjected to LPS at 0.01 mL/g by intraperitoneal injection and tested for anal temperatures 4 h later again. On the 6th day, they were given two subcutaneous injections of Adr at 0.005 mL/g with a four-hour interval while the control group was accordingly injected with 0.9% NS. After fasting for at least 12 h, they were weighed and sacrificed on day 7. We randomly collected orbital blood of six mice in each group first, added 3.2% sodium citrate anticoagulant into their blood (1:9, v/v) immediately, and pipetted plasma by centrifugation to detect coagulation four indices of TT, PT, APTT and FIB. Secondly, serum from the remaining mice in each group was collected for ELISA levels of HSP-70, SOD, NO, TNF-α, IL-1β, IL-6, TXB2, 6-keto-PGF1α, PGE2, ET-1, Mg2+ and Ca2+. Thirdly, all mice needed precise weighing of their spleens and thymuses. In addition, colon tissues were used to detect expression levels of the above 9 common inflammatory proteins by western blotting likewise.
Cholestasis for evaluation of E5
ANIT can cause acute liver injury and has been used to establish the model of cholestatic jaundice in many studies. Through pretesting, we determined its modeling dose of 75 mg/kg for mice. The positive drug was Ursodeoxycholic Acid Capsule in this experiment, which can be cholagogic effectively at the maximum mouse equivalent dosage of 91 mg/kg. Male mice (No. 20,180,006,009,949) were separated into an average of 8 per group. Positive and rhubarb groups were treated by prophylactic administration (0.01 mL/g) once a day for 3 days [3] while the control and model groups were given the same dose of water. On the 4th day, all mice suffered intragastrically 0.01 mL/g of ANIT that dissolved in olive oil except the control group (equal dose of olive oil), and then were administered with corresponding drug liquids once after 6 h. They were given drug liquids again on the fifth night and fasted overnight. On day 6 (i.e., 48 h after ANIT modeling), we weighed them, drawn blood to store serum, and then took the whole livers and gallbladders for exact weighing respectively. Test kits were applied to detect levels of T-SOD, MDA, GSH, Fe3+ in mouse liver and ALT, AST, ALP, GST, GGT, TBIL, DBIL, TBA, Fe3+ in serum. The liver homogenate was prepared as follows: 50 mg of liver tissues of each mouse were added with 0.9% NS (20 µL/mg) and two steel balls, and put into an automatic grinding machine for 3 min. The homogenate was placed on the ice for 30 min and centrifuged at 4 °C to get the supernatant for determination.
BP neural network correlation method
The relative content of each component in 18 rhubarb samples was correlated with 5 integration effects severally using the BP neural network. It mainly consisted of three steps. (1) Data normalization: y = (x - MinValue)/ (MaxValue - MinValue), where x and y were the values before and after conversion, and MaxValue and MinValue were the maximum and minimum values in samples. (2) Data set division: taking all component contents as input and corresponding integration effect values as output, and all of 1 ~ 18 group data as training samples, among which 90% EW-L, ethanol-S and ethanol-L groups were taken as test samples. (3) Quantity-effect correlation: based on BP neural network algorithm, adjusting various parameters to optimize the model for relevant investigation between components and efficacies, so as to reveal the influence of each component on the overall effect.
Statistical analysis
Experimental data were processed by GraphPad Prism 7.0 and SPSS 22.0, and expressed as “mean ± standard deviation (SD)”. One-way analysis of variance or independent sample t-test was carried out to compare the data between groups. Bilateral P value less than 0.05 was considered to be statistically significant.