The correlation between pharmacological activity and contents of eight constituents of Glycyrrhiza uralensis Fisch

Abstract

Licorice (Glycyrrhiza uralensis Fisch. (GUF), Leguminosae) has been extensively applied in traditional Chinese medicine (TCM) to treat diseases, exactly, in almost half of Chinese herbal prescription. However, the relationship between chemical contents and efficacy has not been established, which could evaluate GUF quality. To create a simple and effective quality-evaluation method, 33 batches of GUF from different habitats in China were collected. The correlation between eight constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin and glycyrrhetinic acid) and pharmacological activities (anti-inflammatory, antioxidant and immunoregulatory) was analyzed per the partial least squares regression method. Results showed that eight constituents correlated significantly with the pharmacological activity. The correlation equation modes between pharmacological activity and contents of eight constituents were constructed and verified to be reliable. In GUF extract, the main constituents liquiritin, isoliquiritin and glycyrrhizic acid exhibited positive influence on anti-inflammatory and antioxidant effect with different potent, while the metabolites liquiritigenin and isoliquiritigenin exhibited positive effect on the immunoregulatory activity and glycyrrhetinic acid exhibited positive effect on all the tested activities. Thus, our chemical-efficacy correlation method is reliable and feasible to predict the pharmacological activity based on its eight constituents. It could be powerful in quality control of GUF and provides a useful way for quality evaluation of other medicinal herbs.

1. Introduction

Licorice (roots and rhizomes of Glycyrrhiza, Leguminosae) is a well-known medicinal and edible plant. Three Chinese official licorice including Glycyrrhiza uralensis Fish. (GUF), Glycyrrhiza inflata Bat. and Glycyrrhiza glabra L. are recorded in the Pharmacopoeia of the People's Republic of China. It has been used extensively in Eastern and Western (mainly Glycyrrhiza glabra L.) countries, as a popular medicine and food additives [[1], [2], [3]]. Licorice alone or in combination prescriptions is commonly used in traditional Chinese medicine (TCM) and the world's known medical system, such as European, Russia, Indian and Egyptian medical systems [1] for treatment of respiratory, gastric and liver diseases [3] owing to its pharmacological activities [4], such as anti-inflammatory [5,6], antioxidant [7], immunoregulatory [[8], [9], [10]], antivirus [11], antimicrobial [12], antitussive, expectorant and anti-asthmatic effects [[13], [14], [15]].

GUF is used mostly in China among three licorice. The quality assessment of GUF is difficult due to its chemical and pharmacological complexity. Glycyrrhizin and liquiritin are two markers and controlled compounds of GUF according to the Pharmacopoeia of the People's Republic of China (2020). However, these two ingredients in complex components [[16], [17], [18], [19], [20]] including more than 60 kinds of triterpenoids and about 300 kinds of flavonoids are insufficient to evaluate the GUF quality. The chemical components of GUF are complex and vary from different sources, such as habitats, growing environment and harvest time [21]. Therefore, it is necessary and urgent to explore a reliable and reasonable method for quality assessment of GUF.

In general, the efficacy assay of GUF is considered to be the best indicator to represent its quality. However, it has certain limitations, such as high cost, time-consuming, limited operability, and animal ethics issues [22]. Chemical analysis has been currently employed to quality evaluation due to the ease of practical implementation. Therefore, it might be fascinating if we can combine the efficacy assays and chemical determination, to create a methodology only based on chemical determination to control the quality of GUF. Here, we try to establish a chemical-efficacy correlation approach to evaluate the quality of GUF.

There are many researches focused on the pharmacological activity of single compound of GUF. However, GUF is a complex system containing more than 60 kinds of triterpenoids and 300 kinds of flavonoids [[16], [17], [18], [19], [20]] and cannot be represented by single compound. Compared with the pharmacological activity of single compound, the evaluation of multi-component comprehensive efficacy of GUF is much closer to the reality. Multiple active constituents could more comprehensively reflect the efficacy of GUF rather than a single. Besides the most abundant active compounds glycyrrhizin and liquiritin, recent pharmacological and chemical study [16,19,20,[23], [24], [25]] of GUF showed that triterpenoids, such as glycyrrhetinic acid, the metabolite of glycyrrhizin, and flavonoids including liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, licochalcone A, and glabridin were most studied, possessed significant pharmacological activity and were relatively abundant compounds. Therefore, we selected eight main active constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizin, licochalcone A, glabridin and glycyrrhetinic acid) in this study. GUF and its relevant products have been used widely in China for treating inflammatory diseases due to their pharmacological activities against inflammation, oxidative stress and immunoregulation [4]. Therefore, in this study we focus on its main pharmacological activities (anti-inflammatory, antioxidant and immunoregulatory effect) to evaluate the quality of GUF using the in vivo and in vitro experiments on animals and cells.

To illuminate the contributions of the multiple active constituents to the pharmacological effect of GUF, and to simply predict the pharmacological activity through the more easily determination of contents of 8 constituents, the correlation between the pharmacological activities (anti-inflammatory, antioxidant and immunoregulatory) and contents of eight constituents determined by high performance liquid chromatography (HPLC) was carried out per the partial least squares regression (PLSR) method using 33 batches of GUF from different habitats in China. PLSR is a recent statistical technique that is related to principal component analysis (PCA) and multiple regression. It is to find a linear regression model by projecting the prediction variable (Y) and the observation variable (X) into a new space, calling bilinear factor models. It derives its usefulness when predicting dependent variables (Y) from a great number of independent variables (X), regardless of noisy, collinear data in both X and Y [26]. It is a powerful tool in chemometrics used increasingly in chemistry, engineering and Traditional Chinese medicine [27]. The objective of this study is to establish a feasible and simple method to predict pharmacological activities according to the contents of eight constituents. This strategy could be useful and powerful in quality assessment of the raw materials of GUF.

2. Materials and methods

2.1. Chemicals and reagents

Reference substances of monoammonium glycyrrhizate (CAS 53956-04-0, 95.0%) and glycyrrhetinic acid (CAS 471-54-4, 99.6%) were provided by National Institutes for Food and Drug Control (Beijing, China). Liquiritin (CAS 551-15-5, 95.1%), liquiritigenin (CAS 578-86-9, 96.2%), isoliquiritin (CAS 5041-81-6, 99.1%), isoliquiritigenin (CAS 961-29-5, 95.8%), licochalcone A (CAS 58749-22-7, 95.2%), and glabridin (CAS 59870-68-7, 95.7%) were provided by Beijing Shiji Aoko Biotechnology Co. Ltd (Beijing, China). HPLC-grade acetonitrile was bought from Fisher Scientific agented by Beijing Honghu Lianhe Huagong Chanpin co., Ltd (Beijing, China).

Analytical xylene, urethane and dimethyl sulfoxide (DMSO) were bought from Beijing Chemical Works (Beijing, China). Dexamethasone sodium phosphate injection was obtained from Sino Pharmaceutical Group (Henan, China). Concanavalin A (ConA) was provided by Sigma-Aldrich (Saint Louis, MO, USA). RPMI-1640 medium and penicillin-streptomycin solution (10,000 units/mL-10,000 μg/mL) were provided by HyClone (Logan, UT, USA). Mouse lymphocyte isolation fluid, l-glutathione (GSH), MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide], 0.4% Trypan Blue dye were purchased from Shanghai Macklin Biochemical Technology (Shanghai, China). The 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate (carboxy-H₂DCFDA) was bought from Genecopoeia (Rockville, MD, USA). Fetal bovine serum (FBS), 0.25% trypsin-EDTA (1 × ), 100 × minimum essential medium non-essential amino acid (NEAA) solution and other reagents were purchased from Gibco Life Technologies (Grand Island, NY, USA).

2.2. Animals

Specific pathogen-free male ICR mice (25–30 g) were provided by Beijing Weitonglihua Laboratory Animal Services Center (license: SCXK (JING)-2013-0001). Animals housed under a 12-h light-dark cycle with food and water freely. Zebrafish larvae (3 days after fertilization, AB-type) of uniform size and 2–3 cm in length were obtained from Nanjing Yao Shun Technology (Nanjing, China). Animals were acclimatized to their conditions for ∼1 week before experimentation. Animal studies were done according to rules and guidelines set by the Animal Care and Use Committee of Capital Medical University (Beijing, China). All animals used in this study received humane care.

2.3. Preparation of GUF extract

Thirty-three batches of GUF samples (S1–S33) from different habitats in China (Table 1) were collected in 2016 following the international, national and institutional rules concerning the biodiversity rights. Samples were authenticated by professor Changli Liu (School of Traditional Chinese Medicine within Capital Medical University) and deposited in number AGUF01 to AGUF33 at Beijing Key Laboratory of Traditional Chinese Medicine Collateral Disease Theory Research in Capital Medical University. GUF was crushed into coarse powder and extracted twice with 70% ethanol [28] for 1 h with 10 × and 8 × volumes of solvent, respectively. The two filtrates were combined, concentrated under vacuum (EYELAN-1100), and moved to a water bath at 70 °C to obtain the liquid extract. The latter was dried in a vacuum oven (PM400; Beijing Wuzhou Oriental Technology Development, Beijing, China) to powder. The final GUF ethanol extract (GUFEE) was weighed. The yields of dried extracts as a percentage weight of the GUF samples were in the range of 23.33 ± 1.01% to 42.70 ± 2.96% (Table 1).

Table 1.

The habitat and extraction yield of dried extract as a percentage weight of the Glycyrrhiza uralensis Fisch. samples (mean ± SD, n = 3).

Samples no.	Sample habitat	Wild/Cultivated	Yield of dried extract (%)
S1	Inner Mongolia, China	wild	35.13 ± 1.72
S2	Inner Mongolia, China	wild	31.93 ± 1.78
S3	Inner Mongolia, China	wild	33.03 ± 0.95
S4	Inner Mongolia, China	wild	31.73 ± 1.14
S5	Inner Mongolia, China	cultivated	33.09 ± 1.29
S6	Inner Mongolia, China	cultivated	23.33 ± 1.01
S7	Inner Mongolia, China	cultivated	31.17 ± 4.23
S8	Inner Mongolia, China	cultivated	29.75 ± 6.36
S9	Inner Mongolia, China	cultivated	31.62 ± 0.65
S10	Inner Mongolia, China	cultivated	35.78 ± 0.00
S11	Inner Mongolia, China	cultivated	28.94 ± 2.03
S12	Gansu, China	wild	39.07 ± 3.40
S13	Gansu, China	wild	42.70 ± 2.96
S14	Gansu, China	cultivated	29.50 ± 1.76
S15	Gansu, China	cultivated	28.22 ± 2.41
S16	Gansu, China	cultivated	28.30 ± 0.00
S17	Gansu, China	cultivated	32.80 ± 0.46
S18	Xinjiang, China	wild	33.67 ± 0.21
S19	Xinjiang, China	cultivated	27.20 ± 2.86
S20	Xinjiang, China	cultivated	29.47 ± 2.42
S21	Ningxia, China	wild	31.57 ± 2.06
S22	Ningxia, China	cultivated	33.33 ± 1.37
S23	Shanxi, China	wild	37.07 ± 3.67
S24	Shanxi, China	wild	35.25 ± 6.12
S25	Ukraine	wild	39.46 ± 0.39
S26	Inner Mongolia, China	wild	27.10 ± 0.60
S27	Inner Mongolia, China	cultivated	29.03 ± 1.76
S28	Inner Mongolia, China	cultivated	27.90 ± 1.03
S29	Gansu, China	wild	36.60 ± 0.50
S30	Xinjiang, China	wild	27.90 ± 0.20
S31	Inner Mongolia, China	wild	27.10 ± 0.56
S32	Gansu, China	cultivated	31.80 ± 3.90
S33	Heilongjiang, China	wild	30.30 ± 1.05

2.4. Determination and quantification of 8 constituents of GUFEE by HPLC

2.4.1. Analytical conditions

Fifity mg of GUFEE was dissolved in 10 mL of 70% ethanol. After filtration (0.45 μm), samples were used for HPLC analysis. Liquiritin, liquiritigenin, isoliquiritin, isoliquiritigenin, glycyrrhizate, licochalcone A, glabridin, and glycyrrhetinic acid were determined by Agilent 1100 HPLC using a diode-array UV–vis detector. Analysis was conducted on a Kromasil 100-5 C₁₈ column (250 mm × 4.6 mm, 5 μm) at 1 mL/min. The mobile phases were constituted of water containing 0.05% (v/v) phosphoric acid (A) and acetonitrile (B). A gradient mobile phase elution was used: 17–18% B for 0–13 min, 18–26% B for 13–15 min, 26–27% B for 15–23 min, 27–70% B for 23–85 min. The detection wavelength was 276 nm for 0–16 min, 360 nm for 16–25 min, 250 nm for 25–57 min, and 276 nm for 57–85 min.

2.4.2. Method validation

The precision was assessed by measuring the same sample solution successively for six times. The repeatability was determined by measuring six independently prepared sample solutions. The stability was implemented through determination of one sample solution at 0, 3, 6, 9, 12 h. Sample recovery was carried out with six independently prepared sample solutions containing 25 mg of GUFEE and reference substances of 8 constituents equal to the content of 25 mg of GUFEE. The linearity, LOD and LOQ of eight constituents were determined in mixture of 8 standards in six different concentration ranges.

2.5. Anti-inflammatory activity of GUF on inhibition of xylene-induced ear swelling in mice in vivo

Xylene-induced ear swelling was carried out following a protocol published previously [29]. Mice (n = 8 in each group) were administered (i.g.) with vehicle, GUFEE (4.0 g GUF/kg) according to effective dose equivalent of human of Gancao decoction (31.2 g) recorded in Treatise on Febrile Disease of Zhang Zhongjing in the Han Dynasty or dexamethasone (5 mg/kg) for 3 days. On day-3, 45 min after administration, xylene (30 μL) was smeared uniformly on all surface of the right ear lobe. The left ear was set as the control. After 1 h, mice were injected urethane (1 g/kg, i.p.). Then, 8-mm sections of both ears were cut at the same location and weighed. The degree and inhibition of swelling were calculated using equations shown below:

Degree of swelling = weight of right ear − weight of left ear

Inhibition of swelling (%) = [1 − swelling degree of treatment group/swelling degree of vehicle] × 100 (1)

2.6. Antioxidant activity of GUF on reactive oxygen species (ROS) clearance in zebrafish larvae in vivo

ROS in zebrafish larvae were detected using carboxy-H₂DCFDA staining according to the literature [30,31] with slight modification. Zebrafish were kept in aerated and ultraviolet light-sterilized water at 28 ± 0.5 °C in air incubators (81 M/YCP-50S; Changsha Huaxi Electronic Technology, Hunan, China) in a five-layer zebrafish-feeding system (Far East Instruments, Jiangsu, China) using a 14-h light–10-h dark cycle. They were fed flake food and brine shrimp twice daily. Spawning adult zebrafish were maintained with a male: female ratio of 2:1. Embryos were picked out within 30 min, washed with E3 culture medium (5 mM NaCl, 0.33 mM CaCl₂, 0.17 mM KCl and 0.33 mM MgSO₄) [31] and kept at 28.5 °C in a Petri dish. After 3–4 h fertilization, normally-developed embryos were picked out under a biological stereomicroscope (SZX10; Olympus, Tokyo, Japan) and kept at 28.5 °C in a Petri dish for 3 days to develop into zebrafish.

Normally-developed zebrafishes were transferred to 24-well plates (10 zebrafish per well) and exposed to GUFEE (2.5, 5 and 10 mg/L of GUF) in 2-mL of the embryo culture for 24 h (n = 3) based on our preliminary studies considering literature [32]. The larval were all alive and the GUF at three concentrations is safe to zebrafish. Considering the efficacy and security, 5 mg/L of GUF was used in the in the following test.

The following was performed in the dark. Normally-developed zebrafishes were transferred to 96-well plates, one zebrafish per well. Zebrafishes were exposed to 5 μg/mL of carboxy-H₂DCFDA with test samples (positive control, 100 μM GSH; 5 mg GUFEE/L) or without (model group) (n = 5) in 200-μL embryo cultures for 1 h at 28 °C in air incubators. After washing thrice with embryo culture, the green, fluorescent reactive oxygen species (ROS) signal was detected under a multi-function panel reader (Ex₄₉₅/Em₅₂₉) (SpectraMax M2; Thermo Scientific). ROS clearance was calculated using the following equation:

ROS clearance (%) = (1 − fluorescence of treatment group/fluorescence of model group) × 100 (2)

2.7. Immunomodulatory activity of GUF on the proliferation rete of splenic lymphocytes of mice in vitro

The impact of GUF on the proliferation rate of splenic lymphocytes was undertaken as described previously [33,34] with slight modification. Mice spleens ((B-lymphocyte source) were acquired under aseptic conditions from sacrificed mice by injection of urethane (1 g/kg, i.p.). They were grinded gently with a stylet in cell strainer containing 5 mL of RPMI-1640 to obtain homogenate. After filtration, a single-cell suspension was obtained. Addition of 2 mL of splenic lymphocyte isolation fluid the suspension was centrifuged at 800×g for 30 min at room temperature (RT). The layer of splenic lymphocytes (buffy coat) was suck out, added to 6 mL of RPMI-1640, and centrifuged at 630×g for 10 min at RT. The supernatant was discarded and 5 mL of RPMI-1640 was put into the cells. After centrifugation at 500×g for 10 min at RT, the supernatant was abandoned and 2 mL of RPMI-1640 complete medium (containing 10% FBS, 1% NEAA, 100 U/mL–100 μg/mL penicillin-streptomycin and 5 μg/mL of ConA) was put into the cells. The cells (>95% were alive according to Trypan Blue staining) were cultivated at 37 °C, 5% CO₂. After 72 h incubation, 100 μL of cell suspension (5 × 10⁵ cells/mL) was placed in 96-well plates. RPMI-1640 complete medium served as the blank control. Then, 100 μL RPMI-1640 complete medium containing GUFEE (final concentration 50 μg GUF/mL) according to literature [35] or not (control) was added to the holes (n = 6). After 48 h incubation, 20 μL of MTT solution (5 mg/mL) was put in and cells were incubated for 4 h. The supernatant was discarded and 100 μL of DMSO was put into each hole. Absorbance was read at 570 nm employing a microplate reader (SpectraMax Plus 384; Thermo Scientific, Waltham, MA, USA). The proliferation of splenic lymphocytes was calculated:

Proliferation rete (%) = (A_sample − A_control)/A_control × 100% (3)

2.8. Data analysis

2.8.1. Correlation analyses of efficacy and contents of eight constituents of GUF

Correlation analyses of GUF efficacy (anti-inflammatory, antioxidant and immunomodulatory) and contents of eight constituents were conducted using the PLSR method in Minitab 17.1 (Minitab, State College, PA, USA). The contents of 8 constituents served as the independent variable. The pharmacological activity of GUF served as a dependent variable.

2.8.2. Statistical analysis

Data was analyzed using one-way analysis of variance by SPSS v21.0 (IBM, Armonk, NY, USA). Data was shown as mean ± SD and was significant when p < 0.05.

3. Results and discussion

3.1. HPLC method validation

An effective HPLC detection method of 8 constituents of GUF simultaneously was established. The HPLC fingerprints of 33 batches of GUF samples (S1–S33) were shown in Fig. 1. The specificity of 8 constituents of GUFEE were shown in Figure S1. The method was validated by tests of precision, repeatability, stability, accuracy (sample recovery), linearity, LOD and LOQ. Precision, repeatability and stability was presented by relative standard deviation (RSD %) of eight peaks areas and retention time. The RSD values of eight peaks areas and retention time was less than 3.0% and 0.5%, respectively. The average recovery rate of 8 constituents was 93.2%–105% with RSD values less than 5.0% (Table S1).

Fig. 1.

The HPLC fingerprints of thirty-three batches of GUF samples. Analysis was conducted on a Kromasil 100-5 C₁₈ column (250 mm × 4.6 mm, 5 μm) at 1 mL/min. The gradient mobile phases constituted of water containing 0.05% (v/v) phosphoric acid (A) and acetonitrile (B) were as follows: 17–18% B for 0–13 min, 18–26% B for 13–15 min, 26–27% B for 15–23 min, 27–70% B for 23–85 min. The detection wavelength was 276 nm for 0–16 min, 360 nm for 16–25 min, 250 nm for 25–57 min, and 276 nm for 57–85 min. 1, liquiritin; 2, isoliquiritin; 3, liquiritigenin; 4, isoliquiritigenin; 5, glycyrrhizate; 6, licochalcone A; 7, glabridin; 8, glycyrrhetinic acid.

The calibration curve using the peak (Y) and the concentration (X) was excellently linear of eight constituents. The LOD (mg/L) of liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizate, licochalcone A, glabridin and glycyrrhetinic acid were 0.117, 0.0234, 0.0938, 0.0135, 0.0732, 0.312, 0.0843, 0.0848, respectively. The LOQ (mg/L) of above 8 constituents were 0.391, 0.0781, 0.375, 0.0674, 0.244, 0.781, 0.281, 0.339, respectively (Table S1). Thus, the HPLC method was sufficiently sensitive, precise and accurate for quantitative analysis of 8 constituents of GUF samples simultaneously.

3.2. The contents of eight constituents of GUF

The contents of liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid (the content of glycyrrhizate/1.02078), licochalcone A, glabridin, and glycyrrhetinic acid in 33 batches of GUF samples (S1–S33) were quantified using HPLC method and presented in Table 2. The content of one compound varied greatly from different GUF samples. Glycyrrhizic acid and liquiritin were the most abundant constituents of GUF. Glabridin was only detected in 11 samples and could not be detected in 22 sample among 33 samples of GUF. Liquiritin, isoliquiritin, isoliquiritigenin, glycyrrhizic acid were detected in all tested samples.

Table 2.

The contents of eight constituents (constituent/dried GUF, %) of 33 batches of Glycyrrhiza uralensis Fisch. samples determined by HPLC (mean ± SD, n = 3).

Samples	Liquiritin	Isoliquiritin	Liquiritigenin	Isoliquiritigenin	Glycyrrhizic acid	Licochalcone A	Glabridin	Glycyrrhetinic acid
S1	2.1196 ± 0.2819	0.3381 ± 0.0402	0.0959 ± 0.0394	0.0665 ± 0.0082	3.6889 ± 0.3305	0.1316 ± 0.0770	–	0.0126 ± 0.0112
S2	2.0589 ± 0.3890	0.3200 ± 0.0504	0.0704 ± 0.0198	0.0538 ± 0.0148	3.0066 ± 0.8333	–	–	0.0179 ± 0.0250
S3	1.4817 ± 0.2139	0.2759 ± 0.0747	0.0660 ± 0.0201	0.0257 ± 0.0106	3.9366 ± 0.3068	–	–	0.0799 ± 0.0526
S4	1.1899 ± 0.0425	0.2139 ± 0.0076	–	0.0089 ± 0.0004	2.9781 ± 0.1124	0.1269 ± 0.0210	–	0.0571 ± 0.0133
S5	0.7379 ± 0.2918	0.1327 ± 0.0551	0.0350 ± 0.0094	0.0139 ± 0.0090	2.8380 ± 0.8815	0.6184 ± 0.0365	0.1993 ± 0.1404	–
S6	0.5083 ± 0.0630	0.0690 ± 0.0135	0.0092 ± 0.0059	0.0129 ± 0.0029	1.2411 ± 0.1281	0.2015 ± 0.0237	–	0.0524 ± 0.0244
S7	1.0842 ± 0.3876	0.1662 ± 0.0592	0.0174 ± 0.0020	0.0138 ± 0.0027	1.9892 ± 0.3126	0.1734 ± 0.0082	–	0.0060 ± 0.0167
S8	0.8481 ± 0.0324	0.1954 ± 0.0227	0.0647 ± 0.0177	0.0172 ± 0.0028	2.6220 ± 0.0356	0.3861 ± 0.0147	–	–
S9	0.4177 ± 0.1289	0.0533 ± 0.1289	0.0272 ± 0.0052	0.0488 ± 0.0027	3.3424 ± 0.0248	0.1666 ± 0.0087	0.0252 ± 0.0012	0.0346 ± 0.0129
S10	0.5109 ± 0.0128	0.096 ± 0.0128	0.0534 ± 0.0003	0.0374 ± 0.001	2.9861 ± 0.0283	0.6049 ± 0.0054	0.0736 ± 0.0037	0.0141 ± 0.0014
S11	0.361 ± 0.0117	0.0655 ± 0.0039	0.0249 ± 0.0005	0.0261 ± 0.0005	1.7562 ± 0.0508	0.3092 ± 0.0132	0.063 ± 0.0028	–
S12	2.3603 ± 0.4099	0.4602 ± 0.1138	0.0273 ± 0.0072	0.0320 ± 0.0091	5.5594 ± 0.5184	–	–	0.1030 ± 0.0252
S13	3.4718 ± 0.2074	0.7164 ± 0.0096	0.0559 ± 0.0192	0.0206 ± 0.0023	6.7986 ± 0.5371	–	–	0.0172 ± 0.0156
S14	1.7470 ± 0.3995	0.2784 ± 0.0488	0.0180 ± 0.0166	0.0226 ± 0.0081	3.0571 ± 0.3628	–	–	0.0464 ± 0.0218
S15	0.8506 ± 0.2387	0.1383 ± 0.0183	0.0874 ± 0.0288	0.0291 ± 0.0080	2.5918 ± 0.5909	0.0151 ± 0.0032	–	0.0353 ± 0.0322
S16	0.5179 ± 0.0000	0.0918 ± 0.0000	0.0423 ± 0.0000	0.0217 ± 0.0000	1.7514 ± 0.0000	0.1624 ± 0.0000	–	–
S17	1.0229 ± 0.0342	0.1957 ± 0.0054	0.1705 ± 0.0027	0.0711 ± 0.0002	2.6840 ± 0.0524	0.2186 ± 0.0113	0.0152 ± 0.0016	0.0751 ± 0.0151
S18	0.1244 ± 0.0678	0.0313 ± 0.0124	0.0041 ± 0.0034	0.0442 ± 0.0067	4.6015 ± 0.2255	0.0646 ± 0.0104	0.2278 ± 0.0576	0.0224 ± 0.0150
S19	0.2932 ± 0.0526	0.0513 ± 0.0120	0.0070 ± 0.0032	0.0173 ± 0.0059	1.0348 ± 0.1474	0.4925 ± 0.0997	–	–
S20	0.2154 ± 0.0435	0.0354 ± 0.0137	0.0018 ± 0.0026	0.0222 ± 0.0035	1.9357 ± 0.1703	0.6450 ± 0.2168	0.1194 ± 0.0561	–
S21	0.8045 ± 0.1092	0.1491 ± 0.0202	0.1434 ± 0.0360	0.0448 ± 0.0106	3.2198 ± 0.1473	0.9734 ± 0.0681	–	0.0307 ± 0.0561
S22	0.8712 ± 0.3906	0.1439 ± 0.0603	0.0436 ± 0.0166	0.0165 ± 0.0048	2.5069 ± 0.4807	0.5366 ± 0.1403	0.0462 ± 0.0436	–
S23	2.0320 ± 0.3868	0.2967 ± 0.0338	–	0.0089 ± 0.0020	4.1231 ± 0.1886	0.0733 ± 0.0046	–	0.1169 ± 0.0465
S24	1.9782 ± 0.4745	0.3052 ± 0.0594	0.1855 ± 0.0115	0.0763 ± 0.0131	4.9062 ± 0.8875	0.2247 ± 0.0282	–	0.0896 ± 0.0445
S25	0.6993 ± 0.0262	0.0162 ± 0.0262	0.0284 ± 0.003	0.0406 ± 0.0026	4.8807 ± 0.0736	0.0237 ± 0.0000	0.1204 ± 0.0054	0.0054 ± 0.003
S26	1.2309 ± 0.0458	0.2838 ± 0.0091	0.2108 ± 0.0032	0.0941 ± 0.0032	3.2944 ± 0.0884	0.1313 ± 0.0073	0.0038 ± 0.0007	0.0587 ± 0.0174
S27	0.5907 ± 0.0060	0.1627 ± 0.0042	0.0927 ± 0.0012	0.0635 ± 0.0021	2.4324 ± 0.0421	0.0849 ± 0.0053	0.1629 ± 0.0062	0.0226 ± 0.0013
S28	0.5712 ± 0.0080	0.1266 ± 0.0062	0.066 ± 0.0035	0.0538 ± 0.0013	2.9153 ± 0.0531	0.5233 ± 0.0201	–	0.0194 ± 0.0013
S29	0.1072 ± 0.0071	0.0454 ± 0.0071	0.0312 ± 0.0019	0.0499 ± 0.0010	4.4710 ± 0.0193	0.0553 ± 0.0020	–	0.0603 ± 0.0233
S30	0.5872 ± 0.0094	0.1575 ± 0.0094	0.1689 ± 0.0065	0.1064 ± 0.0052	5.5675 ± 0.1068	0.6244 ± 0.0147	–	0.0525 ± 0.0039
S31	1.2610 ± 0.0000	0.2326 ± 0.0000	0.1459 ± 0.0000	0.0300 ± 0.0000	3.0761 ± 0.0000	0.0781 ± 0.0000	–	0.0823 ± 0.0000
S32	1.3014 ± 0.3622	0.3014 ± 0.0442	0.7262 ± 0.2233	0.1524 ± 0.0473	4.3449 ± 0.3331	0.2032 ± 0.1314	–	0.1753 ± 0.0442
S33	1.1627 ± 0.0000	0.2437 ± 0.0000	0.3707 ± 0.0000	0.0788 ± 0.0000	2.8127 ± 0.0000	0.1076 ± 0.0000	–	0.0969 ± 0.0000

“—“, No detection.

3.3. Correlation analysis of contents of eight constituents and anti-inflammatory effect of GUF (S1–S25) on inhibition of xylene-induced ear swelling in mice

The right ears of mice in the vehicle control group showed significant inflammatory symptoms, containing redness, and swelling and exhibited a significant increase in ear weight. In GUFEE (4.0 g GUF/kg)- and positive control (dexamethasone, 5 mg/kg)-treated groups, xylene-induced ear edema in mice was decreased markedly (p < 0.05) (Fig. 2A1). Notably, the anti-inflammatory effect of S1, S3, S4, S12–S15, S17, S23, S24, S26–S33 was about the same as or stronger (S32) than that of dexamethasone (a well-known anti-inflammatory drug). The swelling inhibition was over 21.4%, and that of half groups was over 55.0% (Fig. 2A2). These data (also showed as Table S2) demonstrated that GUFEE had an excellent anti-inflammatory effect on acute inflammation induced by xylene.

Fig. 2.

Construction (S1–S25) and verification (S26–S33) of the correlation model between the anti-inflammatory indices and eight constituents of 33 batches of GUF A, anti-inflammatory effects of 33 batches of GUF (4.0 g GUF/kg) on xylene-induced ear swelling in mice, A1, swelling degree (g); A2, swelling inhibition (%). BC, blank control (saline); PC, positive control (dexamethasone, 5 mg/kg). Contrasted with blank control, *p < 0.05, **p < 0.01; Contrasted with positive control, #p < 0.05, ##p < 0.01. B, the standardized correlation coefficients between eight constituents and inhibition of xylene-induced ear swelling per PLSR analysis. PLSR, partial least squares regression; LQ, Liquiritin; ILQ, isoliquiritin; LQG, liquiritigenin; ILQG, isoliquiritigenin; GL, glycyrrhizate; LA, licochalcone A; GLB, glabridin; GA, glycyrrhetinic acid. C and D, verification of the PLSR correlation models between anti-inflammatory activity and 8 contituents of GUF in modeling samples (C, S1–S25) and verification samples (D, S26–S33). Y actual versus Y predicted values of the inhibition of xylene-induced ear swelling.

Correlation analysis between anti-inflammatory effect and the contents of eight constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin and glycyrrhetinic acid) of 25 batches of GUF (S1–S25) were done using the PLSR method. The regression coefficient can be used to exhibit the influence of 8 constituents (as independent variable) upon GUF efficacy (as dependent variable). In the PLSR model, if p < 0.05, the constituent was correlated with GUF efficacy. If the coefficient of one constituent index >0, this constituent had a positive effect upon GUF efficacy. As shown in Fig. 2C, six constituents had a positive influence on inhibition of xylene-induced ear swelling model (liquiritin > glycyrrhizic acid > glycyrrhetinic acid > isoliquiritin > isoliquiritigenin > liquiritigenin). Licochalcone A and glabridin exhibited negative influence on anti-inflammatory effect. This is consistent of many literature [4,6,13,28,36], which reported triterpenes such as glycyrrhizic acid and glycyrrhetinic acid, and flavonoids such as liquiritin, isoliquiritin, isoliquiritigenin, and liquiritigenin possessed anti-inflammatory activity. However, licochalcone A [15] and glabridin [36] had no consistent results with its single form. A competition or interaction of constituents might exist in the GUFEE.

The correlation model was constructed by correlating the contents of eight main constituents with inhibition of xylene-induced ear swelling in mice. Results (Table 3) demonstrated that all the eight constituents were correlated significantly (p = 0.000) with anti-inflammatory effect. Liquiritin, glycyrrhizic acid, glycyrrhetinic acid, isoliquiritin, isoliquiritigenin, and liquiritigenin made great contribution on inhibition of xylene-induced ear swelling effect.

Table 3.

Correlation model between pharmacodynamic index and contents of eight constituents of Glycyrrhiza uralensis Fisch.

Y	X	Correlation model	p
A	Content	Y = 3.5499X1 + 16.1992X2 + 28.0386X3 + 73.0419X4 + 2.1336X5 − 11.2747X6 − 22.8849X7 + 82.9789X8 + 32.4052	0.000a
B		Y = 0.7614X1 + 3.3461X2 − 0.6084X3 + 9.5628X4 + 0.3704X5 − 1.6218X6 − 5.7188X7 + 12.9350X8 + 12.2383	0.002a
C		Y = −0.3218X1 − 2.9950X2 + 23.7804X3 + 69.8957X4 − 0.3606X5 − 1.3708X6 − 15.2807X7 + 40.4627X8 + 7.4389	0.000a

It was reported that triterpenes such as glycyrrhizic acid and glycyrrhetinic acid, and flavonoids such as licochalcone A isoliquiritigenin, glabridin, all have been reported to possess anti-inflammatory activity.

^a, p < 0.01; A, Inhibition of xylene-induced ear swelling; B, ROS clearance in zebrafish; C, Proliferation of splenic lymphocytes. X1-X8, the contents of liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin, and glycyrrhetinic acid, respectively.

3.4. Correlation analysis of contents of eight constituents and antioxidant effect of GUF (S1–S25) on ROS clearance in zebrafish larvae in vivo

To study the scavenging capacity of GUFEE against the ROS generation in zebrafish larvae, an oxidation-sensitive fluorescent indicator (carboxy-H₂DCFDA) was utilized. Compared with the positive control (100 μM GSH) with ROS clearance of 18.26 ± 5.08%, ROS clearance in zebrafish larvae of GUFEE samples was not significantly different except for S5, S8 S19 and S20 groups (Fig. 3A). These results suggested that GUFEE could clear ROS in zebrafish larvae and that the scavenging ability of most GUFEE samples was about the same as that of GSH. Our pre-experiment for concentration selection of GUFEE (10, 30, 50, 80, 100, 100 mg/L of GUF) showed that zebrafish embryos exposed to 30–1000 mg/L of GUF were dead while exposed 10 mg/L of GUF were developed to normal zebrafish larvae during the 24 h observation. Finally, 5 mg/L of GUF were used in this experiment considering security of the concentration selection of GUFEE (2.5, 5 and 10 mg/L of GUF) and efficiency. The concentration of GUFEE (5 mg/L of GUF) in this study is lower than literature [32].

Fig. 3.

Construction and verification of the correlation model between the antioxidant indices and eight constituents of 33 batches of GUF A, antioxidant effects of 33 batches of GUF (5 mg GUF/L) on ROS clearance (%) in zebrafish larvae in vivo (mean ± SD, n = 5). PC, positive control (100 μM GSH). Contrasted with positive control, #p < 0.05, ##p < 0.01. B, the standardized correlation coefficients between eight constituents and reactive oxygen species (ROS) clearance in zebrafish larvae per PLSR analysis. PLSR, partial least squares regression; LQ, Liquiritin; ILQ, isoliquiritin; LQG, liquiritigenin; ILQG, isoliquiritigenin; GL, glycyrrhizate; LA, licochalcone A; GLB, glabridin; GA, glycyrrhetinic acid. C and D, verification of the PLSR correlation models between antioxidant activity and 8 contituents of GUF in modeling samples (C, S1–S25) and verification samples (D, S26–S33). Y actual versus Y predicted values of ROS clearance in zebrafish larvae.

Correlation analysis between the antioxidant activity and contents of eight constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin and glycyrrhetinic acid) of 25 batches of GUF (S1–S25) were done. Eight constituents were correlated significantly (p = 0.002) with the antioxidant activity (Table 3). Results (Fig. 3B) showed that five constituents had a positive effect on ROS clearance in zebrafish model (liquiritin > isoliquiritin > glycyrrhizic acid > glycyrrhetinic acid > isoliquiritigenin). Liquiritigenin, licochalcone A and glabridin had negative effect on the antioxidant activity. The correlation model was created (Table 3) with PLSR. Liquiritin, isoliquiritin, glycyrrhizic acid and glycyrrhetinic acid played key role in ROS clearance in zebrafish larvae. This is in accord with most reports [4,6,7,19].

3.5. Correlation analysis of eight constituents and immunoregulatory effect of GUF (S1–S25) on proliferation of splenic lymphocytes of mice in vitro

Splenic lymphocytes reflect the immunocompetence of cells in body directly. Therefore, the influence of GUF on the multiplication capacity of lymphocytes was tested. The proliferation of live T lymphocytes non-specifically stimulated by the mitogen ConA was measured.

Compared with the normal control group, the percentage of splenic lymphocytes in the GUFEE (50 μg/L of GUF) group increased significantly. Proliferation of GUFEE groups (except groups S5 and S22) was >5.0% and that of group S24 reached 18.08 ± 2.50% (Fig. 4A). These results suggested that most GUFEE enhanced the proliferation of splenic lymphocytes induced by ConA significantly.

Fig. 4.

Construction and verification of the correlation model between the immunoregulatory indices and eight constituents of 33 batches of GUF A, immunoregulatory effects of 33 batches of GUF (50 μg GUF/mL) on proliferation of splenic lymphocytes of mice in vivo (mean ± SD, n = 6). B, the standardized correlation coefficients between eight constituents and proliferation of splenic lymphocytes of mice per PLSR analysis. PLSR, partial least squares regression; LQ, Liquiritin; ILQ, isoliquiritin; LQG, liquiritigenin; ILQG, isoliquiritigenin; GL, glycyrrhizate; LA, licochalcone A; GLB, glabridin; GA, glycyrrhetinic acid. C and D, verification of the PLSR correlation models between immunoregulatory activity and 8 contituents of GUF in modeling samples (C, S1–S25) and verification samples (D, S26–S33). Y actual versus Y predicted values of proliferation of splenic lymphocytes of mice.

Correlation analysis between the immunoregulatory activity (the proliferation of splenic lymphocytes) and contents of eight main constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin and glycyrrhetinic acid) of 25 batches of GUF (S1–S25) were done using the PLSR method. The regression coefficient of glycyrrhetinic acid, isoliquiritigenin and liquiritigenin were above zero (Fig. 4B), demonstrating that they had a positive influence on the proliferation of splenic lymphocytes (glycyrrhetinic acid > isoliquiritigenin > liquiritigenin). The regression model was finally created (Table 3). Glycyrrhetinic acid, isoliquiritigenin, and liquiritigenin, the metabolism of glycyrrhizic acid, liquiritin and isoliquiritin, respectively, played a vital role in the proliferation of splenic lymphocytes while glabridin, glycyrrhizic acid, isoliquiritin, licochalcone A, and liquiritin inhibited the proliferation of splenic lymphocytes.

3.6. Verification of the reliability of our regression model

To confirm the reliability of our regression models, the actual pharmacological activities (Fig. 2, Fig. 3, Fig. 4 A) and contents of eight constituents (Table 2) of the other eight batches of GUF (S26–S33) from different habitats (Table 1) were detected. The predicted pharmacological values were calculated from the regression equation (Table 3) in which contents of eight constituents served as an independent variable. Some important parameters for instance, the root mean square error of prediction (RMSEP), explanatory ability (R²Y) and relative error (RE) of actual and calculated pharmacological activity were calculated to verify the reliability of our regression models.

The predicted curves of the PLSR modeling samples (S1–S25) (Fig. 2, Fig. 3, Fig. 4C) and verification samples (S26–S33) (Fig. 2, Fig. 3, Fig. 4D) were established to confirm the reliability and effectiveness of the PLSR models. In the anti-inflammatory model, the explanatory ability R²Y = 0.8551 and RMSEP = 0.055145 (Fig. 2C) for the modeling samples, suggesting that the model was applicable. And R²Y = 0.9916 and RMSEP = 0.016669 (Fig. 2D) for the verification samples, indicating that the established PLSR model was reliable to predict the anti-inflammatory of GUF. The RE (Mean ± SE) of actual and calculated anti-inflammatory of modeling samples (S1–S25) and verification samples (S26–S33) was 10.74 ± 11.37% and 2.45 ± 2.21%, respectively, further indicating that the PLSR model of the anti-inflammatory was appropriate. Therefore, from the anti-inflammatory point of view, the quality of eight verification samples were S32 > S33 > S26 > S31 > S30 > S29 > S28/S27 (Fig. 2C). Intriguingly, although S30 had the highest content of glycyrrhizic acid (Table 2), the most abundant in GUF, it had not the best anti-inflammatory activity among eight verification samples. This might due to the comprehensive effect of constituents not only the mainly one ingredient.

In the antioxidant model, for the modeling samples the explanatory ability R²Y = 0.8192 and RMSEP = 0.019570 (Fig. 3C), for the verification samples R²Y = 0.9087 and RMSEP = 0.004629 (Fig. 3D), demonstrating that the PLSR mode was considerably reliable to predict the antioxidant activity of GUF. Fig. 4C showed the predicted curve of the immunoregulatory PLSR modeling samples (R²Y = 0.8395, RMSEP = 0.015088). The predicted plot of the immunoregulatory PLSR verification samples was depicted in Fig. 4D (R²Y = 0.9816, RMSEP = 0.016315). These results indicated that the established PLSR model was feasible to predict the immunoregulatory activity of GUF. The RE (Mean ± SE) of actual and calculated activity in modeling and verification samples of antioxidant model was 11.67 ± 13.62% and 2.38 ± 2.03%, of immunoregulatory model was 17.32 ± 20.81% and 8.75 ± 8.55%, respectively. These results further suggested that the created PLSR models were properly reliable to predict the antioxidant and immunoregulatory activity of GUF.

This study mainly focused on the correlation between the activity and contents of 8 main constituents and was to find out a powerful method to quality assessment of GUF. As for the dose of GUF in the activity test, we reference the clinical dose (anti-inflammatory), previous reports (immunoregulatory) [35] and pre-experiment (antioxidant). Moreover, the concentrations of constituents in different GUF samples vary greatly. In fact, each of 8 main constituents in 33 different GUF samples did have 33 concentrations (Table 3). As a result, one constituent had 33 doses in every pharmacological activity. Therefore, our study chose one dose of 33 batches GUF to performance. This easy and quick implementation is exactly the advantage of our method comparing with the common IC50.

PLSR is a powerful multivariate method [37] to quantitatively model the complex relationships between independent variable (X) and dependent variable (Y) [26]. It is used to solve many problems that cannot be solved by conventional multiple regression and is appropriate to investigate the quality control of Chinese herbs or TCM overcoming the complexity of chemical compositions and efficacy. It was used successfully to recognize and optimize antimigraine ingredients from Wuzhuyu Decoction [38] and identify quality markers in Chuanxiong Rhizoma and Cyperi Rhizoma herbal pair [39] in our previous work. Here we employed the PLSR method to analyze the correlation between the pharmacological activity and concentration of 8 constituents in GUF.

One important discovery in this study was that a novel chemical-efficacy correlation method to assay the quality of GUF was created. The correlation between contents of eight main constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin and glycyrrhetinic acid) and three prime efficacies (anti-inflammatory, antioxidant and immunoregulatory) of 33 GUF samples from different habits of China (Table 1) was analyzed via PLSR. The correlation modes were constructed and verified. Eight constituents correlated significantly with three pharmacological activities. We can directly use the model equation to predict the anti-inflammatory, antioxidant, and immunoregulatory effect of GUF from its contents of eight constituents, which served as independent variable. It was feasible and powerful in quality evaluation of GUF and provided an accurate, precise, and simple strategy to quality evaluation of other medicinal herbs or plants. Furthermore, an effective HPLC method of detecting 8 constituents simultaneously was established. It could be referenced for its quality control.

The greatest advantage of our correlation modes is that it is simple and closer to reality of the complex systems of GUF extract. We can predict the three pharmacological activities (the anti-inflammatory, antioxidant, and immunoregulatory) from the correlation modes directly based on the contents of eight constituents of GUF without cumbersomely pharmacological tests. The contents of eight constituents of GUF are easier to acquire thanks to the development of related analysis technology. Compared with the single or pure compound research, our research focus on the GUFEE, a comprehensive and complex system of GUF, and evaluation of multi-component (8 constituents) comprehensive efficacy. This is much closer to the reality.

Moreover, the contribution of 8 constituents to pharmacological activity was elucidated. Interestingly, liquiritin, isoliquiritin, glycyrrhizic acid and glycyrrhetinic acid had a key impact on anti-inflammatory and antioxidant effect of GUF. Glycyrrhetinic acid, the metabolism of the main constituent glycyrrhizic acid, exhibited preeminent capacity on all tested activities. Liquiritigenin and isoliquiritigenin, the metabolism of the main constituent liquiritin and isoliquiritin, respectively, displayed outstanding immunoregulatory effect on proliferation of splenic lymphocytes. Licochalcone A and glabridin showed negative impact on tested activities. Some detected activities of 8 constituents were constituent with literature [4,7,13]. However, some reported activities [8] was not detected for the corresponding compounds in this study. The potential reason might be involved in the different system in which the compound existed. In our study eight constituents existed together in GUFEE, a mixture of compounds in GUF extract. While the pure compound was used in literature. This environmental difference might be the reason that the pharmacological effect of each constituent in GUFEE was different from the single compound.

However, another important pharmacological activity of GUF is not implemented in this study due to the time and economic constraints. Although 33 samples of GUF could meet the modeling requirements and were used to establish the models. The samples of GUF are the more, the better and should be expanded in further research to improve the credibility of the model.

PLSR method was employed to analyze the correlation between the pharmacological activity and constituents of GUF. The regression models were constructed to predict the anti-inflammatory, antioxidant and immunoregulatory effect of GUF from its contents of eight constituents. This novel chemical-efficacy correlation method was useful and important to assay the quality of GUF through its simple content determination of 8 constituents simultaneously.

4. Conclusions

The correlation models of eight constituents (liquiritin, isoliquiritin, liquiritigenin, isoliquiritigenin, glycyrrhizic acid, licochalcone A, glabridin and glycyrrhetinic acid) and three pharmacological activities (anti-inflammatory, antioxidant and immunoregulatory) were constructed. Hence, our method is feasible to predict the pharmacological activity of GUF from its contents of 8 constituents. Liquiritin, isoliquiritin and glycyrrhizic acid had positive influence on anti-inflammatory and antioxidant effect with different potent, while their metabolites liquiritigenin, isoliquiritigenin and glycyrrhetinic acid had positive effect on the immunoregulatory activity. Interestingly, glycyrrhetinic acid exhibited preeminent capacity on all tested activities. The novel chemical-efficacy correlation method could be used to assay the quality of GUF. In conclusion, here a novel chemical-efficacy correlation method per PLSR was created to evaluate the quality of GUF. This powerful strategy provides a simple way for quality evaluation of other medicinal herbs or plants.

Author contribution statement

Muxin Gong: Conceived and designed the experiments; Contributed reagents, materials, analysis tools or data.

Rui He: Performed the experiments; Contributed reagents, materials, analysis tools or data; Wrote the paper.

Ting-Ting Ma: Performed the experiments.

Kai-Li Xie: Analyzed and interpreted the data.

Zhi-Min Wang: Contributed reagents, materials, analysis tools or data.

Jing Li: Analyzed and interpreted the data; Contributed reagents, materials, analysis tools or data.

Funding statement

Professor Mu-Xin Gong was supported by National Key R&D Program of China [2017YFC1701900]; National Natural Science Foundation of China [81773860]; the special scientific research of traditional Chinese medicine industry in 2015 [201507002-03-01-04].

Rui He was supported by National Natural Science Foundation of China [81803679].

Jing Li was supported by R&D Program of Beijing Municipal Education Commission [KM201910025021].

Data availability statement

Data will be made available on request.

Additional information

Supplementary content related to this article has been published online at [URL].

Declaration of competing interest

The authors have no conflicts of interest to declare.

Abbreviations

GUF, Glycyrrhiza uralensis Fisch.; HPLC, high performance liquid chromatography; PLSR, partial least squares regression; DMSO, dimethyl sulfoxide; ConA, Concanavalin A; GSH, l-glutathione; MTT, 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide; carboxy-H2DCFDA, 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate; FBS, Fetal bovine serum; NEAA, non-essential amino acid; GUFEE, GUF ethanol extract; ROS, reactive oxygen species; RSD, relative standard deviation; RMSEP, the root mean square error of prediction; RE, relative error.

Appendix B. Supplementary data

The following are the supplementary data related to this article:

figs1.