Spectrum-Effect Relationships of Flavonoids in Glycyrrhiza uralensis Fisch.

Glycyrrhiza uralensis Fisch. is used in large quantities in traditional Chinese medicine. It contains flavonoids, saponins, and polysaccharides, with flavonoids being the main active ingredients. In this study, flavonoids were isolated from the roots of Glycyrrhiza uralensis Fisch. grown in 21 areas in China by water extraction, alcohol precipitation, polyamide resin separation, and other methods. Fingerprints were established by high performance liquid chromatography (HPLC). There were 15 common peaks in the fingerprints by similarity evaluations of the chromatographic fingerprints. The spectrum-effect relationships between the HPLC fingerprints and pharmacological activities of flavonoids in G. uralensis Fisch., including the heat clearing, detoxifying effects, cough relief, and phlegm elimination effects, were assessed by gray relational analysis and partial least squares regression. After HPLC-quadrupole time-of-flight mass spectrometry and standard comparison, these five identified compounds (liquiritin apioside, neoisoliquiritin, licochalcone A, licochalcone B, and licochalcone C) could be used to evaluate licorice quality with regard to its efficacy. This research provides a scientific basis for improving licorice quality and also establishes a model for modernization of traditional Chinese medicines.


Introduction
Glycyrrhizae Radix et Rhizoma, the dried root and rhizome of Glycyrrhiza uralensis Fisch., Glycyrrhiza inflata Bat, or Glycyrrhiza glabra L. is commonly used in traditional Chinese medicine (TCM) [1] and described in the 2015 edition of the Chinese Pharmacopoeia. Glycyrrhiza uralensis Fisch. is one of the species included in the Pharmacopoeia [2]. It has a wide distribution and is planted over large areas in Northeast and Northwest China [3]. Licorice was first recorded in e Shennong Materia Medica and was described as having effects of invigorating the spleen and qi, clearing heat and detoxification, relieving coughing and eliminating phlegm, relieving spasms and pain, and mediating the effects of various medicines [3]. In modern pharmacological studies, many active components of licorice have been observed to have a range of pharmacological activities, including hypocholesterolemic, hypoglycemic, anxiolytic, antimicrobial, antiviral, free radical scavenging, antiulcer, cytotoxic, antitumor, antiallergic, antidiabetic, anticarcinogenic, antioxidant, anti-inflammatory, and hepatoprotective activities. Licorice is widely used in food, medicines, the chemical industry, and other fields [4,5]. At present, traditional formulations of licorice from other countries, such as Japan, Iran, and South Korea, are also widely used [6][7][8]. It is produced in China and is sold in many countries worldwide. Comprehensive development and utilization of licorice resources are expanding; then the market prospects are very broad. e content of total flavonoids and specific flavonoids in licorice from different sources varies greatly [6][7][8].
According to the 2015 edition of the Chinese pharmacopoeia, there are limitations to controlling glycyrrhizin and liquiritin. However, we do not know if their content can reflect their clinical utility and can be used to control the quality of licorice. erefore, it is of great significance to establish a method to evaluate the complexity and integrity of licorice and improve quality standards. Current evaluation models lead to neglect of the interaction effects of multicomponent and multitarget TCMs and do not allow for optimum quality control of licorice. Hence, it is important to establish a method to evaluate complex samples such as licorice.
e spectrum-effect relationship is an effective method to evaluate the quality of TCMs [9][10][11]. Spectrum-effect relationship studies focus on correlations between fingerprint characteristics and pharmacodynamic data. e qualities of TCMs are evaluated using chemical components and biological activity, and the relative contributions of different components to the efficacy are determined. e results can be used to identify components that are most closely related to the pharmaceutical effect and accurately reflect the quality of a TCM.
In this study, the chromatographic fingerprints of 21 different producing areas of licorice were obtained using HPLC following existing quality control methods for crude TCMs. Common peaks were determined using similarity evaluation system software to create chromatographic fingerprints. e compounds for the selected common peaks were identified by HPLC-quadrupole time-of-flight (Q-TOF) mass spectrometry (MS) and comparison with standard samples. Data from experiments investigating the heat clearing and detoxification effects and cough relief and phlegm elimination effects were used to establish spectrumeffect relationships that were assessed using gray relational analysis (GRA) and partial least squares (PLS) regression. Flavonoid components of licorice could be used as indicators for quality control of licorice which were connected with the traditional effects identified.
is research provides a method and scientific data for comprehensively improving licorice quality control.

Reagents and Experimental
Animals. Twenty-one production areas of dried licorice (Table 1)  ). e mice were raised at room temperature (22°C ± 2°C) with a relative humidity of 50%-60% and natural light. For the experiments involving mice, they were randomly divided into a model group, blank group, and licorice treatment group with 10 mice in each group. Disease was induced in the model group but not the blank group, and both groups were given sodium carboxymethyl cellulose (CMC-Na) aqueous solution. Mice in the treatment group were continuously administered 150 mg/kg licorice total flavonoids for 1 week, and those in the blank and model groups were fed CMC-Na once per day. e blank group was not subjected to any experimental treatment.

Chromatography Condition.
e chromatographic separation was performed on a ZORBAX SB-C 18 Column (250 mm × 4.6 mm, 5 μm; Agilent Technologies) maintained at 30°C and using a diode array detector (DAD) with the detection wavelength set to 310 nm. e mobile phase was a mixture of 0.2% glacial acetic acid (A) and acetonitrile (B) with a flow rate of 1.0 mL/min. e sample injection volume was 20 μL and the sample was separated using the following e results were analyzed using Agilent Chem Station.
MS measurements were performed with a 6545 Q-TOF instrument equipped with an electrospray ionization source (Dual AJS ESI+, Agilent Technologies). e drying gas temperature and source temperature were maintained at 350°C and 120°C, respectively. e MS capillary voltage, cone voltage, and frag mentor voltage were fixed at 3500 V, 30 V, and 130 V, respectively. e drying gas flow rate was 10 L/min and the nebulizer pressure was 40 psi. e mass scan range was 50-1000 m/z.

Methodological Validation.
One sample solution was analyzed six times to determine the precision.
e RSD values of the peak area of each major chromatograph were all less than 3.5%, and the RSD values of the relative retention time were all less than 1.8%, indicating good precision of the instrument. A stability study was performed by analyzing a sample at different intervals over 1 day (0, 2, 4, 8, 12, and 24 h). e RSD values of the peak-peak area of each major chromatograph were all less than 3.0%, and the RSD values of the relative retention time were all less than 2.0%, indicating that the sample solution was stable within 24 h. Six sample solutions from the same batch were analyzed to determine the repeatability. e analysis of each sample was repeated three times. e RSD values of each main chromatographic peak area were all less than 5.0%, and the RSD values of retention time were all less than 2.4%, indicating that the method has good repeatability.

Preparation of Liquiritin Standard Solution.
Liquiritin was precisely weighed and dissolved with 70% ethanol in a 10 mL volumetric flask to prepare a solution with a concentration of 1.05 mg/mL.

Preparation of Sample Solutions.
Samples of the licorice flavonoids concentrates were weighed to 30 mg in 10 mL volumetric flasks and 70% ethanol was added. After ultrasonication for 30 mins, each sample was weighed and then stored at 4°C until required for analysis.

Standard Curve of Liquiritin.
e linearity was studied by analyzing liquiritin standard solutions with five different concentrations and determining the absorbance by UV spectrophotometry at 213 nm (after full wavelength scanning, there is the maximum absorption wavelength). Taking the absorbance value (A) as the x-axis, the concentrations of liquiritin were plotted on the y-axis to construct standard curves. e regression equation was y � 1.1299x + 3.2989. e linear range was 0.021-0.105 mg/mL and the correlation coefficient was higher than 0.999.

Determination of Total Flavonoids of 21 Production
Areas. Preparation of sample solutions and method validation for quantitative analysis of total flavonoids content were stated in the supplemental material. e absorbance values for solutions of licorice total flavonoids from the 21 growing areas were determined using the method described in Section 2.3.3.

Preparation of Sample
Solutions. Preparation method was described in Section 2.3.2.

Similarity Evaluation of the Fingerprints.
HPLC results for the samples from the 21 growing areas, including the retention times and peak areas, were exported in AIA ( * .cdf ) format in Chinese Medicine Fingerprint Similarity Evaluation (2004A Version). e median method is used to obtain the result. A representative reference fingerprint was automatically constructed using the median method by comparing the results for the 21 producing areas of licorice flavonoid extracts. Similarity values between the chromatogram for each batch and the reference fingerprint were calculated using software. en, the results were used to identify common peaks for the licorice flavonoids.

Identification of Common Peaks.
Under the conditions described in Section 2.4.2, the mixed standard and sample solutions were analyzed by HPLC-Q-TOF/MS. e spectrum of the mixed standard sample was compared with those of the sample solutions, and common peaks were identified according to the retention times and fragment ions. e flavonoid components for these peaks were inferred.

Cough Relief and Phlegm Elimination Effects
2.5.1. Phenol Red Excretion in Mice Trachea. One hour after the last dose, the mice were intraperitoneally injected with a 5% phenol red normal saline solution. After 30 min, the mice were killed by dislocation. e trachea is separated and flushed with 0.5 mL of a 5% sodium bicarbonate (NaHCO 3 ) solution [9][10][11]. We combine the flushing solution into a small test tube and centrifuge at 2000 rpm for 10 min.

Ammonia-Induced Cough Test.
irty minutes after the last treatment, the mice in the model and treatment groups were exposed to ammonia gas in a 500 mL beaker. e number of times the mice coughed in 2 min was recorded, with abdominal muscle contractions and yawning taken as coughing.

Sulfur Dioxide-Induced Cough Test.
After the last dose, the mice were exposed to sulfur dioxide in a bell jar. For each group, the cough incubation period and the number of times the mice coughed within 3 min were observed and recorded [12].

Heat Clearing and Detoxification Effects
2.6.1. Mouse Acute Paw Swelling Tests. After the last dose, 0.1 mL of a 1% carrageenan solution was subcutaneously injected into the plantar area of each mouse [13]. e degree of swelling was calculated at 0, 1, 2, 3, 4, and 5 h using the following equation: degree of swelling � foot volume at the measurement time − foot volume at 0 h.

Mouse Acute Ear Swelling Test.
After the last dose, the left ears of all the mice were uniformly coated with xylene, except for those in the blank group, which were coated with distilled water. e mice were euthanized 1 h after the ears were coated. e left and right ears of the mice were cut off, and pieces of the ear were taken from the same position with an 8 mm perforator and weighed, and the degree of swelling was calculated using the following equation: swelling � left ear mass − right ear mass.

Data Statistics.
e experimental data are expressed as the means ± SDs. e LSD and Dunnett's T3 test (3) were used for intergroup comparisons.

Statistical Analysis of the Spectrum-Effect Relationships.
Correlation analysis was conducted between the common peak areas and pharmacodynamic data of licorice flavonoids from different regions using the HPLC data by GRA and PLS method. SPSS 24.0 (International Business Machines Corporation, New York, USA) and DPS 7.05 (Zhejiang University, Hangzhou, China) were used to process the data and find common peaks that were significantly related to the pharmacological effects.

Determination of the Total Flavonoids Content in Licorice.
e total flavonoids contents in the 21 samples from different growing areas were calculated from the UV spectrophotometry results (Figure 1). e content of total flavonoids of licorice root was higher than 60% except S16. e content of licorice flavonoids was the highest in S7. Licorice has a complex chemical composition [4,14]. Flavonoids are one of the main active substances in licorice. Modern pharmacological studies have shown that it has multiple activities [15]. Studies have shown that the contents and types of licorice flavonoids from different sources vary greatly [16]. Licorice flavonoids are closely related to the traditional effects of licorice in TCM. In this study, the enrichment method can obtain relatively reliable licorice flavonoids and provide sufficient guarantee for further research.

Chromatographic Fingerprints and Similarity Analysis.
e 21 licorice samples were injected into the HPLC and chromatograms were recorded. e results are shown in Figure 2. e spectrum of sample S1 was selected as the reference spectrum, and the software automatically matched the chromatographic peaks through multipoint correction. A control fingerprint (Figure 3) was generated by the selected median method, and 15 common peaks were analyzed.
e similarity value of the generated control fingerprint (R) was set as one, and the similarities of the characteristic chromatograms of the 21 licorice samples were calculated. e origins of the licorice samples and fingerprint similarities are shown in Table 3. Sixteen samples had similarities greater than 0.9. We suspected that this difference between those samples could be caused by different producing areas of the licorice.

Test Results for Cough Relief and Phlegm Elimination.
According to the tracheal phenol red experiment results (Figure 4), phenol red excretion in the model group was significantly reduced ( ## P < 0.01) compared with the blank group, indicating successful modeling. Compared with the model group, the licorice flavonoids from samples S1-S21 significantly reduced the excretion of tracheal phenol red ( * P < 0.05 and * * P < 0.01).

Cough Relief and Phlegm Elimination Effects.
With GRA Method, 14 peaks were identified with correlation degrees greater than 0.5 in the results of the tracheal phenol red experiment (Table 4). e order of the correlation degrees of these 14 peaks was 8 > 9 > 14 > 12 > 15 > 6 > 4 > 3 > 5 > 13 > 7 > 11 > 10 > 2. ese peaks were closely related to the efficacy of licorice with regard to eliminating phlegm. From the ammonia exposure results (Table 5), 14 peaks were identified to have correlation degrees of greater than 0.5 with the frequency of coughing in mice. e contributions of these peaks were in the order 8 > 14 > 4 > 3 > 5 > 12 > 9 > 7 > 6 > 13 > 11 > 7 > 1 > 10 > 15. ese peaks were closely related to the efficacy of licorice with regard to coughing induced by ammonia. From the results of the sulfur dioxide cough tests in mice (Table 6), correlation degrees of 14 peaks were more than 0.5. ese peaks were in the order 15 > 9 > 14 > 12 > 5 > 4 > 8 > 3 > 11 > 7 > 6 > 10 > 13 > 1. ese peaks were closely related to the efficacy of licorice with regard to coughing induced by sulfur dioxide.
With the PLS regression method, the regression coefficients reflected the contribution of each x to y. e larger the absolute value of the PLS regression coefficient, the greater the contribution of the peak to medical efficacy. Peaks were negatively correlated with the absorbance; that is, as the peak intensity increased, the absorbance value decreased. e remaining peaks were positively correlated with the absorbance. e pharmacodynamic experiment results of the tracheal phenol red test and the common peak areas were fitted by the regression equation Considering the results from the GRA and PLS methods, eight chromatographic peaks (3, 4, 5, 6, 9, 12, 14, and 15) were identified as being correlated with cough relief and phlegm elimination effects.

Heat Clearing and Detoxification Effects.
With GRA method, the results of the carrageenan-induced paw swelling experiments in mice showed that 14 peaks had correlation degrees of greater than 0.5 with swelling (Table 7). e correlation degrees of these peaks were in the order 8 > 12 > 4 > 14 > 5 > 15 > 3 > 6 > 7 > 9 > 11 > 13 > 10 > 1. ese peaks were closely related to the efficacy of the drug with regard to the reduction of paw swelling. e results of the xyleneinduced ear swelling experiments in mice showed that 14 peaks had correlation degrees of greater than 0.5 with ear swelling (Table 8). e correlation degrees of these peaks were in the order 8 > 15 > 5 > 3 > 14 > 4 > 6 > 9 > 12 > 11 > 7 > 13 > 2 > 10. ese peaks were closely related to the efficacy of the drug with regards to the reduction of ear swelling.
In this study, GRA and PLS regression analysis were applied to study the spectrum-effect relationships of the 21 licorice samples. e fingerprint and pharmacological data were analyzed to determine what peaks were closely related to the licorice efficacy. e important components of licorice were identified according to their contributions to the studied pharmacological effects. Finally, five flavonoids (peaks 3, 5, 6, 14, and 15) were identified as being closely related to the cough relief and phlegm elimination effects and heat clearing and detoxification effects.

Identification of Common Peaks.
According to literature reports and expected cleavage of licorice compounds from PubChem and Sci Finder, a licorice database was established. Common peaks of the mixed standard solution and test solutions were subjected to HPLC-Q-TOF/MS in positive ion mode under the above conditions (Section 2.4.2). e analysis and matching were conducted by Agilent Mass Hunter Qualitative Analysis software using reference data, mass spectrometry data, and chromatographic retention data from the literature [21]. After this comparison, the following components were identified: liquiritin, isoliquiritin apioside, isoliquiritin, neoisoliquiritigenin, licochalcone B, isoliquiritigenin, liquiritigenin, formononetin, licochalcone C, and licochalcone A. e five peaks (3, 5, 6, 14, and 15) in the flavonoids fingerprint were closely related to its traditional efficacy, including liquiritin apioside, neoisoliquiritin, licochalcone A, licochalcone B, and licochalcone C (Table 9).

Conclusions
In this study, the HPLC spectrum-effect relationships of the licorice flavonoids for samples grown in 21 areas in China were studied by GRA and PLS regression. Five components of licorice have the largest contributions to the heat clearing and detoxification effects and phlegm elimination and cough relief effects of licorice. ese components could be used for quality control of licorice. Our results provide a data basis for further identification of common peaks and a method for improvement of licorice quality control and drug efficacy evaluations.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that they have no conflicts of interest.

Authors' Contributions
Tingting Li, Shiyao Hua, and Jiahua Ma contributed equally to this work.