A total of 18 batches of CR were obtained from four plant sources, and the specific information is shown in Table 1 . Professor Xian-you Qu from Chongqing Academy of Chinese Materia Medica and senior agronomist Wei Dai from Mianyang Academy of Agricultural Sciences identified all CR samples. Chemical reference substances (CRS), such as magnoflorine, tetrahydricheilanthifolinium, coptisine, epiberberine, columbamine, jatrorrhizine, berberrubine, berberine, palmatine, berlambine (purity ≥98%, HPLC grade), were purchased from Shanghai Yuanye Bio-Technology Co., Ltd. (Shanghai, China). The HPLC-graded methanol, acetonitrile, formic acid, and other chemical reagents were obtained from Fisher Scientific (Pittsburgh, PA, USA). The water was obtained from a Milli-Q water purification system (Millipore, Bedford, MA, USA).

The CRS solutions were accurately weighed and dissolved in methanol. Their concentrations for working solution were as follows: magnoflorine (6.34 μg/mL), tetrahydricheilanthifolinium (5.38 μg/mL), coptisine (36.5 μg/mL), epiberberine (34.7 μg/mL), columbamine (5.82 μg/mL), jatrorrhizine (12.35 μg/mL), berberrubine (4.05 μg/ml), berberine (32.3 μg/mL), palmatine (27.3 μg/mL), berlambine (2.86 μg/mL).

Accurately weighed 0.2 g CR powder and placed it in a stoppered conical flask, accurately added 50 mL of methanol-hydrochloric acid (100: 1) solution to the flask followed by ultrasonic treatment for 30 min. Made up for the mass loss with methanol and centrifuged it. Took 2 mL of filtrate and put it into a 10 mL volumetric flask and set volume with methanol; Then, filtered through a 0.22 μm micropore membrane. The filtrate was used for liquid chromatography tandem mass spectrometry (LC-MS) analysis.

The samples used for the electronic tongue test were prepared as follows: weighed accurately 3.0 g of CR powder and placed in a conical flask, added 50 mL of ultrapure water; Extracted by ultrasonic for 30 min, then centrifuged at 4000 rpm for 5 min. The supernatant was taken for electronic tongue analysis. Potassium chloride and tartaric acid were taken as the reference solution for the electronic tongue. The response values of each sensor of the reference solution were taken as the base point. Since the reference solution contained acid and salt components, the base points for sourness and saltiness were -13 and -6, and the base points for the rest of the tastes were zero. All solutions were stored in a refrigerator at 4°C until analysis.

LC-MS analysis was performed on an Agilent 1290 series ultra-high performance liquid chromatography system tandem 6460 triple quadrupole mass spectrometer (Agilent Technologies, Santa Clara, CA, USA), which was previously reported by our team (He et al., 2022). In brief, the chromatographic separation was carried out at 30°C on a Waters Acquity UPLC CSH C₁₈ column (2.1×100 mm, 1.7 μm, Waters, Milford, MA, USA). A mobile phase consisting of 0.1% acetic acid (A) and acetonitrile (B) in a gradient elution manner as follows: 10-14% B at 0-3 min, 14-16% B at 3-9 min, 16-25% B at 9-13 min, 25-80% B at 13-14 min, 80% B at 14-16 min. The sample injection volume was 2 μL, and the flow rate was 0.35 mL/min. Mass detection was performed on an electrospray ionization (ESI) source. The mass spectrum was chosen in positive mode, and the mass spectra were acquired in multiple reaction monitoring (MRM) mode. The drying gas (N₂) flow rate was 11.0 L/min; the drying gas temperature was 300°C; the nebulizer was 15 psig, and the capillary voltage was 4000 V. The parent ion, daughter ion, fragment voltage, and collision energy were adjusted to obtain the highest abundance of each analyte. The method validation was investigated according to the Chinese Pharmacopoeia (China Pharmacopoeia Commission, 2020), and the results met the requirement of sample chemical analysis.

SA-402B taste analysis system (Intelligent Sensor Technology, Inc., Japan) was adopted to evaluate the taste characteristics of CR in this study. It contained five taste sensors ( Table 2 ) and two reference electrodes (Ag/AgCl).

The electronic tongue was activated, calibrated, and equilibrated to the available conditions; then samples were measured according to the recently reported method (Wang JW. et al., 2024). Briefly, the electronic tongue sensor determined the membrane potential of each sample, collecting one data per second for 30 s with a stirring rate of 1 r/s. It was washed in the reference solution for 3 s and inserted into a new reference solution to test the retort for 30 s. Each sample was measured in parallel four times and analyzed.

The precision of the instrument was investigated. The test solution of sample CC-1 was taken, and the sample was repeated five times using the electronic tongue to measure the response value of each sensor and calculate the relative standard deviation (RSD). The RSD of each sensor response value was less than 2.0%, indicating good precision of the instrument.

The results of tastes and alkaloid contents of CR were subjected to orthogonal partial least squares discriminant analysis (OPLS-DA) as well as cluster analysis using SIMCA 18.0 software (Umetrics AB, Umea, Sweden). Pearson’s correlation analysis between the taste values and alkaloid contents of CR was performed by the Scientific Platform Serving for Statistics Professional (SPSSPRO) platform (Scientific Platform Serving for Statistics Professional, 2021).

The UHPLC-QQQ-MS/MS method successfully determined the contents of ten alkaloids in eighteen batches of CRs from four species. The chromatograms of the samples are shown in ( Figure 1 ), and the content results are shown in Table 3 . Epiberberine was not found in CRs derived from CT and CJ. Berberine is the alkaloid with the highest average content (7.95 ± 2.42%) in all CR samples.

The five sensors of the SA-402B taste analysis system successfully detected eight tastes in 18 batches of CR samples. The taste indices of CR are achieved through quantification and visualization ( Figure 2 ). There were differences in the different taste values for CR. Umami, sourness, and richness were the tastes with CR’s top three response values. The mean value of saltiness for all samples, as well as that of bitterness for CJ, was below the no-taste point of the sensor response value. Unfortunately, the radar chart ( Figure 2 ) cannot distinguish well among the different species of CR, which needs to be further analyzed by statistical methods.

The OPLS-DA of tastes and alkaloid contents for CR were well performed by SIMCA 18.0 software, respectively. Figures 3A, C show that the taste results determined by the electronic tongue can successfully differentiate four species of CR (R2X 0.928, R2Y 0.78, Q2 0.395). Aftertaste-astringency (aftertaste-A), sourness, bitterness, and richness played the key roles (VIP >1). As well as, Figures 3B, D show that the alkaloid contents of CR determined by UHPLC-QQQ-MS/MS can also be used to differentiate CR from the four species (R2X 0.928, R2Y 0.671, Q2 0.57), with jatrorrhizine, columbamine, and magnoflorine playing the key roles (VIP >1).

A hierarchical cluster analysis of tastes and alkaloid contents for CR was performed by SIMCA 18.0 software, respectively. HCA result shows that the taste results ( Figure 4A ) and alkaloid contents ( Figure 4B ) successfully distinguished the four groups, respectively. The grouping results were consistent with those of CR species, which was the same as the results of Clause 3.3.

Pearson’s correlation analysis of tastes and alkaloid contents for CR was carried out using the SPSSRO platform, and a significant correlation was observed in the results in Table 4 . Uncommonly, the bitterness of CR was negatively correlated with jatrorrhizine (P <0.05), while aftertaste-bitterness (aftertaste-B) was positively correlated with berberrubine (P <0.05). The results suggested that the bitterness of CR was mainly manifested as aftertaste-B, indicating the higher the aftertaste-B value, the higher the berberrubine content. In addition, astringency was positively correlated with coptisine and palmatine (P <0.05) and significantly positively correlated with epiberberine and berberine (P <0.01); meanwhile, aftertaste-A was significantly positively correlated with coptisine, epiberberine, berberine, and palmatine (P <0.01), suggesting that astringency and aftertaste-A could be used as quality evaluation indexes of CR. Moreover, saltiness did not correlate with alkaloids, which is consistent with the saltiness result of the electronic tongue in CR.