Green ultrasound‑assisted deep eutectic solvent extraction of flavonoid enzyme inhibitors from Blumea aromatica: process optimization, characterization, and mechanistic insights into α‑glucosidase and tyrosinase inhibition

Abstract

A green ultrasound-assisted deep eutectic solvent (UAE–DES) method was optimised for extracting flavonoid enzyme inhibitors from Blumea aromatica. Optimal conditions (choline chloride-1,4-butanediol 1:3 molar ratio, 43% water content, 50 mL/g liquid-to-solid ratio, 80 °C ultrasound for 48 min) yielded 3.15% total flavonoids, 45.2% higher than 50% ethanol extraction. Scanning electron microscopy confirmed cell wall disruption. The UAE–DES extract showed the strongest enzyme inhibition among all extracts tested (IC₅₀ 35.872 ± 0.294 µg/mL for α-glucosidase, 9.126 ± 0.285 μg/mL for tyrosinase), though the α-glucosidase inhibition was much weaker than acarbose, while tyrosinase inhibition was comparable to kojic acid. Six flavonoids were identified via UPLC-Q-Orbitrap HRMS, including scutellarein and corylin. Molecular docking revealed strong binding affinities (≤ −5 kcal/mol), with corylin showing the highest binding to both enzymes through hydrogen bonds and van der Waals interactions. This approach supports sustainable discovery of natural enzyme inhibitors for antidiabetic and skin-whitening applications.

1. Introduction

Blumea aromatica DC. (Asteraceae) is a traditional medicinal plant widely distributed across southern China and parts of Southeast Asia, including Myanmar, Thailand, and India¹. In Chinese ethnomedicine, its aerial parts have long been used to relieve rheumatic pain, treat eczema, and alleviate joint dysfunction². Phytochemical investigations have primarily focused on its terpenoids, which exhibit anti-inflammatory and immunomodulatory activities^3,4. Our previous work also highlighted the antioxidant potential of its caffeoylquinic acid derivatives⁵. However, limited attention has been paid to the flavonoid components of B. aromatica, despite their recognised role as key bioactive constituents in other Blumea species^6,7. To date, neither the chemical profile nor the bioactivities of these flavonoids have been comprehensively characterised.

Flavonoids are a structurally diverse group of polyphenolic secondary metabolites widely distributed in vegetables, fruits, and medicinal herbs. Due to their hydroxylated aromatic ring structures, they exhibit a broad spectrum of biological activities, including antioxidant, antidiabetic, anti-inflammatory, antiviral, and enzyme inhibitory effects^8,9. Notably, flavonoids can inhibit α-glucosidase, which plays a crucial role in carbohydrate digestion, as well as tyrosinase, the rate-limiting enzyme in melanin biosynthesis. Inhibition of α-glucosidase helps delay glucose absorption and thereby attenuates postprandial hyperglycaemia^10,11, while tyrosinase inhibition is of therapeutic interest for managing hyperpigmentation and is widely applied in cosmetic formulations^12–14. Several studies have reported that specific flavonoids, such as coreopsin, emodin, and luteolin, possess stronger inhibitory activities than standard drugs including acarbose and kojic acid^15,16.

Efficient recovery of flavonoids from plant matrices remains challenging due to their structural complexity and sensitivity to extraction conditions. Conventional extraction techniques using organic solvents, such as ethanol and methanol, often suffer from high energy consumption, low selectivity, and environmental concerns^17–19. As a sustainable alternative, deep eutectic solvents (DESs) have gained increasing attention. These solvents are formed through hydrogen-bond interactions between a hydrogen bond donor and a hydrogen bond acceptor, and they offer advantages such as adjustable polarity, low toxicity, and strong affinity for polyphenolic compounds^20,21. When applied in combination with ultrasound-assisted extraction (UAE), DES systems can significantly enhance mass transfer and solute diffusivity by inducing cavitation, which disrupts plant cell walls and thereby improves flavonoid yield and bioactivity^22,23. Recent reports have demonstrated the potential of UAE–DES systems in extracting phenolics and flavonoids from various plant sources^24–28. Despite increasing interest in green extraction technologies, optimised protocols and mechanistic insights tailored to B. aromatica are still lacking. Furthermore, the inhibitory activities of its flavonoid-rich extracts against key enzymes such as α-glucosidase and tyrosinase remain largely unexplored, limiting their potential therapeutic and cosmetic applications.

This study aimed to establish a sustainable and efficient green extraction strategy for isolating natural flavonoid-based enzyme inhibitors from B. aromatica using UAE–DES technology. By applying response surface methodology (RSM), key extraction parameters were systematically optimised to maximise flavonoid yield while minimising environmental impact. The chemical composition of the extract was characterised via UPLC-Q-Orbitrap HRMS, and microstructural changes in plant tissues were visualised using scanning electron microscopy (SEM) to confirm effective cell wall disruption. The extract’s inhibitory activities against α-glucosidase and tyrosinase were evaluated through in vitro enzyme assays, while molecular docking elucidated the binding interactions of major flavonoid constituents. Overall, this green extraction platform not only supports the efficient recovery of bioactive flavonoids but also provides valuable mechanistic insights into their potential as natural enzyme inhibitors for antidiabetic and skin-whitening applications.

2. Materials and methods

2.1. Plant material

The whole plants of B. aromatica DC. were collected from Dinghu Mountain in Zhaoqing, Guangdong Province, China. The study adhered to institutional, national, and international guidelines for plant research, with approval from the Ethics Committee of Guangdong Pharmaceutical University (No. GDPU-EC-2023–037) and compliance with the Convention on Biological Diversity and Nagoya Protocol. Taxonomic authentication was performed by Associate Researcher Xilong Zheng from the College of Traditional Chinese Medicine Resources at Guangdong Pharmaceutical University. Voucher specimens (No. GDPU-2023–0610) were deposited in the institution’s herbarium for future reference. The plant material was subsequently shade-dried, finely ground to pass through a 50-mesh sieve, and stored under dark conditions to preserve its integrity.

2.2. Chemicals and reagents

Analytical-grade methanol and anhydrous ethanol were purchased from Tianjin Da Mao Chemical Co., Ltd. and Tianjin Fuyu Fine Chemical Co., Ltd., respectively. Distilled water was obtained from Watsons Food & Beverage (Guangzhou, China) Co., Ltd. ChCl, urea, acetic acid, glycerol, lactic acid, oxalic acid, citric acid, 1,4-butanediol, and ethylene glycol were supplied by Guangzhou Chemical Reagent Factory (Guangzhou, China). Sodium nitrite, aluminium nitrate, and sodium hydroxide, all of analytical grade, were purchased from Tianjin Da Mao Chemical Reagent Factory (Tianjin, China). Rutin (≥99%) was provided by Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). α-glucosidase (33.7 U/mg), acarbose (≥ 98%), l-tyrosine, and phosphate-buffered saline (PBS, 0.1 M, pH 6.86) were obtained from Shanghai YuanYe Bio-Technology Co., Ltd. (Shanghai, China). Tyrosinase (≥500 U/mg), kojic acid, and p-nitrophenyl-α-d-glucopyranoside (p-NPG, 98%) were purchased from Shanghai Macklin Biochemical Technology Co., Ltd. (Shanghai, China).

2.3. Instrumentation

The experimental work was carried out using a comprehensive range of instruments sourced from both domestic and international manufacturers. Sample preparation was conducted using a vortex shaker (QL-861, Haimen Qilin Bell Instrument Manufacturing Co., Ltd., Haimen, China), a desktop CNC ultrasonic cleaner (KQ-500 DE, Dongguan Keqiao Ultrasonic Equipment Co., Ltd., Dongguan, China), and a medical centrifuge (L550, Hunan Xiangyi Laboratory Instrument Development Co., Ltd., Hunan, China). Particle size reduction was achieved with a pulveriser (FW100, Beijing Zhongxing Weiye Instrument Co., Ltd., Beijing, China). Precision measurements were performed using an ultra-microbalance (ATY124, Shimadzu Enterprise Management (China) Co., Ltd., Beijing, China) and a magnetic stirrer (DF-101 S, Gongyi Yuhua Instrument Co., Ltd., Gongyi, China). Analytical characterisation included spectrophotometric assays using a multifunctional microplate reader (SYNERGY H1, Agilent BioTek, Winooski, VT, USA), morphological analysis via scanning electron microscopy (JSM-7610F Plus, JEOL Ltd., Tokyo, Japan), and chromatographic separation using a high-resolution LC–MS system (Vanquish Flex UHPLC coupled with Orbitrap Exploris 120 HRMS, Thermo Fisher Scientific, Waltham, MA, USA) equipped with a Hypersil GOLD C₁₈ column (100 mm × 2.1 mm, 1.9 µm).

2.4. Total flavonoid standard curve

The Al(NO₃)₃–NaNO₂ colorimetric method was employed to quantify total flavonoid content²⁹. Rutin (0.0105 g) was dissolved in 60% ethanol and diluted to a final volume of 50 ml, yielding a stock solution with a concentration of 0.21 mg/mL. Aliquots ranging from 0.5 to 6.0 ml of this solution were each mixed with 1 ml of 5% sodium nitrite solution and allowed to stand for 6 min. Subsequently, 1 ml of 10% aluminium nitrate solution was added, followed by another 6-min reaction period. Then, 10 ml of 4% sodium hydroxide solution was added, and the mixture was diluted to 25 ml with 60% ethanol. After standing for 15 min, the absorbance at 500 nm was measured to generate a calibration curve (absorbance versus concentration in mg/mL).

2.5. Preparation of DESs

ChCl (used as the hydrogen bond acceptor, HBA) was combined with various hydrogen bond donors (HBDs), including lactic acid, acetic acid, ethylene glycol, urea, glycerol, 1,4-butanediol, oxalic acid, and citric acid, at specific molar ratios as listed in Table 1. The mixtures were heated to 80 °C and stirred magnetically for 30–120 min until clear and homogeneous liquids were obtained^30,31.

Table 1.

Types of prepared DES.

No.	HBD	Molar ratio	No.	HBD	Molar ratio
DES-1	Lactic acid	1:3	DES-5	Glycerol	1:3
DES-2	Acetic acid	1:3	DES-6	1,4-butanediol	1:3
DES-3	Ethylene glycol	1:3	DES-7	Oxalic acid	1:2
DES-4	Urea	1:3

2.6. Extraction procedure and flavonoid quantification

Approximately 0.1 g of B. aromatica powder was accurately weighed and placed into an extraction vessel. A predetermined volume of DES or conventional solvent was added, and the mixture was subjected to ultrasonic extraction under defined conditions, including temperature, extraction time, and liquid-to-solid ratio. Following extraction, the mixture was centrifuged at 2200g for 10 min, and this step was repeated once. The resulting supernatant was collected for determination of total flavonoid content using the method described above. The extraction yield of total flavonoids was calculated according to the following formula:

$$\mathrm{W}=\left(\frac{\mathrm{C}\times \mathrm{V}\times \mathrm{D}}{\text{1000M}}\right)\times 100\%$$ (1)

where W represents the extraction yield (%), C is the concentration (mg/mL) determined from the standard curve, V is the total volume of the reaction mixture (mL), D is the dilution factor, and M is the mass of the sample (g).

2.7. Optimisation of total flavonoid extraction from B. aromatica

2.7.1. Single-factor experiments

Seven DESs (DES-1 to DES-7) and four conventional solvents (anhydrous ethanol, 50% ethanol, methanol, and water) were evaluated under identical extraction conditions: a molar ratio of 1:2 (hydrogen bond acceptor to donor), 40% water content, extraction temperature of 60 °C, duration of 50 min, and ultrasonic power of 300 W. Among the tested solvents, ChCl/1,4-butanediol exhibited the highest extraction efficiency and was therefore selected for further optimisation. Subsequent single-factor experiments were conducted to investigate the effects of various parameters, including water content (10–50%), solvent-to-solid ratio (10:1–60:1 ml/g), molar ratio of HBA to HBD (1:1 to 1:5), extraction temperature (30–80 °C), and extraction time (10–60 min).

2.7.2. RSM based on Box-Behnken design

Building on the single-factor experiment results, a Box–Behnken design was employed to further optimise the extraction yield of total flavonoids. Three independent variables were selected: water content (A, 10–50%), ultrasonic temperature (B, 40–80 °C), and extraction time (C, 20–60 min). The coded factor levels used in the experimental design are presented in Table 2.

Table 2.

Factor levels design for RSM in total flavonoid extraction from B. aromatica.

Factor	Level
	–1	0	1
A: Water content (%)	10	30	50
B: Ultrasonic temperature (°C)	40	60	80
C: Ultrasonic time (min)	20	40	60

2.8. SEM analysis

To evaluate the effect of the UAE–DES extraction on the plant microstructure, samples of B. aromatica powder were collected both before and after the extraction process. The samples were then freeze-dried for 24 h using a freeze dryer. Afterward, they were sputter-coated with a thin layer of gold and analysed using a scanning electron microscope at magnifications of 500 × and 1000×.

2.9. Enzyme inhibitory activity assays

The α-glucosidase inhibitory activity of B. aromatica extracts from different solvents was evaluated based on the method reported by Gutierrez-Gonzalez et al.³², with slight modifications. The extracts were dissolved in phosphate-buffered saline (PBS) to prepare a series of concentrations. Acarbose, used as the positive control, was prepared in the concentration range of 0.02–1 μg/mL in PBS. The enzyme solution was composed of α-glucosidase (0.5 U/mL) and p-NPG (5 mmol/L), also dissolved in PBS. A total of 200 μL of sample solution was mixed with 400 μL of enzyme solution in a 1.5 ml centrifuge tube and incubated at 37 °C for 10 min. Subsequently, 400 μL of p-NPG solution was added to initiate the reaction, which continued for another 20 min at 37 °C. The reaction was terminated by placing the tubes at −20 °C for 10 min. Absorbance was measured at 405 nm. Inhibitory activity was calculated using four groups: inhibitor group (enzyme + sample), inhibitor blank group (sample only), control group (enzyme only), and blank control group (no enzyme, no sample). The IC₅₀ values were calculated using SPSS 26.0 software based on three parallel measurements.

Similarly, the tyrosinase inhibitory activity was assessed with modifications to the method of Ablat et al.³³. Extracts of B. aromatica in various solvents were dissolved in PBS to prepare a concentration gradient. Kojic acid (1–100 μg/mL, dissolved in 10% DMSO–PBS) was used as a positive control. Tyrosinase and l-tyrosine were dissolved in PBS to obtain final concentrations of 100 U/mL and 0.4 mg/mL, respectively. A reaction mixture containing 200 μL of sample solution, 200 μL of tyrosinase solution, and 400 μL of PBS was prepared in a 1.5 ml centrifuge tube and incubated at 37 °C in a shaking incubator for 10 min. Then, 200 μL of l-tyrosine substrate solution was added to initiate the enzymatic reaction, which continued for 20 min at 37 °C. The absorbance was measured immediately at 475 nm using a microplate reader. The inhibition rate was calculated based on three parallel experiments and the average value was used for analysis.

2.10. UPLC-Q-orbitrap HRMS analysis

2.10.1. Sample preparation

Based on the optimised DES extraction protocol, 200 μL of the B. aromatica extract was transferred to a 2 ml centrifuge tube, followed by the addition of 800 μL of 50% methanol. The mixture was vortexed thoroughly and then filtered through a 0.22 μm syringe filter. The filtered sample was stored at 4 °C until analysis.

2.10.2. Chromatographic conditions

Chromatographic separation was performed using a Vanquish Flex UHPLC system equipped with a Hypersil GOLD C₁₈ column (100 mm × 2.1 mm, 1.9 µm). The column was maintained at 35 °C and the mobile phase consisted of solvent A (acetonitrile with 0.1% formic acid) and solvent B (water with 0.1% formic acid). A gradient elution was applied: 0–5 min, 95–75% B; 15–25 min, 75–5% B; 25–27 min, 5% B; and 27.001–30 min, 95% B, at a flow rate of 0.3 ml/min with an injection volume of 2 μL. The autosampler was kept at 8 °C.

2.10.3. Mass spectrometric conditions

Mass spectrometric detection was carried out using an Orbitrap Exploris 120 high-resolution mass spectrometer operating in both positive and negative electrospray ionisation (ESI) modes. The instrument was set to switch between modes with a positive ion spray voltage of 3.5 kV and a negative ion spray voltage of 2.8 kV. The capillary temperature was maintained at 325 °C and the auxiliary gas (nitrogen) was heated to 350 °C. Mass scans were recorded in the m/z range of 100–1200 with a full-scan resolution of 60,000 and MS/MS (dd-MS²) resolution of 15,000. The sheath, auxiliary, and sweep gas flow rates were set at 50, 8, and 1 arbitrary units, respectively, and collision energies were applied at normalised values of 20%, 40%, and 60%.

2.11. Molecular docking procedure

Molecular docking studies were conducted to investigate the interactions between the identified flavonoids and two key enzymes, α-glucosidase (PDB ID: 3W37) and tyrosinase (PDB ID: 2Y9W). The protein crystal structures were retrieved from the RCSB Protein Data Bank, and the flavonoid structures were obtained from PubChem. Protein preparation was performed using PyMOL by removing water molecules and extracting ligands. Ligand structures were converted from .pdbqt to .mol2 format using Open Babel. In AutoDockTools 1.5.7, ligands were hydrogenated, assigned Gasteiger charges, configured with AD4 parameters, and set with torsional flexibility. Docking calculations were carried out using AutoDock Vina to obtain binding energy values. Binding interactions were visualised using PyMOL and Discovery Studio 2019 Client, producing both 2D and 3D interaction diagrams.

2.12. Data processing

All experiments were performed in triplicate, and the data are presented as the mean ± standard deviation (n = 3). The experimental design for RSM was carried out using Design Expert 11 software (Stat-Ease Inc., Minneapolis, MN, USA). Graphical representations were created with Origin 8.0, and statistical analyses, including one-way analysis of variance (ANOVA) and Pearson correlation tests, were conducted using SPSS version 26.0 (IBM, Armonk, NY, USA). Statistical significance was defined as P < 0.05. The IC₅₀ values for the antioxidant assays, representing the concentration required to achieve 50% inhibition of radical activity, were calculated using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). Mass spectrometric data were processed using Compound Discoverer 3.3 (Thermo Fisher Scientific, USA). The chemical constituents of B. aromatica were systematically identified through peak extraction, alignment processing, and a multi-dimensional identification approach, integrating control analysis, retention time validation, fragmentation information matching, as well as cross-referencing with both in-house and commercial databases (e.g., mzCloud and mzVault). Identification criteria required a mass deviation of <5 ppm and a matching score ≥80.

3. Results and discussion

3.1. Total flavonoid standard curve

To accurately determine the total flavonoid content in the extracts, a calibration curve was established using rutin as the reference compound. A stock solution of rutin at 0.21 mg/mL was prepared and subsequently diluted to yield a series of concentrations ranging from 0.0042 to 0.0504 mg/mL. The absorbance of each dilution was recorded at 500 nm, and a plot of absorbance (y) versus rutin concentration (x, mg/mL) was generated (Figure 1). Linear regression analysis of the data produced the following equation:

$$\mathrm{y}=8.4719\mathrm{x}-0.0023$$ (2)

Figure 1.

Calibration curve for rutin standard solutions.

Among them, with a correlation coefficient of R² = 0.9992. This high degree of linearity confirms that the method is reliable over the tested concentration range and can be used to quantify total flavonoids in subsequent experiments with confidence.

3.2. Single-factor experiments

A series of single-factor experiments were carried out to determine the optimal conditions for extracting total flavonoids from B. aromatica. We systematically evaluated the effects of solvent type, DES molar ratio, water content, ultrasonic time, temperature, and solvent-to-solid ratio on the extraction yield.

3.2.1. Solvent screening and DES selection

Seven DES systems and four conventional solvents were compared under identical extraction conditions (Figure 2). All DESs significantly outperformed the traditional solvents in terms of flavonoid recovery. Among the conventional solvents, 50% ethanol produced the highest yield at 2.17%, whereas absolute ethanol and methanol yielded substantially lower amounts. Pure water demonstrated only slightly better performance than the least effective DES. The ChCl and 1,4-butanediol DES exhibited the highest extraction efficiency, yielding 3.06% total flavonoids, which was approximately 1.4 times higher than that obtained with 50% ethanol (P < 0.05). Therefore, the ChCl and 1,4-butanediol system was selected for all subsequent experiments.

Figure 2.

Effect of different solvents on the extraction yield of total flavonoids from B. aromatica. ^a–iDifferent lowercase letters indicate significant differences among different solvents (P < 0.05).

The superior performance of this DES may be attributed to the multiple hydroxyl groups present in 1,4-butanediol, which form an extensive hydrogen-bond network with ChCl. This interaction likely enhances solvent polarity to a level sufficient for dissolving both flavonoid aglycones and glycosides, while reducing the co-extraction of non-phenolic impurities. Additionally, the moderate viscosity of this DES promotes efficient mass transfer during ultrasonic treatment. In contrast, although water has high polarity, it is less effective at dissolving hydrophobic flavonoid aglycones. On the other hand, absolute ethanol and methanol, while capable of solubilising hydrophobic compounds, tend to co-extract unwanted impurities, thereby reducing the overall yield^31,34.

3.2.2. Effect of DES molar ratio

To optimise the hydrogen-bond donor/acceptor balance, the molar ratio of ChCl to 1,4-butanediol was varied from 1:1 to 1:5 (Figure 3a). The extraction yield rose steadily from a 1:1 ratio, peaking at 3.07% for the 1:3 mixture; further increases in donor content caused a gradual decline in yield. Thus, a 1:3 molar ratio was chosen as optimal. At lower ratios, the DES under-utilizes the donor’s hydrogen-bonding sites, limiting flavonoid dissolution. At higher ratios, the viscosity becomes excessive, attenuating cavitation intensity and slowing diffusion. The 1:3 ratio represents an ideal compromise, providing sufficient donor functionality for flavonoid solubilisation while retaining fluidity for ultrasonic cavitation to disrupt plant cell walls effectively^35,36.

Figure 3.

Influence of various factors on the extraction yield of total flavonoids from B. aromatica. (a) Effect of DES molar ratio; (b) Effect of water content; (c) Effect of ultrasonic time; (d) Effect of ultrasonic temperature; (e) Effect of liquid–soild ratio.

3.2.3. Influence of water content

We next examined the effect of water content in the DES (10–50% v/v) on flavonoid recovery (Figure 3b). Yields increased with water content up to 40%, reaching 3.03%, but decreased when water content rose to 50%. Consequently, 40% (v/v) water was incorporated into the DES for further experiments. Adding water up to this level reduces viscosity and surface tension, which enhances cavitation bubble formation and collapse, thereby improving solvent penetration into plant tissues³⁷. Exceeding 40% water disrupts the eutectic hydrogen-bond network, weakening the DES’s solvation capacity and lowering extraction efficiency.

3.2.4. Effect of ultrasonic time

Ultrasonic time was varied from 10 to 60 min to assess its impact on flavonoid release (Figure 3c). The extraction yield increased with sonication time, reaching a maximum of 3.00% at 50 min. Beyond 50 min, the yield declined slightly, likely because equilibrium between solute desorption and solvent saturation is achieved, and extended ultrasound may cause mild degradation of sensitive flavonoids or re-adsorption onto cell debris³⁸. Therefore, 50 min was selected as the optimal sonication duration.

3.2.5. Effect of ultrasonic temperature

Extraction temperature was evaluated between 30 °C and 80 °C (Figure 3d). Yields rose steadily with temperature, peaking at 3.00% at 80 °C, compared to 2.58% at 30 °C. Considering both energy consumption and extraction efficiency, 80 °C was deemed optimal. Higher temperatures lower DES viscosity and enhance molecular diffusion, thus accelerating mass transfer. Although temperatures above 80 °C might offer marginal yield gains, they entail higher energy costs and risk thermal degradation of flavonoids^39,40.

3.2.6. Effect of liquid–solid ratio

Finally, the solvent-to-material ratio was varied from 10:1 to 60:1 ml/g (Figure 3e). The extraction yield increased with the ratio up to 50:1 ml/g, achieving 3.05%; further increases to 60:1 led to a slight decrease. Consequently, a 50:1 ml/g ratio was chosen. A higher solvent-to-solid ratio enhances the concentration gradient driving mass transfer, but excessive solvent volume disperses ultrasonic energy, reducing cavitation intensity and extraction efficiency^41,42. A 50:1 ratio thus balances a strong concentration gradient with effective ultrasonic cavitation.

3.3. RSM optimization of total flavonoid extraction

3.3.1. Experimental design and results

Based on the outcomes of the single-factor trials, the liquid–solid ratio (50:1 ml/g) and DES molar ratio (ChCl: 1,4-butanediol = 1:3) were held constant. A three-factor, three-level Box–Behnken design was employed, with water content (A, 10–50%), ultrasonic temperature (B, 40–80 °C), and ultrasonic time (C, 20–60 min) as independent variables. The response variable was the total flavonoid yield (Y, %). The design matrix and observed yields are presented in Table 3.

Table 3.

Response surface design and results for the extraction of total flavonoids from B. aromatica.

Test Number	A: Water content (%)	B: Ultrasonic temperature (°C)	C: Ultrasonic time (min)	Y: Total flavonoid yield (%)
1	10	80	40	2.481
2	10	60	60	2.390
3	30	40	20	2.471
4	30	60	40	2.864
5	50	40	40	2.675
6	30	40	60	2.585
7	30	60	40	2.957
8	50	80	40	3.070
9	30	60	40	2.961
10	30	80	60	3.021
11	10	60	20	2.191
12	10	40	40	2.265
13	30	60	40	2.927
14	30	60	40	2.968
15	50	60	20	2.753
16	30	80	20	2.718
17	50	60	60	2.860

3.3.2. Analysis of variance of the regression model

Multiple quadratic regression on the data yielded the following model:

$$\mathrm{Y}=2.94+0.2539\mathrm{A}+0.1618\mathrm{B}+0.0904\mathrm{C}+0.0447\text{AB}-0.0230\text{AC}+0.0473\text{BC}-0.2315{\mathrm{A}}^2-0.0812{\mathrm{B}}^2-0.1554{\mathrm{C}}^2$$ (3)

Analysis of variance (Table 4) shows that the model is highly significant (P < 0.0001) with a non-significant lack-of-fit term (P = 0.7481). The coefficient of determination (R² = 0.9817) and adjusted R² (0.9597) differ by less than 0.02, indicating excellent fit and predictive capability. A low coefficient of variation (1.37%) and a precision ratio of 31.63 (>4) further confirm the reliability of the experimental data. From the ANOVA results, the linear terms A, B, and C and the quadratic terms A², B², and C² all have P < 0.01, exerting very significant effects on Y. The interaction terms AB and BC are significant (P < 0.05), whereas AC is not. The influence of the factors on extraction yield ranks as A > B > C.

Table 4.

Analysis of variance (ANOVA) for response surface model.

Source	Sum of squares	Degrees of freedom	Mean square	F value	p value	Significance
Model	1.2	9	0.1331	96.3	<0.0001	**
A	0.5156	1	0.5156	372.96	<0.0001	**
B	0.2093	1	0.2093	151.4	<0.0001	**
C	0.0653	1	0.0653	47.26	0.0002	**
AB	0.008	1	0.008	5.79	0.047	*
AC	0.0021	1	0.0021	1.53	0.2559	NS
BC	0.0089	1	0.0089	6.46	0.0386	*
A 2	0.2256	1	0.2256	163.15	<0.0001	**
B 2	0.0278	1	0.0278	20.08	0.0029	**
C 2	0.1017	1	0.1017	73.6	<0.0001	**
Residual	0.0097	7	0.0014	–	–	–
Lack-of-fit	0.0023	3	0.0008	0.4214	0.7481	NS
Pure error	0.0074	4	0.0018	–	–	–
Total	1.21	16	–	–	–	–
R 2	0.9817	–	–	–	–	–
R2adj	0.9597	–	–	–	–	–
Precision	31.6256	–	–	–	–	–

NS: Not significant

^*Significant (P < 0.05)

^**Highly significant (P < 0.01)

3.3.3. Response surface interaction analysis

Three-dimensional response surfaces and contour plots (Figure 4) illustrate the interactions among factors. The AB interaction (water content × temperature) and BC interaction (temperature × time) exhibit elliptical contours and steep gradients, indicating significant synergistic effects. In contrast, the AC interaction (water content × time) shows nearly circular contours and a relatively flat surface, consistent with its non-significant ANOVA result.

Figure 4.

Interaction effects of factors on the extraction yield of total flavonoids from B. aromatica. (a) Interaction between Factors A and B; (b) Interaction between Factors B and C; (c) Interaction between Factors A and C.

3.3.4. Model validation and verification

The validation experiments confirmed that the optimised UAE–DES protocol is both robust and practically applicable. Under the rounded conditions (water content 43%, ultrasound temperature 80 °C, ultrasound time 48 min, liquid–solid ratio 50:1 ml/g, ChCl: 1,4-butanediol = 1:3), three replicate runs yielded an average total flavonoid extraction rate of 3.15%, closely matching the model’s predicted value of 3.137%. These results highlight the superiority of the ChCl/1,4-butanediol DES over conventional solvents. Compared with 50% ethanol extraction (2.17%), the optimised UAE–DES method achieved a 45.2% higher total flavonoid yield, demonstrating significantly enhanced solubilisation and recovery of polyphenolic compounds from B. aromatica. This enhancement stems from the DES’s tailored polarity and hydrogen-bonding network, which, under ultrasonic cavitation, facilitate deeper solvent penetration and more efficient disruption of plant cell walls. Such performance improvements are consistent with literature reports that UAE–DES approaches can markedly outperform traditional organic solvent systems in flavonoid extraction, underscoring their potential for sustainable, high-efficiency natural product isolation^43,44.

3.4. SEM analysis

To assess the impact of the UAE–DES protocol on the structural integrity of B. aromatica cells, SEM observations were carried out on both untreated powder and UAE–DES treated residue (Figure 5). In the untreated material (Figures 5a and 5b), cell walls appear smooth and continuous, with polygonal shapes and clear intercellular boundaries, indicating that conventional drying and milling preserved the native architecture without visible damage. After UAE–DES treatment (Figures 5c and 5d), the powder surface is markedly altered: cell walls show extensive fragmentation, and the formerly smooth surface becomes uneven and porous. High-magnification images (Figure 5d) reveal micron-scale perforations and fissures penetrating the wall matrix. These structural changes result from the combined action of the low-viscosity, hydrogen-bonding enriched DES penetrating wall components and ultrasonic cavitation, in which alternating high and low pressure cycles form bubbles that collapse violently, generating shear forces and microjets that rupture cell structures. These observations are in agreement with the findings of He et al. and Wang et al., confirming that DES–UAE treatment effectively disrupts the plant cell matrix and thereby promotes the release of intracellular flavonoids and other bioactive compounds^27,28.

Figure 5.

SEM of B. aromatica powder before and after DES extraction. (a) Before extraction (×500); (b) Before extraction (×1000); (c) After extraction (×500); (d) After extraction (×1000).

3.5. In vitro α-glucosidase and tyrosinase inhibition by different B. aromatica extracts

The inhibitory activities of B. aromatica extracts prepared by various ultrasound-assisted methods against α-glucosidase and tyrosinase were evaluated and expressed as IC₅₀ values (Table 5). The DES–UAE extract exhibited the strongest inhibition, with an α-glucosidase IC₅₀ of 35.872 ± 0.294 µg/mL and a tyrosinase IC₅₀ of 9.126 ± 0.285 µg/mL, significantly outperforming the water (UAE-W), 50% ethanol (UAE-E50), absolute ethanol (UAE-E) and methanol (UAE-M) extracts (P < 0.05). Among the conventional solvents, UAE-E showed the weakest activity, with IC₅₀ values of 112.361 ± 0.335 µg/mL for α-glucosidase and 24.578 ± 0.251 µg/mL for tyrosinase. Compared to pharmaceutical standards, the DES–UAE extract’s α-glucosidase inhibition was lower than that of acarbose (IC₅₀ = 0.053 ± 0.003 µg/mL), whereas its tyrosinase inhibition approached that of kojic acid (IC₅₀ = 8.352 ± 0.273 µg/mL). This suggests that DES–UAE selectively enriches flavonoids and other bioactives with potent melanogenesis-inhibitory effects.

Table 5.

In vitro α-glucosidase and tyrosinase inhibition by B. aromatica extracts (IC₅₀, µg/mL, mean ± SD, n = 3).

Extraction Method	α-glucosidase inhibition IC50	Tyrosinase inhibition IC50
UAE-DES	35.872 ± 0.294b	9.126 ± 0.285b
UAE-W	69.145 ± 0.247d	14.893 ± 0.334c
UAE-E50	52.361 ± 0.335c	16.578 ± 0.251d
UAE-E	112.361 ± 0.335f	24.578 ± 0.251e
UAE-M	98.490 ± 0.279e	27.234 ± 0.318f
acarbose	0.053 ± 0.003a	–
kojic acid	–	8.352 ± 0.273a

^※a–e Different lowercase letters in each column indicate significant differences (P < 0.05).

Numerous studies have demonstrated that DES extracts possess superior stability and enhanced biological activity compared with traditional solvent systems^35,45. In line with this, our results reveal that flavonoids extracted from B. aromatica using DES combined with ultrasound exhibit significantly stronger α-glucosidase and tyrosinase inhibitory activities than those obtained by water, ethanol or methanol. This not only confirms the high extraction efficiency of DES but also underscores its advantage in preserving or even enhancing the bioactivity of natural products. Overall, these findings demonstrate that UAE–DES is a green and highly effective strategy for obtaining B. aromatica extracts with dual antidiabetic and skin-whitening potential. By contrast, conventional solvent systems yielded extracts with markedly lower enzyme inhibition, highlighting the critical role of solvent choice and extraction technology in maximising the bioactive yield from plant materials.

3.6. Identification of flavonoid compounds by UPLC-Q-orbitrap HRMS

The chemical profile of the optimised B. aromatica UAE–DES was characterised using UPLC-Q-Orbitrap high-resolution mass spectrometry. Six flavonoid compounds were unambiguously identified based on accurate mass measurements, retention times, and MS² fragment patterns (Table 6). These include scutellarein, nobiletin, pectolinarigenin, tangeretin, eupatilin, and corylin. Notably, none of these flavonoids have been previously reported in B. aromatica, providing new insight into its phytochemical basis and supporting further investigation of their contribution to the observed bioactivities. The six flavonoids identified in B. aromatica have been extensively studied in various medicinal plants and exhibit a rich spectrum of bioactivities: scutellarein possesses anticancer, anti-inflammatory, antioxidant, anti-obesity, and vasodilatory effects^46,47; pectolinarigenin demonstrates pronounced antioxidant, anti-inflammatory, antidiabetic, and antitumor activities⁴⁸; tangeretin inhibits tumour cell proliferation by modulating JAK/STAT and caspase-3 signalling pathways and inducing cell-cycle arrest^49,50; eupatilin shows strong anti-inflammatory, antioxidant, and anticancer properties and can alleviate Helicobacter pylori CagA-induced gastric inflammation⁵¹; nobiletin exerts anti-inflammatory, neuroprotective, and lipid-lowering effects, making it a promising candidate for metabolic and neurodegenerative disorders^52,53; and corylin has been shown in other sources to offer bone-protective and anti-inflammatory benefits^54–56.

Table 6.

Flavonoid compounds identified in B. aromatica using UPLC-Q-Orbitrap HRMS.

No.	RT/min	Compound	Molecular formula	Error/ppm	m/z	Ion mode	Fragments (m/z)	Type
1	11.05	scutellarein※	C15H10O6	2.34	287.05568	[M + H]+	229.02884; 117.06807; 161.01282; 86.93504 64.97443	flavone
2	19.77	nobiletinΔ	C21H22O8	0.19	403.13882	[M + H]+	388.18008; 373.09155; 229.02466; 203.05656; 157.79492	flavone
3	20.09	pectolinarigenin※	C17H14O6	0.55	315.08651	[M + H]+	229.04643; 186.01588; 168.00548	flavone
4	20.54	tangeretin※	C20H20O7	0.56	373.12839	[M + H]+	343.08203; 274.09253; 229.02863; 155.66515	flavone
5	20.64	eupatilin※	C18H16O7	−1.59	345.09622	[M + H]+	329.06555; 287.05533 281.04486; 229.03526	flavone
6	21.91	corylin※	C20H16O4	0.05	321.11215	[M + H]+	279.06512; 251.06950; 223.07532; 195.08052; 137.02336	isoflavone

^※First identified in plants of the genus Blumea.

^ΔFirst discovered in B. aromatica.

3.7. Molecular docking results

Molecular docking was performed to investigate the binding affinities and interaction modes of the six identified flavonoids with tyrosinase (PDB ID: 2Y9W) and α-glucosidase (PDB ID: 3W37). In general, a lower binding energy suggests a stronger interaction between the ligand and the target protein⁵⁷. The calculated binding energies are summarised in Table 7. Among the six compounds, corylin exhibited the strongest binding to both enzymes, with energies of −8.2 kcal/mol for tyrosinase and −9.0 kcal/mol for α-glucosidase. In the tyrosinase–corylin complex (Figure 6a), one hydrogen bond forms between the corylin hydroxyl group and the side chain of LYS-180 (bond length 5.2 Å), supplemented by multiple van der Waals contacts involving THR-197, ASN-174, PRO-175, GLN-44, and ALA-45. In the α-glucosidase–corylin complex (Figure 6b), corylin establishes two hydrogen bonds with ASP-666 and ARG-676 (bond lengths 2.3 Å and 2.5 Å, respectively), along with van der Waals interactions with GLU-792, THR-662, and LEU-663. This multi-site binding network underlies corylin’s high complex stability. Other flavonoids, also exhibited binding energies below −5 kcal/mol for both enzymes, suggesting they too can form stable enzyme–ligand complexes.

Table 7.

Docking binding energies of six flavonoids with α-glucosidase and tyrosinase.

Compound	α-Glucosidase binding energy (kcal/mol)	Tyrosinase binding energy (kcal/mol)
scutellarein	–8.0	–7.0
eupatilin	–7.4	–7.2
tangeretin	–7.3	–7.0
corylin	–9.0	–8.2
nobiletin	–6.7	–7.0
pectolinarigenin	–7.7	–6.9
acarbose	−8.0	–
kojic acid	–	−5.6

Figure 6.

Molecular docking results of corylin with α-glucosidase (PDB ID: 3W37) and tyrosinase (PDB ID: 2Y9W). (a) 3W37–corylin complex; (b) 2Y9W–corylin complex. For each panel, the left image shows the overall 3D binding conformation, the centre image shows the enlarged 3D local interaction view, and the right image presents the 2D interaction diagram (red: amino acid residues; blue: compound).

Recent work by Chang et al. has validated corylin’s anti-melanogenic and antidiabetic potential in cellular and tissue models. They demonstrated that corylin isolated from Pueraria lobata aerial parts effectively reduces melanin synthesis and tyrosinase activity without cytotoxicity, downregulates MITF and related melanogenic enzymes, and inhibits α-glucosidase to disrupt glycosylation processes. In a UVB-stimulated three-dimensional human skin model, corylin significantly reduced pigmentation, confirming its ability to lighten hyperpigmented skin. These findings align with our docking results, where corylin’s strong binding to the active sites of both tyrosinase and α-glucosidase provides a molecular explanation for its dual skin-whitening and antidiabetic effects⁵⁸. Together, the molecular docking data and literature evidence position corylin and the other identified flavonoids as key contributors to the pronounced α-glucosidase and tyrosinase inhibition observed in the B. aromatica DES–UAE extract. These results implicate corylin, scutellarein, eupatilin, and the other identified flavonoids as likely contributors to the α-glucosidase and tyrosinase inhibition observed in the UAE–DES extracts, providing a molecular-level rationale for the dual antidiabetic and skin-whitening activities of B. aromatica.

4. Conclusions

A green and efficient UAE–DES extraction method was developed for the targeted recovery of enzyme-inhibitory flavonoids from B. aromatica. Optimal extraction conditions using choline chloride and 1,4-butanediol (1 to 3 molar ratio) with 43% water content, a solvent-to-solid ratio of 50 ml/g, ultrasound at 80 °C for 48 min yielded a total flavonoid content of 3.15%, significantly higher than that obtained with conventional solvents such as ethanol and methanol. Scanning electron microscopy confirmed extensive disruption of plant cell walls, enhancing the release of intracellular compounds. UPLC-Q-Orbitrap HRMS analysis identified six flavonoids from B. aromatica, and molecular docking revealed strong binding affinities of these compounds to α-glucosidase and tyrosinase. In particular, corylin showed high binding energies of −9.0 kcal/mol and −8.2 kcal/mol, respectively, supporting the observed potent inhibitory activities. This study demonstrates the potential of DES-based extraction as a green and effective strategy for discovering natural enzyme inhibitors with therapeutic and cosmetic applications.

Glossary

Abbreviations

ANOVA

Analysis of variance

ChCl

Choline chloride

DES

Deep eutectic solvent

HBA

Hydrogen bond acceptor

HBD

Hydrogen bond donor

HRMS

High-resolution mass spectrometry

IC₅₀

Half maximal inhibitory concentration

PBS

Phosphate-buffered saline

RSM

Response surface methodology

SEM

Scanning electron microscopy

UAE

Ultrasound-assisted extraction

UHPLC

Ultra-high performance liquid chromatography

UPLC-Q-Orbitrap

Ultra-performance liquid chromatography coupled with quadrupole-orbitrap high-resolution mass spectrometry

Funding Statement

This research was supported by the following funds: Science and Technology Projects in Guangzhou (E33309); Operation and Maintenance of the Yunfu Medicinal Plant Germplasm Resource Bank (Garden) in Guangdong Province under the 2023 Provincial Seed Industry Revitalisation Special Fund Project (2023-NBH-21–001); the Guangzhou Basic Research Program Fund Project (No. 202102020630); the Research Fund Project of Guangdong Provincial Administration of Traditional Chinese Medicine (No. 20201199); and the 2023 College Students’ Innovation and Entrepreneurship Training Program (No. 202310573012).

Disclosure statement

The authors report no conflicts of interest.

Data availability statement

The datasets generated during the current study are available from the corresponding authors (Hongfeng Chen, h.f.chen@scbg.ac.cn; Xilong Zheng, zhengxl2020@gdpu.edu.cn) on reasonable request.