Ultrasound-assisted supramolecular solvent extraction with enzymatic pretreatment of Celastrus orbiculatus roots: Process optimization, enrichment, and antibacterial activity

Graphical abstract

Highlights

• •Ultrasound-assisted enzyme-SUPRAS extraction was used to extract triterpenoids, promising plant-derived antimicrobials.
• •Optimized processing parameters using PBD and BBD experiments.
• •Ultrasound-assisted enzyme-SUPRAS extraction method significantly outperforms single methods in terms of extraction yield.
• •Macroporous resin was employed for the purification of the Celastrus orbiculatus crude extract.
• •Enriched extracts had significant antibacterial activity.

Abstract

Ultrasound offers a greener and more efficient alternative to conventional extraction methods, requiring less solvent and shorter processing times. This study developed an ultrasound − assisted enzyme − supramolecular solvents (SUPRAS) extraction (UAESE) method to synergistically extract two key triterpenoids from Celastrus orbiculatus roots. In this study, ultrasound-assisted extraction was performed in a 30 L ultrasonic cleaning bath operating at a frequency of 40 kHz. Based on single-factor experiments, Plackett-Burman design (PBD) and Box-Behnken design (BBD) were employed to optimize the extraction parameters. The optimal integrated process involves: ultrasound-assisted enzymatic pretreatment with 6% enzyme at pH 5 and 41℃ for 50 min; formulation of a SUPRAS containing 5% octanoic acid, 30% ethanol, and 60% aqueous phase at pH 4; and final ultrasonic extraction at 40℃ for 70 min with 640 W power and a liquid–solid ratio of 12 mL/g. The optimized UAESE method achieved a significantly higher extraction yield compared to other extracts. Then scanning electron microscopy (SEM) was employed to explore the extraction mechanism of UAESE. Subsequent enrichment of the UAESE crude extract with an AB-8 resin yielded a purified product containing 13.545 ± 0.542% pristimerin (PRI). Furthermore, the enriched extract showed significant activity against methicillin-resistant Staphylococcus aureus (MRSA), primarily attributed to celastrol (CEL)and PRI. In conclusion, the proposed method offers an innovative, efficient, and environmentally sustainable approach to extracting diverse triterpenoids from plant materials.

1. Introduction

Foodborne diseases pose a significant global health concern. In China, Staphylococcus aureus is a leading cause of foodborne disease [1]. The misuse of antibiotics, despite their fundamental role in treating bacterial infections, has led to widespread resistance in Staphylococcus aureus. In recent years, plant extracts have been considered to have antimicrobial properties against foodborne pathogens due to their multiple bioactive components (such as polyphenols, saponins, organic acids, flavonoids, and terpenoids) and have been gradually applied in the food field [2], [3]. Celastrus orbiculatus Thunb., a member of the Celastraceae family, is predominantly distributed in regions south of the Yangtze River in China. Furthermore, the species is also found in Russia, India, North Korea, and Japan [4]. As a commonly used folk herbal medicine in China, all parts of Celastrus orbiculatus (roots, stems, leaves, and fruits) can be used for medicinal purposes [5], with effects including clearing away heat, removing toxic substances, eliminating pathogenic wind and dampness, promoting blood circulation, and subduing swelling. Root bark extracts of Celastrus orbiculatus are potent, broad-spectrum inhibitors of bacterial growth, effective against strains such as Bacillus subtilis, Staphylococcus aureus, and Escherichia coli [6], [7]. Celastrol and pristimerin, two friedelane-type triterpenoids, are prominent bioactive constituents primarily derived from the root bark of Celastraceae plants [8]. Recent studies have isolated these compounds from Celastrus hypoleucus, a species phylogenetically close to Celastrus orbiculatus, and demonstrated their notable antifungal activity, underscoring their broad-spectrum bioactivity[9]. Nzier et al.[10] obtained pristimerin from a hexane extract of Salacia crassifolia roots and purified it via column chromatography, establishing it as a promising natural product with potent efficacy against Staphylococcus aureus. Yuan et al. [11] reported that celastrol exhibits broad-spectrum activity against Gram-positive bacteria, including strong inhibitory effects against diverse clinical and standard strains of MRSA. To fully leverage the potential of Celastrus orbiculatus, this study focuses on its root bark, centering on the isolation and analysis of key triterpenoid constituents. Unfortunately, for extraction process development, the characteristically low solubility and bioavailability of triterpenoids pose a central obstacle [12], limiting both the yield obtainable from Celastrus orbiculatus and the translatability of the isolates into effective applications.

The efficient liberation and solubilization of target compounds from dense plant matrices poses a persistent challenge in natural product extraction. Conventional methods like Soxhlet extraction face growing criticism for their high energy consumption and extensive use of volatile organic solvents [13]. Current strategies to address these limitations focus on two fronts: enhancing mass transfer via physical or biological cell disruption (such as ultrasound or enzymes), and employing designer solvents with superior affinity and selectivity. Ultrasound-Assisted Extraction (UAE) utilizes the cavitation, mechanical, and thermal effects generated by ultrasonic waves to enhance solvent penetration into plant matrices [14]. As ultrasound propagates through a medium, it creates periodic compressions and expansions, leading to the nucleation and implosive collapse of microscopic cavities [15]. This collapse generates localized spikes in temperature and pressure, which force the solvent deeper into the sample matrix, thereby accelerating analyte transfer into the solvent [16]. UAE has been extensively investigated for obtaining plant bioactive compounds due to its inherent merits, including operational simplicity, high efficiency, energy savings, and environmental friendliness [17]. Enzyme-Assisted Extraction (EAE) employs exogenous enzymes like cellulase, ligninase, pectinase, and protease to degrade plant raw materials [18]. Ultrasound-assisted enzymatic extraction (UAEE) is an enhanced method that uses ultrasonic energy to boost the efficiency and yield of enzymatic reactions [19], largely by promoting substrate accessibility and enzyme activity. Tran et al. [20] reported that UAEE enhanced the recovery of triterpenoids from Ganoderma lucidum compared to conventional maceration. The extract obtained by this combined method demonstrated potent antibacterial and anticancer activities, indicating that the ultrasonic and enzymatic techniques acted synergistically to improve both extraction yield and bioactivity. Guo et al. [21] demonstrated that UAEE markedly enhanced both the extraction yield and the antioxidant activity of polysaccharides obtained from Ziziphus jujuba Mill.

Supramolecular solvents (SUPRAS) can be defined as nanostructured, micellar aggregates. They are produced when amphiphilic compounds spontaneously self-assembly in an aqueous environment, creating a distinct phase with tailored extraction properties [22]. As a new type of green solvent, SUPRAS possess excellent properties including simple preparation, environmental friendliness, designable structure, and low solvent consumption. The tunable affinity of SUPRAS, arising from their composition, allows efficient extraction of diverse bioactive phytochemicals [23]. SUPRAS extraction is an innovative technique that enables the efficient and selective enrichment of target compounds in an environmentally benign manner [24], [25]. Wang et al [26]. demonstrates that combining a hexafluorobutanol − hexanol SUPRAS with ultrasound-assisted extraction enables the efficient extraction of fangchinoline and tetrandrine from the roots of Radix Stephaniae tetrandrae, with benefits including faster processing, lower toxicity, and reduced environmental impact compared to traditional organic solvents. This integration positions ultrasound-assisted extraction with SUPRAS as an advanced green technique, forming the basis for the present study on the targeted extraction of active ingredients.

This study assessed the suitability of ultrasound-assisted enzyme-SUPRAS extraction (UAESE) for extracting triterpenoids from Celastrus orbiculatus. To improve the efficiency of active component extraction, process parameters were optimized using a sequential strategy of Plackett-Burman (PBD) and Box-Behnken (BBD). Additionally, a comparative analysis with alternative extraction methods was performed. An adsorption step using resin was then applied to enrich the target bioactive compounds. Finally, the antimicrobial efficacy of the UAESE − enriched extract was assessed through in vitro tests. This study aims to establish an integrated extraction methodology for plant-derived bioactive compounds that aligns with the principles of green chemistry, operational efficiency, and sustainable processing. This could potentially improve the effectiveness of traditional medicines and support the theoretical development of new natural antibacterial agents derived from plants.

2. Materials and methods

2.1. Materials and chemicals

Celastrus orbiculatus Thunb. was purchased from Guangzhou Zhixin Pharmaceutical Co., Ltd. (Guangdong, China). The root bark of Celastrus orbiculatus was dried in an electric blast drying oven (GZX-9240MBE) at 50°C for 48 h. The dried root bark was crushed (FW100 high-speed universal pulverizer) and sieved (80-mesh) to obtain the Celastrus orbiculatus powder.

Cellulase (Cel, 400 U/mg) was purchased from Meilun Biotechnology (Dalian, China); octanoic acid (purity ≥ 99%) and formic acid (chromatographic grade) were purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (Shanghai, China); celastrol (CEL) and pristimerin (PRI) standards (all with purity ≥ 98%) were obtained from Chengdu Plant Standardization Biotechnology Co., Ltd. (Chengdu, China); lipophilic Nile Red (purity ≥ 99%) and D101, AB-8 and NKA-9 macroporous resin were supplied by Shanghai Yuanye Biotechnology Co., Ltd. (Shanghai, China); hydrophilic sodium fluorescein (purity ≥ 99%) was purchased from Shanghai Macklin Biochemical Technology Co., Ltd. (Shanghai, China); methanol (chromatographic grade) was obtained from Beijing Diama Technology Co., Ltd. (Beijing, China); all chemicals, including citric acid, Na₂HPO₄, and absolute ethanol, were of analytical purity and sourced from Tianli Chemical Reagents Ltd. (Tianjin, China).

2.2. Screening and characterization of SUPRAS reagents

200 mg of Celastrus orbiculatus powder was evenly divided into two parts and placed in 2 mL centrifuge tubes. 1 mL of SUPRAS-1 (octanoic acid) and SUPRAS-2 (decanoic acid) were added to the two centrifuge tubes respectively. The mixture was first vortex-mixed for 60 s; then performed ultrasound-assisted SUPRAS extraction at 40°C and 840 W for 50 min. Following centrifugation at 8000 rpm for 5 min, the resulting supernatant was collected for further examination. The two target components in Celastrus orbiculatus were quantified by high-performance liquid chromatography (HPLC), with each treatment replicated three times. SUPRAS formation began by dissolving long-chain acids (5%, v/v) in ethanol (35%, v/v). The aqueous phase (60%, v/v) was then introduced to initiate self-assembly. After vortex mixing (60 s) and centrifugation (3500 rpm, 5 min), the system separated into a lower equilibrial phase (EqS) and an upper SUPRAS phase (Fig. 1).

Fig. 1.

Schematic picture of the SUPRAS forming procedure.

Microstructural analysis was performed with an inverted fluorescence microscope (Leica DM-18, Germany) to observe the architecture of the SUPRAS droplets. A mixture of octanoic acid 5% (v/v) and ethanol 35% (v/v) was injected into aqueous phase infused with either the hydrophobic dye Nile Red or the hydrophilic dye sodium fluorescein, which served as fluorescent probes to identify the SUPRAS type [27]. Furthermore, Fourier transform infrared spectroscopy (FTIR) was used to obtain spectra for pure octanoic acid, pure ethanol, their 1:1 (v/v) mixture, and the octanoic acid − ethanol SUPRAS. Spectra were recorded from 500 to 4000 cm^–1 at 20 scans per minute to enhance the signal-to-noise ratio.

2.3. Phase diagram and phase ratio

To define the SUPRAS formation zone, we constructed a phase diagram for the SUPRAS system at room temperature, following the method of Li et al. [28] with slight modifications. The octanoic acid–ethanol stock mixture was prepared by dissolving octanoic acid (1.8% − 9.0%, v/v) in ethanol (10% − 40%, v/v). Then, water was added to the octanoic acid–ethanol mixture to achieve a final total volume of 40 mL. After vortexing for 60 s and centrifuging at 3500 rpm for 5 min to induce phase separation, we immediately recorded the system's phase behavior. Each experiment was conducted in triplicate to ensure reproducibility.

Based on the phase diagram, a series of 4 mL SUPRAS was prepared within the stable phase separation region. To investigate the effects of composition on the phase ratio (defined as Vs/Vt, the ratio of the SUPRAS upper phase volume (Vs) to the total system volume (Vt)), a series of SUPRAS was prepared under two sets of conditions. When studying the effect of octanoic acid, its concentration was varied from 3% to 7% (v/v) while maintaining ethanol at a constant 35% (v/v). Conversely, when studying the effect of ethanol, we varied the ethanol content from 16% to 36% (v/v) while maintaining a constant octanoic acid concentration of 5% (v/v). All experiments were conducted in triplicate under otherwise identical conditions.

2.4. Ultrasound-assisted enzyme-SUPRAS extraction

2.4.1. HPLC determination of CEL and PRI

The qualitative and quantitative detection of CEL and PRI in the root bark extract of Celastrus orbiculatus was performed using a Waters e2695 HPLC system (Waters, America) with a C18 column (250 mm × 4.6 mm, 5 μm). The mobile phase consisted of methanol (A) and 0.1% aqueous formic acid (B). Separation was achieved using the following gradient program: 87% − 95% A from 0 − 19 min, followed by a return to 87% A from 19 − 20 min. Chromatographic conditions were as follows: flow rate, 1.0 mL/min; injection volume, 10 μL; column temperature, 30°C. At a wavelength of 425 nm, the HPLC chromatograms of CEL (retention time: 12.21 min) and PRI (retention time: 15.95 min) are shown in the figure (Fig. 4A) by comparison with the corresponding standards. The internal standard method was utilized to establish calibration curves, and the linear regression equations for CEL and PRI were calculated. Furthermore, the HPLC method for quantifying CEL and PRI was validated by evaluating its linearity, precision, accuracy, and reproducibility (as indicated in Tables S1 and S2). Three aliquots of each sample were collected for analysis.

Fig. 4.

HPLC chromatographic diagram of standards (CEL and PRI) and extracts by ESU (A); The impact of optimization factors on the yields: percentage of ethanol in SUPRAS (B), the pH of the water used in the synthesis of SUPRAS (C), enzyme concentration (D), enzymolysis pH(E), enzymolysis time (F), enzymolysis temperature (G), extraction time (H), extraction temperature (I), ultrasonic power (J), and liquid–solid ratio (K). Different letters indicate significant data differences between different groups (P < 0.05).

2.4.2. Single − factor experiment for UAESE

The extraction procedure for Celastrus orbiculatus consisted of two sequential steps assisted by ultrasound: enzyme pretreatment followed by SUPRAS treatment, as illustrated in Fig. 2. Specifically, a sample of the plant powder (1.0 g) was prepared in a 15 mL centrifuge tube for extraction. The pH of the mixture, prepared in a Na₂HPO₄-citric acid buffer system with a liquid–solid ratio of 10 mL/g, was adjusted to target values (3.5, 4.0, 4.5, 5.0, and 5.5) using either NaOH or HCl. Then, various concentrations (2%, 4%, 6%, 8%, and 10%) of cellulase (Cel) were added to the mixture, which was subsequently vortexed for 60 s. The enzyme pretreatment mixture was then subjected to ultrasound-assisted enzymatic hydrolysis under varying conditions of temperature (35°C − 55°C, in 5°C steps) and time (10, 30, 50, 70, and 90 min). Subsequently, we terminated the enzymatic reaction by heating the mixture at 100°C for 5 min. Centrifugation of the enzymatically pretreated sample (5000 rpm, 5 min) yielded a solid pellet, which was then collected. The enzyme-pretreated precipitate was mixed with SUPRAS at varying liquid–solid ratios (10, 20, 30, 40, and 50 mL/g). The SUPRAS consisted of a fixed 5% (v/v) octanoic acid, alongside two variable components: ethanol (15%, 20%, 25%, 30%, and 35%, v/v) and aqueous solutions of different pH (2, 3, 4, 5, 6). Extraction was performed using an ultrasonic cleaning bath (SB-800DTD, Ningbo Scientz Biotechnology Co., Ltd., China; 30 L tank volume: 500 × 300 × 200 mm) operating at 40 kHz with a maximum power of 840 W. The process was systematically investigated across a range of temperatures (20°C − 60°C, in 10°C steps), ultrasound power (440–––840 W, in 100 W increments) is defined as the received electrical input power and time intervals (10–––90 min, in 20 min intervals). Following a final centrifugation (5000 rpm, 5 min), the supernatant was collected, and its volume was measured and recorded. Using the comprehensive evaluation value (CEV) of the final extract as an index, the process variables were systematically selected and optimized to determine the optimal single-factor conditions.

Fig. 2.

The extraction process of UAESE..

2.4.3. Calculation of comprehensive evaluation value

The Analytic Hierarchy Process (AHP) is a research method that integrates qualitative and quantitative analyses to assign weights in solving intricate multi-criteria problems [29], [30]. Nevertheless, AHP is prone to relatively subjective judgments, which is considered its main drawback. By contrast, the Criteria Importance Through Inter-criteria Correlation (CRITIC) method determines weights objectively by quantifying both the variability and the conflict between criteria [31]. To this end, a combined AHP-CRITIC weighting method was developed by integrating the two approaches. This integrated approach systematically evaluated the UAESE process and identified its optimal parameters.

2.4.4. Screening of significant factors by PBD

When a response is influenced by numerous variables, it is crucial to identify the significant factors, guided by preliminary single-factor experiments. PBD, a two-level experimental design method, can effectively filter out significant factors from numerous variables using a limited number of experiments, and then determine the optimal combination of factors for further verification and optimization. To screen for significant variables, we used a PBD to evaluate 10 factors affecting the extraction yield of Celastrus orbiculatus, including percentage of ethanol in SUPRAS (A), the pH of the water used in the synthesis of SUPRAS (B), enzyme concentration (C), enzymolysis pH(D), enzymolysis time (E), enzymolysis temperature (F), extraction time (G), extraction temperature (H), ultrasonic power (J), and liquid–solid ratio (K).

2.4.5. Analysis of optimization conditions by BBD

Based on the PBD, we employed response surface methodology (RSM), using a BBD, to optimize four key parameters for Celastrus orbiculatus extraction, including enzymolysis pH (A), enzymolysis temperature (B), extraction temperature (C), and liquid–solid ratio (D). A three-level BBD was constructed in Design-Expert 13, where the CEV served as the response variable. Subsequently, the model's accuracy and goodness-of-fit were assessed through Analysis of Variance (ANOVA). Finally, following triplicate replication of all experimental groups for reliability, the optimal extraction process conditions were determined.

2.4.6. Determination of ultrasonic power via calorimetry

Following the experimental approach described by Bizzi et al. [32], the energy delivered to the solution was determined via calorimetry. The same ultrasonic extraction setup was employed, with the reaction vessel placed inside a thermally insulated calorimetric jacket positioned at the center of the ultrasonic cleaning bath to minimize environmental heat exchange. The initial temperature rise was monitored at room temperature using a DM‑1047 digital thermometer (DELIXI, China), immersed in the solution and maintained at the mid‑height of the liquid column. After recording the initial steady‑state temperature without ultrasound, ultrasonic irradiation was initiated and maintained at a fixed power setting for a defined duration, while the temperature was recorded continuously over time. Triplicate experiments were performed to obtain averaged values. The delivered ultrasonic power was calculated according to the following equation:

$${\mathrm{P}\mathrm{o}\mathrm{w}\mathrm{e}\mathrm{r}}_{\mathrm{U}\mathrm{S}}\left(\mathrm{W}\right)={\mathrm{C}}_{\mathrm{P}}\bullet \mathrm{M}\bullet \left(\frac{\mathrm{d}\mathrm{T}}{\mathrm{d}\mathrm{t}}\right)$$ (1)

where Cp is the specific heat capacity of water (4.2 J/g), M is the mass of water (g), and dT/dt represents the rate of temperature increase per second.

2.5. Comparison of extraction methods and SEM analysis

This experiment aimed to evaluate the effect of the UAESE and to explore its extraction mechanism. We compared the UAESE with four alternatives: UAEE, ultrasound − assisted SUPRAS extraction (UASE), ultrasound − assisted 30% (v/v) ethanol extraction, and absolute ethanol ultrasound extraction. Except for the type of solvent, the extracts of the latter four methods were obtained using the optimal extraction process of the UAESE. In addition, when multiple samples are processed in a single run, the energy consumption per sample is calculated by dividing the total energy requirement of the device by the number of samples treated concurrently [33]. The energy consumption of each extraction method was evaluated using the following formula:

$$\mathrm{E}(\mathrm{W}\mathrm{h})={\mathrm{P}\mathrm{o}\mathrm{w}\mathrm{e}\mathrm{r}}_{\mathrm{U}\mathrm{S}}\bullet \mathrm{t}$$ (2)

where E is the energy consumption (Wh), Power_US represents the ultrasonic power delivered into the solution (W), and t is the extraction duration (h).

Scanning electron microscopy (SEM, Hitachi SU8010, tungsten filament) was used to examine the powder microstructure, including particle dimensions, surface morphology, and geometric form. Microstructural comparison at 5000 × magnification was made between the raw material and samples subjected to different extraction methods.

2.6. Enrichment of CEL and PRI by macroporous resin adsorption chromatography

2.6.1. Resin screening

To identify the optimal resin, we selected three macroporous adsorption resins as candidates, namely D101 (non-polar), AB-8 (weakly polar), and NKA-9 (polar) resins. The macroporous resins were pretreated based on a literature method, with minor adjustments [34], [35]. Briefly, the resins were first immersed in absolute ethanol (at least double their volume) for 24 h. They were then thoroughly washed with distilled water until the eluent became colorless, sequentially treated with 5% HCl (4 h) and 5% NaOH (4 h), and finally rinsed with distilled water to neutrality between and after each treatment. Afterward, in 25 mL conical flasks, 1.0 g of pretreated resin was combined with 10 mL of the Celastrus orbiculatus extract at its initial concentration (C₁). Adsorption equilibrium was achieved by agitating the flasks at 25°C and 150 rpm for 24 h. Following filtration of the adsorbed solutions, the filtrate volume (V₁) was measured, and the concentrations of CEL and PRI (C₂) were detected. Desorption was performed by adding 25 mL of absolute ethanol to the mixture and then shaking (150 rpm) at 25°C for 24 h. Following vacuum filtration, the volume of the desorption solution (V₂) was recorded, and the mass concentrations of CEL and PRI (C₃) were determined. The adsorption and desorption rates [36] are defined by Equations (2), (3), respectively.

$$\mathrm{A}\mathrm{d}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{R}\mathrm{a}\mathrm{t}\mathrm{e}=\left(C_1-C_2\right)/C_1\times 100\%$$ (3)

$$\mathrm{D}\mathrm{e}\mathrm{s}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{R}\mathrm{a}\mathrm{t}\mathrm{e}=\left(C_3\times V_2\right)/\left(C_1\times V_1-C_2\times V_2\right)\times 100\%$$ (4)

C₁, C₂, C₃: Concentrations of CEL and PRI (mg/mL) before adsorption, after adsorption, and desorption solution, respectively; V₁: Adsorption solution volume (mL); V₂: Desorption solution volume (mL).

2.6.2. Experiment on static adsorption kinetics

2.0 g of pretreated resin was weighed into a 50 mL conical flask, with three parallel replicates prepared per group. Then, 20 mL of the crude extract, whose concentration had been predetermined, was introduced into the flask. The sample was subsequently shaken (150 rpm) at 25°C for 2 h. Samples were collected from the incubated mixture at 20 min intervals to determine the residual concentrations of CEL and PRI.

2.6.3. Dynamic adsorption and desorption experiments

Dynamic adsorption and desorption experiments for CEL and PRI were performed using a glass column (1.5 cm i.d. × 30 cm height). We packed the pretreated AB-8 resin into a glass column to a final bed volume of 15 mL. A 100 mL volume of crude extract of known concentration was loaded at 1.5 mL/min, and the effluent was collected as ten 10 mL fractions. The concentrations of CEL and PRI in each fraction were determined individually. The column was first eluted with 6 bed volumes (BV, 1 BV = 15 mL) of distilled water. A stepwise ethanol gradient (25%, 50%, 75%, equal volumes of each) was then applied at 1.0 mL/min. Subsequently, the target eluate containing CEL and PRI was pooled. After vacuum concentration, the contents of CEL and PRI in the eluate were quantified.

2.7. Antimicrobial activity

2.7.1. Detection of Minimum inhibitory concentrations (MICs) and Minimum bactericidal concentrations (MBCs)

With reference to the standards of the Clinical and Laboratory Standards Institute (CLSI) [37], the broth microdilution method was employed to assess the MIC of CEL, PRI, the purified and enriched Celastrus orbiculatus extract, and a mixture of CEL and PRI at a ratio equivalent to that in the purified extract. Then, we employed 0.1% (w/v) resazurin to assess bacterial viability. The assay was conducted in U-bottom 96-well plates. For the MIC assays, the Staphylococcus aureus ATCC BAA-1717 (USA300) (MRSA) strain was harvested during the logarithmic growth phase and diluted in sterile Mueller-Hinton (MH) broth to a final concentration of 10⁶ CFU/mL. Two control groups were established: a negative control with sterile medium only and a positive control containing the bacterial suspension. For MBC determination, we transferred 100 μL aliquots from clear wells onto sterile TSA and incubated at 37°C for 24 h. All experiments were conducted with three independent replicates.

2.7.2. SYTO 9/PI staining assay of MRSA

The MRSA suspensions treated with different drugs were double-stained using a SYTO 9/PI staining for visual assessment of antibacterial effects. The specific method was as follows: different concentrations of drugs (1/2 MIC PRI, 1/2 MIC CEL, 1/2 MIC PRI + 1/2 MIC CEL, and the purified enriched extract) were added to the bacterial suspensions in PBS (10⁸ CFU/mL). Meanwhile, bacteria treated with PBS alone constituted the control group. The samples were then kept static at 37°C for 12 h. Samples were stained with propidium iodide (PI) and SYTO 9 (1.0 μL each), then incubated in the dark for 15 min. Following staining, the bacterial were PBS-washed and prepared for imaging with a Leica inverted fluorescence microscope.

2.8. Statistical analysis

Data are expressed as means ± standard deviation (n = 3) from at least three independent experiments. Statistical analysis was conducted with GraphPad Prism 8.0.2 for single-factor experiments, Minitab 21 for Plackett-Burman designs, and Design-Expert 13 for regression analysis and response surface optimization. Data were analyzed by one-way ANOVA, with statistically significance set at P < 0.05.

3. Results and discussion

3.1. SUPRAS screening

SUPRAS exhibits structural adjustability, as its properties are governed by the composition of amphiphilic compounds, thereby establishing a direct structure–function relationship that determines extraction efficiency. Based on cost and environmental considerations, this study employed octanoic acid and ethanol as a sustainable substitute for conventional hazardous solvents like tetrahydrofuran. The extraction rates for the two target components from Celastrus orbiculatus by the two types of SUPRAS are shown in Fig. S1A. SUPRAS-1 (octanoic acid) had higher extraction rates of CEL and PRI than SUPRAS-2 (decanoic acid), among which the extraction rate of PRI was the highest 5.16 mg/g. The variation in extraction efficiency arises from differences in the solvent system's hydrogen-bonding profile. This alters key physicochemical properties − including viscosity, polarity, and density − thereby affecting the solvent's capacity to recover target compounds [38]. Consequently, SUPRAS-1 was selected for subsequent experiments.

3.2. Phase diagram

Initially, upon exceeding their critical aggregation concentration, the amphiphilic molecules spontaneously form reverse micelles or vesicles [39]. Subsequently, under specific external conditions (pH, temperature, solvent, electrolyte), these nanostructures further organize into larger aggregates that ultimately undergo coacervation and separate from the bulk solvent [40]. This work examined the influence of octanoic acid and ethanol concentrations on the phase behavior and formation of SUPRAS. Fig. S1B illustrates the phase diagram of the octanoic acid–ethanol-water ternary system. In the phase diagram (Fig. S1B), SUPRAS formation occurs in the region below the curve, while a homogeneous solution exists above it. A higher octanoic acid content expands this SUPRAS-forming region but necessitates more ethanol. Based on this analysis, a fixed octanoic acid concentration of 5% (v/v) was selected for subsequent experiments.

3.3. Characterization of SUPRAS

3.3.1. Phase ratio

The phase ratio quantifies the efficiency of octanoic acid in forming the SUPRAS phase in the ethanol–water system. Fig. S1C presents the effects of the addition amounts of octanoic acid on the phase ratio. At a fixed ethanol concentration (30%, v/v), the phase ratio exhibited a linear increase with octanoic acid content (3% − 7%, v/v) under SUPRAS-forming conditions. With octanoic acid fixed at 5% (v/v), the phase ratio initially increased with ethanol concentration over the 15–––32% (v/v) range. Beyond a threshold ethanol content, the phase ratio decreases sharply, accompanied by a progressive blurring of the phase interface (Fig. S1D). Increasing the ethanol content reshapes the hydrogen-bonding interactions between octanoic acid, ethanol, and water molecules [41]. This shifts the composition of the resulting SUPRAS, increasing its water-to-ethanol ratio while decreasing the octanoic acid fraction [42]. The excess ethanol then dissolves into the lower aqueous phase, leading to a convergence in density between the two phases and thereby blurring the phase interface.

3.3.2. The formation of SUPRAS

As depicted in fig. S2, the SUPRAS phase displays the color of the dye, while the water phase does not. The simultaneous extraction of both hydrophobic (Nile red) and hydrophilic (fluorescein sodium) dyes into the SUPRAS phase confirms its dual affinity and enrichment capability. The generated droplets were examined using an inverted fluorescence microscope for further structural confirmation of the SUPRAS. Obviously, most of the sodium fluorescein was encapsulated within the spherical micellar aggregates. (green fluorescence, Fig. 3C), with no fluorescence detected on the exterior. Conversely, Nile Red partitioned onto the surfaces of the spherical micelles rather than penetrating their interiors (red fluorescence, Fig. 3B). These results support a structural model of SUPRAS as a reverse micelle system, with molecules self-assembling into aggregates that possess a hydrophilic core and a hydrophobic exterior. The above results have also been confirmed in previous studies on SUPRAS [43]. In addition, optical microscopy revealed that the SUPRAS comprised numerous spherical droplets (Fig. 3A), consistent with prior characterizations in the literature [44].

Fig. 3.

LSCFM images of octanoic acid–ethanol-water SUPRAS. The conditions of SUPRAS synthesis: octanoic acid = 5% (v/v), ethanol = 30% (v/v); Fluorescence dye: Nile Red (B) and fluorescein sodium (C).

3.3.3. Mechanism of SUPRAS formation

FTIR analysis of octanoic acid revealed key molecular features: the strong C-H stretching vibration at 2926 cm^–1 corresponds to its hydrocarbon chain, while the carbonyl stretching at 1707 cm^–1 characterizes the carboxyl head group (Fig. S3). After dissolving octanoic acid in ethanol, the characteristic peaks of SUPRAS remained present. However, substantial broadening was observed for the ν(O-H) peak of ethanol at 3322 cm^–1. Additionally, the intensity and shape of the C-H stretching vibrations (around 2926 cm^–1) were altered. These spectral changes indicate the presence of hydrogen-bonding interactions within the ethanol-octanoic acid system. Besides, the significant broadening of the ν(O-H) and ν(C=O) peaks in SUPRAS indicates the formation of more hydrogen bonds. The enhanced hydrogen bonding limits solvation of the acid's polar head groups, which drives tighter molecular packing and leads to the growth of reverse micellar aggregates. The attenuation of the ν(O-H) and ν(C=O) stretching peaks indicates restricted vibration of octanoic acid's carboxyl groups due to hydrogen bonding. This intermolecular interaction drives octanoic acid to spontaneously organize into reverse micelles.

3.4. Results of single-factor experiments

3.4.1. Effect of the ethanol percentage and pH

Initially, we investigated the effect of ethanol concentration on the extraction efficiency, using the SUPRAS. Fig. 4B demonstrates how the ethanol concentration used in SUPRAS synthesis influences the extraction efficiency of active components. At 30% (v/v) ethanol, the CEV was highest among all tested groups. This condition also yielded the highest PRI extraction (4.16 mg/g). The extraction rates of the target components decreased to varying degrees when the ethanol content deviated from the optimum of 30% (v/v). This result is attributed to the increase in SUPRAS formation with higher ethanol content, which dilutes the target compounds and thereby reduces extraction efficiency [45]. The observed increase in extraction rate at 15% (v/v) ethanol is likely due to the enhanced solubility of low-polarity triterpenoids in the SUPRAS, which at this point contains the highest proportion of octanoic acid. To ensure extraction efficiency and minimize solvent consumption, 30% (v/v) ethanol was determined to be the optimal extraction condition.

The effect of pH on extraction efficiency was further investigated by evaluating values across a range from 2 to 6. The maximum CEV was achieved at a pH of 4, indicating a strong dependence of extraction efficiency on aqueous phase acidity (Fig. 4C). At moderately acidic pH, triterpenoids remain predominantly protonated, facilitating their extraction via hydrogen bonding and hydrophobic interactions into the SUPRAS phase. However, further pH elevation promotes ionization of the carboxyl groups, enhancing their hydrophilicity and thus reducing partition into the SUPRAS. Consequently, an optimal pH of 4.0 was identified and employed for the SUPRAS synthesis.

3.4.2. Effect of enzyme concentration

Natural active compounds in plants are usually enclosed by cell walls, which, due to their tight structure, complex composition, and high mechanical strength, both support and protect the cells, acting as a natural barrier to bioactive substances. Cellulase (Cel) degrades cellulose by cleaving its β-1,4-glycosidic linkages [46]. This compromises the integrity of the cell wall, facilitating the release of intracellular constituents. As exhibited in Fig. 4D, the highest CEV value was achieved at 6% Cel, showing a positive correlation with enzyme concentration from 2% to 6%. However, beyond 6% Cel concentration, the extraction rate decreased progressively. An optimal cellulase concentration maximizes extraction by effectively disrupting the cell wall. Beyond this optimum, increased viscosity hampers mass transfer and molecular diffusion, thereby reducing the extraction yield [47]. From the perspective of cost saving, the optimal Cel concentration was selected as 6%.

3.4.3. Effects of enzymolysis pH, time and temperature

The efficiency of ultrasound-assisted enzymatic extraction is influenced by critical process parameters such as pH, temperature, and time [48]. Studies have shown that the suitable pH range for Cel to hydrolyze cellulose is 4.0–6.0 [49]. Therefore, the system pH should be maintained at the enzyme's optimum throughout enzymolysis to ensure maximal catalytic efficiency. As shown in Fig. 4E, the CEV peaks at pH 5.0, which corresponds to the established optimum for Cel activity and confirms its critical role in achieving efficient extraction. The observed decline in CEV and extraction efficiency as the pH rose to 6.0 can be attributed to the deviation from Cel's optimal pH (5.0), which diminishes enzymatic activity and consequently reduces extraction yield. Consequently, the optimal pH for enzymolysis was identified to be 5.0.

The content of active components changes significantly over the course of an enzymatic reaction. As depicted in Fig. 4F, the CEV increased with prolonged enzymolysis, peaking at 50 min before declining. The results indicated that the prolonged enzymatic treatment did not enhance efficacy but instead reduced the yield of target constituents. This trend was also evident in the ultrasound-assisted β-cyclodextrin extraction of bioactive compounds from Forsythia suspensa [50]. Additionally, enzymatic activity is highly sensitive to temperature. While a temperature increase within the optimal range enhances enzyme activity, deviation from this range reduces it, but excessively high temperatures induce structural changes in the enzyme, which impairs hydrolysis and degrades bioactive compounds. Fig. 4G depicts the dependence of reaction efficiency on enzymatic hydrolysis temperature. Within the range of 35℃ to 45℃, the CEV indicates that Cel activity exhibited an increasing trend, reaching its peak at 45°C. However, the CEV at 45°C did not differ significantly from that at 40°C (P > 0.05), indicating that Cel activity is maximized at 40°C. Moreover, beyond a specific threshold, the enzyme denatures, leading to a loss of catalytic efficiency [51]. Therefore, based on enzymolysis efficiency and actual production conditions, the optimal enzymatic hydrolysis conditions were selected to be 40°C for 50 min.

3.4.4. Effects of ultrasound-assisted extraction time and temperature

The yield of bioactive compounds was significantly influenced by extraction time (Fig. 4H). The CEV increased from 3.04 to 3.61 as ultrasonic time rose from 10 to 70 min, but decreased upon further extension. The rise in CEV is attributed to ultrasound-induced cell wall disruption via mechanical vibration and cavitation, which enhances the release of active components. Furthermore, during ultrasonication, the reverse micelle structure of SUPRAS diffuses cells more efficiently and interacts with the target analyte [52]. However, prolonged extraction can promote the dissolution of impurities, reducing the purity of the extract. Therefore, optimal ultrasound-assisted extraction required a duration of 70 min.

Fig. 4I illustrates the relationship between ultrasound-assisted extraction temperature and the extraction yield. The CEV rose with temperature across a 20°C − 40°C range, reaching a maximum of 3.69. Further temperature increase led to a reduction in extraction value. This is a phenomenon similar to that observed for triterpenoids from Chaenomeles speciosa leaves in ultrasound-assisted extraction [53]. The reason is that as the temperature increases during ultrasonication, the intensified thermal motion of molecules synergizes with ultrasound's mechanical vibrations to facilitate the dissolution of triterpenoid components, thus boosting extraction efficiency. Nonetheless, thermal degradation of the target compounds at excessively high temperatures can explain the observed reduction in content. Then, 40°C was chosen for extraction, which gave the highest extraction efficiency.

3.4.5. Effect of ultrasonic power

The CEV correlated significantly with the applied ultrasound power intensity (Fig. 4J). The CEV reached its peak value of 4.43 at an ultrasonic power of 640 W. At this power level, ultrasound may induce intense cavitation, along with the formation of micro-jets and shock waves. Appropriately elevated ultrasonic power facilitates cell wall disintegration. This rupture frees the target compounds, thereby enhancing their dissolution and facilitating faster mass transfer within the solvent [54]. However, a slight decrease in efficiency was observed at the ultrasonic power of 740 W. This is presumably because excessive cavitation generated an overabundance of bubble layers, or rapid ultrasonic heating promoted the degradation and transformation of the target compounds. Based on energy consumption and extraction efficiency, this study identified 640 W as the optimal ultrasonic power.

3.4.6. Effect of the liquid–solid ratio

As depicted in Fig. 4K, the liquid–solid ratio significantly affected the yield of Celastrus orbiculatus. The CEV peaked at 4.03 at a liquid–solid ratio of 20 mL/g. The lower efficiency at reduced liquid–solid ratios is due primarily to limited solvent − material contact area, which hinders triterpenoid dissolution and diffusion. An increased liquid–solid ratio provides a greater contact area between the raw material and solvent, which facilitates more complete mass transfer and dissolution of the target compounds, thereby enhancing the extraction efficiency [55]. The observed decline in CEV occurs once triterpenoid extraction is largely complete, as any additional solvent merely dilutes the compounds and raises operational costs. Therefore, the optimal liquid solid ratio was 20 mL/g.

3.5. Evaluation of extraction efficiency based on AHP-CRITIC model

The AHP-CRITIC approach was applied for weighting factor determination in assessing comprehensive extraction performance across varied conditions. In this experiment, the extraction rates of CEL and PRI, as well as the priority order of phase ratio were utilized as weight indicators. Through subjective analysis, the following priority order was established: extraction rate of CEL > extraction rate of PRI > phase ratio. According to the AHP theoretical judgment matrix (Table S3), we constructed a pairwise comparison matrix and a set of relative scores was given (Table S4). The weight coefficients for the CEL extraction rate, PRI extraction rate, and the phase ratio were 0.6491, 0.2790, 0.0719, respectively. The calculated consistency index (CI) was 0.032, with a random consistency index (RI) of 0.525. The results showed a random consistency ratio (CR) of 0.062, which is below the threshold of 0.10. This indicates that the consistency test was satisfied, confirming the accuracy and consistency of the weighting coefficients.

Subsequently, the CRITIC method was applied to calculate the weighting factors for secondary indicators. The original experimental data were normalized using the SPSSPRO (https://www.spsspro.com) system, after which a CRITIC analysis was performed. This analysis yielded objective weight coefficients of 0.0192, 0.9103, and 0.0705 for the CEL extraction rate, PRI extraction rate, and phase ratio, respectively. The AHP-CRITIC method was applied to combine expert judgment with data-driven analysis, generating the comprehensive weighting coefficients [56]. The weighting coefficient, W, was determined from the following equation: $W=W_{\text{AHP}}{W}_{\text{CRITIC}}/\sum W_{\text{AHP}}{W}_{\text{CRITIC}}(5)$ (where W, W_AHP, and W_CRITIC denote the subjective weight, the objective weight, and the final comprehensive weight, respectively). The final weight coefficients for the extraction rates of CEL and PRI, as well as the phase ratio, were determined as 0.0464, 0.9349, and 0.0187, respectively. Subsequently, a comprehensive evaluation value was determined using formula to guide the optimization process: $\mathrm{C}\mathrm{E}\mathrm{V}=0.0464X_1+0.9349X_2+0.0187X_3(6)$ (where CEV denotes the comprehensive evaluation value; X₁ and X₂ represent the extraction yields of CEL and PRI, respectively; and X₃ is the phase ratio).

3.6. PBD to screen for significant factors

The complexity of optimizing the UAESE, with its many factors, necessitated a PBD design to identify key variables for simplification. Table 1, Table 2, Table 3 list the factor levels, experimental design and results, and ANOVA, respectively. The statistical significance of the model (P < 0.05, Table 3) confirms the reliability of the results. R² = 0.9512 and R²_adj = 0.8969, indicating that the predicted value of the regression model can achieve an accuracy of 89.69%. The coefficient of variation (CV) is a key indicator of experimental reliability. The obtained CV of 2.35% in the PBD experiment thus confirms its high reliability. The correlation between the CEV and the factors can be expressed by the equation: $\mathrm{Y}=4.14+0.0125\mathrm{A}-0.0020\mathrm{B}+0.0.0384\mathrm{C}+0.0900\mathrm{D}-0.0041\mathrm{E}+0.1144\mathrm{F}-0.0098\mathrm{G}+0.1487\mathrm{H}-0.0227\mathrm{J}+0.3100\mathrm{K}$

Table 1.

Variables and levels of Plackett-Burman design for Celastrus orbiculatus extraction.

Variables	Coded symbols	Levels
		−1	+1
Percentage of ethanol in SUPRAS	A	25	35
Water pH in SUPRAS	B	3	5
Enzyme concentration	C	4	8
Enzymolysis pH	D	4.5	5.5
Enzymolysis time	E	30	70
Enzymolysis temperature	F	40	50
Extraction time	G	50	90
Extraction temperature	H	30	50
Ultrasonic power	J	540	740
Liquid-solid ratio	K	10	30

Table 2.

Different combinations of PBD variables and the values of CEV.

Run order	Variables										Indicator
	A	B	C	D	E	F	G	H	J	K	CEV
	(%)	−	(%)	−	(min)	(℃)	(min)	(℃)	(W)	(mL/g)
1	+1	−1	+1	+1	−1	−1	−1	−1	+1	−1	3.730
2	+1	+1	+1)	−1	−1	+1	−1	−1	−1	+1	4.170
3	−1	+1	−1	+1	+1	+1	−1	−1	−1	−1	3.796
4	−1	−1	+1	+1	+1	−1	−1	+1	−1	−1	4.079
5	+1	−1	+1	−1	−1	+1	+1	+1	−1	−1	4.096
6	+1	+1	−1	−1	+1	+1	+1	−1	+1	−1	3.758
7	+1	+1	−1	+1	+1	−1	−1	+1	+1	+1	4.455
8	+1	+1	+1	+1	−1	−1	+1	+1	−1	+1	4.503
9	−1	+1	+1	+1	−1	+1	+1	−1	+1	−1	3.783
10	+1	−1	+1	+1	+1	+1	−1	−1	+1	+1	4.382
11	−1	+1	+1	−1	+1	+1	−1	+1	−1	+1	4.450
12	+1	−1	−1	+1	+1	+1	+1	+1	−1	−1	4.115
13	−1	+1	+1	−1	+1	−1	+1	+1	+1	−1	3.853
14	−1	−1	−1	+1	−1	+1	+1	+1	+1	+1	4.568
15	−1	−1	+1	−1	+1	−1	+1	−1	+1	+1	3.876
16	−1	−1	−1	−1	−1	+1	−1	+1	+1	+1	4.312
17	+1	−1	−1	−1	+1	−1	+1	−1	−1	+1	3.860
18	+1	+1	−1	−1	−1	−1	−1	+1	+1	−1	3.722
19	−1	+1	−1	+1	−1	−1	+1	−1	−1	+1	4.156
20	−1	−1	−1	−1	−1	−1	−1	−1	−1	−1	3.668

Table 3.

Results of the Plackett-Burman design.

Source	Sum of squares	df	Mean square	F-Value	P-Value
Model	1.60	10	0.16	17.53	0.0001***
A	0.0031	1	0.003	0.34	0.57
B	0.0001	1	0.0001	0.01	0.93
C	0.01	1	0.01	1.43	0.26
D	0.16	1	0.16	17.71	0.002**
E	0.0003	1	0.0003	0.04	0.85
F	0.12	1	0.12	12.73	0.01*
G	0.002	1	0.002	0.21	0.66
H	0.44	1	0.44	48.33	< 0.0001***
J	0.01	1	0.01	1.13	0.32
K	0.85	1	0.85	93.37	< 0.0001***
Residual	0.08	9	0.01
Cor Total	1.69	19
R2	0.9512	Radj2	0.8969

Note: ***P < 0.001; **P < 0.01; *P < 0.05.

The Pareto chart of standardized effects visually represents the significant influencing factors. As presented in Fig. S4, four factors significantly influenced the CEV (P < 0.05): enzymolysis pH (D), extraction temperature (H), enzymolysis temperature (F), and liquid–solid ratio (K), and these factors accounted for more than 80% of the total effect. The Pareto principle states that roughly 20% of factors account for 80% of outcomes in most systems, highlighting the need to prioritize these key factors for greater efficiency [57]. Therefore, in the subsequent BBD experiment optimization stage, the enzyme concentration (C), ultrasonic power (J), percentage of ethanol in SUPRAS (A), extraction time (G), enzymolysis time (E), and the pH in the synthesis of SUPRAS (B) were fixed at 6%, 640 W, 30%, 70 min, 50 min, and pH 4.0, respectively.

3.7. Process optimization using BBD

Based on the PBD experiment, four significant influencing parameters, namely enzymolysis pH (A), enzymolysis temperature (B), extraction temperature (C), and liquid–solid ratio (D), were selected. The BBD, which comprised 29 experimental runs, was further adopted to perform response surface optimization with CEV as the evaluation index. Table 4 shows the independent and dependent variables used in the RSM design. We employed a quadratic response surface model to optimize the process conditions and to examine the relationships between the variables and responses. The models were statistically significant (P < 0.05) with an F value of 23.23, with no significant lack of fit. With R² = 0.9587 and R²_adj = 0.9175, the model demonstrates a high degree of accuracy in fitting the experimental data (Table 5). The response surface diagram depicted the variations in CEV across the four variables. We used multiple regression analysis to fit the data and develop a second-order polynomial equation. The relationship between the independent and response variables was expressed by the following equation:.$\mathrm{Y}=3.63-0.0686\mathrm{A}-0.1712\mathrm{B}+0.0473\mathrm{C}-0.1148\mathrm{D}-0.0178\mathrm{A}\mathrm{B}-0.0110\mathrm{A}\mathrm{C}+0.1119\mathrm{A}\mathrm{D}+0.0467\mathrm{B}\mathrm{C}+0.0260\mathrm{B}\mathrm{D}-0.0240\mathrm{C}\mathrm{D}-0.1850{\mathrm{A}}^2-0.1110{\mathrm{B}}^2-0.3147{\mathrm{C}}^2-0.0356{\mathrm{D}}^2$

Table 4.

Independent variables, their levels for the BBD, and the responses obtained.

Run order	Variables				Indicator
	Enzymolysis pH (A)	Enzymolysis temperature (B)	Extraction temperature (C)	Liquid-solid ratio (D)	CEL	PRI	CEV
	−	(℃)	(℃)	(mL/g)	(μg/g)	(mg/g)	−
1	4.5(−1)	40(−1)	40(0)	20(0)	88.58	3.88	3.63
2	5.5(+1)	40(−1)	40(0)	20(0)	81.44	3.51	3.45
3	4.5(−1)	50(+1)	40(0)	20(0)	81.92	3.59	3.32
4	5.5(+1)	50(+1)	40(0)	20(0)	77.59	3.39	3.07
5	5.0(0)	45(0)	30(−1)	10(−1)	79.85	3.63	3.40
6	5.0(0)	45(0)	50(+1)	10(−1)	78.60	3.71	3.50
7	5.0(0)	45(0)	30(−1)	30(+1)	75.57	3.44	3.17
8	5.0(0)	45(0)	50(+1)	30(+1)	75.35	3.41	3.18
9	4.5(−1)	45(0)	40(0)	10(−1)	84.54	3.94	3.71
10	5.5(+1)	45(0)	40(0)	10(−1)	85.96	3.53	3.30
11	4.5(−1)	45(0)	40(0)	30(+1)	77.74	3.43	3.25
12	5.5(+1)	45(0)	40(0)	30(+1)	70.27	3.52	3.29
13	5.0(0)	40(−1)	30(−1)	20(0)	79.26	3.55	3.30
14	5.0(0)	50(+1)	30(−1)	20(0)	66.32	3.05	2.86
15	5.0(0)	40(−1)	50(+1)	20(0)	85.73	3.64	3.41
16	5.0(0)	50(+1)	50(+1)	20(0)	74.12	3.42	3.16
17	4.5(−1)	45(0)	30(−1)	20(0)	66.82	3.31	3.12
18	5.5(+1)	45(0)	30(−1)	20(0)	68.95	3.30	3.12
19	4.5(−1)	45(0)	50(+1)	20(0)	75.81	3.39	3.17
20	5.5(+1)	45(0)	50(+1)	20(0)	72.70	3.34	3.13
21	5.0(0)	40(−1)	40(0)	10(−1)	92.16	4.05	3.77
22	5.0(0)	50(+1)	40(0)	10(−1)	78.43	3.45	3.39
23	5.0(0)	40(−1)	40(0)	30(+1)	86.64	3.82	3.54
24	5.0(0)	50(+1)	40(0)	30(+1)	76.77	3.48	3.26
25	5.0(0)	45(0)	40(0)	20(0)	87.18	3.87	3.63
26	5.0(0)	45(0)	40(0)	20(0)	88.66	3.97	3.69
27	5.0(0)	45(0)	40(0)	20(0)	88.41	3.85	3.60
28	5.0(0)	45(0)	40(0)	20(0)	87.23	3.86	3.60
29	5.0(0)	45(0)	40(0)	20(0)	91.68	3.89	3.65

Table 5.

Analysis of variance of regression equation for Box-Behnken Design.

Source	Sum of squares	df	Mean square	F-Value	P-Value	Significant
Model	1.43	14	0.10	23.23	< 0.0001	***
A	0.06	1	0.06	12.85	0.003	**
B	0.35	1	0.35	80.00	< 0.0001	***
C	0.03	1	0.03	6.10	0.03	*
D	0.16	1	0.16	35.98	< 0.0001	***
AB	0.001	1	0.001	0.29	0.60	−
AC	0.001	1	0.001	0.11	0.75	−
AD	0.05	1	0.05	11.38	0.005	**
BC	0.01	1	0.01	1.99	0.18	−
BD	0.003	1	0.003	0.62	0.45	−
CD	0.002	1	0.002	0.52	0.48	−
A2	0.22	1	0.22	50.49	< 0.0001	***
B2	0.08	1	0.08	18.16	0.001	**
C2	0.64	1	0.64	146.06	< 0.0001	***
D2	0.01	1	0.01	1.87	0.19	−
Residual	0.06	14	0.004
Lack of fit	0.06	10	0.01	4.46	0.08	−
Pure error	0.01	4	0.001
Cor total	1.49	28
R2	0.9587	Radj2	0.9175

Note: ***P < 0.001; **P < 0.01; *P < 0.05; −P > 0.05.

Fig. 8 displayed the interactions among variables and the maximum value's location via a 3D response surface (Fig. 5A-F) and a 2D contour diagram (Fig. 5a-f). Within the study range, CEV showed significant sensitivity to A, B, C, and D (P < 0.05), which was consistent with the findings of analysis of variance. A significant interaction between factors A and D (P < 0.05) was evident from the pronounced curvature of the 3D response surface and the elliptical contour lines (Fig. 5C). However, consistent with the ANOVA, the interaction terms of AB, AC, BC, BD, and CD were not significant (P > 0.05), as visually evidenced by the flat 3D response surfaces and concentric circular contours. The BBD model predicted the following optimal conditions: enzymolysis pH 5.0, enzymolysis at 41.21°C, extraction at 40.03°C, and a liquid–solid ratio of 11.70 mL/g, with a corresponding CEV of 3.79. The optimization, for practical laboratory feasibility, yielded the following parameters: liquid-to-solid ratio, 12 mL/g; enzymolysis temperature, 41°C; extraction temperature, 40°C. To validate the BBD model, three independent experiments were conducted under these adjusted conditions. According to Table 6, experimental results agreed closely with model predictions, yielding a mean CEV of 3.808 ± 0.003 (RSD = 0.07%) under optimized conditions. These results confirm the reliability and effectiveness of the model.

Fig. 8.

Anti-MRSA activity of CEL and PRI in vitro. MICs of CEL and PRI against MRSA(A); MICs of the standards mixture and the purified extract against MRSA(B); MBCs of CEL and PRI against MRSA(C); MBCs of the standards mixture and the purified extract against MRSA(D). PC: positive control (MH containing MRSA without drug); NC: negative control (MH containing drug without MRSA); BC: blank control (MH only).

Fig. 5.

The 3D surface plots (A-F) and contour plots (a-f) showing interactions of any two variables on CEV. A and B (A); A and C (B); A and D (C); B and D (D); Cand B (E); C and D (F).

Table 6.

Validation results of the best extraction conditions.

Extraction yield		CEV	RSD (%)	CEV mean value	CEV predicted value
CEL (μg/g)	PRI (mg/g)
72.445	4.068	3.808	0.07	3.808	3.787
57.642	4.066	3.806
62.855	4.072	3.811

3.8. Comparison of different extraction methods

3.8.1. Comparison of CEL and PRI yields

Based on the yields of CEL and PRI, we evaluated the synergistic efficacy of the UAESE method against three single extraction methods and conventional organic solvent extraction. According to Table 7, the UAESE outperforms the other single extraction methods in terms of CEL and PRI yields. This result confirms the synergy of combining enzyme pretreatment, SUPRAS, and ultrasound, which stems from the complementary and enhanced roles each component plays in the extraction process. Enzymes initiate the process by destroying the cell wall, thereby facilitating the liberation of intracellular constituents. Ultrasound then enhances this effect through cavitation-induced cell wall disruption and by generating turbulence that improves mass transfer. Concurrently, SUPRAS contributes by enhancing the solubility of the target compounds and promoting solvent penetration, which collectively accelerates the overall extraction efficiency. The energy delivered to the solution and the calculated energy consumption under ultrasonic power settings of 640 W and 840 W are presented in Table 7. Under the experimental conditions (ultrasonic frequency: 40 kHz), the measured dissipated acoustic power were 11 W and 15 W, respectively, representing the actual energy delivered to the solution. The corresponding energy consumption for each extraction method was further calculated as 12.50 Wh, 12.83 Wh, and 25.33 Wh, respectively. Furthermore, the established UAESE method was compared with other conventional extraction techniques reported in the literature. As summarized in Table 8, the proposed UAESE method achieved the highest extraction efficiency for PRI, approximately three times greater than that of the methanol-based UAE method. Moreover, this approach employs the environmentally friendly SUPRAS as an alternative to traditional organic solvents, operates under milder temperature conditions, and requires substantially shorter extraction times compared with reflux extraction. Although the extraction yield for CEL was relatively lower, the method enables the simultaneous extraction of both CEL and PRI while significantly enhancing the PRI yield, demonstrating its potential for efficient and sustainable extraction of bioactive compounds from Celastrus orbiculatus.

Table 7.

Comparison results of different extraction methods.

Extraction method	Extraction yield		Total extraction yield (mg/g)	Ultrasonic power delivered into the solution (W)	Energy consumption/sample (Wh)
	CEL (μg/g)	PRI (mg/g)
UAEE	−	−	−	15	12.50
UASE	84.65	5.40	5.48	11	12.83
Ultrasound − assisted 30% (v/v) ethanol extraction	−	−	−
Absolute ethanol ultrasound extraction	71.90	5.02	5.09
UAESE	91.23	5.53	5.62	15 (UAEE)	25.33
				11 (UASE)

Table 8.

Comparison between the novel UAESE process and other processes.

Reference	Matrix	Time	Temper-ature	Solvent	Yield (mg/g)		Extraction method
		(min)	(°C)	(type, volume)	CEL	PRI
[64]	Celastrus orbiculatus roots	3 times, 3 h each	78	95% Ethanol	0.90	NR	Reflux extraction
[65]	Celastrus orbiculatus	40	52	Methanol, 40 mL/g	NR	1.843	Ultrasound assisted extraction
[66]	Celastrus monospermus	20	46	Ethanol, 30 mL/g	3.116	NR	Ultrasound assisted extraction
[67]	Celastrus monospermus fruit	3 times, 2 h each	80	45% Ethanol, 8 mL/g	0.916	NR	Reflux extraction
[68]	Celastrus aculeatus roots	35	35	95% Ethanol, 8 mL/g	NR	0.014	Ultrasound assisted extraction
This work	Celastrus orbiculatus roots	50 (UAEE)	41 (UAEE)	SUPRAS, 12 mL/g	0.091	5.53	UAESE
		70 (UASE)	40 (UASE)

Note: NR stands for Not Reported.

3.8.2. SEM analysis

The synergistic effect of the combined enzyme-SUPRAS and ultrasound approach was further evidenced by distinct morphological changes in the plant powder before and after extraction. Before extraction, the surface of the medicinal material was smooth and undamaged (Fig. 6A). Following UAEE, the plant surface exhibited cracks of varying degrees compared with that before extraction (Fig. 6B). As illustrated in Fig. 6C, after ultrasound − assisted 30% (v/v) ethanol extraction, the cell wall showed a small degree of breakage, and the surface of the cell tissue was slightly damaged. As shown in Fig. 6D, the surface of the plant tissue treated with UASE showed signs of rupture and slight erosion. Ultrasonic extraction with absolute ethanol induced more substantial breakage to the plant surface than that with 30% (v/v) ethanol (Fig. 6E), potentially due to the greater damage inflicted by the higher ethanol concentration. In addition, the sample surface exhibited severe erosion and a large number of tissue fragments following UAESE (Fig. 6F). This phenomenon can be attributed to a synergistic mechanism: the enzyme disrupted the cell tissue, while the SUPRAS extended this damage through erosion. The implosion of cavitation bubbles under ultrasound generates intense turbulence and pressure shocks. These disturbances propagate strong collisions within the sample matrix, while associated thermal effects disperse micropores through molecular motion and diffusion, thereby inducing surface abrasion, dissolution, and structural damage [58]. In summary, the UAESE method effectively extracts key triterpenoids from Celastrus orbiculatus.

Fig. 6.

SEM images (5000 × ) of untreated powder powder (A) in different extraction methods; enzyme-assisted ultrasonic (B); 30% ethanol-assisted ultrasonic (C); SUPRAS-assisted ultrasonic(D); ethanol-assisted ultrasonic (E); ESU (F).

3.9. Enrichment of CEL and PRI

3.9.1. Selection of macroporous resins

The polarity, particle size, average pore size, and specific surface area of macroporous resins significantly influence their capacity to adsorption and desorption triterpenoids. In this study, three types of resins, D101, AB-8, and NKA-9, were evaluated using static adsorption–desorption experiments under identical conditions to identify the optimal resin. As can be seen from Fig. 7A, compared with D101 and NKA-9 resins, AB-8 resin showed better performance in adsorption rate and desorption rate. The static adsorption rates of AB-8 resin for CEL and PRI were 48.78% and 28.36%, respectively. This is because adsorption by macroporous resins is primarily a physical process whose performance is largely governed by the polarity of the resin and the physicochemical properties of the target compounds [59]. The triterpenoids from Celastrus orbiculatus are effectively adsorbed by AB-8 macroporous resin due to their comparable polarity and the resin's suitably large pore size. Furthermore, the AB-8 resin also showed superior desorption performance, with a PRI desorption rate of 35.97% that surpassed those of D101 and NKA-9, facilitating a more efficient release of the adsorbed triterpenoids. Based on this, AB-8 resin served as the adsorbent for purifying Celastrus orbiculatus triterpenoids.

Fig. 7.

Static adsorption and desorption ratio based on CEL and PRI by different resins (A); Static adsorption kinetic curve (B); Dynamic breakthrough curves (C); Gradient elution curve (D).

3.9.2. Static adsorption kinetics curve of resin

Kinetic analysis, commonly used to elucidate adsorption mechanisms, examines how adsorption rate changes over time. This approach pinpoints the critical equilibrium time. Considering the effect of pH on macroporous resin adsorption performance, we evaluated the adsorption and desorption of AB-8 resin for CEL and PRI under various pH conditions (Fig. S5). The static adsorption kinetics of AB-8 resin for triterpenoids were investigated at 25°C using a crude extract from Celastrus orbiculatus. Fig. 7B shows that the adsorption of CEL and PRI by AB-8 is not linearly related but multi-stage. The adsorption rate of CEL and PRI onto AB-8 resin increased rapidly within the first 20 min, slowed between 20 and 80 min, and stabilized after 80 min, approaching adsorption equilibrium. The observed kinetics are explained by the adsorption mechanism: a rapid initial stage governed by surface diffusion, a slower phase as compounds occupy internal pores and deplete sites and concentration gradients, and a final approach to equilibrium beyond 80 min as saturation is reached [60]. In summary, AB-8 is a fast-adsorbing resin, characterized by its high initial adsorption rate.

3.9.3. Dynamic adsorption and desorption of AB-8 resin

During dynamic adsorption on macroporous resin, breakthrough of triterpenoids occurs once the binding sites on the resin become saturated. The dynamic adsorption curve is also known as the breakthrough curve. The breakthrough point, which corresponds to the saturated sample volume, is defined as the moment when the effluent concentration of the target compounds reaches 10% of the initial concentration. The breakthrough curve indicated that 15 mL of AB-8 resin successfully processed 50 mL of a known concentration of Celastrus orbiculatus crude extract without breakthrough at a flow rate of 1.5 mL/min. (Fig. 7C). When the sample loading volume is further increased, CEL and PRI begin to break through significantly, indicating that the breakthrough point has been reached. Based on the breakthrough curve, a sample volume of 50 mL was determined to be optimal.

Following sample adsorption, hydrophilic alcohols such as ethanol and acetone are typically used for desorption. In this study, considering elution efficiency, energy consumption, cost, and food safety, ethanol is the most suitable eluent. As shown in Fig. 7D, a higher ethanol concentration resulted in a greater desorption capacity for Celastrus orbiculatus triterpenoids. A higher ethanol concentration (75%) markedly enhanced desorption compared to 50%, with PRI achieving 95.53% and CEL reaching 58.50%. This improvement is due to the higher ethanol concentration, which reduces the eluent's polarity. The resulting closer polarity match between the eluent, the AB-8 resin, and the triterpenoids promotes the diffusion of CEL and PRI into the eluent, thereby improving the desorption rate. To remove impurities that co-elute with triterpenoids, the resin was sequentially eluted with distilled water, followed by 25% and 50% ethanol, prior to the final elution with 75% ethanol to enrich CEL and PRI [61]. The final enriched product was obtained by collecting the target fraction (Fig. 7D), followed by concentration under reduced pressure and vacuum drying. The enrichment process achieved both a high PRI recovery of 95.53% and a dramatic increase in purity from 0.191 ± 0.004% to 13.545 ± 0.542%, confirming the efficacy of the separation method.

3.10. Evaluation of antibacterial activity

3.10.1. Antimicrobial activity of CEL and PRI against MRSA

The in vitro antibacterial activities of CEL and PRI against MRSA were evaluated by determining their MICs. Both CEL and PRI showed significant antibacterial activity against MRSA, especially CEL. The MIC ranges were 0.5–4 μg/mL for CEL and 8–32 μg/mL for PRI (Fig. 8A) Additionally, we established the MBCs for CEL and PRI against MRSA. The MBCs of CEL and PRI against MRSA were 8 μg/mL and 128 μg/mL, respectively (Fig. 8C) Our findings on PRI’s biological activity corroborate those reported by Gonzalez et al [62]. Furthermore, the MIC determinations demonstrated that both the purified enriched extract and the standard mixture exhibited antibacterial activity against MRSA (Fig. 8B). The compounds exhibiting significant antibacterial activity against MRSA were identified as CEL and PRI (Table S5). The MBC assay confirmed that the purified extract and the standard mixture displayed comparable efficacy against MRSA, aligning with the previous results. As shown in Fig. 8D, the MBC differs between the standard mixture and purified extracts, which may be attributed to the presence of additional antibacterial constituents in the purified extracts beyond the target compounds. Notably, a synergistic effect was observed when CEL and PRI were used in combination, significantly improving their bactericidal activity against MRSA.

3.10.2. SYTO 9/PI staining observation of MRSA

We further assessed the antibacterial efficacy of the CEL and PRI combination using a SYTO 9/PI staining assay. SYTO 9 and PI bind strongly to DNA, making them effective nucleic acid stains for bacterial viability assays [63]. When used in combination, SYTO 9 (green) labels all bacteria, while PI (red), which is impermeant to intact membranes, enters only damaged cells and quenches the green fluorescence, resulting in red staining. SYTO 9 has excitation/emission maxima at 480/500 nm. In contrast, PI fluoresces at 490 nm (excitation) and 635 nm (emission). As shown in Fig. 9, the control group exhibited strong green fluorescence, and almost no red fluorescence was seen, confirming that the majority of bacterial cells were in a healthy state with intact membranes. Similarly, treatment with 1/2 MIC PRI or CEL alone produced predominantly green and only minimal red fluorescence, which aligns with the time-kill curve data in supplementary material 2.2. In contrast, treatment with the combination of PRI and CEL, both at 1/2 MIC, or with the purified enriched extract resulted in a marked predominance of red over green fluorescence. This signal indicates widespread loss of membrane integrity, confirming potent bactericidal activity. These results confirm that both the CEL and PRI combination and the purified extract exhibit antibacterial activity.

Fig. 9.

Fluorescent images of bacteria stained with SYTO 9 and PI.

4. Conclusions

This study developed a novel, sustainable UAESE method for the efficient extraction of triterpenoids from Celastrus orbiculatus roots. Following single-factor experiments, PBD and BBD were used to optimize the extraction parameters. The optimal extraction process was established as follows: the SUPRAS consisted of 5% (v/v) octanoic acid, 30% (v/v) ethanol, and 65% (v/v) aqueous solution at pH 4, enzymatic hydrolysis was performed with 6% enzyme at pH 5.0 and 41°C for 50 min, and ultrasound-assisted extraction was performed at 40°C for 70 min with 640 W power and a liquid–solid ratio of 12 mL/g. It is noteworthy that this study focused on optimizing the extraction process for two key active triterpenoids, CEL and PRI. Future work colud investigate the influence of the optimized parameters on other minor constituents in the extract. Compared with other methods, the UAESE method yielded extraction yields of 91.23 μg/g for CEL and 5.53 mg/g for PRI, significantly outperforming single extraction methods and traditional organic solvents. Furthermore, fluorescence microscopy and FTIR revealed the reverse micelle structure of SUPRAS. SEM analysis showed that the enzyme and SUPRAS acted synergistically. CEL and PRI were purified and enriched from the Celastrus orbiculatus crude extract on an AB-8 resin column. Finally, in vitro antibacterial assays demonstrated that CEL, PRI, and their combination all displayed significant activity against MRSA, with the combination showing a distinct additive effect. Notably, at the same concentration, the monomer mixture and the Celastrus orbiculatus enriched solution exhibited similar antibacterial efficacy against MRSA. The demonstrated anti-MRSA activity and additive effect of the combination of CEL and PRI support the promise of the Celastrus orbiculatus purified extract as a lead formulation, justifying its further development into natural preservatives. In summary, the novel extraction method holds significant promise as an efficient and environmentally sustainable extraction technique. This advance enables innovative strategies to plant bioactive compound extraction and accelerates the development of natural, effective plant-derived antibacterial agents.

CRediT authorship contribution statement

Qingmiao Li: Writing – original draft, Data curation. Yu Zhang: Supervision. Jingyou Hao: Supervision. Jing Su: Supervision. Xueying Chen: Supervision. Yanhua Li: Writing – review & editing. Yanyan Liu: Writing – review & editing, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: