Optimization and antioxidant evaluation of clean and efficient recovery of polyphenols from Phellodendron amurense waste leaves via ultrasound-assisted extraction

Abstract

The production of edible and medicinal plants results in significant biomass waste, which can be a rich source of bioactive compounds. This study optimized the extraction of polyphenols from Phellodendron amurense leaf waste using ultrasound-assisted extraction (UAE), with modeling via both response surface methodology (RSM) and artificial neural networks (ANN). ANN exhibited superior predictive accuracy. Under optimal UAE conditions (60 min ultrasonic time, 1:20 g/mL w/v solid-to-liquid ratio, 60 % v/v ethanol concentration, and 190 W ultrasonic power), the highest total polyphenol yield of 28.66 ± 0.07 mg GAE/g DW was achieved. The resulting extracts exhibited strong antioxidant activity in cellular assays. A total of 25 polyphenolic compounds, including luteolin, isorhamnetin, and kaempferol, were identified by liquid chromatography paired with mass spectrometry (LC-MS/MS), and molecular docking predicted their interaction with collagen I (COL I) as a potential antioxidant mechanism. These findings support the value-added utilization of P. amurense leaf waste and provide a foundation for its application in antioxidant nutraceuticals and plant-based pharmaceuticals.

Keywords: Antioxidant; Green chemistry; Phellodendron amurense Rupr. waste leaves; Polyphenols; Ultrasound-assisted extraction.

Graphical abstract

Fig. 1

Post-training optimal architecture for the ANN and GA optimization stages.

Fig. 2

Effects of various extraction parameters on polyphenols yield. (A) Schematic representation of the ultrasonic-assisted extraction process. The influence of (B) ultrasonic time, (C) solid to liquid ratio, (D) ethanol concentration, (E) and ultrasonic power on polyphenol yield. All data are expressed as mean ± standard deviation (SD), n = 3. Statistical significance was evaluated by comparing each treatment group with the center point condition, **p < 0.01, *p < 0.05.

Fig. 3

Three-dimensional response surface diagrams of the influence of different factors interaction on TPC. (A) Interaction between solid-to-liquid ratio and ultrasonic time, (B) Interaction between ethanol concentration and ultrasonic time, (C) Interaction between ultrasonic power and ultrasonic time, (D) Interaction between ethanol concentration and solid-to-liquid ratio, (E) Interaction between ultrasonic power and solid-to-liquid ratio, (F) Interaction between ultrasonic power and ethanol concentration.

Fig. 4

Contour plots depicting the interactive effects of various factors on TPC. (A) Interaction between solid-to-liquid ratio and ultrasonic time, (B) Interaction between ethanol concentration and ultrasonic time, (C) Interaction between ultrasonic power and ultrasonic time, (D) Interaction between ethanol concentration and solid-to-liquid ratio, (E) Interaction between ultrasonic power and solid-to-liquid ratio, (F) Interaction between ultrasonic power and ethanol concentration.

Fig. 5

The results of ANN model display best validation performance (A), training state (B), and regression plot (C).

Fig. 6

RSM and ANN model comparison. Impacts of matching across all datasets (A), the link between real values and values projected by ANN and RSM (B), and the GA optimization process (C).

Fig. 7

Antioxidant capacity analysis of UG and OG. (A) Schematic diagram of the cell experiment. Effect of UG (B) and OG (C) on the viability of HSF cells. (D) The ROS activation in cells, original magnification: 400×. (E) Detection the secretion of T-AOC, SOD, GSH, and MDA in cells. All data are expressed as mean ± S.D, n = 6. **p < 0.01 vs. MG group. ^#p < 0.05, ^##p < 0.01 vs. UG group.

Fig. 8

The effects of COL I expression and apoptosis in H₂O₂-induced senescent HSFs. (A) The fluorescence intensity image of COL I was analyzed by immunofluorescence, original magnification: 400×. (B) The mRNA levels of COL I, MMP-1, SOD, and TIMP-1 in cells. All data are expressed as mean ± S.D, n = 6. **p < 0.01 vs. MG group. ^#p < 0.05, ^##p < 0.01 vs. UG group.

Fig. 9

Determination of chemical constituents of polyphenols in PLE by ultrasound-assisted method. (A) Workflow for the analysis of chemical composition. (B) Total Ion Chromatogram (TIC) in positive ion mode. (C) TIC in negative ion mode. (D) Correlation analysis between the polyphenolic compounds and mRNA expression levels associated with antioxidant capacity.

Fig. 10

Molecular models of the first five compounds and the most relevant target proteins (COL I). (A) Luteolin. (B) Isorhamnetin. (C) Kaempferol. (D) Catechin. (E) Apigenin.