Ultrasound-Assisted Deep Eutectic Solvent-Based Extraction of Polysaccharides from Okra: Optimization by Response Surface Methodology and Artificial Neural Network Modeling

Abstract

Plant-derived polysaccharides are critical bioactive compounds; however, conventional extraction methods are often inefficient, energy-intensive, and may compromise their bioactivity. Ultrasound-assisted deep eutectic solvent (UA-DES) extraction offers a greener alternative by integrating acoustic cavitation with tunable solvent properties; however, optimization remains complex due to the interaction of multiple processing variables. This study reports a novel application of ultrasound-assisted deep eutectic solvent (UA-DES) extraction for okra polysaccharides (OPs), with process optimization using response surface methodology (RSM) and artificial neural network (ANN) modeling to identify optimal conditions and clarify nonlinear extraction behavior. Among the tested DES systems, choline chloride-citric acid (CCA) exhibited the highest extraction performance. Single-factor experiments and RSM identified sonication time and liquid-solid ratio as key variables. The ANN model achieved higher predictive accuracy than RSM and captured nonlinear and synergistic parameter interactions that were not evident in traditional response surfaces, providing deeper insight into process behavior. Under optimized conditions (2 h, 80 °C, 190 W, 60 mL/g), UA-DES extraction produced 23.56 % OPs and 80.75 % DPPH• scavenging activity, representing 94 % higher yield and 28 % greater antioxidant activity than hot-water ultrasonic (HWU) extraction. UA-DES-derived OPs contained higher contents of uronic acids, total sugars, and glucans, and uniquely included arabinose absent in HWU extracts. Structural analyses revealed pyranose configurations, amorphous crystallinity, and porous microstructures, which collectively contribute to improved solubility and bioactivity. Overall, UA-DES extraction using CCA provides an eco-efficient strategy for producing high-value okra polysaccharides. The integrated RSM-ANN framework enables precise optimization and enhanced mechanistic understanding, supporting UA-DES as a scalable, green technology for the production of functional polysaccharides.

Keywords: Antioxidants; Bioactivity; Cavitation; Hydrogen bonding; Microstructure.

Graphical abstract

Fig. 1

Flow chart illustrating the ultrasound-assisted deep eutectic solvent (UA-DES) extraction of okra polysaccharides (OPs). Fresh okra pericarps were stored at − 20 °C, freeze-dried, ground, and sieved (0.22 mm) to obtain okra powder, which was stored at 4 °C under dark conditions before extraction. Six types of deep eutectic solvents (DESs) were prepared by combining choline chloride (ChCl, hydrogen bond acceptor, HBA) with various hydrogen bond donors (HBDs) at a 1:2 M ratio: citric acid (CCA), urea (CCU), glycerol (CCG), malic acid (CCM), ethylene glycol (CCE), and 1,4-butanediol (CCB). Each DES mixture was heated and stirred at 80 °C and 300 rpm for 2 h, followed by the addition of 30 % (v/v) deionized water and further stirring for 1 h to enhance solvent homogeneity. Pretreated okra powder was mixed with each DES at a liquid-to-solid ratio of 30 mL/g and subjected to ultrasound-assisted extraction (UAE) for 4 h at 80 °C and 380 W sonication power. After extraction, the mixture was centrifuged (7000 rpm, 20 min, 4 °C), and the supernatant was collected for polysaccharide recovery. Extracts were filtered, concentrated, and precipitated with 95 % ethanol for 48 h. The precipitated polysaccharides were collected by centrifugation, washed, and freeze-dried to obtain purified okra polysaccharides (OPs).

Fig. 2

Effects of extraction solvents and ultrasound-assisted extraction parameters on the yield and antioxidant activity of okra polysaccharides. (a) Polysaccharide yield with different extraction solvents. (b) DPPH radical scavenging activity with different extraction solvents. (c) Effect of sonication time on polysaccharide yield. (d) Effect of extraction temperature on polysaccharide yield. (e) Effect of sonication power on polysaccharide yield. (f) Effect of liquid-to-solid ratio on polysaccharide yield. Data are expressed as mean ± SD (n = 3). Different letters above the bars indicate significant differences among groups, as determined by one-way ANOVA followed by Tukey’s post hoc test. Statistical significance was considered at p < 0.05. OPs Experimental conditions: OPs concentration of 5 mg/mL, DPPH^• concentration was 0.1 mM.

Fig. 3

Response surface plots showing the interactive effects of extraction parameters on polysaccharide yield and DPPH radical scavenging activity of okra polysaccharides. (a–f) Interaction effects of extraction temperature, sonication time, sonication power, and liquid-to-solid ratio on OPS yield. (g–l) Interaction effects of the same parameters on DPPH radical scavenging activity.

Fig. 4

Response surface plots predicted by the ANN model showing the interactive effects of extraction parameters on polysaccharide yield and DPPH radical scavenging activity of okra polysaccharides. (a–f) Interaction effects of extraction temperature, sonication time, sonication power, and liquid-to-solid ratio on OPS yield. (g–l) Interaction effects of the same parameters on DPPH radical scavenging activity.

Fig. 5

Evaluation of the TRAINLM (Levenberg–Marquardt algorithm) ANN model based on regression plots for training, validation, testing, and overall datasets in predicting extraction yield and DPPH radical scavenging activity of okra polysaccharides. (a–d) Regression plots of OPs yield in the training, validation, testing, and overall datasets. (e–h) Regression plots of DPPH radical scavenging activity in the training, validation, testing, and overall datasets.

Fig. 6

Comparison of RSM and ANN models for predicting extraction yield and DPPH radical scavenging activity of okra polysaccharides. (a–b) Correlation between actual and predicted values of yield and DPPH^• activity using the RSM model. (c–d) Correlation between actual and predicted values of yield and DPPH^• activity using the ANN model.

Fig. 7

HPLC chromatograms of monosaccharide composition in OPs obtained by UA-DES and HWU. (a) Mixed monosaccharide standards (mannose, rhamnose, glucuronic acid, galacturonic acid, glucose, galactose, and arabinose). (b) OPs obtained by UA-DES. (c) OPs obtained by HWU.

Fig. 8

Structural characterization of OPs obtained by HWU and UA-DES. (a) FTIR spectra of OPs showing characteristic absorption bands. The broad peak around 3348 cm⁻¹ corresponds to O–H stretching vibrations of hydroxyl groups, while peaks near 2933–2880 cm⁻¹ are attributed to C–H stretching of aliphatic chains. The signals observed between 1700 and 1600 cm⁻¹ represent C=O stretching of ester or carboxyl groups, and those in the range of 1200–1000 cm⁻¹ are related to C–O–C or C–O stretching vibrations, indicating the presence of polysaccharide structures. (b) XRD patterns of OPs showing broad diffraction peaks centered at 2θ ≈ 12.5° and 21.6°, characteristic of amorphous biopolymeric materials. The disappearance of sharp crystalline peaks confirms the non-crystalline nature of OPs extracted by both HWU and UA-DES. (c) SEM micrographs showing the surface morphology of OPs obtained by HWU and UA-DES. OPs extracted by HWU (top row) and UA-DES (bottom row) were observed under different magnifications to reveal surface morphological features. Images were captured at (i) 500 × magnification, showing the overall particle structure and fracture surfaces; (ii) 2000 × magnification, highlighting the surface roughness and aggregation; and (iii) 30000 × magnification, providing detailed views of the microstructural porosity and fine surface texture.

Fig. 9

(a) DPPH radical scavenging activity of OPs obtained by UA-DES and HWU. OPs samples extracted by UA-DES and HWU were tested at concentrations ranging from 1 to 10 mg/mL. DPPH concertation was 0.1 mM. Data are expressed as mean ± SD (n = 3). Different letters above the data points indicate significant differences among concentrations within the same extraction method, as determined by one-way ANOVA followed by Tukey’s post hoc test. Statistical significance was considered at P < 0.05. (b) Proposed mechanisms of deep eutectic solvent coupled with ultrasound-assisted (UA-DES) extraction of OPs. In this system, choline chloride/citric acid-based DES penetrates the plant matrix and disrupts hydrogen-bonding networks, while ultrasound cavitation generates bubbles and shear forces that mechanically break cell walls. The combined effects facilitate solvent infiltration and promote the release of intracellular polysaccharides and phytochemicals. The citric acid component contributes to pH modulation and chelation of metal ions, thereby enhancing polysaccharide solubilization and extraction efficiency. Additionally, free radical generation during cavitation may further increase solvent penetration and extraction performance.