Artificial neural network model-based optimization of Polygonum perfoliatum L. Polysaccharide ultrasonic-assisted...

Polygonum perfoliatum L. (PP) is utilized in Miao medicine for treated herpes zoster, gynecological inflammation and poisonous snakebites. However, research on its polysaccharides with therapeutic potential remains limited. In this study, polysaccharides from PP were extracted via ultrasonic-assisted extraction (UAE), with parameters were optimized through response surface methodology (RSM) and artificial neural network (ANN) modeling, using the transfer rate and purity of crude polysaccharides as indicators. Optimal conditions were: liquid-to-solid ratio 35 mL/g, three 29-min cycles, 72 ℃ temperature, and 415 W ultrasonic power. This achieved a comprehensive score of (96.96 ± 6.23) %. The prediction results of the RSM and the two ANN models were compared. Ultimately, the extraction conditions predicted by the genetic algorithm-back propagation (GA-BP) neural network were identified as providing the optimal outcome. Following Sephadex purification, fraction S-PPP3 (79 kDa) was isolated. Monosaccharide composition analysis indicated that S-PPP3 is composed of six distinct monosaccharides: L-arabinose, D-galacturonic acid, D-galactose, D-glucose, D-mannose, and L-Rhamnose. Structural analysis further revealed that the main backbone of S-PPP3 primarily consists of repeating units of → 4)-β-D-Galp-(1 → and → 4,6)-β-D-Galp-(1 → . Side chains are attached at the C-6 position of the → 4,6)-β-D-Galp-(1 → residues and incorporate structural motifs such as T-α-L-Araf-(1 → 5)-α-L-Araf-(1 → 5)-α-L-Araf-(1→, T-α-D-Manp-(1 → 3)-β-D-Glcp-(1 → 4)-α-D-GalAp-(1→, and T-α-L-Rhap-(1 → 2,4)-α-L-Rhap-(1 → 2,4)-α-L-Rhap-(1 → . Furthermore, Additionally, the anti-inflammatory activity of S-PPP3 was evaluated in lipopolysaccharide (LPS)-stimulated RAW264.7 macrophages, demonstrating significant efficacy. These findings highlight the potential of S-PPP3 as a functional food or anti-inflammatory therapeutic agent.