Kewei Chen

Background/Objectives: Type 2 diabetes mellitus (T2DM) is a multifactorial metabolic disorder characterized by chronic hyperglycemia, oxidative stress, protein glycation, low-grade inflammation, and dysregulation of signaling pathways, including those involving nuclear factor kappa B (NF-κB). Isoflavones from Fabaceae species exhibit diverse biological activities and may represent multifunctional candidates for addressing the interconnected mechanisms underlying T2DM. This study performed a consensus in silico pharmacodynamic screening of twelve representative Fabaceae isoflavones to prioritize compounds with predicted antiglycating, deglycating, antidiabetic, antioxidant, and NF-κB-related anti-inflammatory activities. Methods: Twelve representative Fabaceae isoflavones were evaluated using a multimethod computational workflow. Antiglycating and deglycating activities were predicted using IT Microcosm, antidiabetic and hypoglycemic activities were assessed using Prediction of Activity Spectra for Substances (PASS), antioxidant potential was evaluated using similarity-based prediction and quantum-chemical calculations, and NF-κB-related anti-inflammatory activity was investigated using similarity-based prediction combined with molecular docking. Molecular docking was applied specifically to the NF-κB-related anti-inflammatory analysis and did not contribute to the hypoglycemic/antidiabetic prediction. The individual predictions were integrated using a consensus scoring strategy to prioritize compounds across multiple pharmacodynamic domains. Results: The investigated isoflavones demonstrated moderate but broad predicted multifunctional activities. Malonylgenistin, biochanin A, and formononetin showed the most favorable predicted antiglycating profiles. Genistin, acetyldaidzin, acetylglycitin, and puerarin ranked highest in terms of predicted antidiabetic activity, whereas genistein exhibited the strongest predicted antioxidant profile. Daidzin and genistin received the highest integrated NF-κB-related computational rankings, although these predictions apply to intact glycosides rather than their expected in vivo hydrolysis products. Conclusions: This study presents a consensus computational framework for the comparative prioritization of Fabaceae isoflavones based on their predicted multifunctional pharmacodynamic profiles. The results provide a rational basis for selecting candidate compounds for future biochemical, cell-based, pharmacokinetic, and in vivo investigations, although it is emphasized that the present findings are based exclusively on computational predictions and require experimental validation.

To better align infant formula lipid design with human milk, effective methods are needed to identify suitable lipid sources. Here, human milk from different lactation stages was used as the reference to evaluate 8 woody oil plants based on phospholipid classes, phospholipid fatty acid composition, and sn-2 fatty acid distribution. A stage-specific multimodal fusion model was developed to assess overall similarity. Xanthoceras sorbifolium Bunge phospholipids were most similar to colostrum and transitional milk phospholipids, whereas Juglans regia L. phospholipids were most similar to mature milk phospholipids. Xanthoceras sorbifolium Bunge, Juglans regia L., and Elaeagnus mollis Diels phospholipids consistently ranked among the top three in similarity scores across all three stages of human milk. Feature importance analysis showed that stage-specific similarity was jointly determined by phospholipid classes, fatty acid composition, and sn-2 distribution. This framework supports woody oil screening and lipid design for infant formula.

The development of high-performance, binder-free, and mechanically flexible self-supported electrodes is crucial for advancing sustainable hydrogen evolution reaction (HER) technologies. However, HER electrodes with high activity and long-term stability are difficult to construct, primarily because of the binder dependency of conventional powder catalysts and the limited adaptability of rigid metal substrates. Herein, we report an effective strategy for fabricating flexible self-supported electrodes by employing renewable silk fabric as a nitrogen- and carbon-rich precursor, in combination with the synergistic incorporation of multiple non-noble metals (Mo, Fe) and nonmetal heteroatoms (N, P). Through a streamlined impregnation–carbonization process, multiphase active sites were efficiently integrated and strongly coupled within the silk-derived carbon framework. Experimental results demonstrate that the synergistic interactions among these multicomponents considerably optimize HER kinetics. The optimized electrode (Sample 12) delivers a low overpotential of 122.60 ± 2.00 mV at 10 mA·cm–2, together with a high Faradaic efficiency of 96.7%, representing a clear improvement over our previous single-component ammonium-molybdate-loaded system. More remarkably, this electrode exhibits outstanding long-term operational robustness. After continuous operation of the optimized electrode at a large current density of 100 mA·cm–2 for 50 h, its overpotential increases by only 22.08 mV, markedly outperforming commercial Pt/C, which degrades by 148.33 mV, and other silk-derived HER electrodes. Furthermore, comprehensive multiscale characterizations (FESEM, TEM, specific surface area analysis, XPS, XRD, and wettability measurements) were utilized to reveal the mechanism underlying the enhanced HER performance. This work offers a valuable design paradigm for developing low-cost, efficient, and durable industrial-grade flexible electrodes toward large-scale hydrogen production.