Jialiang Guo

Influenza is a highly transmissible airborne disease that causes acute febrile respiratory illness, ranging from mild fatigue to fatal respiratory failure. Neuraminidase (NA) serves as a primary drug target for both the prevention and treatment of influenza. In this study, an efficient strategy integrating an at line nanofractionation (ANF) platform with Global Natural Products Social Molecular Networking (GNPS) was employed for the rapid discovery of neuraminidase inhibitors (NAIs) from the traditional Chinese medicine (TCM) Smilax china L. High throughput screening of a self constructed TCM extract library revealed that the ethanolic extract of Smilax china L. exhibited over 90% inhibitory activity against oseltamivir sensitive neuraminidase (NAS), with an IC50 value of 88.78 ± 13.26 μg/mL. The NAS based ANF platform was then applied, and 21 potential NAIs were identified based on concentration responsive inverse peaks. By combining GNPS molecular networking with manual interpretation (mass errors < 5 ppm), a total of 97 compounds were annotated, covering six major classes, including flavonoids (47), cinnamic acid derivatives (20), polyphenolic compounds (16), and others (14). Among these, the 21 bioactive constituents included chlorogenic acid, astilbin, procyanidin B, and smilaside A. Molecular docking and dynamic simulations indicated that these compounds exert inhibitory activity mainly through interactions with key residues in the NAS active site, including the 150 loop, the Arg cluster (residues 118, 292, 371), and Glu119. Notably, the 150 loop region is relatively conserved, offering potential to overcome drug resistant mutations. The integrated ANF GNPS platform established in this study provides an efficient and reliable technical approach for the rapid screening and structural annotation of bioactive constituents from TCM. It lays a solid chemical foundation for the discovery of novel NAIs.

Biofilm-associated antimicrobial resistance creates an urgent need for label-free optical methods that can read out subtle microstructural heterogeneity without staining or culture-based labeling. Here, we present polarization information distribution spectroscopy (PIDS) as a Mueller matrix-based statistical polarization fingerprinting strategy for bacterial biofilms. Rather than treating stochastic spatial fluctuations as imaging noise, PIDS quantifies the distributional behavior of Mueller-matrix-derived polarimetric basis parameters (PBPs) and converts these statistics into microstructure-sensitive optical fingerprints. Using a sterile-medium control and four bacterial biofilm models, including Pseudomonas aeruginosa, Escherichia coli, Staphylococcus aureus, and Streptococcus mutans, we compared raw 36-dimensional PIDS descriptors with feature representations optimized by element contribution analysis and redundancy-aware compression. Feature-contribution weighting sharpened inter-group separation by emphasizing polarization channels with stronger discriminative contribution, whereas compression reduced feature dimensionality while retaining the most informative optical structure. As a downstream validation readout, a fixed ANN classifier showed that the optimized representation improved biofilm differentiation, with the compressed 21-dimensional representation reaching 94.0% independent test accuracy. These results position PIDS primarily as a Mueller-matrix-derived statistical optical framework, with classification serving as a practical validation of feature separability rather than the central contribution.

How artificial intelligence (AI) can be rationally applied to antibacterial drug discovery and development to overcome antimicrobial resistance (AMR) remains insufficiently defined, particularly across different resistance mechanisms and translational stages. The global emergence and dissemination of AMR are compromising conventional antibacterial therapies, while slow antibiotic discovery, limited target diversity, and reliance on single-mechanism drug design further increase the risk of inadequate treatment options. With strengths in multi-omics integration, complex pattern recognition, and structure-guided molecular design, AI offers a promising approach to support next-generation antibacterial development. This review focuses on AI-related strategies for antimicrobial development against bacterial resistance and provides a systematic overview of advances in target identification, molecular design, activity prediction, and clinical translation. Rather than proposing a new theoretical framework, the review introduces a pragmatic synthesis tool for evaluating where AI adds value and where important limitations remain, considering application objectives, data properties, validation depth, biological interpretability, and translational feasibility. Major challenges are highlighted, including data heterogeneity, clonal bias, limited external validation, discrepancies between in vitro activity and in vivo therapeutic efficacy, and the absence of regulatory approval for any AI-discovered or AI-designed antibacterial agent. Correspondingly, future directions are delineated, including multimodal data integration, explainable AI methodologies, validation through synthetic biology, and coordinated international regulatory efforts. Overall, this review provides a structured overview of how AI can be integrated with rational antibacterial design while considering both its potential and current limitations for addressing multidrug-resistant pathogens.

The pharmacokinetic properties of drugs are significantly influenced by their interactions with cell membranes. Phosphatidylserine (PS) as one of the crucial membrane phospholipids is present abundantly in biological cells. In this study, a PS‐mimicking phospholipid membrane polymer monolithic column was developed using the synthetic compound 2‐methylacryloxydodecyl phosphatidylserine (MDPS). The physicochemical properties and chromatographic performance of the poly(MDPS‐co‐ethylene dimethacrylate (EDMA)) monolithic column were evaluated through scanning electron microscopy (SEM) and nano‐high‐performance liquid chromatography (HPLC). The retention mechanism was investigated by eluting the acidic, neutral, and basic compounds under the same gradient conditions, and the performance of the poly(MDPS‐co‐EDMA) monolithic column was compared with that of a poly(12‐methacryloyl n‐dodecylphosphocholine (MDPC)‐co‐EDMA) monolithic column. The poly(MDPS‐co‐EDMA) column demonstrated effective separation capabilities for small peptides, protein enzymatic hydrolysis polypeptides, and pharmaceutical compounds. Furthermore, the column was also employed to predict the blood–brain barrier (BBB) permeability of drugs, and the correlation coefficient of 0.73 indicates the promising potential of poly(MDPS‐co‐EDMA) monolithic column for predicting the ability of drugs to cross the BBB.

Background and Aim: The enzyme/protein immobilized monolithic capillary combined with liquid chromatography-mass spectrometry is an efficient screening strategy for the corresponding agonist/antagonist. LasR is the potential therapeutic target since it plays a vital role in the colonization and invasion of Pseudomonas aeruginosa (P. aeruginosa). Therefore, reagents that inhibit LasR may be effective against P. aeruginosa. To screen and find LasR inhibitors rapidly, a LasR-immobilized monolithic capillary column was prepared and characterized. Methods: Firstly, the recombinant LasR protein was prepared in E. coli. Then, the LasR protein was immobilized to the surface of poly (glycidyl methacrylate-co-poly(ethylene glycol)diacrylate)-ethylenediamine monolithic column. The affinity and stability of prepared column was also evaluated. Furthermore, the prepared column was applied to fishing LasR inhibitor in Scutellaria baicalensis Georgi extract. The interaction of the screening compound to LasR was confirmed through molecular docking. Results: The recombinant active LasR protein was prepared in E. coli. After purification and validation, a comparative ligand fishing monolithic column was prepared through immobilizing LasR to the surface of the poly (glycidyl methacrylate-co-poly(ethylene glycol)diacrylate)-ethylenediamine through amidation reaction. The LasR was successfully immobilized to the monolithic column characterizing by Fourier transform infrared spectroscopy and scanning electron microscopy. The activity of immobilized LasR was reserved as it has affinity to the nature ligand 3-oxo-C₁₂-HSL and stablied within 24 h at 4 °C. In the Scutellaria baicalensis Georgi extract, baicalein was screened as a potential LasR inhibitor. The molecular docking results and the in vivo evaluation confirmed the activity of baicalein. Conclusion: The proposed LasR immobilized monolithic column is a viable strategy in screening LasR inhibitors. It can be considered as a possible alternative to traditional methods for screening LasR inhibitors as drug candidates against P. aeruginosa.