A Bioinspired Three-Dimensional High-Curvature Nano-Interface Integrated Microfluidic Chip for Small Extracellular Vesicles Enrichment and Machine Learning-Assisted Prostate Cancer Precision Diagnosis.

The efficient and unbiased isolation of small extracellular vesicles (sEVs) from complex biological fluids remains a major obstacle for clinical diagnostics. Here, we report a bioinspired microfluidic chip that integrates a three-dimensional high-curvature TiO2 nano-interface (3D Hic-TiO2) with a biotin-modified artificial insertion peptide (BAIP) modification for rapid enrichment of sEV. Bowl-shaped TiO2 nanospheres fabricated via electrospray provide topological nanotraps that match the size and curvature of sEVs, enabling efficient size-selective capture. Coupled with BAIP-mediated membrane affinity and herringbone-induced chaotic mixing, the BAIP-TiO2-Chip achieved >90% capture efficiency within 5 min. Redox-responsive BAIP variants enabled mild release of intact sEVs for downstream analysis. When the BAIP-TiO2-Chip was applied to plasma samples from clinical prostate cancer (PCa) patients, mass spectrometry-based proteomic profiling revealed 110 differentially expressed sEV-associated proteins, including candidates involved in immune regulation and cell adhesion. In parallel, simultaneous quantification of PSA and PSMA mRNAs in sEVs could be achieved. Assisted by machine learning, a boosted decision tree model achieved 80% diagnostic accuracy in distinguishing PCa from benign conditions and healthy donors. This work presents a versatile platform for sEV isolation, enabling both transcriptomic mRNA analysis and proteomic profiling, and provides new molecular insights into PCa for improved early diagnosis.