CHAI-sEV: A Programmable Hairpin-Driven Cas9 Platform with AI Integration for Dual-Protein Profiling and Intelligent Classification of Tumor-Derived Small Extracellular Vesicles

Abstract

Multiplexed protein profiling of tumor-derived small extracellular vesicles (TsEVs) requires signal-conversion strategies that are sensitive, low-background, and orthogonal. Collateral-cleavage CRISPR assays based on Cas12a or Cas13a provide efficient amplification but can compromise single-pot multitarget detection, whereas the sequence-specific cleavage of Cas9 remains underexplored for converting vesicle-surface protein recognition into orthogonal amplified outputs. Here, we report CHAI-sEV, a programmable Cas9-mediated trigger-release strategy based on PAM-bearing hairpin-locking probes (PHLPs) for dual-protein profiling of TsEVs. Each PHLP integrates Cas9 recognition and signal conversion into a single molecular substrate, in which sgRNA-guided Cas9 cleavage unlocks a sequestered CHA activator to initiate orthogonal signal amplification. After EGFR- or PD-L1-binding aptamers label TsEVs, engineered aptamer tails recruit specific Cas9-sgRNA complexes to the vesicle surface, enabling protein-specific PHLP activation and amplified fluorescence readout. This Cas9-programmed trigger-release mechanism enables dual-channel TsEV protein detection without relying on nonspecific collateral cleavage. Under optimized conditions, CHAI-sEV achieved detection limits of 162 and 328 particles/μL for EGFR and PD-L1. In plasma samples, CHAI-sEV differentiated lung cancer patients from benign-lung-disease patients and healthy controls and showed decreased EGFR and PD-L1 signals in post-treatment samples. As proof-of-concept extensions, the CHAI-sEV was further coupled to an IGZO-FET electrical readout and an exploratory machine-learning analysis combining dual-protein signals with clinical features. Overall, PHLPs provide a substrate-design principle that converts Cas9 cleavage into amplification-compatible and orthogonally multiplexable biosensing outputs, enabling sensitive and multiplexed protein profiling of TsEVs and offering potential for extension to diverse molecular targets.