Fig. 5: Glycan profiling of extracellular vesicles.

a Schematic of the STAR glycan assay. A dual-probe STAR architecture was developed on a microfluidic platform. It contained GQDs (green arrow) peripherally dispersed and RQDs (red arrow) centrally encapsulated within the MOF as the working probes and reference probes, respectively. b Fluorescence changes. In the presence of target glycan, the specific binding of lectin-oxidase mediates in situ generation of hydrogen peroxide, which selectively quenches the fluorescence of the peripheral working probes, while leaving the central reference probes in the same STARs unaffected. Typical fluorescence spectra showed intensity changes (I_G and I_R) before and after the assay. c Robust performance in the presence of interfering agent. Only the STAR assembly accurately revealed the target glycan concentration, amidst the interfering agent (Fe³⁺, 1 µM). d Detection sensitivity of the assay. The limit of detection was assessed by titrating a known quantity of the target glycoprotein (transferrin) and measuring the associated signal response. The detection limit for ELISA was independently assessed based on fluorescence signals. e Correlation between STAR and ELISA measurements, when measured with six lectins. R², Pearson’s correlation coefficient. f Glycan profiling in EVs. Left: comparison of glycan profiles between EVs and their parent cells. ∆Response = Response_EV – Response_Cell. Right: comparison of glycan profiles between EVs derived from different cell origins. All EV measurements were performed with an equal vesicle concentration (5 × 10⁸/ml) through the multiplexed STAR platform. Datasets were normalized to their respective highest. For inter-sample comparisons, multiple pairs of samples were analyzed by two-tailed Student’s t-test, and the resulting p values were adjusted for multiple hypothesis testing using Bonferroni correction. P values are shown and p < 0.05 is determined as significant. NS not significant; *p < 0.05, **p < 0.01, ***p < 0.001. g STAR analysis of 25 lectin markers in clinical cancer ascites (n = 12). Hierarchical clustering of patient profiling data classified the patients into two populations and the glycan expressions into two clusters. h Principal component analysis of Cluster 1 glycans for prognosis classification. Ellipses were drawn at 95% confidence. All measurements were performed in triplicate and the data are displayed as mean ± SD in d–f, and as mean in g, h. arb. units, arbitrary units. Source data are provided as a Source Data file.