Electrokinetically enhanced label-free plasmonic sensing for rapid detection of tumor-derived extracellular vesicles

Abstract

of rare circulating extracellular vesicles (EV) from early cancers or different types of host cells requires extremely sensitive EV sensing technologies. Nanoplasmonic EV sensing technologies have demonstrated good analytical performances, but their sensitivity is often limited by EVs' diffusion to the active sensor surface for specific target EV capture. Here, we developed an advanced plasmonic EV platform with electrokinetically enhanced yields (KeyPLEX). The KeyPLEX system effectively overcomes diffusion-limited reactions with applied electroosmosis and dielectrophoresis forces. These forces bring EVs toward the sensor surface and concentrate them in specific areas. Using the keyPLEX, we showed significant improvements in detection sensitivity by ∼100-fold, leading to the sensitive detection of rare cancer EVs from human plasma samples in 10 min. The keyPLEX system could become a valuable tool for point-of-care rapid EV analysis.

Keywords: Cancer; Electrokinetics; Extracellular vesicles; Label-free detection; Plasmonic sensing.

Fig. 1.. Electrokinetically enhanced yield of plasmonic sensing (KeyPLEX) for rapid, label-free detection of extracellular vesicles (EVs).

A. Schematic illustration of the KeyPLEX system. AC potentials were applied between a nanoplasmonic chip and a transparent indium tin oxide (ITO) electrode to generate electokinetic forces, pushing EVs toward the sensing surface. The plasmonic sensing surface was functionlized by capture antibodies immobilized via a layer of polyethylene glycol (PEG). Specific EV capture was detected by measuring a spectral shift of surface plasmon resonance in a reflection mode. B. A photograph of the KeyPLEX system. A microfluidic channel was used to deliver samples to the plasmonic sensor and wash out unbound EVs. Gold and ITO electrodes were connected to a signal generator by copper tapes. A 10x objective (0.3 NA) was used for reflective spectral measurements. C-D. Scanning electron micrographs (SEMs) of the KeyPLEX chip. The chip was made of a hybrid structure with micropatterns of 40 μm × 40 μm SiN square arrays with a 40 μm gap between squares for electrokinetic force generation (C) and nanowell arrays with 200 nm well diameter with 500 nm periodicity for plasmonic sensing. E. Fluorescence images of the sensing surface before (E) and after applying an electrical potential, showing fluorescently labeled EVs with green fluorescence proteins (GFPs) accumulated on the sensing surface. F. A spectral shift of reflection reflectance spectrum upon EV binding to the sensing surface in a label-free manner. A microfluidic channel was used to deliver EV samples to the sensing area and wash out unbound EVs after incubation (Fig. 1B). We made a KeyPLEX chip in a hybrid structure; SiN micropatterns (40 μm square with 40 μm gap between squares, Fig. 1C) were made to generate and direct electroosmosis and DEP forces toward gold sensing areas made of periodic nanowelll arrays (Fig. 1D). We used interference lithography for periodic nanowell patterns (200 nm well diameter and 500 nm periodicity) and conventional optical lithography for micropatterns. This allows us to fabricate KeyPLEX chips in a wafer scale (Fig. S2).

Fig. 2.. Characterization of electrokinetically enhanced EV detection.

A. EV binding kinetics to the anti-CD63 antibody-immobilized gold nanowell surface with (active) and without external potentials (passive). B. Incubation time optimization. After applying AC fields for 1 min, additional EV incubation times between 0 and 9 min were added for EV binding to the nanowell surface. The spectral shift saturated after 4 min additional incubation and reached the maximum value after 9 min. C. Spectral shifts between active and passive modes were compared with titrating input EV counts.

Fig. 3.. Molecular profiling of EVs from ovarian cancer and benign cell lines.

Spectral shifts shows higher levels of CD24 and EpCAM in EVs from ovarian cancer cell lines (OVCAR3, A; OV90, B; CaOV3, C) compared to those in EVs from benign cell line (TIOSE4, D). The signal was normalized by the signals of CD63 (putative EV marker). Error bars represent standard deviation from triplicate measurements while individual values are shown as dots.

Fig. 4.. Tumor-derived EV detection in human plasma samples.

(A) EVs isolated from plasma samples of 8 patients (P1–5: ovarian cancer patients and N1–3: 3 health donors) were analyzed for CD63, CD24, EpCAM signals via active and passive modes of KeyFLEX. (B) Heatmaps showing ovarian cancer marker signals (CD24 and EpCAM) measured in active (left column)and passive (right column) modes. The signal values are calculated by the linear sum of the two markers for both active and passive modes.