Advanced Extracellular Vesicle Isolation: A Hybrid Electrokinetic-Tangential Flow Filtration Approach for Improved Yield, Purity, and Scalability

Abstract

As extracellular vesicles (EVs) become increasingly important in diagnostics and therapeutics, achieving both improved purity and yield during isolation remains a critical challenge. Conventional techniques often suffer from the coisolation of nonvesicular particles and soluble proteins, limiting their clinical and research utility. In response, we introduce ExoTFF, a hybrid isolation technology that sequentially integrates electrokinetic filtration (ExoFilter) with size-exclusion tangential flow filtration (TFF) to deliver unprecedented performance gains through an iterative, synergistic mechanism. In the ExoTFF system, the sample is repeatedly circulated through an electrokinetic mesh filter and TFF until the liquid is removed. This recirculating flow gradually eliminates contaminants, while the electrokinetic filter continuously captures EVs as the sample is purified. Finally, any residual impurities in the TFF unit are completely removed via a dead volume elimination process. The complementary actions of these two distinct separation mechanisms double EV recovery rates and reduce impurity levels by 80% compared to conventional TFF, culminating in an impressive 800% improvement in the purity ratio. In proof-of-concept experiments, ExoTFF processed 10 mL of plasma within 10 min, efficiently depleting albumin and high-density lipoprotein (HDL) while achieving superior EV recovery. To further explore scalability, an automated ExoTFF system processed 500 mL of sample in 50 min, maintaining consistent yield and purity. The ability to sustain performance across different scales highlights ExoTFF's potential for both laboratory research and industrial-level EV production. Beyond biological applications, this platform also offers broad applicability for the isolation of negatively charged nanoparticles, demonstrating its potential impact across multiple nanotechnology-driven fields.

1

Hybridization of electrokinetic mesh filtration (ExoFilter) and size-based Tangential Flow Filtration (TFF). (a) Schematic of electrokinetic assisted mesh flow filtration with TFF, (b) dual mechanisms of ExoTFF consisting electrokinetic filtration and size exclusion filtration, (c) electrokinetic mechanism of ExoFilter for the selective capture of EVs over smaller contaminant particles, (d) operation window of ExoTFF in ζ-potential vs size domain, (e) Venn diagram of the hybrid ExoTFF system, which combines key features of ExoFilter and TFF, (f) protocol of ExoTFF consisting of sample filtration, elution and recovery processes.

2

Characterization of EVs isolated from blood plasma using ExoTFF. (a) SEM image of an eluted EV, (b) TEM image of an eluted EV, (c) particle size distribution and concentrations of EVs, (d) western blot analysis of EV markers and contaminant proteins across five isolation methods (UC, SEC, TFF, ExoFIlter and ExoTFF), (e) ELISA assay results for albumin, ApoA1, ApoB100, and CD9. (*p < 0.05, ***p < 0.005), (f) cellular uptake of EVs into human dermal fibroblasts (HDFs). (g) Cytotoxicity test of EVs in HDF after 72 h.

3

Comparative analysis of EV isolation methods (TFF, ExoFilter, ExoFilter→TFF, and ExoTFF) using plasma samples. (a) Particle concentration, (b) protein concentration of isolated EVs, (c) ELISA assays for a specific EV marker (CD81), (d) quantification of EV-derived miRNAs (miR-142 and let-7a) using RT-PCR, (e) ELISA assays for specific HDL markers (ApoA1), (f) comparison of Purity ratios for various methods. (*p < 0.05, **p < 0.01, ***p < 0.005).

4

Comparison of EV isolation methods across various biofluids (CCM, urine, saliva). For protein concentration analysis (a, c, e), ExoTFF showed statistically significant reductions in contaminating protein levels compared to all other methods (***p < 0.005). For particle concentration analysis (b, d, f), statistical significance was determined between specific groups using one-way ANOVA followed by Tukey’s post hoc test (*p < 0.05, **p < 0.01, ***p < 0.005; NS = not significant).