# Magnetically Sculpted Microfluidics for Continuous-Flow Fractionation of Cell Populations by EpCAM Expression Level

**Authors:** Zhenwei Liang, Xiaolei Guo, Xuanhe Zhang, Yiqing Chen, Chuan Du, Yuan Ma, Jiadao Wang

PMC · DOI: 10.3390/mi17010009 · 2025-12-22

## TL;DR

This paper introduces a microfluidic system that uses magnetic fields to separate cells based on their surface protein expression levels, enabling precise cell fractionation for research and assays.

## Contribution

The novel contribution is a magnetic-field design strategy using soft magnetic strips to enable expression-level-dependent cell sorting in continuous-flow microfluidics.

## Key findings

- The system successfully partitions cells into four EpCAM-related subgroups (high, medium, low, near-negative).
- The sorted fractions maintain high cell recovery (>90%) and viability (98.2 ± 1.3%).
- The magnetic interface design is optimized using a COMSOL–MATLAB framework and a force-equivalent metric.

## Abstract

Continuous-flow separation of magnetically labeled cells according to surface-marker expression levels is increasingly needed to study phenotypic heterogeneity and support downstream assays. Here, we present a microfluidic platform that uses spatially engineered soft magnetic strips (SMS) to sculpt lateral magnetic deflection fields for quantitative, label-guided cell fractionation. Under a uniform bias field, the SMS generates controllable magnetic gradients within the microchannel, producing distinct lateral velocities among EpCAM-labeled tumor cells that carry different Dynabead loads, which indirectly report membrane protein expression. Multi-outlet collection converts these “race-based” trajectory differences into discrete expression-level-resolved fractions. A COMSOL–MATLAB framework and a force-equivalent metric |(H·∇)H| are used to optimize key structural parameters of the magnetic interface, including strip thickness, width, and vertical spacing from the flow channel. Three journey nodes at 1.5, 3, and 9 mm along the flow path define a three-stage cascade that partitions MDA-MB-231, Caco-2, and A549 cells into four EpCAM-related magnetic subgroups: high (H), medium (M), low (L), and near-negative (N). Experiments show that the sorted fractions follow the expected expression trends reported in the literature, while maintaining high cell recovery (>90%) and viability retention of 98.2 ± 1.3%, indicating compatibility with downstream whole-blood assays and culture. Rather than introducing a new biomarker, this work establishes a quantitative magnetic-field design strategy for continuous microfluidic sorting, in which the spatial configuration of soft magnetic elements is exploited to implement expression-level-dependent fractionation in next-generation magneto-fluidic separation systems.

## Linked entities

- **Proteins:** EPCAM (epithelial cell adhesion molecule)

## Full-text entities

- **Genes:** EPCAM (epithelial cell adhesion molecule) [NCBI Gene 4072] {aka Ber-Ep4, BerEp4, DIAR5, EGP-2, EGP314, EGP40}
- **Diseases:** tumor (MESH:D009369)

## Figures

6 figures with captions in the complete paper: https://tomesphere.com/paper/PMC12844116/full.md

---
Source: https://tomesphere.com/paper/PMC12844116