A1 - Physical Profiling Distinguishes Fibroblast Biological States Beyond Molecular Signatures and Identifies Conserved Migratory Phenotypes Across Tissues

Introduction: : Cancer-associated fibroblasts (CAFs) are essential constituents of the tumor microenvironment, where they actively regulate tumor progression, metastasis, and therapeutic responses. However, the functional heterogeneity of CAFs remains incompletely understood, posing challenges to the precise classification of their phenotypes. Conventional molecular classifications often fall short of capturing the dynamic and mechanical heterogeneity of CAF populations. In this study, we introduce a physical phenotype-based framework that integrates morphological, mechanical, and dynamic properties extracted from live-cell imaging. We aimed to identify CAF subtypes with enhanced migratory characteristics across different tissue origins, using pancreatic fibroblasts cocultured with cancer cells and patient-derived breast CAFs.

Materials and

Methods: : Primary human pancreatic fibroblasts (Cell Biologics) were activated toward to CAFs by coculture with two pancreatic cancer cell lines: Capan2, characterized by low invasiveness, and MiaPaCa2, exhibiting high invasiveness in the coculture dish. Breast CAFs (BioIVT) were isolated from invasive ductal carcinoma patients and sorted into PDPN⁺ and PDPN⁻ subpopulations using fluorescence-activated cell sorting (FACS) based on the expression of CAF biomarkers (FAP, PDPN, PDGFRβ, and MHC II). Cells were cultured on 2 kPa polyacrylamide gels for one day before imaging. For breast CAFs, gels were embedded with fluorescent beads to facilitate traction force microscopy (TFM). Phase contrast time-lapse imaging of pancreatic fibroblasts was performed over 12-hour periods with 10-minute intervals. Traction force maps of breast CAFs were obtained using bead reference images after trypsinization. Cell segmentation was achieved using a fine-tuned Cellpose model and YOLOv12, enabling the extraction of morphological, mechanical, and dynamic features. Feature selection was performed via factor analysis, and principal component analysis (PCA) and clustering were applied. Morphological features primarily identified cross-tissue phenotypes, while mechanical and dynamic features supported discrimination of biological conditions. Additionally, RT-qPCR arrays targeting 23 extracellular matrix (ECM)- and fibrosis-related genes were used for gene expression profiling to compare with morphological clustering results.

Results, Conclusions, and Discussions:: Morphology-based clustering distinguished six distinct morphological clusters (MCs). Tissue origins were generally well separated, with the exception of pancreatic fibroblasts cocultured with MiaPaCa2, which exhibited highly heterogeneous distributions across multiple clusters. Pancreatic fibroblasts from other conditions grouped together, characterized by a higher proportion of MC2. Breast CAFs also showed similar MC distribution patterns across their subpopulations (FAP+ PDPN- MHC II- PDGFRβ- and FAP+ PDPN+ MHC II- PDGFRβ-)
To further evaluate the biological discrimination capacity of physical phenotypes, clustering based on dynamic and mechanical features was performed, achieving higher resolution compared to RT-qPCR array analysis targeting ECM and fibrosis genes. According to the Shannon entropies of MCs within dynamic clusters (DCs) and mechanical clusters (MECs), MC6 exhibited a particularly notable pattern: cells consistently demonstrated high motility, weak traction forces, and a smooth, less actin distribution, with elongated morphologies and reduced solidity.
MC6 was weakly conserved across tissue origins but relatively enriched in pancreatic fibroblasts cocultured with MiaPaCa2, suggesting a link between the migratory phenotype and interaction with highly invasive cancer conditions. The consistency between morphology, traction force, and dynamic features underscores the relevance of morphology-based clustering for identifying migratory-prone fibroblast phenotype.
These findings highlight the potential of physical phenotype profiling to reveal conserved functional states among heterogeneous fibroblast populations. This approach offers a pragmatic and scalable strategy for advancing the understanding of fibroblast contributions to tumor progression and may contribute to the development of precision therapeutic strategies targeting the tumor microenvironment.

Acknowledgements and/or References (Optional):: This research was supported by a grant of the Korea-Us Collaborative Research Fund(KUCRF), funded by the Ministry of Science and ICT and Ministry of Health & Welfare, Republic of Korea (grant number: RS-2024-00468873). This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (2021R1A2C300840813). This research was supported by the Korean Fund for Regenerative Medicine (KFRM) grant funded by the Korea government (the Ministry of Science and ICT, the Ministry of Health & Welfare) (22B0401L1).