Single-cell transcriptomics identify mechanical-memory–associated cell states in metastatic HR+ breast cancer

Mechanical memory has recently emerged as an important concept in tumor mechanobiology, reflecting the ability of cancer cells to retain and integrate past mechanical cues to guide future behaviors. However, a systematic definition of mechanical memory–related genes and their functional implications across cancers remains lacking. We curated literature-supported mechanotransduction and mechanical memory pathways to construct a 79-gene Mechanical Memory Signature (MMS). MMS expression patterns and prognostic relevance were evaluated across 32 cancer types using TCGA datasets, followed by focused analyses in breast cancer. Single-cell RNA sequencing datasets containing primary and liver metastatic hormone receptor–positive (HR+) breast cancer cells were used to assess MMS activity at single-cell resolution. Pseudotime and transcriptional regulatory analyses were performed to define MMS-associated cellular states. A three-dimensional stiffness-tuned collagen mechanical memory model was employed for experimental validation. MMS expression was broadly elevated in aggressive tumors and predicted unfavorable survival outcomes, with the strongest association observed in breast cancer. Single-cell integration revealed MMS-high tumor cell clusters exhibiting a progressive rise in MMS activity along pseudotime, suggesting acquisition and persistence of mechanical memory during metastatic evolution. Within these clusters, RELA was identified as a central transcriptional node strongly correlated with MMS activity and linked to cytoskeletal remodeling and ECM-regulatory genes, including PFN1, CFL1, RHOA, TIMP1, and MMP14. In 3D collagen cultures, matrix stiffening markedly increased RELA and MMP14 expression, while pharmacological activation of the RELA–RhoA axis further amplified this effect even under soft matrix conditions. Our study revealed RELA as a key mediator of mechanical memory–driven metastatic behavior in ER+ breast cancer, providing a mechanistic framework and actionable targets for understanding and potentially disrupting mechanically informed tumor progression.