Cancer-associated fibroblasts (CAFs) have distinct roles within the tumor microenvironment, which can impact the mode and efficacy of tumor cell migration. CAFs are known to increase invasion of less-aggressive breast cancer cells through matrix remodeling and leader–follower dynamics. Here, we demonstrate that CAFs communicate with breast cancer cells through the formation of contact-dependent tunneling nanotubes (TNTs), which allow for the exchange of cargo between cell types. CAF mitochondria are an integral cargo component and are sufficient to increase the 3D migration of cancer cells. This cargo transfer results in an increase in mitochondrial ATP production in cancer cells, whereas it has a negligible impact on glycolytic ATP production. Manually increasing mitochondrial oxidative phosphorylation (OXPHOS) by providing extra substrates for OXPHOS fails to enhance cancer cell migration unless glycolysis is maintained at a constant level. Together, these data indicate that tumor–stromal cell crosstalk via TNTs and the associated metabolic symbiosis is a finely controlled mechanism by which tumor cells co-opt their microenvironment to promote cancer progression and may become a potential therapeutic target.

Cancer cell migration is highly heterogeneous, and the migratory capability of cancer cells is thought to be an indicator of metastatic potential. It is becoming clear that a cancer cell does not have to be inherently migratory to metastasize, with weakly migratory cancer cells often found to be highly metastatic. However, the mechanism through which weakly migratory cells escape from the primary tumor remains unclear. Here, utilizing phenotypically sorted highly and weakly migratory human breast cancer cells, we demonstrate that weakly migratory metastatic cells disseminate from the primary tumor via communication with stromal cells. While highly migratory cells are capable of single cell migration, weakly migratory cells rely on cell-cell signaling with fibroblasts to escape the primary tumor. Weakly migratory cells release microvesicles rich in tissue transglutaminase 2 (Tg2) which activate murine fibroblasts and lead weakly migratory cancer cell migration in vitro. These microvesicles also induce tumor stiffening and fibroblast activation in vivo and enhance the metastasis of weakly migratory cells. Our results identify microvesicles and Tg2 as potential therapeutic targets for metastasis and reveal a novel aspect of the metastatic cascade in which weakly migratory cells release microvesicles which activate fibroblasts to enhance cancer cell dissemination.

Background/Objectives: Cancer cell migration is a hallmark of cancer and is associated with metastasis. While large-scale functional screens have identified regulators of migration, less is known about how intrinsic transcriptional heterogeneity drives highly migratory phenotypes within individual cancer models or whether these transcriptional changes are conserved across different models of varying tissues of origin. This study aims to define shared and cell line-specific transcriptional programs associated with cancer cell migration and analyze their relevance to patient datasets. Methods: Five cancer cell lines across three cancer types (breast, colorectal, and melanoma) were subjected to transwell-based migratory sorting to isolate highly and weakly migratory subpopulations. Bulk RNA sequencing, differential gene expression analysis, Gene Ontology (GO) enrichment, and upstream regulator prediction were performed. Public tumor datasets were analyzed to evaluate gene expression and its association with patient survival. Results: EVA1A was consistently upregulated in all highly migratory (HM) subpopulations. Multiple GO terms were enriched across all cell lines, often driven by distinct gene signatures, indicating convergence at the level of biological processes despite transcriptional divergence. TEAD4 was predicted as an upstream regulator, and increased TEAD4 nuclear localization was observed in four of the five HM subpopulations. EVA1A and TEAD4 expression were elevated in tumors relative to normal tissues, with cancer type-dependent survival outcomes. Conclusions: Migratory selection was accompanied by extensive transcriptional change within each model, yet across five models spanning three tissue types these changes converged on shared biological processes rather than shared genes. Migration-associated phenotypes may therefore be better defined by pathway-level than single-gene analyses, and the clinical relevance of regulatory nodes such as TEAD4 appears conditional on cancer type.