Blood vessels are a hideaway for disseminated breast tumor cells

From the Ghajar lab, Human Biology Division

Breast cancer is the second-largest cause of cancer-related death for women in the US. Although improved preventive screening strategies frequently catch the tumor early enough to allow for a curative treatment, about ten percent of women diagnosed with invasive breast cancer relapse five years or more after surgery and adjuvant treatment. Tumor cells that disseminate (disseminated tumor cells or DTCs) from the primary tumor and persist at distant locations despite adjuvant chemotherapy are thought to be the main source of relapses. As a postdoctoral fellow in Mina Bissell’s laboratory, Cyrus Ghajar demonstrated that the perivascular niche (PVN) regulated breast tumor dormancy. Now an assistant member in the Public Health Sciences division’s Translational Research Program and in the Human Biology division, his lab hypothesized that the PVN could also be responsible for DTC resistance to chemotherapy. Their results were published last month in the journal Nature Cell Biology.

Microscopic observation of disseminated tumor cells (DTCs, green) lining vascular structures (red) in the bone marrow (grey is for all nuclei).
Microscopic observation of disseminated tumor cells (DTCs, green) lining vascular structures (red) in the bone marrow (grey is for all nuclei). Illustration provided by Dr. Arko Dasgupta

To mimic adjuvant chemotherapy after surgery in a mouse model, the authors implanted fluorescently-labelled breast tumor cells in the mammary fat pads of syngeneic mice and let the tumors grow for two and a half weeks. They surgically resected the tumor and started chemotherapy (doxorubicin and cyclophosphamide or paclitaxel) one week after, for a total period of five weeks. To assess the presence of DTCs in the bone marrow (BM), femur bones were analyzed by immunofluorescence, and the location of DTCs relative to endothelial cells (ECs) of blood vessels was assessed. Patrick Carlson and colleagues demonstrated that DTCs localize closer to blood vessels when chemotherapy is administered, although the overall DTC burden was reduced. This suggested that chemotherapies select for perivascular DTCs. In other words, only DTCs in the PVN are resistant to the treatment. In cell culture, chemotherapeutic treatment of breast tumor cells co-cultured with BM stromal cells alone or combined with fluorescently-labelled ECs confirmed the in vivo observations. Tumor cells co-cultured with BM stroma underwent apoptosis upon chemotherapeutic treatment whereas co-culture with BM stroma and ECs protected tumor cells from chemotherapy-induced cell death. This cell culture model is highly relevant, as one of the first authors, Dr. Candice Grzelak, comments: “Utilizing our organotypic culture model that faithfully mimics the in vivo breast cancer dormancy niche was essential to complete this research project. It provides a wonderful platform for proof of concept studies that can then be followed up in vivo.”

The authors wanted then to tackle one of the most important questions of the field. Dr. Arko Dasgupta explains: “There is a long-held belief that dormant cells are chemoresistant because they don’t cycle.” Indeed, as chemotherapy targets proliferative cells, it is possible that the quiescent state of the DTCs rather than the microenvironment is responsible for the resistance to treatment. To answer this question, the Ghajar lab engineered breast tumor cells designed to co-express markers for quiescence and apoptosis. When co-cultured with BM stroma, chemotherapy killed cells that were mostly non-quiescent. However, when BM stroma and ECs were present in the culture, both quiescent and non-quiescent cells were equally resistant to chemotherapy. In addition, stimulation of DTC proliferation by Insulin-like Growth Factor 1 did not increase the chemosensitivity of DTCs on microvascular niche cultures. For the team, it was a clear demonstration that PVN-induced DTC chemoresistance was independent to the cell cycle status.

To understand the molecular mechanisms behind such interesting observations, the researchers sequenced the transcriptome of the BM stroma versus BM stroma plus ECs and identified integrin signaling as highly enriched in the BM stroma plus ECs. Using their co-culture model and a chemoresistance assay, they tested a variety of antibodies blocking different integrins and identified Integrins β1 and αvβ3 as key players in PVN-mediated tumor cell chemoresistance. They identified two ligands of these integrins, von Willebrand Factor (VWF) and vascular cell adhesion molecule 1 (VCAM1), that are expressed in the vascular endothelium. Downregulating VWF expression by shRNA or functionally blocking VCAM1 with an antibody both chemo-sensitized tumor cells, demonstrating that direct interaction between tumor cell integrins and endothelial ligands is involved.

Although their culture model mimics the bone marrow microenvironment, the authors sought to assess the therapeutic potential of integrin blockade in vivo. Going back to the above-mentioned mouse model for adjuvant (after resection) dose-dense chemotherapy, they demonstrated that downregulation of both integrins β1 and αv in tumor cells lead to decreased DTC burden in the BM. However, to assess the metastatic potential of these DTCs with or without functional integrin signaling, the authors had to switch to a more metastasis-permissive model using immunodeficient mice. Cardiac injection of human tumor cells was followed five days later by treatment with chemotherapy with or without antibodies targeting human Integrins β1 and αvβ3.  When integrin signaling was blocked, DTC burden in the BM was reduced by over 90%, and metastasis-free survival was increased. Thus, disruption of the PVN-mediated chemoprotection by integrin signaling blockade has the potential to prevent recurrence in metastatic breast cancer.

Dr. Dasgupta summarizes: “Here we have shown that it is a direct interaction between dormant cells and their microenvironment that enables the cells to survive chemotherapy regardless of cell cycle status. Additionally, we show that if we disrupt this interaction (between dormant cells and their microenvironment) then we can chemosensitize them both in our culture models and in mice.” He acknowledges that “this manuscript is focused mainly on the bone marrow perivascular niche; which begs the question as to whether the same mechanisms exist in the perivascular niches of other tissues when it comes to breast cancer chemoresistance. To this end our lab is actively developing organotypic models to study other tissues that harbor these dormant cells that become metastatic lesions.” Dr. Ghajar is confident that this work may translate to other diseases: “We think these results and the phenomena we describe is applicable to several cancers, especially those showing a dormancy phenotype or show association with the perivascular niche. However, this needs to be empirically tested and we hope that our paper gives impetus to others to develop cancer specific organotypic cultures and find ways to target residual diseases.”


This work was supported by the start-up funds provided by the Fred Hutchinson Cancer Research Center, by the Cuyamaca foundation, by the Department of Defense Breast Cancer Research Program, by the Breast Cancer Research Foundation, by the National Breast Cancer Coalition’s Artemis Project for Metastatic Prevention and the National Institutes of Health.

Cancer Consortium faculty members Drs Cyrus Ghajar and Peter Nelson contributed to this research.

Carlson P, Dasgupta A, Grzelak CA, Kim J, Barrett A, Coleman IM, Shor RE, Goddard ET, Dai J, Schweitzer EM, Lim AR, Crist SB, Cheresh DA, Nelson PS, Hansen KC, Ghajar CM. 2018 Targeting the perivascular niche sensitizes disseminated tumour cells to chemotherapy. Nat Cell Biol. 2019 Jan 21. doi: 10.1038/s41556-018-0267-0.