A Discovery That Could Help Predict Which Cells Will Become Cancerous

A catalog of breast cells has been created to find the origins of cancer and predict which cells could become cancerous.

Breast tissue changes dramatically throughout a woman's life, making it particularly difficult to find a biomarker for sudden changes that could lead to cancer. Associate Professor Camila dos Santos and her team at Cold Spring Harbor Laboratory (CSHL) identified thousands of types of normal human and mouse breast cells and created a catalog of breast cells. The new catalog redefines healthy breast tissue, giving scientists an opportunity to investigate cancer development or other changes in breast tissue by identifying where these changes originate.

Any breast cell can become cancerous. Dos Santos says: "Breast cancer risk can only be understood by first understanding normal breast cells. So when we think about preventive and even targeted treatments, we are preventing the right types of cells from turning into cancer cells."

Figure-1: Mammalian tissue consists of various cells, including the epithelial (pink), immune (cyan) and fat (black) cells visible in this cross-section of mouse breast tissue.

The traditional way of creating a catalog of cells is to track a few specific genes related to a cell's function. To classify cells more comprehensively, dos Santos's team turned to single-cell RNA sequencing, a relatively new technique. The team studied more than 15,000 cells in mouse and human breast tissue and tracked the activity of several genes. Samantha Henry, a graduate student from dos Santos's laboratory at Stony Brook University, described the study by saying: "We really created a catalog of many genes for each group of cells to identify them better."

Cell Diversity in Breast Tissue

One group of cells typically contains estrogen receptor-positive cells. However, dos Santos's team found that this group could be divided into several subgroups, each of which could respond differently to treatments. Analyzing noncancerous breast tissue from nulliparous women (women who had never given birth), the team obtained a set of human breast tissue cells (hTM) representing clusters of breast epithelial cells (EPCAM+), myeloid cells (SERPINE1+), T lymphocytes (CD3E+), B lymphocytes (MS4A1+), endothelial-like cells Claudin 5 (CLDN5) and fibroblast-like cells (MYLK+) (Figure-2A).

Figure-2: A) UMAP plot of the cluster distribution of total noncancerous breast tissue from nulliparous women (n=5, hTM). B) UMAP plots showing the expression levels of EPCAM, SERPINE1, CD3E, MS4A1, CLDN5 and MYLK mRNAs in hTM clusters.

Figure-3: A) Dot plot and dendrogram branching showing the average and percentage of expressed genes that distinguish and classify clusters of epithelial cell lineages (hECs). B) Diffusion map projection showing a Monocle-informed transcriptional trajectory arrangement of hEC clusters. Arrows are shown to suggest that a possible portion of the cells' bipotential progenitor fate (hEC3) leads to myoepithelial progenitors (hEC5) and predicted luminal progenitors (hEC2 and hEC4).

Working with CSHL Professor of Computational Biology Adam Siepel and his team, Santos's team also showed that mouse and human breast cells have similar genetic profiles and why studies in mice may be relevant to humans. Marygrace Throusdell, a computational scientist in the dos Santos laboratory, said: “We always find ourselves struggling with clear classifications of certain cell types. Sometimes I see every gene there and then think that something else will emerge.”

The laboratory is currently working to distinguish normal genetic changes from those associated with cancer. Dos Santos said she hoped the list of breast cell types would show which types of cells are involved and guide research toward future treatment options.