The mission of the Human Cell Atlas is to create comprehensive reference maps of all human cells to describe and define the cellular basis of health and disease. Dr. Kai Kessenbrock, Assistant Professor of Biological Chemistry at UC Irvine, is a Co-Principal Investigator on the Human Breast Cell Atlas Project, which seeks to map all cell types and states present in human breast tissue with single-cell and spatial resolution.
Dr. Kessenbrock joined us for a recent Nature webinar to share how his team discovered unique cellular niches within the breast tissue microenvironment by complementing single-cell RNA sequencing data with CODEX spatial proteomic data. We’ve summarized some of the highlights from his presentation.
No cell is an island
Cells are embedded into a complex three-dimensional microenvironment consisting of the extracellular matrix as well as other cell types, which may form tight cell-cell interactions with the cell of interest. The breast epithelial system consists of a ductal epithelial network embedded into adipose rich tissue within the breast. This system contains several distinct regions of interest, including lobular units, connective tissue, ducts, and adipose tissue.
Dr. Kessenbrock’s lab is studying breast tissue at various stages: during normal homeostasis, early tumorigenesis, and cancer immunity, when immune cells are inhibited by the tumor microenvironment and tumor cells begin to metastasize.
Single-cell level analysis allows us to identify and characterize previously unrealized distinct cell populations.
The team previously used bulk-level analysis tools, including Western blot, quantitative PCR, and bulk RNA-sequencing. These methods enable the analysis of tens of thousands to millions of cells, but provide a population average that may miss underlying cell states in a subset of the cell populations. “Single-cell level analysis allows us to identify and characterize previously unrealized distinct cell populations,” said Dr. Kessenbrock.
Dr. Kessenbrock’s goal is not only to profile each individual cell type on a single cell level, but also to understand where in the tissue they’re located. The strategy they’ve used so far involves isolating cells and performing single-cell RNA sequencing to look at the way the cells are maintained in the tissue. However, the team is also interested in learning about lineages within the mesenchymal cell types that are locally maintained within the tissue.
Building a spatially resolved single-cell multi-omics atlas of the human breast
The breasts undergo quite a drastic reconstruction during the lifecycle of human individuals from puberty, through pregnancy, lactation, involution, and menopause. These are very distinct phases of rearrangement of the tissue. Thus, it’s important to keep a close eye on each individual sample being profiled, Dr. Kessenbrock noted. Ultimately, the goal of the project is to build a more precise reference atlas, which allows researchers to determine when a tissue deviates from normal, improving early detection of breast cancer.
Single-cell RNA sequencing makes it possible to capture most of the important cell types in breast tissue. After analyzing their single-cell RNA-seq data, however, Dr. Kessenbrock’s team noticed a lack of mature adipocytes. They concluded that most mature adipocytes would not survive the harsh dissociation process involved in single-cell RNA sequencing. To combat this issue, the team used single-nucleus RNA sequencing, which enabled the capture of adipocytes and other previously undetected immune cells, including mast cells, which are known to reside within the ducts of the epithelium.
The combination of single-cell and single-nucleus RNA sequencing provided a data set which comprehensively covered all cell types in the breast. Based on the markers defined via single-cell transcriptomics, the team built a 34-antibody panel in collaboration with Akoya using the CODEX platform for multiplex immunofluorescence-based spatial profiling of breast tissue.
Qualitative assessments of CODEX imaging data
CODEX imaging of breast samples revealed the four distinct regions of interest: lobular units, connective tissue, ducts, and adipose tissue.
The image below zooms in on one of the regions containing a lobular unit. Markers for luminal, epithelial, and basal cells indicate where the epithelium is located in relation to surrounding stroma. Using single-cell RNA sequencing data, the team observed a number of distinct cell types and cell states within the epithelium.
One of the basal cell states they identified was marked by high levels of Keratin 14 along with myoepithelial markers. By visualizing KRT14 on the tissue, it can be inferred that this cell state likely corresponds to the ductal basal epithelium compartment, and less so within the lobular epithelium.
The team was able to qualitatively assess a number of other cell populations using CODEX images, which Dr. Kessenbrock covered during his presentation. But qualitative assessment alone was not enough. The team went on to further quantify and analyze the imaging data.
We’re combining the power of spatial analysis with the power of single-cell RNA sequencing.
Identifying cellular neighborhoods and interactions within the human breast
In collaboration with Akoya, Dr. Kessenbrock’s team applied the Stardist pipeline to perform segmentation, capturing nuclear and cytoplasmic markers and turning imaging data into a cell by protein matrix. The structure of the data returned was similar to that returned by single-cell RNA sequencing, noted Dr. Kessenbrock.
Unbiased dimensionality reduction and clustering of the data resulted in a number of clusters that need to be labeled. Dr. Kessenbrock’s team is currently working on automated labeling algorithms to pull in RNA sequencing information to automatically label clusters. After cells were segmented and labeled, each cell had its own identity and two-dimensional coordinates within the tissue sample, making it possible to quantitatively analyze the multiplex immunofluorescence data.
Looking back at the regions of interest in breast tissue, Dr. Kessenbrock’s team was now able to ask questions about cellular density and proportion of cell types within those regions and across the whole tissue.
The team found that ducts and lobules in the breast have the highest cell density and diversity. A Voronoi analysis, shown below, illustrates the structure of the cellular ecosystems within the tissue. Dr. Kessenbrock aptly compared the high cellular density and diversity present in lobular units to the high population density and diversity in cities and metropolitan areas, while ducts correspond to “interstate highways”. Connective and adipose tissue display the same low density and diversity expected in suburban and rural areas.
Dr. Kessenbrock’s team performed cellular proximity analysis on the data to further define, in an unbiased way, the patterns in which breast tissue is composed. One way to do this is via Delauney triangulation, a graph-based method of determining which cells are geographically close to one another. Each cell has its nearest neighbors, and this analysis makes it possible to characterize typical neighboring cells. The team was able to build connective maps based on this proximity analysis.
Another useful method for identifying “cellular neighborhoods” was defined in a recent paper by Dr. Garry Nolan’s lab at Stanford University. Dr. Kessenbrock’s team applied this to breast tissue, identifying 14 distinct cellular neighborhoods. To understand the how cells communicate with one another within these neighborhoods, the team reconstructed mechanisms of cellular interactions between cell types using the CellChat pipeline.
“We’re combining the power of spatial analysis with the power of single-cell RNA sequencing, where we have rich information about ligands and receptors being expressed tailored to these niches and cellular neighborhoods defined using spatial analysis,” said Dr. Kessenbrock. Through this multi-omics approach the Human Breast Cell Atlas is developing a comprehensive map of the distinct cell populations, pathways, lineage hierarchies, and cellular neighborhoods in human breast tissue.
To learn more about the Human Breast Cell Atlas project, watch Dr. Kessenbrock’s full presentation on-demand.