The Importance of Segmentation in Spatial Biology

In a highly interconnected world, clear communication and information across the globe are critical. The same is true at the tissue and cellular levels. Cells communicate through chemical signals such as peptides, cytokines, chemokines, or exosomes. When a cell sends a signal, another cell with the proper receptor receives it and acts accordingly.

Cells talk to each other. Examples include metabolites in the blood triggering a cell’s receptors to cause the release of glucagon, cancer cells communicating with the surrounding healthy tissues to ensure their growth, metastasis, and drug resistance, or even viral proteins blocking interferon release thus preventing an efficient immune response. Everything that lives communicates. Not only is the conversation important, like in the case of cancer cells or viruses invading vital organs, knowing what cells are talking is just as critical.

Tumor signaling is made easier with spatial segmentation
In a dialogue, knowing who is talking is critical for understanding the conversation.

Context is Essential

For over a century, pathologists have understood that the context of expression is critical when delivering diagnostic information based on protein expression in tissues through techniques like immune histochemistry (IHC) used by clinicians to inform diagnosis and therapies. In IHC, antibodies specifically bind to target protein markers in the tissue. The antibodies are conjugated to a fluorescent or enzymatic dye that the pathologist can visualize under a microscope. Staining techniques highlight tissue morphology, informing the pathologist not only about what is expressed but also where it is located within the tissue.

All techniques that deliver a message within its context have limited multiplexing abilities. With the advent of genomics and proteomics, we now know that communication networks within and across cells are complex and involve hundreds to thousands of players.

Therefore, untangling the complexity of cellular relationships is a matter of how many proteins, transcripts, or genes one can visualize and measure within its cellular context.

Why Segmentation? Because Strength is in Numbers…and Location

In spatial biology, segmentation is the further section of a marker-defined area within a defined region of interest (ROI).

The GeoMx® Digital Spatial Profiler builds upon the importance of histology by allowing users to select biologically relevant regions of interest (ROI) based on the appropriate morphology markers for the tissue. The staining patterns not only guide researchers on ROI selection but also enable segmentation to isolate the expression patterns of specific compartments and cell types within a tissue. GeoMx® Digital Spatial Profiler does this by combining standard immunofluorescence techniques with digital optical barcoding technology, enabling multiplexed spatial analysis of RNA and proteins. DSP’s software contains a segmentation tool to focus the analysis on specific areas of the sample. To summarize the process, the samples are stained with a large panel of biological probes that carry a unique DNA barcode and fluorescently labeled antibodies or RNA in situ probes (morphology markers) to explore the morphology of the tissue visually or computationally. Users can define their regions of interest by outlining important biological regions or defining regions to contour based on cell or tissue morphology. Then each region can be further subdivided or segmented into more ROIs through fluorescent intensity of the morphology markers. For example, a “tumor” segment can be generated by finding and segmenting an area within the region of interest that is positive for a PanCK marker. Another segment can be the Immune Tumor Micro-Environment, defined as the area negative for PanCK and positive for CD45. Now the user can compare hundreds or thousands of RNA or protein expressions between the tumor and its microenvironment.

Segmentation and masking at work with GeoMx: now the user can see the molecular conversation between cells
Segmentation and masking at work with GeoMx: now the user can see the molecular conversation between cells

In other words, now we can witness the dialogue between cancer and immune cells to understand why some immune cells react to the tumor cells while others stay quiet and allow the tumor to grow.

Segmentation at work

In this study presented at the 2020 AACR Meeting, FFPE tissue sections from colorectal cancer (CRC) samples were analyzed to investigate tumor immunomodulation. Regions of interest were selected inside and outside the tumor invasive margin, focusing on tumor or stromal regions with high numbers of CD3+ tumor-infiltrating lymphocytes (TILs). Within each ROI, the researcher team used a custom segmentation strategy to isolate CD3+ cells and then highlight regions near these cells by using custom-defined compartments or segmentation defined by extending multiple contours around the initially selected lymphocytes to determine differences in the local environment of each population. Such segmentation and masking strategy revealed that regions neighboring TILs had up-regulated oncogenic pathways. In contrast, TILs specifically had increased expression of cytolytic pathway genes and several coinhibitory and costimulatory checkpoint genes. The authors provided evidence that segmentation enables pathway differential expression analyses and exploration of critical interactions between neighboring cell types while retaining the spatial context.

In a seminal study conducted by Christopher Mason’s group at Cornell University, the team studied the “cartography” of three structural components of the lungs – vascular, airway, and alveolar regions – in patients who died from COVID-19 and compared them with healthy and non-viral lung disease tissues. GeoMx DSP segmentation capability allowed them to map the infection in these tissues, discover disruptions of cell-to-cell interactions, and analyze the impact of the disease across entire tissues for the first time.

If you are interested in learning more about GeoMx and the power of segmentation, you can find more information in our webinars, product bulletins, manuals, and blogs.

For Research Use Only. Not for use in diagnostic procedures.

By Laura Tabellini Pierre
For research use only. Not for use in diagnostic procedures.