New technology developed by Stanford Medicine researchers automatically identifies cell types and provides a comprehensive view of how cells interact with their environment in intact tissues. The tool, which draws on gene expression and spatial data, allows researchers to compare how changes in cell interactions contribute to the progression of cancer and other diseases, and provides a view of how immune cells interact with cancer cells in a tumor.
The research appears in Cell.
Ten years ago marked the introduction of a technique called CyTOF — a type of mass cytometry that measures up to 40 intracellular and surface protein markers from a single cell at once. The new technology, called CODEX, provides the same depth of expression data as CyTOF but adds spatial information to it.
“We’re able to accurately label and identify cell types within intact tissue, which has not been done before since the cells are so crowded and overlap,” said Nikolay Samusik, PhD, research scientist in microbiology and immunology, who shares first authorship of the paper with Yury Goltsev, PhD, a senior research scientist in the same department. “We pioneered new computational models of image analysis to look at how the cells are grouping together in the tissue, and we made some interesting discoveries."
They compared tissue from the spleens of healthy mice with spleen tissue from mice with lupus, an autoimmune inflammatory disease, systematically examining cell interactions in healthy tissues and during the progression to disease state.
One surprising observation was that each of the 24 cell types they identified was statistically more likely to neighbor cells of the same type. Another finding was that cell types express different levels of biomarkers depending on where they are positioned in the tissue and what cell types surround them. Tissue-specific and microenvironmental impact on gene expression has never been shown on such a large scale and with such rigorous statistical analysis, Goltsev said.
Many of the clinical applications of CODEX focus on analyzing tissue from human disease, said Garry Nolan, PhD, professor of microbiology and immunology and senior author of the paper.
It can also be used for diagnostics. For example, clinicians could look at 10 markers of interest at once in one tissue sample instead of having to stain 10 separate samples, one per biomarker, according to Samusik.
The method could also be used for cancer immunology, Goltsev said. For example, CODEX can show how immune cells interact with cancer cells inside the tumor. “One of the biggest hurdles of immunotherapy is the toxicity associated with autoimmune response,” Goltsev said. “Knowing which cells are affected and which ones make the effects is very important.”
The work took about four years of refining and troubleshooting — a prototype part made out of Legos remains on display in the lab as a testament to their early efforts. "We had a lot of interesting and creative engineering solutions from everyone who was involved," Samusik said.
Some scientists believe that automated diagnostics and automated pathology reports are the wave of the future, and Goltsev said that using multidimensional analysis of cell environment is the ticket to those possibilities. But the biggest accomplishment of the project, he said, is that it created a new tool to fuel future discoveries: The group used the new technology to create large data sets that have been shared with the scientific community for examination and further analysis.
Image by Frank Tranfaglia