MIT's IRIS model reads the signals that tell stem cells what to become

Fingerprints left by signaling pathways

As an embryo grows from a small cluster of stem cells, those initially unspecialized cells become brain, liver or muscle cells and organize into tissues and organs. Which type a cell becomes depends on which genes are turned on or off. According to the researchers, that decision is not made by cells alone: they constantly send and receive chemical signals to and from their neighbors, which help them learn where they are, what developmental stage they are in and what they should become.

These messages travel through multi-step sequences called signaling pathways, which translate external signals into specific changes in gene activity. Retracing the instructions a cell has received would help scientists understand how tissues develop and how the process goes wrong in disease. But scientists have long assumed that the effects of signaling pathways vary widely across cell types, which would mean mapping each pathway separately in each cell type, a painstaking process.

The new study, led by MIT assistant professor of biology and Whitehead Institute member Pulin Li and graduate student Nicholas Hutchins, found that each signaling pathway leaves a unique "fingerprint", a distinctive pattern of gene activity reflecting the signals the cell has encountered. These fingerprints are consistent across different cell types for the same pathway. Scientists can therefore reconstruct signaling histories across many cell types instead of mapping each one separately.

How IRIS was trained and tested

IRIS is a neural network-based model, a system designed to recognize patterns in complex data. It examines a cell's overall gene activity and estimates which signaling pathways were likely "on" at specific times during development.

Li and Hutchins trained IRIS on a large experimental dataset that measured how thousands of human embryonic stem cells responded to dozens of combinations of six major signaling pathways at multiple stages of development. This produced an atlas of how signaling combinations influence cell behavior.

The team then tested the model on single cells from mouse embryos during gastrulation, a stage when cells are rapidly branching into different fates. According to the researchers, IRIS accurately predicted when and where specific signaling pathways would activate in cells destined to become heart, gut, muscle and spinal cord tissue.

Li compared the approach to voice recognition. "Think of voice recognition systems like Siri, which are trained mainly in English, but then use that training to help them recognize other languages," she says. "This is called transfer learning, and this is why IRIS can work across many different cell types."

A prediction confirmed in the lung

The researchers also used IRIS to identify the signals needed to create a cell type critical for lung development. The model predicted that activating a specific signaling pathway would encourage lung-specific development, and experiments in mouse embryos confirmed it.

"In these ways, IRIS is helping us decode the language cells use to talk to each other at a much faster rate than we could realistically achieve through experiments," Hutchins says.

What it could mean for medicine and research

The team says that once researchers know the pattern of signals that drives a stem cell to become a specific cell type, they can recreate those signals to control stem cell fate in a lab or medical setting. That could accelerate stem cell engineering for regenerative medicine and improve the creation of organoids, miniature 3D models that mimic real organs, used to study disease mechanisms and test new drugs.

For the lung, more accurate lab-grown models would allow study of diseases such as asthma, lung cancer and pulmonary fibrosis, in which lung tissue becomes scarred, often without a known cause. Researchers could then test potential therapies and ultimately design regenerative treatments to repair damaged tissue.

The researchers also say the approach opens the possibility of comprehensively mapping the signaling histories of every cell inside a mouse or human embryo at an unprecedented scale, replacing the need to experimentally test every pathway in every possible cell type.