How artificial intelligence found the words to kill cancer cells

Immune cell illustration cancer cell

Cancer is a disease characterized by the abnormal growth and division of cells in the body. Tumors can affect any part of the body and can be benign (noncancerous) or malignant (cancerous), and spread to other parts of the body through the bloodstream or lymphatic system.

A predictive model has been developed that enables researchers to encode instructions for cells to carry out.

scientists in University of California, San Francisco (UCSF) And IBM Research Created a virtual library of thousands of “command statements” for cells using machine learning. These “sentences” are based on sets of “words” that instruct engineered immune cells to continually find and eliminate cancer cells.

This research was recently published in the journal Sciencesis the first time that advanced computational techniques have been applied in a field that has traditionally progressed through trial-and-error experiments and the use of pre-existing rather than synthetic molecules to engineer cells.

The advance is allowing scientists to predict what elements — natural or synthesized — must be included in a cell to give it the precise behaviors required to respond effectively to complex diseases.

“This is a vital shift in the field,” said Wendell Lim, PhD, Byers Distinguished Professor of Cellular and Molecular Pharmacology, who directs the Cell Design Institute at UCSD and led the study. “Only by having this power of prediction can we get to a place where we can quickly design new cell therapies that carry out the required activities.”

Learn the molecular words that make up cellular command sentences

Much of therapeutic cell engineering involves selecting or creating receptors that, when added to a cell, will enable it to perform a new function. Receptors are molecules that line the cell membrane to sense the external environment and provide the cell with instructions on how to respond to environmental conditions.

Placing the right receptor on a type of immune cell called a T cell can reprogram it to recognize and kill cancer cells. These so-called chimeric antigen receptors (CARs) have been effective against some types of cancer but not others.

Lim and lead author Kyle Daniels, a researcher in Lim’s lab, focused on the part of the receptor found inside the cell, which contains chains of[{” attribute=””>amino acids, referred to as motifs. Each motif acts as a command “word,” directing an action inside the cell. How these words are strung together into a “sentence” determines what commands the cell will execute.

Many of today’s CAR-T cells are engineered with receptors instructing them to kill cancer, but also to take a break after a short time, akin to saying, “Knock out some rogue cells and then take a breather.” As a result, the cancers can continue growing.

The team believed that by combining these “words” in different ways, they could generate a receptor that would enable the CAR-T cells to finish the job without taking a break. They made a library of nearly 2,400 randomly combined command sentences and tested hundreds of them in T cells to see how effective they were at striking leukemia.

What the Grammar of Cellular Commands Can Reveal About Treating Disease

Next, Daniels partnered with computational biologist Simone Bianco, Ph.D., a research manager at IBM Almaden Research Center at the time of the study and now Director of Computational Biology at Altos Labs. Bianco and his team, researchers Sara Capponi, Ph.D., also at IBM Almeden, and Shangying Wang, Ph.D., who was then a postdoc at IBM and is now at Altos Labs, applied novel machine learning methods to the data to generate entirely new receptor sentences that they predicted would be more effective.

“We changed some of the words of the sentence and gave it a new meaning,” said Daniels. “We predictively designed T cells that killed cancer without taking a break because the new sentence told them, ‘Knock those rogue tumor cells out, and keep at it.’”

Pairing machine learning with cellular engineering creates a synergistic new research paradigm.

“The whole is definitely greater than the sum of its parts,” Bianco said. “It allows us to get a clearer picture of not only how to design cell therapies, but to better understand the rules underlying life itself and how living things do what they do.”

Given the success of the work, added Capponi, “We will extend this approach to a diverse set of experimental data and hopefully redefine T-cell design.”

The researchers believe this approach will yield cell therapies for autoimmunity, regenerative medicine, and other applications. Daniels is interested in designing self-renewing stem cells to eliminate the need for donated blood.

He said the real power of the computational approach extends beyond making command sentences, to understanding the grammar of the molecular instructions.

“That is the key to making cell therapies that do exactly what we want them to do,” Daniels said. “This approach facilitates the leap from understanding the science to engineering its real-life application.”

Reference: “Decoding CAR T cell phenotype using combinatorial signaling motif libraries and machine learning” by Kyle G. Daniels, Shangying Wang, Milos S. Simic, Hersh K. Bhargava, Sara Capponi, Yurie Tonai, Wei Yu, Simone Bianco and Wendell A. Lim, 8 December 2022, Science.
DOI: 10.1126/science.abq0225

The study was funded by the National Institutes of Health. 

Leave a Reply

Your email address will not be published. Required fields are marked *