Human-induced pluripotent stem cells (hiPSCs) generated from a person’s own adult cells can grow into complex organs that help scientists test drugs or even transplant into patients. However, directing stem cells to form functional organs in the lab remains challenging.
In a study published in the journal Cell Systems, researchers from Gladstone Institutes, in collaboration with Boston University (BU), described using machine learning to better understand how to use CRISPR-Cas9 gene-editing tools to control iPSC organization.
By coaxing these stem cells into forming specific arrangements, the researchers believe they could create functional organs for research or therapeutic purposes.
Researchers have managed to develop iPSCs into many different cell types, but not necessarily functional 3D organs, mainly because they have struggled to manipulate the spatial patterns of stem cells, which define the tissues they eventually grow into. Some have resorted to 3D printing, but it isn’t always successful, as cells often migrate away from their printed locations.
“Despite the importance of organization for functioning tissues, we as scientists have had difficulty creating tissues in a dish with stem cells,” Ashley Libby, a co-first author of the new study, said in a statement. “Instead of an organized tissue, we often get a disorganized mix of different cell types.”
The researchers previously showed that knocking down two genes, ROCK1 and CDH1, affected the layout of iPSCs in lab dishes. The proteins they encode help regulate interactions between cells, making them ideal candidates to alter the cellular organization of an iPSC group.
But there are so many variables that make testing all the combinations by human almost impossible, including the timing and level of each gene knockdown, the duration and the proportion of cells to work on, and more. So they turned to machine learning for help.
RELATED: Growing transplantable arteries from stem cells
They used a CRISPR-Cas9 gene-editing system that could be triggered by adding the antibiotic doxycycline. To help link changes to specific arrangements of the iPSCs, the cells were also engineered to fluoresce in different colors when they lost ROCK1 or CDH1.
Researchers at Gladstone tested different doses and timing of gene blockade. How changes in cell subpopulations affected the observed pattern was captured, and the BU computational scientists fed the results to a machine learning algorithm, which was hence trained to classify patterns according to their similarity and infer ways of how ROCK1 and CDH1 affect iPSC organization.
“Our machine-learning model allows us to predict new ways that stem cells can organize themselves, and produces instructions for how to recreate these predictions in the lab,” the study’s co-first author Demarcus Briers said in a statement.
The model simulated specific experimental conditions—such as when, where and how to add drugs to the iPSCs—that could yield unique patterns in silico. Then, the team put those suggested conditions to test.
It was successful. The researchers were able to generate concentric circles to two layers of stem cell populations in a bull’s-eye pattern, they reported.
“We've shown how we can leverage the intrinsic ability of stem cells to organize,” Todd McDevitt, the study’s senior author, said in a statement. “This gives us a new way of engineering tissues, rather than a printing approach where you try to physically force cells into a specific configuration.”
RELATED: Nose drop with adult stem cells restores sense of smell in mice
Stem cells are a key venue for regenerative research, either for studying disease and potential treatment or for transplant. Last year, scientists from the University of Edinburgh used 3D scaffolds made of polycaprolactone to carry embryonic stem cells and iPSCs, and successfully generated functional liver tissues that help diseased mice break down the amino acid tyrosine. A research team at the Morgridge Institute for Research recently used a drug called RepSox to help iPSCs form better smooth muscle cells as building blocks for functional arteries.
For the Gladstone-BU team, the researchers are planning to expand the model to include other genes to get an even wider pool of possible cell configurations. On top of that, rather than just making flat patterns as in this study, their goal is to design 3D shapes or organs.
“We're now on the path to truly engineering multicellular organization, which is the precursor to engineering organs,” said McDevitt. “When we can create human organs in the lab, we can use them to study aspects of biology and disease that we wouldn't otherwise be able to.”