Cells extracted from a frog embryo can be sculpted to create new shapes and carry out unique functions in a structure that’s not-quite-organism and not-quite-machine, researchers report today (January 13) in PNAS.
“My first reaction to the article was: Holy moly, this is potentially huge,” Pamela Lyon tells The Scientist in an email. Lyon is a cognitive biologist at the University of Adelaide who was not involved in the study but is collaborating on a different project with Michael Levin, a developmental biologist at Tufts University and a senior author on the study.
The researchers designed and built so-called xenobots that could locomote across the bottom of a petri dish. They also designed structures that could manipulate and transport other objects. When several designs were housed together, they began to exhibit “collective behaviors” such as orbiting one another or temporarily binding.
“This is just the tip of the iceberg,” says Levin. “The goal of this is not to make frog cell robots. In the end, the goal of this is to understand the very big questions about the relationship between genomes and anatomy and to move forward regenerative medicine, robotics, and A.I.”
Evolutionary robotics is a new strategy that some scientists use to probe questions about the function of biological systems and others use to design and test the abilities of possible robotic structures. Josh Bongard, a coauthor and a computer scientist studying evolutionary robotics at the University of Vermont and his team first started working with Levin on a shared grant from DARPA. As they became acquainted with each other’s work, Bongard shared several of the designs of simple, cell-based robots that his algorithm had created with Levin and his team.
Bongard’s designs were only theoretical, but to his surprise, a microsurgery expert and staff scientist in Levin’s lab, Douglas Blackiston, took one of the plans and actually built it using frog embryo cells.
Up until then, says Bongard, “we didn’t realize that they were able to actually take computer-designed blueprints and build a design in reality. That really shifted the focus of the project.”
The question became, “cells that normally will build a frog—what else could they build?” says Levin.
Bongard’s team developed an evolutionary algorithm. Its job was to determine what shape the cardiac and skin cells from the frog should take in order to move across the bottom of a petri dish.
“The idea here is kind of like if you had a whole bunch of people that were rowing but they’re all rowing at their own their own frequency,” says Bongard. “Could you design the shape of a boat and put people in the boat so that the boat moves in a straight direction?”
First, the program generates a bunch of random shapes made up of the two building blocks—heart cells from the frog embryo that contract and skin cells from the embryo that remain passive. Then it performs its own version of natural selection. It begins testing the designs in simulations to see which ones are best at completing the desired task. Ineffective forms are deleted. The most efficient designs undergo further simulations in different environments, allowing the algorithm to select the most robust versions—those that can maintain the desired function in several environments and withstand random noise.
Then Levin’s team extracts cells from the frog embryos, allows them to grow in “a pile,” and eventually carves them into the shape the algorithm established, like a sculptor building a statue—except this statue is made of living cells rather than stone.
Levin imagines that in the future, such structures could be used to build machines that could repair themselves, like tissue does when its damaged. The machines could collect microplastics in the ocean without adding their own pollution as they degrade in the saltwater, improve drug delivery, or create completely new treatments for disease.
In addition to these practical applications, Levin says, developmental biologists will be interested in what these structures can teach us about how cells organize.
Still, the work is in its earliest stages. David Ackley, a computer scientist at the University of New Mexico who was not involved in the study, says, “this is a marvelous proof-of-concept demonstrating how the artificial evolution of mathematical ‘soft robots’ can guide the designs of functional biological systems.”
S. Kriegman et al., “A scalable pipeline for designing reconfigurable organisms,” PNAS,