
The new method uses engineered Bacteria and AI to sense and record environmental signals
(Nanowerk News) Researchers in Professor Tal Danino’s Biomedical Engineering lab brainstormed several years ago on how they could engineer and apply pattern-forming bacteria naturally. There are many species of bacteria, such as Excellent proteus (P. mirabilis), which arrange themselves into defined patterns on a solid surface visible to the naked eye. These bacteria can sense several stimuli in nature and respond to these cues by “swarming”—the highly coordinated and rapid movements of the bacteria are supported by their flagella, long tail-like structures that cause whip-like movements to help propel them.
For inspiration, Danino’s team at Columbia Engineering, who have extensive experience using synthetic biology methods to manipulate bacteria, discussed where else they could find similar patterns in nature and what they could do. They noted how tree rings recorded the age of trees and climate history, and that sparked their idea to implement them P. mirabilis rings as a recording system. They were also interested in applying artificial intelligence (AI) to characterize the different features of bacterial colony patterns, an approach they realized could then be used to decode the engineered patterns.
“For us, this is an untapped opportunity to create natural recording systems for certain cues,” said Danino, a member of Columbia’s Data Science Institute (DSI).
In a new study, published in Natural Chemical Biology (“Engineered bacterial swarm patterns as a spatial record of environmental inputs”), the researchers worked with P. mirabilis, commonly found in soil and water and occasionally in the human intestine, is known by its well-marked-looking colony pattern. When bacteria are grown on solid growth medium Petri dishes, they alternate between the bacterial growth phase, which makes dense circles visible, and the bacterial movement, called a “swarming” movement, which expands the colony outward.

The team engineered the bacteria by adding what synthetic biologists call “genetic circuits”—systems of genetic divisions, logically arranged to make the bacteria behave in a desired way. The engineered bacteria sensed the presence of the inputs the researchers selected — from temperature to sugar molecules to heavy metals like mercury and copper — and responded by changing their swarming ability, which markedly altered the output pattern.
Working with Andrew Laine, Percy K. and Vida LW Hudson Professor of Biomedical Engineering and DSI member and Jia Guo, assistant professor of neurobiology (in psychiatry) at Columbia University’s Irving Medical Center, the researchers then applied deep learning — a state-of-the-art AI technique — to decode the environment from patterns, the same way scientists look at the rings on a tree trunk to understand its environmental history. They used models that could classify patterns holistically to predict, for example, the concentration of sugars in a sample, and models that could describe or “segment” edges in patterns to predict, for example, how many times the temperature changed as the colony grew. .
Advantages of working with P. mirabilis is that, compared to many typical patterns of engineered bacteria, the original P. mirabilis patterns are visible to the naked eye without expensive visualization technology and form on solid agar substrates that are durable and easy to work with. This property increases the potential for implementing systems as sensor readouts in a variety of settings. Using deep learning to interpret patterns can allow researchers to extract information about the concentration of input molecules even from complex patterns.
“Our goal was to develop this system as a low-cost detection and recording system for conditions such as pollutants and toxic compounds in the environment,” said Anjali Doshi, lead author of the study and a recent PhD graduate of Danino’s lab. “To our knowledge, this work is the first study in which a naturally pattern-forming bacterial species has been engineered by a synthetic biologist to modify its native swarming abilities and function as a sensor.”

Such work could help researchers better understand how native patterns form, and more than that, could contribute to other areas of biotechnology beyond the field of sensors. Being able to control bacteria as a group rather than as individuals, and control their movement and organization within a colony, could help researchers construct living materials on a larger scale, and aid the Danino lab’s parallel goal of engineering bacteria to live “smart” lives. therapeutic, by allowing better control of the behavior of bacteria in the body.
This work is a novel approach to constructing macro-scale bacterial recorders, extending the framework for engineering the behavior of emerging microbes. The team next plans to build on their system by engineering the bacteria to detect a wider range of pollutants and toxins and moving the system to safe “probiotic” bacteria. Ultimately, they aim to develop devices to implement recording systems outside of the lab.