Microorganisms can be modified to detect a range of substances, including contaminants or soil elements. However, in the majority of instances, these cues can solely be perceived by examining the cells through a microscope, rendering them unsuitable for widespread implementation.
Employing a novel approach that prompts cells to synthesize molecules that produce distinct color combinations, engineers at MIT have demonstrated that they can interpret these bacterial cues from distances up to 90 meters. Their research could pave the way for the creation of bacterial sensors applicable in agriculture and other fields, potentially monitored by drones or satellites.
“It’s an innovative method of extracting information from the cell. If you’re directly adjacent to it, nothing can be seen with the naked eye, yet from several hundred meters away, utilizing specialized cameras, you can acquire the data when it activates,” states Christopher Voigt, leader of MIT’s Department of Biological Engineering and the principal author of the recent study.
In a publication released today in Nature Biotechnology, the scientists demonstrated their ability to modify two distinct bacterial types to produce molecules that emit unique light wavelengths within the visible and infrared light ranges, detectable by hyperspectral cameras. These reporting molecules were associated with genetic circuits that recognize proximate bacteria; however, this method could also be integrated with any current sensors, such as those for arsenic or other pollutants, according to the researchers.
“An appealing aspect of this technology is its versatility — you can easily integrate whichever sensor you prefer,” remarks Yonatan Chemla, an MIT postdoctoral researcher and one of the lead authors of the study. “There’s no reason any sensor couldn’t work alongside this technology.”
Itai Levin PhD ’24 is also a leading author of the paper. Additional contributors include former undergraduate researchers Yueyang Fan ’23 and Anna Johnson ’22, as well as Connor Coley, an associate professor of chemical engineering at MIT.
Hyperspectral imaging
There are various methods to modify bacterial cells for the purpose of detecting specific chemicals. Most of these methods function by linking the detection of a substance to an output like green fluorescent protein (GFP). While these techniques are effective for laboratory experiments, such sensors cannot be measured from considerable distances.
For extended-range detection, the MIT team conceived a strategy to engineer cells to produce hyperspectral reporter molecules, effectively detectable by hyperspectral cameras. These cameras, which were first developed in the 1970s, can assess the quantity of each color wavelength within any specified pixel. Instead of appearing solely red or green, each pixel holds information about hundreds of distinct light wavelengths.
Currently, hyperspectral cameras find applications in areas such as radiation detection. In the vicinity of Chernobyl, these cameras have been employed to gauge subtle color alterations produced by radioactive metals within the chlorophyll of plants. Additionally, hyperspectral cameras are used to identify signs of malnutrition or the invasion of pathogens in vegetation.
This work inspired the MIT team to investigate whether they could modify bacterial cells to generate hyperspectral reporters upon detecting a target molecule.
For a hyperspectral reporter to maximize its utility, it should have a spectral signature characterized by peaks across various light wavelengths, facilitating easier detection. The researchers conducted quantum calculations to forecast the hyperspectral signatures of approximately 20,000 naturally occurring cell molecules, enabling them to pinpoint those exhibiting the most distinctive patterns of light emission. Another vital factor is the count of enzymes that would need to be incorporated into a cell for it to produce the reporter — a characteristic that will differ among various cell types.
“The ideal molecule is one that stands out significantly from all others, making it easily detectable, and necessitates the fewest number of enzymes for its production within the cell,” Voigt clarifies.
In this investigation, the team identified two molecules that were optimally suited for different bacterial strains. For a soil microorganism known as Pseudomonas putida, they utilized a reporter named biliverdin — a pigment generated from the degradation of heme. For an aquatic microorganism called Rubrivivax gelatinosus, they employed a variant of bacteriochlorophyll. For each bacterium, the researchers integrated the enzymes required to produce the reporter into the host cell and connected them to genetically modified sensor circuits.
“You could introduce one of these reporters to a bacterium or any cell possessing a genetically encoded sensor within its genome. Thus, it could react to metals, radiation, toxins in the soil, nutrients, or anything else you wish for it to respond to. The output would then be the synthesis of this molecule, which can be sensed from a distance,” Voigt notes.
Long-distance sensing
In this research, the team interconnected the hyperspectral reporters with circuits crafted for quorum sensing, which enable cells to recognize other adjacent bacteria. They have also demonstrated, in subsequent work, that these reporting molecules can be associated with sensors for various chemicals, including arsenic.
During the trials of their sensors, the researchers housed them within boxes to maintain containment. These boxes were situated in fields, deserts, or atop buildings, with the cells emitting signals detectable by hyperspectral cameras mounted on drones. The cameras require around 20 to 30 seconds to scan the field of view, followed by computer algorithms analyzing the signals to determine the presence of the hyperspectral reporters.
In this study, the researchers documented imaging from a maximum range of 90 meters but are currently working on extending these distances.
They foresee the potential for these sensors to be implemented for agricultural tasks, such as assessing nitrogen or nutrient levels in soil. For these applications, the sensors could also be tailored to function within plant cells. Detecting landmines is another prospective use for this type of sensing.
Prior to deployment, the sensors would require regulatory endorsement from the U.S. Environmental Protection Agency, along with the U.S. Department of Agriculture if utilized in agriculture. Voigt and Chemla have been collaborating with both agencies, the scientific community, and other relevant stakeholders to clarify what questions need addressing before these technologies can receive approval.
“We’ve been quite active over the past three years aiming to comprehend the regulatory landscapes and assess safety concerns, risks, and benefits associated with this kind of technology,” Chemla states.
The investigation was financed by the U.S. Department of Defense; the Army Research Office, a division of the U.S. Army Combat Capabilities Development Command Army Research Laboratory (the funding supported the engineering of environmental strains and the optimization of genetically-encoded sensors and hyperspectral reporter biosynthetic pathways); and Israel’s Ministry of Defense.