Beneficial soil microbes could replace some of the fertilizers and pesticides that are often over-applied. But microbes are motile: they can spread beyond the target crop, where their benefits are diluted or could even cause unintended ecological change. A research team at Stanford and the University of Delaware set out to solve this issue of non-specificity by with a more directed approach by engineering bacteria to “switch on” only near a specific, engineered plant1. We explore this exciting story here.

Building a Private Channel Between a Plant and a Bacterium
The targeted approach this research team wanted to test has three parts: a signal molecule, a plant that sends the signal, and a group of soil bacteria that receive it. The signal is a single molecule, O-methyl-L-tyrosine (OMY). OMY is a noncanonical amino acid, meaning it falls outside the 20 standard amino acids cells use to build proteins. Notably, OMY has not been detected in agricultural soils, which makes it a good candidate for a signal that soil microbes won’t mistake for something else.
The scientists then needed a way to send the signal out through the plant’s roots. To do this, they chose Arabidopsis thaliana, a small, fast-growing flowering plant in the mustard family that researchers widely use as a model in plant biology. They added the gene for a single enzyme, MfnG, to the plant’s genome. MfnG converts the common amino acid tyrosine into OMY. They found that Arabidopsis seedlings grown in a nutrient solution released about 3.1 µM OMY into the liquid around their roots. The plant was ready to deliver the signal. Finally, the researchers needed a species to receive the signal and close the loop. Bacillus subtilis, a common model soil bacterium, was a natural choice because it is well characterized and has established tools for genetic engineering2.
Finally, soil bacteria don’t typically respond to OMY, so the team had to engineer them to. They added genetic components that let B. subtilis incorporate OMY into a protein at an “amber” stop codon, a three-letter genetic signal that normally tells the cell to end a protein at that spot. The team placed this codon in a target gene. Without OMY, protein production halts at the codon. With OMY, the cell reads through it and makes the full-length protein.
Switching the Reporter: From Fluorescence to NanoLuc® Luciferase
With a sender and a receiver in place, the team needed a way to see the conversation. The target gene could be any gene, so they chose a reporter gene, one whose protein produces a measurable signal, so the bacteria would signal only when OMY was present. The question was which reporter would be sensitive enough to detect the small amounts of OMY a plant releases?
The team first used the fluorescent protein mNeonGreen (commercially avaialble, but see this paper for the original research3) as the reporter. mNeonGreen responded to OMY in a dose-dependent way, with 2-fold induction at 1 µM and 20-fold at 1 mM, but the cells also glowed substantially without OMY because translation did not always stop at the single stop codon. To reduce background and widen the dynamic range, they switched to NanoLuc® Luciferase and tested reporters with one or two amber stop codons. Luminescence in the presence of OMY was similar across the variants, but the second stop codon cut background about 55-fold, giving a 3,100-fold change between bacterial cultures with and without OMY. That result reflects NanoLuc® Luciferase and the stop-codon design together, not the enzyme alone.
Next, they tested whether plant-made OMY could switch on the reporter. Arabidopsis seedlings were placed on an agar matrix containing the bacteria and imaged after four days. Luminescence appeared at the roots of OMY-producing plants, about 3.2-fold above the surrounding baseline, compared with 1.4-fold for wild-type plants. When engineered and wild-type seedlings grew side by side, the light concentrated around the engineered ones. A control reporter that doesn’t depend on OMY glowed across the entire plate regardless of plant genotype, so the pattern reflects the OMY signal rather than a plate artifact.
The approach also worked in tomato and poplar roots and in a wild root-colonizing strain of B. subtilis, which needed a different number of stop codons than the lab strain. The same lesson applies to anyone designing a reporter: a sensitive, low-background readout determines what you can detect, and reporter design is worth testing in each system. This is proof-of-concept work, and testing in non-sterile soil with diverse native microbes remains future work. Together, these results are exciting for researchers looking to uncover underground traits in soil communities.
Read the full paper in Nature Biotechnology.
Learn more about NanoLuc® Luciferase here.
- Zhong, V., Jones, M. A., Cabales, A., Gevorgyan, A., Inckemann, R., Johnson, A. A., Karunadasa, S. S., Forti, A. M., Xu, S.-L., Kunjapur, A. M., & Brophy, J. A. N. (2026). Genetic code expansion enables plant-directed control of bacterial activity. Nature Biotechnology. https://doi.org/10.1038/s41587-026-03360-5 ↩︎
- Kang, S.-Y., Pokhrel, A., Bratsch, S., Benson, J. J., Seo, S.-O., Quin, M. B., Aksan, A., & Schmidt-Dannert, C. (2021). Engineering Bacillus subtilis for the formation of a durable living biocomposite material. Nature Communications, 12, Article 7133. https://doi.org/10.1038/s41467-021-27467-2 ↩︎
- Shaner, N. C., Lambert, G. G., Chammas, A., Ni, Y., Cranfill, P. J., Baird, M. A., Sell, B. R., Allen, J. R., Day, R. N., Israelsson, M., Davidson, M. W., & Wang, J. (2013). A bright monomeric green fluorescent protein derived from Branchiostoma lanceolatum. Nature Methods, 10(5), 407–409. https://doi.org/10.1038/nmeth.2413 ↩︎