How can we program cells to make targeted decisions in response to complex signals, like logic circuits in a computer? An interdisciplinary team from two research groups at the Center for Synthetic Biology at the Technical University of Darmstadt has developed a new approach: an RNA-based gene switch. The results were published in a magazine Nucleic acid research.
This switch is based on so-called riboswitches, short sections of messenger RNA (mRNA) that can respond to specific small molecules (‘ligands’). When the ligand binds to the “switch,” the RNA changes shape, blocking ribosomes that would normally make proteins according to instructions encoded in the mRNA.
Riboswitches are particularly attractive for synthetic biology because they function without additional proteins, are very small, and require little energy to produce within cells. This makes them ideal tools for synthetic gene regulation.
A research team at Darmstadt University of Technology has now combined two such riboswitches. The result is a switch that allows the simultaneous evaluation of two different molecular signals. Lead author Dr. Daniel Kelvin, a researcher at the Center for Synthetic Biology at the Technical University of Darmstadt, demonstrated that seamlessly linking two riboswitches allows the creation of gene switching elements with two different inputs.
Computer “functions” in living cells
“We use these RNA-based dual-input switches to implement computer-like logical functions in living cells,” Kelvin says. “To do this, we built a combination of two riboswitches that function like a Boolean NAND gate.”
NAND gates are fundamental components of digital electronics. Generates an “off” signal only when both inputs are active at the same time. Otherwise, the signal remains “on”.
Translated into biology, this means that gene expression (the production of the protein encoded by the gene) is switched off only when two different ligands bind to the riboswitch at the same time. If one of the two ligands is missing, the gene remains active. This behavior is complex and has never been observed in nature. Furthermore, the number of possible sequence variants increases exponentially with sequence length.
Lab experiments combined with AI
This made building a hybrid NAND riboswitch a major challenge. To identify suitable variants, the research team combined laboratory experiments with artificial intelligence techniques. The researchers first generated thousands of variants of the RNA switch. The researchers then tested how these variants responded to different combinations of ligands in the lab. The results served as training data for the computer program.
“Deep learning models predict which RNA variants perform the NAND function best. Our optimization algorithm, based on Bayesian optimization, selectively selects new candidates and learns from each experiment,” explains Eric Kubaczka, also a researcher at the Center for Synthetic Biology and co-author of this publication.
Using this approach, the team was able to identify several significantly improved RNA switches after testing just 82 variants. The best candidates had very distinct “on” and “off” states.
Biosensors for medical and environmental monitoring
With a new hybrid riboswitch and an AI-based design approach, a team led by Professor Beatrix Süß (Synthetic Biology Center, Synthetic RNA Biology Group) and Professor Heinz Kepl (Synthetic Biology Center, Self-Assembling Systems Group) at the University of Tokyo provides a way to design biological circuits in a more targeted way. Many other logical functions can be built from NAND gates, so living cells may learn to make more complex decisions in the future. For example, producing a substance only when a particular combination of nutrients or signaling molecules is present.
This may also enable the development of biosensors for medical and environmental monitoring. For example, sensors that detect specific metabolic states, identify signs of tumors, or report environmental toxins in specific combinations.
This project shows how biology and artificial intelligence are coming together, and how machine learning can help discover new functional RNA elements never produced by nature itself.
reference: Kelvin D, Kubaczka E, Karava M, Koeppl H, Suess B. Iterative design of NAND hybrid riboswitches using deep-batch Bayesian optimization. Nucleic acid research. 2026;54(5):gkag145. doi: 10.1093/nar/gkag145
This article has been reprinted from the following material: Note: Materials may be edited for length and content. Please contact the citation source for more information. You can access our press release publishing policy here.
