Deoxyribozymes, tools for precision sensing and machine learning-guided discoveries

Deoxyribozymes represent a new frontier in biotechnology. They show that DNA is much more than a storage molecule for genetic information: it can also function as a catalytic machine. Unlike traditional enzymes made of proteins, such DNA-based tools are stable, easy to produce in large quantities, and can be evolved in a lab to perform different chemical tasks we need. Their potential is vast, ranging from low-cost diagnostic kits to high-speed screening tools that can facilitate drug discovery. As our ability to understand and predict their behavior continues to improve, these versatile molecules are becoming a cornerstone for future personalized medicine and environmental monitoring.

IOCB Prague: Deoxyribozymes, tools for precision sensing and machine learning-guided discoveries: Researchers described a new assay for DNA-editing enzymes in which the substrate in the reaction is a chemiluminescent deoxyribozyme called Supernova. ACS Chem. Biol. 2026, 21, 5, 941-950

IOCB Prague: Deoxyribozymes, tools for precision sensing and machine learning-guided discoveries: Researchers also systematically investigated the specificities of self-phosphorylating deoxyribozymes that convert the coumarin substrate 4-MUP into a fluorescent product using biochemical assays, single-step selections, and machine learning. ACS Chem. Biol. 2026, 21, 5, 951-959

In another advance, researchers in the Curtis Group used a fluorescent deoxyribozyme named Aurora to explore how these molecular tools can be taught to recognize specific chemical targets. Using a combination of artificial evolution and machine learning, the team successfully identified specific mutations that change Aurora’s preference for one substrate over another. Machine learning is particularly powerful here because it allows researchers to predict how millions of different DNA variants will behave without having to test each one individually in a lab. This approach helps us understand the fundamental rules of how DNA molecules interact with the world, which is essential for designing more precise medicines and diagnostic tools.

These results are described in two recent publications:

Chemiluminescent Deoxyribozyme Sensors for DNA-Editing Enzymes

Jakubec, M.; Svoboda, M.; Kurfürst, J.; Svehlova, K.; Veverka, V.; Curtis, E. A.  

ACS Chem. Biol. 2026

http://doi.org/10.1021/acschembio.5c00927

licensed under CC-BY 4.0

Abstract

DNA-editing enzymes such as those in the APOBEC family of cytidine deaminases play important roles in both normal and pathogenic function, while engineered enzymes offer exciting new possibilities for genome editing. Despite their importance, widely used assays for DNA-editing enzymes are time-consuming and expensive. Here, we describe a new assay for DNA-editing enzymes in which the substrate in the reaction is a chemiluminescent deoxyribozyme called Supernova. Editing alters the sequence of Supernova, which results in a change in catalytic activity and light production. By analyzing a data set of Supernova variants previously identified by selection and high-throughput sequencing, it was possible to generate sensors with a wide range of specificities. Sensors were also developed for APOBEC3A, a cytidine deaminase which converts C to U in single-stranded DNA and RNA. These include a turn-off sensor that produces light 14-fold slower after incubation with recombinant APOBEC3A than in its absence, and a turn-on sensor that generates light 10-fold faster after incubation with APOBEC3A than in its absence. Assays that use these sensors are faster and less expensive than existing ones, and should be particularly useful for applications such as high-throughput screening.

Probing the Specificity of Fluorescent Deoxyribozymes Using Single-Step Selections and Machine Learning

Král’ová, Z.; Isler, L.; Volek, M.; Jandová, M.; Kurfürst, J.; Curtis, E. A. 

ACS Chem. Biol. 2026 

http://doi.org/10.1021/acschembio.5c00969

licensed under CC-BY 4.0

Abstract

The ability of proteins and nucleic acids to form specific binding sites for ligands is critical for biological function, and methods to modulate biochemical specificity are important for fields such as enzyme engineering and drug design. Here, we systematically investigated the specificities of self-phosphorylating deoxyribozymes that convert the coumarin substrate 4-MUP into a fluorescent product using biochemical assays, single-step selections, and machine learning. Activity assays using a panel of 20 catalytic motifs and 10 substrates that generate different types of signals when they are dephosphorylated revealed that these deoxyribozymes are extremely specific for 4-MUP. To identify mutations that change specificity, we constructed a library based on a self-phosphorylating fluorescent deoxyribozyme called Aurora. A series of single-step selections yielded variants that react with 4-MUP and the structurally similar substrate diFMUP, but not with the more distinct substrates pNPP and ELF. Pairwise analysis of sequences in the 4-MUP and diFMUP data sets revealed four mutations that modulate Aurora specificity. The effects of these mutations were confirmed using biochemical assays and could be predicted using models developed by machine learning. Taken together, our results show how single-step selections can be used to identify mutations that change the specificity of a deoxyribozyme. They also highlight how machine learning can be used to model complex data sets from in vitro selection experiments.