Transcription factor–based biosensors provide a powerful and modular platform for detecting molecular signals in living cells. Their architecture—linking ligand sensing to transcriptional control through distinct ligand-binding and DNA-binding domains—makes them ideal programmable measurement tools, yet their engineering has been limited by the small set of transcription factors with well-characterized and compatible domains. Within our biosensor research group, we are developing a unified pipeline to enable systematic domain swapping across transcription factor families. Curated databases of annotated transcription factor sequences are embedded using protein language models to derive design rules governing allosteric communication. Unsupervised generative models then propose new-to-nature hinge domains, which are subsequently screened to identify variants capable of functionally coupling ligand-binding and DNA-binding domains within but also across families. These synthetic hinges are integrated into hybrid transcription factors using scalable biofoundry assembly and high-throughput screening, rapidly identifying functional chimeras. This workflow expands the accessible biosensor design space, enabling programmable, custom transcription factor biosensors for applications in biotechnology, metabolic engineering, environmental monitoring, and autonomous biological control.
Paepe et al. (Thu,) studied this question.