AI-designed proteins turned into molecular switches

C&I Issue 5, 2026

BY MARIA BURKE

An international team has used AI to create smart proteins that switch on when they detect a chosen target and can operate inside living bacterial cells. The research opens the way to a new generation of low-cost biosensors for medicine, environmental monitoring and biotechnology.

‘Until recently, protein engineers were mostly limited to adapting natural proteins found in biology,’ says lead author Kirill Alexandrov from the Queensland University of Technology, Australia. ‘That gave us only a small set of starting options and made it very difficult to design new sensors on demand. Our study shows that AI-designed proteins can be turned into effective molecular switches, greatly expanding what protein engineers can build.’

The researchers used machine learning-designed binding proteins as artificial receptors which responded to small molecules, peptides and proteins (Z. Guo et al, Nat Biotechnol (2026). DOI: 10.1038/s41587-026-03081-9). They then connected them to enzymes that produce easily measurable outputs such as colour changes, light emission and electrical signals, making the switches suitable for different types of sensing technologies.

The study showed that the switches could operate in living cells, an important step towards future synthetic biology applications. It also demonstrated the potential practical utility of this approach by developing bioelectronic devices capable of quantifying steroid hormones and tested them using Escherichia coli cells engineered to have steroid-dependent antibiotic resistance.

Alexandrov claims the work challenges a long-held idea in protein science.

‘It was widely believed that sensing proteins had to undergo large shape changes to function as switches. We found that these artificial receptors do not need a dramatic structural rearrangement. Instead, binding of the target molecule subtly changes how the protein moves, and that is enough to turn activity on. That gives us new insight into how natural protein regulation works and provides a powerful new strategy for designing useful biosensors.’