When it comes to underwater adhesives, biological organisms offer a variety of role models. Mussels, mushrooms, bacteria and many other life forms produce sticky proteins that you can use to adhere to wet surfaces under extreme conditions. With the help of data mining and machine learning, researchers have now identified protein sequences that offer particularly good adhesive properties. From this, they have developed new super adhesives that are many times stronger than their natural models, also adhere immediately under water and can also be replaced without residue and can also be removed again.
Be it in surgery or in deep -sea research: adhesives that are reliably adhering to wet surfaces are required in many areas. But the development of such hydrogels is difficult, because unlike solid materials, their properties can hardly be predicted on the basis of their structure. Instead, researchers have so far have to rely on experiments and errors – a time -consuming and expensive process, at the end of which it can be found that none of the tested materials has the desired properties.
Natural models
A team around Hongguang Liao from the Hokkaido University in Japan has now found a way to develop functional hydrogels with the help of artificial intelligence – in this case powerful underwater adhesives. Natural role models served as the basis: The researchers searched a data set with the sequences of 24,707 adhesive proteins that naturally occur in organisms such as bacteria, mushrooms, mussels and snails and help them adhere to wet surfaces. “Despite their diversity, these proteins have common sequence patterns that provide valuable knowledge for the development of underwater adhesives,” explain the researchers.
These common sequence patterns, i.e. small sections within the proteins that are responsible for the adhesive effect, identified Liao and his team in a first step. With this knowledge, they synthesized 180 hydrogels, which contained exactly these adhesive sequences. In experimental tests, it was then shown that more than half of the hydrogels developed in this way were actually better adhered to than comparable previous adhesives. In the next step, the researchers trained artificial intelligence with the data of these 180 hydrogels. In a total of three round mechanical learning, they had the AI develop new, optimized underwater adhesives. The strongest candidates from each of these rounds synthesized them and subjected him to further experimental tests.
Strong, durable and multiple usable
And indeed: all three adhesives proposed by the AI showed a glue power many times stronger than all previously available hydrogels for underwater liability. With an adhesive strength of more than one megapascal, a stamp -sized piece of the newly developed hydrogels could hold an average adult. The hydrogels are also immediately liable on different surfaces, can be replaced again without residue and, in experiments, also retained their adhesion in experiments, as Liao and his colleagues report. Water of different salt content is also not a problem for the adhesives.
In order to demonstrate the durability, the researchers stuck a rubber duck on a wet stone in the sea burn: the adhesive resistance easily resisted the salty waves. In another experiment, the researchers locked a two centimeter leak in a water pipe from which water splashed with high pressure, with a barely handle -sized piece of hydrogel. The patch was immediately stuck and kept the hole tight after five months. In another test, the team mice implanted a piece of the hydrogel under the skin and demonstrated so that the material is also biocompatible.
(Video: Gong et al., Nature 2025)
“Super adhesive hydrogels such as this, which are strongly adhering to irregular and wet surfaces, could be of great importance for many biomedical applications, including prosthesis coatings and portable biosensors. Such hydrogels can also be useful in industrial and environmental contexts,” writes Laura Russo from the University of Milano-Bicocca, which was not involved in the study, in one Accompanying comment in the journal Nature. “In addition, Liao’s approach to the development of functional hydrogels is versatile and could be adapted for other types of functional soft materials.”
Source: Hongguang Liao (Hokkaido University, Japan) et al., Nature, DOI: 10.1038/S41586-025-09269-4
