Preventing damage due to the formation of ice crystals, both during freezing but perhaps even more so during thawing, is relevant for all kinds of materials that need to function or remain intact at sub-zero temperatures. Natural ice-binding proteins are therefore a favourite source of inspiration for developing new forms of antifreeze compounds, that are, for example, suitable for the preservation of donor organs.  

Thanks to ice-binding proteins, numerous organisms can survive in conditions that would normally lead to freezing (and thus, death). Well-known examples are fish living in the arctic seas, where temperatures can drop below zero. The protective effect of ice-binding proteins relies in part on their ability - as the name suggests - to bind ice crystals and thus inhibit the (re)crystallisation process. This ensures that incipient crystals do not grow into large and sharp forms that destroy soft tissues. 

Wide variation  

However, these natural proteins are not suitable for practical purposes due to a variety of reasons (limited stability, costly to produce, not compatible with the specific application, etc.), but structural elements and functional sequences can serve as a basis for new molecules with the same functionality. However, structure-activity relationships are difficult to unravel for this group, because there are many different ice-binding proteins that exhibit a wide variety of structures. Some have distinct secondary structural elements such as alpha-helices or beta-solenoids (folded ‘ribbons’), while others are inherently disordered, or are heavily glycosylated. So far, simulations have shed some light on the molecular geometry of ice-binding proteins that allow them to bind to certain ice crystal surfaces.   

Using these insights as a starting point, Rob de Haas, PhD student in the Physical Chemistry and Soft Matter group of Renko de Vries (Wageningen University), opted for de novo design of a new ice-binding structure. For this, he took the simplest and best-studied protein, wfAFP, from the type-1 IBP family, as a template. This small protein of 37 residues consists of a single, almost straight alpha-helix with the sequence TXXXAXXXAXX as a repeating element, where X is often also alanine. 

Undertwist 

Interestingly, the helix in wild type wfAFP, which occurs in the American winter flounder, is slightly distorted; there is a tiny undertwist. As if the helix has been ‘screwed back’ a bit. De Haas decided to focus on that aspect in his new design. He started with a perfectly twisted alpha helix as part of a bundle with two more helices without ice-binding activity that serve as a scaffold to keep the structure of the ice-binding helix intact. He then designed four variants, each with slightly different undertwist in the functional helix.   

By producing these four virtual ‘proteins’ in bacteria, their activity was evaluated. This proved successful - in the presence of the helix bundles, the ice crystals remained significantly smaller than without. According to the researchers, the just-not-perfect rotation of the helix ensures that the threonines, which are crucial for ice-binding activity, are all aligned, which creates the perfect positioning to act on the ice crystals.   

The value of these results, according to the team, lies mainly in the new insights into the role of the secondary structure of ice-binding proteins. With this approach, they show that de novo design based on natural templates is a meaningful way to arrive at novel ice-binding molecules. Given the huge collections of nature ice-binding proteins, this offers a range of possibilities for developing new, potential anti-freeze, compounds.  

Robbert de Haas, et al., De novo designed ice-binding proteins from twist-constrained helices, PNAS (2023)