The redesign of antifreeze proteins: an improved procedure

Everyone has at least once experienced on a hot summer day the melting of their ice cream. To slow down this process, Unilever started introducing antifreeze proteins in their ice cream about ten years ago. Originally, they were derived from ocean pout, but for this industrial process genetically manipulated brewer’s yeast is used to produce higher quantities of these proteins. This animal-friendly method allows Unilever to produce ice cream with a reduced fat content, without a change in the taste, that will retain their solid structure longer. Additionally, with the reduced fat content, it is possible to have a higher fruit content, making the new ice creams even healthier.

Upon further research it was observed that antifreeze proteins, derived from different organisms such as rye-grass and extremophile bacteria, contain a linear structure consisting of repeating units that coordinates on ice crystal lattices. The presence of these repeating units is interesting for the biotechnology for several reasons, but mainly because it gives the possibility to modify these proteins in such a way that identical repeating units are obtained. Once this is done, repeating units can easily be added or deleted, resulting in a set of proteins that only vary in the amount of repeating units. Such a set of proteins has specific purposes in, i.e., the coordination of metal arrays, as the length can be easily modified. For this reason, the conducted lab work focused on these proteins.

During the first part of this research a procedure was developed that uses the sequences of antifreeze proteins as an input to computationally generate the desired proteins. The procedure helps to identify the different repeats, which will be aligned and compared. Via computational methods it was then possible to generate putative ancestral sequences for these repeats, as they probably all share the same evolutionary origin, but diversified due to mutations over the course of time. The different ancestral sequences were assessed on their properties and the most promising repeat sequence was chosen. As such, proteins can be designed with a varying number of identical repeats in its core.

However, no proteins were obtained after multiple attempts to bring the proteins to expression. After investigating the possible reasons as to why no proteins were obtained, alterations were made to the design algorithms and procedure. These changes resulted in an improved procedure that would now combine two repeats and use these to generate multiple alternating repeats instead of multiple identical repeats. This increase in diversity may have a stabilising effect on the protein. After comparing the newly obtained sequences with the initial sequences, it was clear that the new sequences were improved as they clearly had a better design score. However, due to a lack of time these improvements could not be tested experimentally, but only computationally.

In the future, these improved designs will be validated experimentally and, if a set of proteins is obtained, it is possible to assess these proteins based on their antifreeze activity as well. If they still contain their antifreeze activity, they can be compared to the original protein and its activity. If the new proteins exceed this activity they may even find new purposes besides the coordination of metal arrays.