A protein chain folds into a working tool in a fraction of a second, and almost never makes a mistake. For a long time, it was thought that evolution specifically polishes amino acid sequences so that a protein not only works but also easily finds its unique shape. New work turns this idea on its head.
Researchers built a computer model: a flat lattice, four types of amino acids, and the "function" was defined as the ability to form a specific local pattern, an active site, at an environmental temperature T. Selection was strictly based on this ability, with no mention of whether the protein should fold at all. Sequences were sampled using multicanonical Monte Carlo, covering the entire fitness range, and for each sequence they constructed energy and free-energy landscapes.
The result was surprising. At a "warm" environmental temperature, highly fit sequences spontaneously acquired a funnel-shaped energy landscape — the very one that guides the chain to a single native fold, like water flowing to the center of a funnel. Moreover, the number of allowed final forms for such proteins was extremely limited. But in the cold, the picture broke down: the energy landscape became rugged and glass-like, and structural diversity sharply increased. The protein seemed frozen in indecision, unable to choose a single form.
The roundabout brings us back to the beginning. A protein chain folds without error not because evolution commanded it to "be stable." It folds because it must work under thermal jiggling. The local functional requirement, maintaining the shape of the active site, subtly organizes the entire global architecture. Foldability turns out to be not a goal but a quiet thermodynamic echo of the main task.
as written by the authors arXiv:2608.01143v1
Proteins perform biological functions by folding into specific native structures that are stabilized by funnel-like energy landscapes shaped through evolution. We investigate whether such foldable proteins can emerge solely from selection for function, without any direct selection for foldability. Using a two-dimensional lattice protein model with four amino-acid types, we define protein function as the equilibrium probability that a prescribed local structure forming an active site is realized at environmental temperature $$T$$, and use this probability as the evolutionary fitness. Amino-acid sequences spanning the entire fitness range are sampled by multicanonical Monte Carlo method and their energy and free-energy landscapes are constructed from the equilibrium conformational ensemble. We find that only within an appropriate range of environmental temperatures do high-fitness sequences spontaneously acquire funnel-like energy landscapes, despite foldability never being included in the fitness definition. High-fitness sequences are furthermore supported by only a limited number of native conformations. At low temperatures, in contrast, high-fitness sequences exhibit rugged, glass-like energy landscapes and much greater structural diversity. These results suggest that foldable proteins need not be direct targets of evolution, but can emerge as a thermodynamic consequence of maintaining function under thermal fluctuations, whereby local functional constraints organize the global fold and give rise to funnel-like energy landscapes.
The energy funnel of a protein resembles a landscape where a ball rolls into a hole by itself: give the chain a slight push, and it rolls down to a single native form.