Proteins, chains of amino acids, perform vital cellular functions, from catalyzing reactions to regulating the cell cycle. To function, many require a specific three-dimensional structure, whose dynamics are shaped by natural selection. This leaves traces in protein sequences, such as the frequency of certain amino acids in specific positions or combinations of amino acids that appear together with unusual frequency.
These evolutionary correlations are crucial for studying protein structure and have been fundamental for tools like AlphaFold. However, the underlying evolutionary processes still generate debate. The new work proposes a model where natural selection prioritizes protein stability, predicting amino acid combinations similar to those found in natural proteins.
The model considers two challenges: stability to prevent unfolding and the prevention of incorrect alternative structures, a process known as 'negative design against misfolding'. Some amino acid combinations are favored because they hinder the adoption of erroneous structures, leaving a detectable trace in the sequences.
The study, which analyzed 37,108 protein chains from the Protein Data Bank, confirms that the model's predictions reproduce a good part of the observed correlations, especially in positions that form contacts in the three-dimensional structure. Interestingly, the most intense correlations occur in surface positions, suggesting the influence of functional factors and molecular interactions.
Furthermore, global correlations are identified that do not directly depend on the structure, reflecting changes in selective pressures or mutation rates. Amino acids like cysteine, key in disulfide bonds, show strong correlations. Groupings related to metabolic cost, electrical charge, or genetic code characteristics are also observed.
Overall, the results indicate that protein correlations arise from structural stability, avoidance of misfolding, and other evolutionary pressures. The study provides a theoretical framework for understanding the information contained in protein sequences and the evolutionary strategies to maintain their functionality.
