Protein folding

Proteins are made as chains of amino acids joined by peptide bonds. Interactions among the amino-acid side chains and with the surroundings cause the chain to adopt a particular three-dimensional conformation, often described as its native state. That structure helps determine what the protein can do, such as catalyse a reaction, bind another molecule, or provide structural support. Many proteins contain several structural levels and some consist of multiple chains assembled together.

Folding is not usually a single, rigid sequence of steps. A protein can pass through partially folded intermediates and may reach its functional structure along more than one route. Interactions among residues, solvent, temperature, pH, and other conditions shape the process. In 1961, Christian Anfinsen's work on ribonuclease supported the principle that, for a protein under suitable conditions, its amino-acid sequence contains the information needed to determine its native conformation. This does not mean every protein folds unaided inside a cell: the crowded cellular environment and other factors can make folding difficult.

Molecular chaperones are helper proteins that interact with newly made or non-native proteins. Depending on the system, they can shield exposed regions, limit inappropriate aggregation, assist folding or refolding, and help direct proteins that cannot be rescued toward degradation. Chaperones generally do not dictate a final structure as a permanent template; rather, they help proteins reach or maintain appropriate states. Cells therefore balance protein production, folding, repair, and removal as part of protein quality control, also called proteostasis.

A misfolded protein may lose its usual activity or form aggregates that disrupt cellular processes. Misfolding and aggregation are associated with several diseases, though the mechanisms and contributions differ and are not simply explained by one common pathway. Researchers investigate folding with biochemical and structural experiments, spectroscopy, microscopy, and computational models. Artificial-intelligence systems such as AlphaFold have greatly advanced prediction of many protein structures from sequence, but a predicted structure is not a complete account of a protein's movements, interactions, cellular context, or folding pathway.