The protein is produced by expression from bacteria which are grown on minimal medium supplemented with small amounts of 15NH4Cl and 13C-labelled glucose as well as labelled and unlabelled amino acids. The idea is that only those amino acids which are added in labelled form become labelled in the protein. Unfortunately, this may not always work as desired, since the E.Coli metabolism and catabolism causes a degree of interconversion between amino acids. Thus, it is not possible to create a sample with any combination of labelled amino acids. The situation can be improved somewhat by using auxotrophic bacterial strains or incorporating enzyme inhibitors. However, if complete control over the incorporation of amino acids is required, then cell-free methods must be used.
A cheaper way of labelling only certain amino acids, often called reverse labelling, involves expression from bacteria which are grown on minimal medium supplemented with 15NH4Cl and 13C-labelled glucose as well as unlabelled amino acids. This supresses the labelling of these amino acids and only those which have not been added unlabelled will be synthesised by the bacteria using the 13C-glucose as the carbon source. Again, a certain amount of scrambling may occur.
Amino acid specific labelling is usually used in order to reduce the spectral overlap and be able to monitor certain amino acids without interference from other signals. In some cases is may also be used to help with assignments processes.
L.P. McIntosh and F.W. Dahlquist (1990) Quart. Rev. Biophys. 23 1-38. (Link to Article)