Amino acid-selective segmental isotope labeling of multidomain proteins for structural biology

Current solution NMR techniques enable structural investigations of proteins in molecular particles with sizes up to several hundred kDa. However, the large molecular weight of proteins in such systems results in increased numbers of NMR signals, and the resulting spectral overlap typically imposes limitations. For multidomain proteins, segmental isotope labeling of individual domains facilitates the spectral interpretation by reducing the number of signals, but for large domains with small signal dispersion, signal overlap can persist. To overcome limitations arising from spectral overlap, we present a strategy that combines cell-free expression and ligation of the expressed proteins to produce multidomain proteins with selective amino acid-type labeling in individual domains. The bottleneck of intrinsically low cell-free expression yields of precursor molecules was overcome by introducing new fusion constructs that allowed milligram production of ligation-competent domains labeled in one or multiple amino acid types. Ligation-competent unlabeled partner domains were produced in vivo, and subsequent domain ligation was achieved by using an on-column strategy. This approach is illustrated with two multidomain RNA-binding proteins, that is, the two C-terminal RNA-recognition motifs of the human polypyrimidine tract-binding protein, and two highly homologous helix-turn-helix domains of the human glutamyl-prolyl-tRNA synthetase.