Production of disulphide-bonded domains suitable for NMR structure determination : application to the SRCR domains of the lymphocyte receptor CD5
This work describes the development of a systematic methodology to overcome two of the difficulties commonly encountered when expressing eukaryotic domains in bacterial hosts, namely the failure to obtain folded protein in vivo and the low solubility of the expression product. This methodology made possible the production of samples of the first scavenger receptor cysteine rich (SRCR) domain of human CD5, /zCD5dl, with properties suitable for multidimensional NMR studies. The SRCR domains of /zCD5 express in bacteria as insoluble aggregates. The aggregates were purified in order to perform the folding in vitro. Optimal conditions for folding were found using a novel systematic screen based on a fractional factorial design. In vitro folding yields were assessed using RP-HPLC and non-denaturing PAGE. The attainment of /zCD5dl protein samples of sufficiently high concentration to perform multidimensional NMR was achieved by performing rational apolar-to-polar mutations selected by analysis of a multiple sequence alignment. Eight single residue mutants were engineered and expressed. Four of them had better in vitro folding yields than the wild type and a double mutant was constructed by combination of the best behaved single mutations. This double mutant was used to determine the structure of the domain. NMR experiments at 298 K showed that some regions of /zCD5dl undergo conformational exchange on a microsecond to millisecond timescale, hampering the assignment of the resonance signals. Increasing the temperature to 318 K was found to greatly enhance the quality of the NMR spectra and enabled the assignment of more than 95% of the resonances. The solution structure of /*CD5dl was determined using standard interproton distance and dihedral angle-restrained molecular dynamics protocols. Forty percent of the residues were found to be in structurally well-defined regions, including all of the regular secondary structure features found for other members of SRCR superfamily. The remaining residues of the polypeptide appear to be distinctly less well ordered.