Using codon optimization, chaperone co-expression, and rational mutagenesis for production and NMR assignments of human eIF2α

Producing a well behaved sample at high concentration is one of the main hurdles when starting a new project on an interesting protein. Especially when one attempts to overexpress a eukaryotic protein in bacteria, some difficulties are encountered, such as low expression level, low solubility, or even lack of folded structure. Overexpression in prokaryotic systems is highly desirable for cost-effective production of different isotope-labeled samples needed for NMR studies. Here we describe generally applicable methods for obtaining highly concentrated protein samples efficiently. This approach was developed as we tried to produce a NMR-suitable sample of the 35 kDa human translation initiation factor eIF2alpha, a protein that expresses poorly in E. coli and has very low solubility. First, an E. coli codon-optimized gene was synthesized on a thermal cycler, which increased the expression level by a factor of two. Second, we used co-expression of bacterial chaperone proteins, which largely increased the fraction of correctly folded protein found in the soluble phase. Third, we used rational mutagenesis guided by both the sequence alignment among homologues and the homology of one domain to a known fold for improving solubility and stability of the target protein by tenfold. Combining all these methods made it possible to produce from a one-liter preparation a 0.5 mM sample of human eIF2alpha that showed well-resolved NMR spectra and enabled nearly complete assignment of the protein. These methods may be generally useful for studies of other eukaryotic proteins that are otherwise difficult to express and exhibit poor solubility.