Multiple-site labeling of proteins with unnatural amino acids.

The advent of efficient systems using genetic encoding for the site-specific incorporation of unnatural amino acids (UAA) into proteins has opened countless new possibilities for studying the structure, dynamics, and interactions of proteins. In particular, orthogonal pairs of amber suppressor tRNA (MjtRNA) and tyrosyl-tRNA synthetase (MjTyrRS) of Methanocaldococcus jannaschii have been evolved that specifically recognize amber stop codons for the incorporation of over 40 different UAAs. Although these systems can produce mutant protein with yields as high as for the wildtype, the protein yields are strongly context dependent. Thus, the presence of a single amber stop codon in the gene of a target protein can lead to unacceptably poor expression yields. Many efforts have been directed at biasing the competition between the Escherichia coli release factor RF1 that recognizes the amber stop codon and the suppressor tRNA in favor of the production of full-length protein. Thus, the incorporation of UAAs can be enhanced by omission of RF1 from a cell-free (CF) synthesis system reconstituted from the individually purified enzyme components. CF systems were also used to eliminate RF1 with anti-RF1 antibodies, or by deploying an RNA aptamer against RF1. Unfortunately, all these approaches are affordable only for small-scale sample preparations, and protein yields can be compromised by antibodies, while aptamers still yield truncated protein as the predominant product. Heat-shock inactivation of a thermosensitive mutant of RF1 in a low-yield E. coli CF system increased the incorporation efficiencies of UAAs to at most 75% (< 50% in most cases), while the fidelity of translation was compromised by prolonged heat treatment. Depletion of tagged RF1 from a cell-free S30 extract by affinity chromatography was reported, but the identity of the tag was not revealed and anti-RF1 antibodies were required for complete elimination. In a different approach, truncation of the ribosomal protein L11 was shown to weaken the binding of RF1, but the resulting protein yields with UAAs were only moderately enhanced. Initial efforts to produce RF1-deficient E. coli strains led to compromised strains that depended on unnatural amino acids for growth (making protein expression expensive) or strains that were not entirely independent of RF1. Recently, an enhanced version of release factor RF2 enabled the production of an RF1-free E. coli DH10b strain and the incorporation of UAAs at multiple sites in vivo, but the protein yields obtained with UAAs were significantly reduced compared with those obtained with the natural amino acid (tyrosine). In view of the cost of many of the most attractive UAAs and the difficulty to tailor the concentrations of the aminoacyl-tRNA synthetase (RS) and suppressor tRNA in in vivo expression systems to the requirements of different UAAs and incorporation sites, we developed a continuous exchange cell-free (CECF) system that allows facile, inexpensive, and complete removal of the release factor RF1 from an S30 extract derived from the widely used high-yielding and protease-deficient E. coli strain BL21 Star (DE3) (Invitrogen). The approach relies on replacing wild-type RF1 by a mutant with a C-terminal affinity tag consisting of three consecutive chitin-binding domains (RF1-CBD3) for selective removal by filtration through a chitin column after production of an S30 extract in the usual way. The chitin-binding-domain tag allows the removal of RF1 under conditions that maintain the full activity of the S30 extract and at the same time delivers dramatically improved incorporation yields of difficult UAAs at difficult positions, suppresses the production of truncated protein, and allows the incorporation of UAAs at multiple sites in the same protein. Conveniently, the modified strain is fully compatible with protein expression from pET vectors, which are the most frequently used vectors for protein production in structural biology. The yields of wild-type protein obtained with chitintreated (S30 ) and untreated (S30) extracts were indistinguishable from the yields obtained with S30 extracts prepared from the original BL21 Star (DE3) strain (Figure S3 in the Supporting Information). To assess and optimize the expression yields with UAAs without having to purify the proteins, we used the MjtRNA/MjTyrRS pair evolved for incorporation of the fluorescent UAA l-(7-hydroxy-coumarin-4-yl)ethylglycine (Hco). Optimization employed the West Nile virus NS2B-NS3 protease (WNVpro), a 27 kDa protein that is also an established drug target. Optimization of the concentration of aminoacyl-tRNA syn[*] Dr. K. V. Loscha, A. J. Herlt, Dr. R. Qi, Dr. T. Huber, Dr. K. Ozawa, Prof. G. Otting Research School of Chemistry, The Australian National University Canberra ACT 0200 (Australia) E-mail: gottfried.otting@anu.edu.au Homepage: http://rsc.anu.edu.au/~go/