Enzyme stabilisation due to incorporation of a fluorinated non-natural amino acid at the protein surface

We have previously engineered E. coli transketolase (TK) enzyme variants that accept new substrates such as aliphatic or aromatic aldehydes, and also with improved thermal stability. Irreversible aggregation is the primary mechanism of deactivation for TK in the buffers used for biocatalysis, and so we were interested in determining the extent to which this remains true in more complex media, crude cell lysates or even in vivo. Such understanding would better guide future protein engineering efforts. NMR offers a potential approach to probe protein structure changes, aggregation, and diffusion, and19F-labelled amino acids are a useful NMR probe for complex systems with little or no background signal from the rest of the protein or their environment. Here we labelled E. coli TK with two different19F probes, trifluoromethyl-L-phenylalanine (tfm-Phe), and 4-fluoro phenylalanine (4 F-Phe), through site specific non-natural amino acid incorporation. We targeted them to residue K316, a highly solvent exposed site located at the furthest point from the enzyme active sites. Characterisation of the19F-labelled TK variants revealed surprising effects of these mutations on stability, and to some extent on activity. While variant TK-tfm-Phe led to a 7.5 degrees C increase in the thermal transition midpoint (Tm) for denaturation, the TK-4 F-Phe variant largely abolished the aggregation of the enzyme when incubated at 50 degrees C19. F-NMR revealed different behaviours in response to temperature increases for the two TK variants, displaying opposite temperature gradient chemical shifts, and diverging motion regimes, suggesting that the mutations affected differently both the local environment at this site, and its temperature-induced dynamics. A similar incubation of TK at 40-55 degrees C is also known to induce higher cofactor-binding affinities, leading to an apparent heat activation under low cofactor concentration conditions. We have hypothesised previously that a heat-inducible conformational change in TK leads to this effect1. H-NMR revealed a temperature-dependent re-structuring of methyl groups, also at 30-50 degrees C, which may be linked to the heat activation. While our kinetic studies were not expected to observe the heat activation event due to the high cofactor concentrations used, this was not the case for TK-4 F-Phe, which did appear to heat activate slightly at 45 degrees C. This implied that the mutations at K316 could influence cofactor-binding, despite their location at 47 & Aring; from either active site. Such long-distance effects of mutations are not unprecedented, and indeed we have previously shown how distant mutations can influence active-site loop stability and function in TK, mediated via dynamically coupled networks of residues. Molecular dynamics simulations of the two19F containing variants similarly revealed networks of residues that could couple the changes in dynamics at residue K316, through to changes in active site dynamics. These results independently highlight the sensitivity of active-site function to distant mutations coupled through correlated dynamic networks of residues. They also highlight the potential influence of surface-incorporated probes on protein stability and function, and the need to characterise them well prior to further studies.