55HIGHLIGHTS 2020

Robust w-Transaminases by Computational Stabilization of the Subunit Interface, Q. Meng (a), N. Capra (a), C.M. Palacio (a), E. Lanfranchi (a), M. Otzen (a), L.Z. van Schie (a), H.J. Rozeboom (a),

A.-M.W.H. Thunnissen (a), H.J. Wijma (a) and D.B. Janssen (a), ACS Catal. 10, 2915- 2928 (2020); http://doi.org/10.1021/ acscatal.9b05223. (a) Biotransformation and Biocatalysis,

Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen (The Netherlands)

[1] G. Kiss et al., Angew. Chem. Int. Ed. 52, 5700-5725 (2013). [2] H.J. Wijma et al., Methods Mol. Biol. 1685, 69-85 (2018). [3] D. Koszelewski et al., Trends Biotechnol. 28, 324-332 (2010). [4] T. Börner et al., ACS Catal. 7, 1259-1269 (2017).

PRINCIPAL PUBLICATION AND AUTHORS

REFERENCES

Fig. 41: The crystal structure of PjTA-R6. a) Surface representation of the PjTA-R6 dimer with different colours for the two subunits, and showing the mutated residues and PLP co-factors as spheres. b) Close-up view of the M128F mutation and its

interacting residues at the dimer interface. The phenyl rings of Phe128 and Phe113 are ideally positioned for a T- shaped π−π interaction. The mesh represents the 2Fo-Fc electron density map for the residue side chains, calculated at 1.85 Å

resolution and contoured at 1σ.

Unfortunately, these enzymes commonly suffer from poor operational stability due to dynamic instability of their quaternary structure, resulting in PLP cofactor leakage, local unfolding, protein precipitation and permanent loss of activity [4]. The FRESCO protocols were applied to stabilise an (S)-selective omega-transaminase from Pseudomonas jessenii (PjTA) (Figure 40). Starting from the crystal structure of wild-type PjTA, these produced a virtual library of 226 potentially stabilising PjTA variants containing point mutations distributed on the protein surface and at the subunit interface of the protein dimer. Subsequent experimental screening of the mutant library revealed a particularly high success rate for the interface mutations (19 hits among 34 mutants), confirming the importance of the subunit interface for stability, and emphasising the effectiveness of the FRESCO approach. After rationally combining the confirmed stabilising mutations, two robust variants with four and six mutations were obtained which, in comparison to wild- type PjTA, displayed an increase in apparent melting temperature (ΔTmapp) of 18°C and 23°C, respectively. These two variants (PjTA-R4 and PjTA-R6) were also five-fold more active at their optimum temperatures and more tolerant to

co-solvents and high concentrations of the amine donor isopropylamine, allowing the production of (S)-1-phenylethylamine (>99% enantiomeric excess) from acetophenone (100 mM) with 92% yield (24% yield with the wild-type PjTA). High- resolution crystal structures of PjTA-R4 and PjTA-R6 were determined with X-ray diffraction data recorded in-house and with an EIGER X 4M detector at beamline ID30A-3 (Figure 41). Examination of the crystal structures provided clear explanations for the stabilising effects of the designed mutations, resulting from improved hydrogen bonding, an additional salt-bridge interaction, increased hydrophobic interactions, introduction of π-stacking, reduced exposure of hydrophobic surface and redistribution of electrostatic surface charge. An unexpected stabilising effect was deduced for mutation I154V, which eliminates conformational strain that is present in the wild- type PjTA. The identification of this stabilisation mechanism, which is rarely reported, underlines the power of the folding energy calculations and molecular dynamic routines implemented in the FRESCO workflow. The results further show how computational redesign of subunit interfaces in multimeric enzymes may help to create stable biocatalysts for use in green chemistry.