(309c) Improvement of Amino Ester Hydrolase (AEH) Thermostability through Rational Design | AIChE

(309c) Improvement of Amino Ester Hydrolase (AEH) Thermostability through Rational Design


Lagerman, C. - Presenter, Georgia Institute of Technology
Joe, E., Georgia Institute of Technology
Grover, M., Georgia Tech
Rousseau, R., Georgia Institute of Technology
Bommarius, A., Georgia Institute of Technology
Amino ester hydrolase (AEH) catalyzes the synthesis of semisynthetic beta-lactam antibiotics and is a potential alternative to Pen G acylase (PGA). While AEH is more active and selective than PGA towards the synthesis of targets with (R)-phenylglycyl side chains, its substrate specificity is limited, its biophysical behavior is complex, and it deactivates rapidly at temperatures > 25oC.

We set out to improve the utility of AEH by improving its overall stability. While we had succeeded in stabilizing wild-type AEH from Xanthomonas campestris via site-directed mutagenesis [1], we now employ machine-learning based techniques, such as FireProt [2] and PROSS [3], to improve its thermal stability. In addition, we employed differential scanning fluorimetry (DSF) to study unfolding, back reflection to study AEH aggregation, and analytical ultracentrifugation (AUC) to study AEH’s oligomericity behavior and improve its deactivation kinetics.

The effects of several single mutations were studied. Mutating increasingly larger numbers of residues, up to 30% of the residues, led to dramatic stabilization, as measured by DSF via the melting temperature Tm (Figure 1). However, most variants were found to be inactive. Only the (re) discovery of a Ca2+-binding site in X. campestris AEH reliably recovered activity, ranging from small to sizable. Supplementing Ca2+ to the protein solution also increased the stability of AEH, which suggests inadequate Ca2+ is supplied during expression for stable AEH folding.

The development of AEH towards a biocatalyst useful in large-scale synthesis serves as an example of the need to explore various techniques and pathways and not just to rely on a single development path, whether via machine learning or experimental protein engineering.


[1] JK Blum, WD Ricketts, AS Bommarius, J. Biotechnol. 2012, 160, 214-221

[2] D Bednar, K Beerens, E Sebestova, J Bendl, S Khare, R Chaloupkova, Z Prokop, J Brezovsky, D Baker, J Damborsky, PLOS Comput. Biol. 2015, 11(11), e1004556

[3] A Goldenzweig, M Goldsmith, SE Hill, O Gertman, P Laurino, Y Ashani, O Dym, T Unger, S Albeck, J Prilusky, RL Lieberman, A Aharoni, I Silman, JL Sussman, DS Tawfik, SJ Fleishman, Mol. Cell. 2016, 63(2), 337-346