Open Access

The synthesis of valuable chemicals via traditional chemical methods can be often outperformed by the use of enzymes because of their excellent chemo-, regio- and stereoselectivity in aqueous solvents at ambient temperatures. On the other hand, enzymes often suffer from several limitations that hamper their industrial application. Protein engineering is commonly applied to overcome these limitations although the generation and the validation of mutants is often a laborious process that may not lead to the desired results within reasonable time frames. This thesis focuses on engineering the enantioselectivity and the substrate scope of industrially relevant enzymes, such as esterases and transaminases. Semi-rational protein engineering was employed to identify improved variants for the synthesis of valuable chemicals ensuring a reduced screening effort. Compared to previous works, 3DM’s applicability was extended to the study of correlated mutations and proved effective in the acceleration of the comprehension and in the mutation of these enzymatic scaffolds. Semi-rational approaches require an extensive amount of information such as protein structures, reaction mechanisms, previous mutational experiments reported in literature and a considerable amount of amino acid sequences from similar proteins to analyze amino acid distributions and correlated mutations. Here, we have exploited 3DM as a tool that can combine all this wealth of information: 3DM is a convenient solution to retrieve and integrate information simplifying decision making in the planning of a semi-rational mutant library since in 3DM’s multiple sequence alignments (MSA) is summarized Nature’s screening process for alternative variants. Furthermore, naturally evolving enzymes often require mutations at more than one position for the acquisition of a new property. Such mutations generate patterns that are recognized by the 3DM algorithm, which creates networks that can be investigated to design strategies that aim to improve the property of interest. Finally, these correlated mutations are connected to the mutations described in publications covered in the PubMed database, thus helping to investigate the role certain positions might play in the network. Article I shows that it is possible to improve the enantioselectivity of an esterase towards a highly symmetrical substrate while drastically reducing the screening effort. This was achieved through the creation of libraries that limit the variants to those identified in the 3DM alignment. Article II shows that networks of correlated mutations are composed of positions that may cluster around a function. These functions can be investigated because 3DM connects the positions in the network to their related publications. In this article, a mutant of the esterase PFE-I from Pseudomonas fluorescens was generated having increased enantioselectivity in the hydrolysis of important target compounds. Article III suggests that the in silico modelling software YASARA, combined with the use of the 3DM database, can further reduce the screening effort: it was possible to identify a hot-spot because both the 3DM database and YASARA docking studies, indicated its importance. This led to a further improved enantioselectivity of the enzyme variant identified in Article II. Article IV shows how MSA may be used to get structural insights into the catalytic properties of enzymes with documented activity. The study of the patterns observed in a large subfamily alignment allowed the definition of the structural determinants important for the substrate recognition in amine transaminases. Article V and VI apply the knowledge acquired for the improvement of the substrate scope in the amine transaminase from Vibrio fluvialis.

The focus of this thesis is the engineering and analysis of the enantioselectivity of esterases using 3-phenylbutyric acid (3-PBA) as model substrate. An ultra high throughput assay for identification of enantioselective esterases has been developed, based on the combination of in vivo selection and flow cytometry. The in vivo selection medium consists of a couple of pseudo-enantiomers of 3-PBA; one enantiomer is coupled to glycerol (GE), and hydrolysis of this substrate will enable cell survival. The other enantiomer is coupled to the toxin 2,3-dibromopropanol (BE), the hydrolysis of this substrate will cause cell death. Thus, cell survival is a function of the enantioselectivity of the enzyme expressed. The pseudo-enantiomeric substrates are structurally similar to allow selection for enantioselectivity instead of selection for enzyme substrate affinity. Next, esterase BS2 was chosen as negative control to establish the selection system since it hydrolyses both pseudo-enantiomers with low enantioselectivity (E~3 and 1, respectively). High enantioselective esterases towards 3-PBA: esterases PestE and CL1 (E > 100, both (R)-selective) were identified in a screening and used as positive controls. Further, the hyperthermophilic esterase PestE was crystallized. After elucidation of the enzyme structure, the high enantioselectivity of the enzyme towards 3-PBA could be explained by molecular modelling. The optimal concentration of the pseudo-enantiomeric substrates was set to be 5 mM for GE (higher concentrations were toxic) and 20 mM for BE (lower concentrations did not completely inhibit bacterial growth). The in vivo selection system was established together with the identification of a flow cytometric method to differentiate bacterial physiological status. The combination of Syto9 and PI was chosen as staining technique, because it allowed differentiation of the viable and the dead cell populations, and of these from the background. After viability detection by flow cytometry was established, esterases PestE and BS2 were cultivated in selection ((R)-GE and (S)-BE) and anti-selection medium ((S)-GE and (R)-BE). Clear differences in the culture viability depending on the enantioselectivity of the enzyme expressed appeared: cells expressing the (R)-enantioselective PestE could proliferate in selection medium, but could not proliferate in anti-selection medium. Cells expressing the non-selective BS2 did not grow in any media. Further, cultures containing mixtures of BS2/PestE or BS2/CL1 expressing cells were incubated in selection and anti-selection medium, and the viable clones were detected by flow cytometry analysis, sorted out and plated on agar. When the mixtures were incubated in selection medium, enrichment of the (R)-selective enzyme (PestE or CL1) over the non-selective enzyme (BS2) was observed. When the enzyme mixtures were incubated in anti-selection medium, very few colonies grew on agar, indicating that cell survival was a function of enzyme enantioselectivity. The successfully developed assay was used to identify variants with increased enantioselectivity in a mutant library of esterase PFEI (E ~ 3, (R)-selective) created by saturation mutagenesis. After library expression, 108 clones were in vivo selected and analyzed by flow cytometry. The viable cells were sorted out and plated on agar. The 28 resulting colonies were transferred to one microtiterplate and their activity and enantioselectivity (Eapp) was investigated using p-nitrophenyl derivatives. Four interesting mutants were identified: Table 1. Enantioselectivity of the in vivo selected mutants. Mutant Eapp[a]Etrue[b]Etrue[c]Etrue[d]Etrue[e] Mutations C4 80 4 4 3 1 V121I, F198G, V225A E7 >100 2 n.d. 3 n.d. V121S E8 2 25 16 50 >100 V121S, F198G, V225A F5 5 13 15 18 80 F121I, F198C [a] with separate (R)- or (S)-enantiomers of p-nitrophenyl-3-phenylbutanoate. [b] towards GE with cell lysate or [c] pure enzyme. [d] towards Et-3-PB with cell lysate or [e] pure enzyme. n.d. not determined. The mutants were purified and activity and enantioselectivity were determined in kinetic resolutions towards Et-3-PB and GE (Table 1). Mutants identified as highly enantioselective in the Eapp-assay (C4 and E7) were low selective in kinetic resolutions. On the contrary, mutants E8 and F5, which showed low enantioselectivity towards p-nitrophenyl-3-phenylbutanoate, hydrolyzed the 3-phenylbutyric esters with good to excellent enantioselectivities. This confirms that Eapp values can differ much from Etrue values as “you get what you screen for”, and supports that the here described method is very suitable for identification of enantioselective esterases. In this PhD thesis a novel strategy for identification of enantioselective esterases has been developed. This method allows a very high throughput (≥ 108 mutants/day) and opens the bottleneck of variant analysis, which exists in protein engineering technology.