Generation of amine dehydrogenases with increased catalytic performance and substrate scope from ε-deaminating L-Lysine dehydrogenase

Abstract

Amine dehydrogenases (AmDHs) catalyse the conversion of ketones into enantiomerically pure amines at the sole expense of ammonia and hydride source. Guided by structural information from computational models, we create AmDHs that can convert pharmaceutically relevant aromatic ketones with conversions up to quantitative and perfect chemical and optical purities. These AmDHs are created from an unconventional enzyme scaffold that apparently does not operate any asymmetric transformation in its natural reaction. Additionally, the best variant (LE-AmDH-v1) displays a unique substrate-dependent switch of enantioselectivity, affording S- or R-configured amine products with up to >99.9% enantiomeric excess. These findings are explained by in silico studies. LE-AmDH-v1 is highly thermostable (T_m of 69 °C), retains almost entirely its catalytic activity upon incubation up to 50 °C for several days, and operates preferentially at 50 °C and pH 9.0. This study also demonstrates that product inhibition can be a critical factor in AmDH-catalysed reductive amination.

Fig. 1

Model of the active site of the ε-deaminating L-lysine dehydrogenase from G. stearothermophilus. a Schematic representation of the reactions undergone by the natural substrate 1b. b Model of the active site containing its natural substrate 1c. Residues H181, Y238 and T240 (in cyan) appear to stabilise the alpha NH₃⁺ group, whereas R242 appears to interact with the COO⁻ of the substrate. Residues V172, F173 and V130 create a hindered hydrophobic environment. All highlighted residues are amenable targets for protein engineering studies

Fig. 2

Initial screening with LE-AmDH variants. a General scheme of the biocatalytic reductive amination performed by LE-AmDH variants. A catalytic amount of NAD⁺ (1 mM) was applied. The reducing equivalents, as well as the nitrogen source originated from the buffer of the reaction: HCOONH₄/NH₃ (pH 9.0, 2 M). The substrate concentration was 10 mM. AmDH and Cb-FDH were used in final concentrations of 90 μM and 19 µM, respectively. Reaction volume: 0.5 mL; reaction time: 48 h; temperature: 30 °C; agitation on an orbital shaker at 170 rpm. b Group B: substrates used for screening of the LysEDH variants reported here. Group A: additional substrates tested with LE-AmDH-v1. c Heatmap of the screening outcome, in which Group B substrates were tested to reduce screening effort (results expressed in conversion). d List of mutations introduced in the LysEDH scaffold

Fig. 3

Optimisation studies employing LE-AmDH-v1. Reaction conditions: 0.5 mL final volume; buffer: HCOONH₄/NH₃ 2 M pH 7.0–9.5; 170 rpm on orbital shaker; [substrate] = 10–100 mM; [NAD⁺] = 1 mM; [LE-AmDH-v1] = 90 μM; [Cb-FDH] = 19 μM; reaction time: 1–48 h (a–b) and 48 h (c–i) a Influence of the temperature (20 to 60 °C) on the reductive amination of 8a in HCOONH₄/NH₃ 2 M pH 9.0 buffer. b Influence of the pH values using HCOONH₄/NH₃ buffer at 50 °C on the reductive amination of 8a. c–i Effect of the substrate concentration on conversions (left y-axis; columns) and productivities (right y-axis; dashed lines with symbols) at two different temperatures: 30 and 50 °C). Source data are provided as a source data file. Data are based on single measurements

Fig. 4

Product inhibition studies employing LE-AmDH-v1 and Ch1-AmDH. a Overall scheme of the kinetic experiments, in which 8a is used as substrate and (R)-8b as competitive inhibitor. b Inhibitory effect (IC₅₀¹⁵) of the amine product (R)-8b at 15 mM of substrate concentration 8a. c and d Primary double reciprocal plots for competitive inhibition according to Lineweaver-Burk analysis for LE-AmDH-v1 and Ch1-AmDH, respectively. e Secondary plots of the slopes obtained from the primary Lineweaver-Burk plots vs. inhibitor concentrations. The absolute value of the intercept with the x-axis provides the competitive inhibition constant (K_I).Source data are provided as a source data file. Data points of Fig. 4b are the average of n = 2 independent measurements, but the deviation is so low that the error bar is smaller than the dots; the only visible error bar is for the pink line at k_appca. 5. Figure 4c, d are based on single measurements. Figure 4e is a secondary plot derived from Fig. 4c, d

Fig. 5

Pro-chiral preferences of LE-AmDH-v1. a Total average hydride/pro-chiral carbon distance over a minimum of 6 MD simulations and average under the distance threshold (3 Å). The percentiles indicate how often the average distance was under the given threshold. For details, see Supplementary Methods. b Superposition of the natural substrate and 8a in the active site of the wild type enzyme. The natural intermediate 1c is depicted in magenta, whereas the iminium of acetophenone (8c) is depicted in yellow. The mutation point (F173A) is marked in red and the cofactor NADH is depicted in green. It can be observed that once the mutation F173A is introduced, the aromatic ring of 8c gets accommodated into the newly created hydrophobic cavity, which was previously occupied by the phenyl group of F173; that produces an inversion of the chiral preferences of this mutant towards aromatic substrates and towards substrates with (relatively) short hydrophobic chains. Source data are provided as a source data file. Error bars represent the standard deviation of n = 6 independent experiments