Screening and characterization of a diverse panel of metagenomic imine reductases for biocatalytic reductive amination

Abstract

Finding faster and simpler ways to screen protein sequence space to enable the identification of new biocatalysts for asymmetric synthesis remains both a challenge and a rate-limiting step in enzyme discovery. Biocatalytic strategies for the synthesis of chiral amines are increasingly attractive and include enzymatic asymmetric reductive amination, which offers an efficient route to many of these high-value compounds. Here we report the discovery of over 300 new imine reductases and the production of a large (384 enzymes) and sequence-diverse panel of imine reductases available for screening. We also report the development of a facile high-throughput screen to interrogate their activity. Through this approach we identified imine reductase biocatalysts capable of accepting structurally demanding ketones and amines, which include the preparative synthesis of N-substituted β-amino ester derivatives via a dynamic kinetic resolution process, with excellent yields and stereochemical purities.

Fig. 1. A workflow and approximate time frame for generating a metagenomic library with examples of how this platform was used to expand the biocatalytic toolbox as applied to iReDs.

Various environmental niches were sampled (1) to obtain DNA from the environment without the need for pre-culturing of the organism. DNA was sequenced through a Mi-Seq Illumina platform. Sequence reads (contigs) were assembled, processed and deposited to the relevant database (2). Query sequences of known enzymes were used to identify putative sequences, which were then cloned and expressed heterologously in E. coli (3). Enzymes were arrayed into mictotitre plates for screening (4), which then enabled expansion of the biocatalytic scope across various avenues, which included synthetic scopes and operational properties (bottom). bp, base pairs.

Fig. 2. 384 iReDy-to-go screening of amine 2b combined with biotransformation data for the reductive amination of 2 with b mapped phylogenetically to show the overall distribution of activity.

a, A heatmap representative of IREDy-to-go screen of 2b in which conversion (%, determined by a GC–flame ionization detector) for the reductive amination of 2 and b are overlaid to each well for each IRED. IREDy-to-go data generated by measuring the end point read at λ = 490 nm is also shown behind the data generated by a GC–flame ionization detector as a heat map. A deeper red colour represents a higher absorbance. b, The distribution of activity and stereoselectivity mapped phylogenetically across all 384 IREDs for the reductive amination of 2 with b shown as bars with a ■red–white gradient in which darker red represents a higher conversion up to >99%. The multi-value bar chart represents the % conversion (as the total height), with the percentage of each enantiomer for the reduction of 4 to 3 with grey representing % (S)-3 ▶ and light purple denoting % (R)-3 ▶. The expression is represented by the outer blue gradient, in which low expression is shown in light blue ● and good expression in dark blue ●. The parental organism of each IRED (if known) is highlighted through tip labels, clades and branches (centre): plantae IREDs ■, fungal IREDs ■, human IREDs ■, cyanobacteria IREDs ■ and bacterial IREDs ■ and ■.

Fig. 3. High-throughput characterization employing the colorimetric screen, a chart showing the substrates presented to the colorimetric screen alongside the number of definitive enzyme hits; the method for the number of hits calculated is given in Supplementary Section 4.5.

The IRED-mediated oxidation of the target amine is shown (top) in which the hydride is delivered to the cofactor for regeneration, which leads to subsequent reduction of 5 to 6 mediated by the diaphorase to generate a red colorimetric change. The number of enzyme hits is shown above the structure of the corresponding compound. The red colour scale in each tile is representative of the number of hits generated through the screen with the given substrate. A scale for the number of hits is also shown below the chart. A variety of amines was screened, which included five-, six- and seven-membered heterocycles (1, 3, 7, 9, 19, 20 and 21), linear, allylic and benzylic amines (2a, 2b, 2c, 2f, 10d, 11e, 12f, 13a, 13b, 13f, 13j, 13k, 15a, 22a, 23d, 24k and 25b), N-substituted β-amino ester derivatives (26b, 27b, 28b and 29b) and several APIs (14d, 16g and 18i).

Fig. 4. Analytical scale reductive aminations.

Reductive aminations of 2, 12, 13, 15, 27, 30, 31, 32, 33 and 34 with a, b, d, e, f, i, k and l. IRED catalyst conversions and e.e. (brackets) were determined by GC and are given below the product. APIs are i , ii , iii , iV , V , Vi and Viii . See Supplementary Table 6.1 for the assignment of (A) and (B). * indicates a chiral centre. ND, not determined.

Fig. 5. Preparative-scale asymmetric reductive aminations of β-keto esters.

Products of the preparative-scale synthesis shown with corresponding IRED with isolated yield. For the complete figure, which includes % conversion and % e.e., see Supplementary Figure 5.6.1. The e.e. and d.r. were determined by chiral GC analysis. NaPi, sodium phosphate. (A)- and (B)- represent the absolute configuration of the product.