Rapid Screening of Diverse Biotransformations for Enzyme Evolution

Abstract

The lack of label-free high-throughput screening technologies presents a major bottleneck in the identification of active and selective biocatalysts, with the number of variants often exceeding the capacity of traditional analytical platforms to assess their activity in a practical time scale. Here, we show the application of direct infusion of biotransformations to the mass spectrometer (DiBT-MS) screening to a variety of enzymes, in different formats, achieving sample throughputs equivalent to ∼40 s per sample. The heat map output allows rapid selection of active enzymes within 96-well plates facilitating identification of industrially relevant biocatalysts. This DiBT-MS screening workflow has been applied to the directed evolution of a phenylalanine ammonia lyase (PAL) as a case study, enhancing its activity toward electron-rich cinnamic acid derivatives which are relevant to lignocellulosic biomass degradation. Additional benefits of the screening platform include the discovery of biocatalysts (kinases, imine reductases) with novel activities and the incorporation of ion mobility technology for the identification of product hits with increased confidence.

Figure 1

Overview of the DiBT-MS workflow for screening biocatalytic reactions and identifying improved enzyme variants. Initially, the reactions (whole cell or purified enzyme) are performed within Eppendorf tubes or 96/384-well plates (1) before spotting (0.5 μL) of the quenched reaction onto a nylon membrane grid (see photograph, Figure S1) prepared for DiBT-MS analysis (2). The membrane containing dried spotted reactions is subjected to DiBT-MS analysis where necessary with ion mobility (IM) separation (3), and the resulting mass- and mobility-resolved heat maps were assessed for wells with the highest product detection (4). The sample remaining in the corresponding 96/384-well plate or Eppendorf tube can then be extracted for DNA sequencing to identify the mutations responsible for improved activity (5).

Figure 2

Representative results of DiBT-MS screening of diverse biotransformations. (a) DiBT-MS heat map obtained upon screening 11 purified kinase enzymes with 11 monosaccharide substrates. In total 15 substrates were screened, with the full results including reaction conversions as determined by ¹⁹F-NMR given in Figures S3 and S4. (b) Resulting heat maps for imine reductase (IRED) reactions performed in a 384-well metagenomic plate. The heat maps indicate areas in which starting material (red) and product (green) are present on the membrane with each m/z value detected simultaneously. (c) DiBT-MS heat map results for PAL whole cell reaction screening. 10 enzyme variants were screened against 15 cinnamic acid substrates for conversion to the corresponding phenylalanine derivative. Reaction conversions obtained by HPLC-(UV) analysis for each of these reactions are given in Figure S12. (d) Table to show the increase in throughput achieved for three specific reaction types when employing DiBT-MS as the primary screening technique in comparison to alternative techniques, wherein the longer time denotes discovery, and the shorter time indicates an optimized assay.

Figure 3

(a) Mass-resolved heat map indicating the limit of detection of six different concentrations of galactose-1-phosphate using a pixel size of 500 μm × 500 μm and a rate of 2000 μm/s. The intensity of each pixel within each concentration spot has been summed and plotted versus the concentration of that spot. The graph shows good linearity between the concentration of each spot and the summed pixel intensity over the range 5–100 μM. Graph insert illustrates saturation of the compound. (b) DiBT-MS analysis of 12 kinase biotransformations at varying pixel sizes and stage speeds. The resulting throughput per sample has been noted. A 44 s/sample throughput has been highlighted as being the optimal conditions for high-throughput screening without compromising data quality. (c) Heat maps obtained when analyzing IRED reactions in 5 different wells of a metagenomic plate both with (right) and without (left) traveling wave ion mobility (TWIMS) enabled within the SYNAPT mass spectrometer. Two visualizations of each heat map are given to show IM filtering of the substrates’ second ¹³C isotope ion from the major product ion (m/z 208) within the heat maps.

Figure 4

(a) Homology-based active site model of AL-11 enzyme highlighting the amino acid residues conferring selectivity and reactivity from the structure of AvPAL (PDB: 5LTM) with selectivity residues highlighted (blue). The post-translationally modified 3,5-dihyro-5-methylene-4H-imidazol-4-one (MIO) ring and the neighboring polar side chains are shown (black). Residues circled (purple) were selected for reduced degenerate codon library (RBT) creation. (b) DiBT-MS screening results for library A (L148RBT/L196RBT), library B (N199RBT/I400RBT), and library C (N196RBT/N199RBT) in a 96-well plate format as indicated. (c) Analytical scale biotransformations using 5m as a substrate with AL-11 and variants identified during DiBT-MS screening. Conversion values (%) determined by reverse-phase HPLC.