Multiplexed Assessment of Promiscuous Non-Canonical Amino Acid Synthase Activity in a Pyridoxal Phosphate-Dependent Protein Family

Abstract

Pyridoxal phosphate (PLP)-dependent enzymes afford access to a variety of non-canonical amino acids (ncAAs), which are premier buildings blocks for the construction of complex bioactive molecules. The vinylglycine ketimine (VGK) subfamily of PLP-dependent enzymes plays a critical role in sulfur metabolism and is home to a growing set of secondary metabolic enzymes that synthesize γ-substituted ncAAs. Identification of VGK enzymes for biocatalysis faces a distinct challenge because the subfamily contains both desirable synthases as well as lyases that break down ncAAs. Some enzymes have both activities, which may contribute to pervasive mis-annotation. To navigate this complex functional landscape, we used a substrate multiplexed screening approach to rapidly measure the substrate promiscuity of 40 homologs in the VGK subfamily. We found that enzymes involved in transsulfuration are less likely to have promiscuous activities and often possess undesirable lyase activity. Enzymes from direct sulfuration and secondary metabolism generally had a high degree of substrate promiscuity. From this cohort, we identified an exemplary γ-synthase from Caldicellulosiruptor hydrothermalis (CahyGS). This enzyme is thermostable and has high expression (~400 mg protein per L culture), enabling preparative scale synthesis of thioether containing ncAAs. When assayed with l-allylglycine, CahyGS catalyzes a stereoselective γ-addition reaction to afford access to a unique set of γ-methyl branched ncAAs. We determined high-resolution crystal structures of this enzyme that define an open-close transition associated with ligand binding and set the stage for future engineering within this enzyme subfamily.

Keywords: Biocatalysis; competition reactions; methionine analogs; noncanonical amino acids; sequence similarity network analysis.

Figure 1.

Metabolic pathways involving enzymes from the vinylglycine ketimine (VGK) subfamily. A) γ-synthases produce γ-substituted amino acids and γ-lyases break down γ-substituted amino acids. B) The direct sulfuration (DS) pathway is shown with a blue solid arrow and uses a homocysteine γ-synthase. C) The forward transsulfuration (TS) pathway (red) uses a cystathionine γ-synthase (CGS) to synthesize cystathionine (Cth), while reverse TS (gray) uses a cystathionine γ-lyase (CGL) to breakdown Cth. D) Secondary metabolic pathways are shown with a solid green arrow and features many VGK enzymes.

Figure 2.

Sequence similarity network (SSN) analysis of the vinylglycine ketimine (VGK) subfamily. 40 homologs were chosen to be screened and are denoted by diamonds. A) 37,793 sequences were compressed into 5,263 nodes that share the same length and >70% identity. An alignment score of 125 (~50 %ID) show the Met γ-lyases separate into a distinct cluster, but other functions appear interspersed based on annotations. B) 13,073 sequences (1479 nodes) of the Hcy γ-synthases and secondary metabolic γ-synthases cluster from panel A at an alignment score of 176 (~70%ID). C) 3879 sequences (500 nodes) of the Hcy γ-synthases and secondary metabolic γ-synthases cluster from panel B at an alignment score of 187 (>70% ID). Despite this high degree of stringency, unique catalytic functions do not resolve into separate clusters.

Figure 3.

Assessment of cystathionine (Cth) γ-lyase activity among select homologs. Reactions were run using 1.125 mM Cth (red) and 25 μM enzyme at pH 8.0 for 1 h in triplicate. Substrate and product quantitation was performed by UPLC-MS. αKB is shown in gray.

Figure 4.

Substrate multiplexed screening (SUMS) with a mixture of nucleophiles. A) In SUMS, activity is measured for multiple competing substrates simultaneously from whole-cells expressing a VGK homolog in duplicate. The products are mass resolved and the corresponding product profiles retain kinetic and specificity information. B) Reaction scheme with a mixture of nucleophiles. O-acetyl-l-homoserine (OAHS) was the pro-electrophile and was provided at 5 mM; each nucleophile was provided at 1 mM. Reactions used whole-cells grown in a 96-well plate or 25 μM enzyme were allowed to proceed for 1 h at pH 8.0 in duplicate and triplicate respectively. C) Product profiles for the 40-homolog library. Products of the corresponding Met analogs are colored to correspond to the nucleophile shown in panel B. In the case of lyase activity, α-ketobutyrate formation is shown in gray. The distribution of those products is indicated by the relative height of the stacked bar chart. D) Product profiles of a trio of enzymes. Products of the corresponding Met analogs are colored to correspond to the nucleophile shown in panel B. In the case of lyase activity, α-ketobutyrate formation is shown in gray. The total product concentration is given with a diamond (black) and the distribution of those products is indicated by the relative height of the stacked bar chart.

Figure 5.

SUMS with a mixture of pro-electrophiles and nucleophiles. A) Reaction scheme for double SUMS for γ-synthase and γ-lyase activity. Each pro-electrophile was provided at 5 mM; each nucleophile was provided at 1 mM. Reactions used 25 μM enzyme and were allowed to proceed for 1 h at pH 8.0 in triplicate. B) Pro-electrophile conversion profiles. Unreacted starting material was measured via UPLC-MS. The bars are colored to correspond to pro-electrophile as shown in panel A. The distribution of pro-electrophile conversion is indicative by the relative height of the colored bars. C) Product profiles for each Met analog (colored) or αKB (grey) were measured in parallel with the data in panel B and the colors correspond to the products shown in panel A. The total product concentration is given with a diamond (black) and the distribution of those products is indicated by the relative height of the stacked bar chart

Figure 6.

Biocatalytic preparative scale reactions of Met analogs with CahyGS. One mmol γ-substitution preparative scale reactions with CahyGS for the production of Met analogs. Catalyst loadings were either 1% w/v wet whole cell weight (^a) or 1% w/v of a heat-treated and clarified lysate (^b). Reactions were run overnight at 37 °C. Isolated yields are provided for the products from all nucleophiles tested in SUMS, except for Cys.

Figure 7.

γ-Addition into L-allylglycine catalyzed by CahyGS. A) Proposed γ-addition mechanism B) One mmol γ-addition preparative scale reactions with CahyGS for the production of γ-methyl Met analogs. Catalyst loading was 1% wet whole cell weight (^a). Products from all nucleophiles tested in SUMS, except for Cys, were isolated with varying yields. The d.r was determined via NMR and e.r was determined via Marfey’s derivatization. The small molecule X-ray crystal structure of the γ-addition product with the cyclopentanethiolate nucleophile shows a 2S,4R configuration

Figure 8.

Structural analysis of CahyGS. Open (gray) and closed (blue) conformation of CahyGS. The internal aldimine (E(Ain)) structures were solved for both conformations at 2.3 and 1.5 Å resolution, respectively. The closed conformation has a helix shifted by 9.4 Å, encapsulating the bound 2-ethanesulfonic acid (MES) buffer molecule to Arg403 in the active site.