Secondary Amine Catalysis in Enzyme Design: Broadening Protein Template Diversity through Genetic Code Expansion

Abstract

Secondary amines, due to their reactivity, can transform protein templates into catalytically active entities, accelerating the development of artificial enzymes. However, existing methods, predominantly reliant on modified ligands or N-terminal prolines, impose significant limitations on template selection. In this study, genetic code expansion was used to break this boundary, enabling secondary amines to be incorporated into alternative proteins and positions of choice. Pyrrolysine analogues carrying different secondary amines could be incorporated into superfolder green fluorescent protein (sfGFP), multidrug-binding LmrR and nucleotide-binding dihydrofolate reductase (DHFR). Notably, the analogue containing a D-proline moiety demonstrated both proteolytic stability and catalytic activity, conferring LmrR and DHFR with the desired transfer hydrogenation activity. While the LmrR variants were confined to the biomimetic 1-benzyl-1,4-dihydronicotinamide (BNAH) as the hydride source, the optimal DHFR variant favorably used the pro-R hydride from NADPH for stereoselective reactions (e.r. up to 92 : 8), highlighting that a switch of protein template could broaden the nucleophile option for catalysis. Owing to the cofactor compatibility, the DHFR-based secondary amine catalysis could be integrated into an enzymatic recycling scheme. This established method shows substantial potential in enzyme design, applicable from studies on enzyme evolution to the development of new biocatalysts.

Keywords: Artificial Enzyme; Genetic Code Expansion; Organocatalysis; Protein Engineering; Secondary Amine Catalysis.

Figure 1

Components used in the design of the protein‐hosted secondary amines. A) The structures of the unnatural amino acids (UAAs) used include D‐prolyl‐L‐lysine (1), L‐prolyl‐L‐lysine (2) and L‐thiazolidine‐L‐lysine (3, thioproline). B) The X‐ray crystal structure of the Lactococcus multidrug resistant regulator LmrR (PDB: 3F8F). Two monomers are shown in red and light red. Residues targeted for UAA incorporation are highlighted in cyan. C) The X‐ray crystal structure of E. coli dihydrofolate reductase (DHFR) with NADPH bound in the active site (PDB: 1RA1). Residues targeted for UAA incorporation are highlighted in yellow.

Figure 2

Deconvoluted ESI mass spectra for the wild‐type LmrR (calculated to be 14657 g/mol) and its variants, in which Val15, Asp19, Met89 and Phe93 were individually replaced with UAA 1, 2 or 3. All 13 proteins were found to have the N‐terminal methionine removed. For variants incorporated with 2, profound peaks that match closely to the hydrolysis of the L‐prolyl group were observed (red fonts).

Figure 3

Assessment of the catalytic efficiency in the transfer hydrogenation reaction. A) Conversion of the model reaction catalyzed by the LmrR variants using BNAH as determined by GC‐MS (see SI). The model organocatalytic transfer hydrogenation reaction contained cinnamaldehyde (0.68 mM, 1 equiv.) and LmrR variants (68 μM, 10 mol %), and then BNAH (1.36 mM, 2 equiv.) was added and stirred for 18 h in PBS buffer (pH 7.0, 10 % methanol) at 25 °C. The estimated turnover (k _cat) and Michaelis (K _M) constants at 25 °C for LmrR‐Phe93‐1 (50 μM) using 1 mM of cinnamaldehyde (4 a) and various concentrations of BNAH (200–5000 μM) were reported. B) Conversion of the model reaction catalyzed by the DHFR variants using NADPH as determined by GC‐MS. The model organocatalytic transfer hydrogenation reaction contained cinnamaldehyde (4 a, 0.52 mM, 1 equiv.) and DHFR variants (51 μM, 10 mol %), and then NADPH (1.04 mM, 2 equiv.) was added and stirred for 18 h in PBS buffer (pH 7.0, 5 % methanol) at 25 °C. The estimated turnover (k _cat) and Michaelis (K _M) constants at 25 °C for DHFR‐Ala7‐1 (50 μM) using 1 mM of cinnamaldehyde (4 a) and various concentrations of NADPH (40–500 μM) were reported. The template experiments (A) and (B) were tested alongside with the wild‐type (WT) proteins, UAA 1 and a negative (−) control where the reaction was performed without any protein or catalyst. Each reaction was performed in triplicate and the mean value (±standard deviation) is shown.

Figure 4

Proposed catalytic cycle of the organocatalytic transfer hydrogenation reaction by DHFR‐Ala7‐1 based on the iminium trapping and kinetic isotope effect (KIE) experiments (see also Figures S6–S8 and S11). The residue of UAA 1 forms an iminium ion with the α,β‐unsaturated carbonyl substrate, and hydride transfer occurs from the pro‐R position of NADPH to C_β of the iminium intermediate. A KIE (k ^NADPH/k ^NADPD) of 1.1±0.2 was measured using NADPH and [4R‐²H]‐NADPH (NADPD) during catalysis.

Figure 5

Substrate scope analysis of DHFR‐Ala7‐1. The DHFR variant (5.1 μM, 10 mol %) was introduced with the indicated α,β‐unsaturated carbonyl compound (4 a–4 i, 0.51 mM, 1 equiv.) and NADPH (3.4 mM, 5 equiv.) in PBS buffer at pH 7.0 for 48 h. Each reaction was performed in triplicate and the mean yield (±standard deviation) reported. Conversion (%) was estimated by use of ¹H NMR spectroscopy as previously described (see Table S7 and Figure S18 in SI).[ ²³ , ²⁴ , ²⁵ ] *Over 50 % of the conversion resulted in unidentified byproduct as indicated by ¹H NMR spectroscopy. **The product was detectable by GC‐MS but not NMR spectroscopy. ***20 % of the substrate was converted into an unidentified product as revealed by ¹H NMR spectroscopy. See Section 17 in Supporting Information for corresponding data.

Figure 6

A) Free energy profile (potential of mean force, PMF, in kcal/mol) computed at AM1/MM level of theory for interconversion between the pro‐R and pro‐S orientation within the active site of DHFR‐Ala7‐1. B) Evolution of the donor‐acceptor distance (DAD) during substrate rotation. C) The relative position of NADPH cofactor and iminium intermediate in the active site of DHFR‐Ala7‐1 illustrating pro‐R and pro‐S orientation used as initial geometries for 500 ns of unbiased MM MD simulations. D) Substrate‐protein interaction energies (electrostatic plus Lennard‐Jones; in kcal/mol) for residues located within 6 Å distance from the C_β of the substrate.

Figure 7

Coupling of the organocatalytic DHFR‐Ala7‐1 and the enzymatic glucose‐6‐phosphate dehydrogenase (G6PDH) reactions. A) The organocatalytic transfer hydrogenation reaction of cinnamaldehyde 4 a (1 mM) by DHFR‐Ala7‐1 (10 mol %) was driven by the enzymatic G6PDH reaction (50 nM), which oxidizes glucose‐6‐phosphate (2 mM) to the corresponding lactone with the associated NADPH regeneration. B) Product conversions were estimated by a GC‐FID assay similar to previously described (see SI). C) The total turnover number for NADPH refers to the ratio of mole of product formed to mole of NADPH used. Each reaction was performed in triplicate and the mean reported.