Stereoselective Biosynthesis of Bichiral Amino Alcohols from Diketone via Difunctionalization

Abstract

Chiral amino alcohols are vital in many biologically active molecules and natural products and are crucial building blocks for the pharmaceutical industry. However, the chemosynthesis of chiral amino alcohols, particularly bichiral amino alcohols, faces challenges related to chemoselectivity and enantioselectivity. Here, a dual-enzyme cascade pathway for the synthesis of bichiral amino alcohols through difunctionalization, exhibiting both high chemo- and stereoselectivity, was proposed for the first time. Among them, NADPH-dependent meso-diaminopyrimidate dehydrogenase from Proteus vulgaris (PvDAPDH) exhibited (R)-selective reductive amination activity, converting ketones into chiral amines; natural glucose dehydrogenase from Bacillus megaterium (BmGDH) not only facilitated NADPH regeneration but also catalyzed the reductive hydroxylation of ketones to chiral alcohols. Second, the rational design of the rate-limiting enzyme PvDAPDH generated the mutant M4 with increased specific activity and catalytic efficiency (k _cat/K _m) for 16a up to 3.2- and 10.3-fold, respectively, compared to the wild type. Additionally, the yield of mutant M4 for 1a–16a increased by 1.1- to 37-fold compared to the wild type. Finally, two high-value bichiral amino alcohols (14b and 16b) were asymmetrically synthesized from the corresponding bietone via a one-pot dual-enzyme cascade, exhibiting excellent stereoselectivity (dr ≥ 98:2). These findings provide a potential biosynthetic pathway for the green synthesis of bichiral amino alcohols and complement existing synthetic methods.

Introduction

Chiral amino alcohols are ubiquitous structural motifs present in many biologically active molecules and natural products, as well as being crucial intermediates in pharmaceutical synthesis, for example, antibacterial drugs, anticancer drugs, anti-HIV drugs, and drugs used to treat Alzheimer’s disease (Scheme a). −

1. Importance of Chiral Amino Alcohols and Synthetic Approaches .

^a (a) Importance of chiral amino alcohols, (b) chemosynthesis of chiral amino alcohols, (c) biosynthesis of chiral amino alcohols, and (d) stereoselective sequential synthesis of bichiral amino alcohols using PvDAPDH and bifunctional BmGDH.

Various chemosynthetic methods have been developed, including Sharpless asymmetric amino hydroxylation of olefins, , reduction of amino/hydroxy ketones, − cross-coupling of imines with carbonyl compounds, and asymmetric ring opening of epoxides (Scheme b). , However, the chemoselectivity resulting from the construction of two polar bonds in the synthesis of amino alcohols required additional protection and deprotection steps. Although the transformations involving radical intermediates were compatible with common functional groups, achieving enantioselectivity remained a challenge due to the lack of suitable radical acceptors and reliable stereocontrol. ,

Biosynthesis is an alternative due to its high selectivity and mild conditions for the synthesis of chiral amino alcohols (Scheme c). The enzymes that catalyzed the asymmetric monofunctionalization of monoketones included ω-transaminases (ω-TAs), , imine reductases (IREDs) or amine dehydrogenation enzymes (AmDHs) for the asymmetric reductive amination of hydroxyl ketones, − as well as ketone reductases (KREDs) or carbonyl reductases for the asymmetric reductive hydroxylation of amino ketones. , In addition, engineered cytochrome c had been reported to difunctionalize olefin into monochiral amino alcohols. 4-Aminocyclohexanol was synthesized using aminotransferase and ketone reductase in one pot, achieving a titer of 40 mM with excellent cis-selectivity (cis/trans = 99:1) but poor trans-selectivity (cis/trans = 3:7). An engineered AmDH catalyzed the asymmetric reductive amination of α-hydroxy ketones to produce (S)-chiral amino alcohols, with remarkable conversion rates and enantiomeric excess >99%. A mutant mh13 capable of synthesizing (R)-chiral amino alcohols was obtained by directed evolution of GsAmDH from Geobacillus stearothermophilus with excellent enantioselectivity (enantiomeric excess (ee) > 99%) but narrow substrate scope and poor catalytic activity. Furthermore, engineered IREDs synthesized (R)-N-substituted 1,2-amino alcohols with excellent stereoselectivity (ee > 99%), but primary amino alcohols were not being synthesized. Native amine dehydrogenase and AspRedAm can catalyze the reductive amination of various cyclic ketones to synthesize chiral amines and their derivatives but with no catalytic activity against hydroxyketone substrates. , The RasADH/photoredox catalyzed the radical decarboxylative coupling reaction of amino acids and carbonyls to synthesize chiral 1,2-amino alcohols with high enantioselectivity (ee ≥ 92% S), but harsh reaction conditions limit its wide application. Therefore, the installation of multiple functional groups and the construction of two or more stereocenters in one building block need to be further solved.

In this study, we reported an effective solution to synthesize high-value bichiral amino alcohols with (R)-enantiomers through a dual-enzyme cascade system via difunctionalization of biketone (Scheme d). PvDAPDH exhibited extended (R)-selective reductive amination activity from ketone acid to general ketone for the first time, with excellent stereoselectivity (ee > 99%). Natural BmGDH had not only the function of cofactor regeneration but also the function of reductively hydroxylated ketones. Second, rational design guided by the catalytic mechanism mutant PvDAPDH^M4 obtained increased catalytic activity toward ketone substrates. Finally, two high-value-added bichiral amino alcohols were asymmetrically synthesized via the one-pot dual-enzyme cascade using inexpensive diketone substrates. These findings may provide a potential biosynthetic pathway for the green production of high-value bichiral amino alcohols.

Results and Discussion

Discovery of Bichiral Amino Alcohol Biosynthesis

Our previous study found that the rational design of PvDAPDH can extend the substrate from aliphatic ketone acids to aromatic ketone acids, thereby reducing amination to generate the corresponding d-amino acids. To further expand the substrate scope of PvDAPDH, we investigated the potential for catalyzing the reductive amination of generalized ketones to produce R-chiral amines. The reductive amination activity of aromatic, aliphatic, and alicyclic ketones was evaluated utilizing a purified dual-enzyme cascade system comprising PvDAPDH and glucose dehydrogenase from Bacillus megaterium (BmGDH), which is frequently employed for NADPH regeneration. Fortunately, some alicyclic ketones (2a–7a, 10a–11a, 14a, and 16a) were highly stereospecifically reduced to the corresponding R-chiral amines (ee ≥ 99%) but with lower activity (Figure and Figure S1). Among them, 11a showed the highest activity, with a conversion rate of 19.9%. To our surprise, alicyclic monoketones (2a, 5a, 7a, 8a, and 10a) were mainly asymmetrically reduced to chiral alcohols (Figure S2). More interestingly, the inexpensive and readily accessible alicyclic diketones 14a and 16a were successfully asymmetrically reduced to generate high-value dichiral amino alcohols 14b and 16b (Figure ).

1.

The catalytic performance of the BmGDH/PvDAPDH cascade system. The products were determined via chiral high-performance liquid chromatography (HPLC) and gas chromatography (GC) using 20 mM ketone substrate, 1 mM NADP⁺, 200 mM NH₄Cl, 50 mM d-glucose, and purified PvDAPDH/BmGDH in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. All experiments were conducted in triplicate. N.D.: not detected.

To investigate how the dicarbonyl group in 16a is asymmetrically reduced to produce dichiral amino alcohol by the dual-enzyme cascade system, we conducted detailed control experiments. Using 16a as the substrate, 3.1% chiral amino alcohol 16b was detected only in the presence of both PvDAPDH and BmGDH. (Figure and Table S2, entry 1). Using 16a as substrate, when only under the catalysis of BmGDH, the products 16c and 16d of asymmetric reductive hydroxylation were detected (Figure and Table S2, entries 2 and 4). In contrast, when only under the catalysis of PvDAPDH, no product of asymmetric reductive amination or hydroxylation was detected with 16a as the substrate (Figure b and Table S2, entry 3). However, the asymmetric reductive amination product 16b was detected with 16c as a substrate (Figure a and Table S2, entry 5). Together, PvDAPDH exhibits the ability to perform asymmetric reductive amination of ketone carbonyl to produce (R)-chiral amines, while BmGDH not only regenerates the NADPH cofactor but also has the function of asymmetric reductive hydroxylation of ketone carbonyl to generate (R)-chiral alcohols.

2.

The functional verification of the BmGDH/PvDAPDH cascade system. (a) Chiral HPLC was used for production identification. (b) GC was used for production identification. The products were determined via chiral HPLC and GC using 10 mM 16a or 16c, 1 mM NADP⁺, 100 mM NH₄Cl, 50 mM d-glucose, and purified BmGDH/PvDAPDH in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. *NADPH was added to the reaction system. See Table S2 for the details.

Identification and Optimization of Bichiral Amino Alcohol Biosynthesis

Two possible synthetic pathways from 16a to 16b were hypothesized: route A and route B (Figure a). To further elucidate the synthetic pathway of 16b, we analyzed the reaction process curve (Figure b). As substrate 16a was consumed, intermediate 16c gradually accumulated. Subsequently, intermediate 16c was partially converted into product 16b and byproduct 16d, and a small amount of intermediate 16e was also generated. In addition, no 1,4-cyclohexanediamine products were detected during the reaction (Figure b). Therefore, substrate 16a first underwent reductive hydroxylation by BmGDH to form intermediate 16c and subsequently underwent asymmetric reductive amination by PvDAPDH to produce 16b, following synthetic route A (Figure a).

3.

Identification and optimization of the synthetic pathway of 16b. (a) Identification of the synthetic pathway for 16b. (b) The reaction process curve of 16b. The products were determined using 20 mM 16a, 1 mM NADP⁺, 200 mM NH₄Cl, and 50 mM d-glucose and purified BmGDH/PvDAPDH in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. (c) Optimization of the pathway enzyme. The specific activity was determined using 1 mM NADP⁺, 50 mM d-glucose, and four purified enzymes in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. The titer of the 16c was determined using 10 mM 16a, 1 mM NADP⁺, 50 mM d-glucose, and four purified enzymes in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. (d) The activity of 16b was determined using four purified enzymes and PvDAPDH. The titer of the 16b was determined using 10 mM 16a, 1 mM NADP⁺, 100 mM NH₄Cl, 50 mM d-glucose, four purified enzymes, and PvDAPDH in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. All experiments were conducted in triplicate.

To optimize the synthesis pathway of 16b, we screened enzymes for cofactor regeneration and asymmetric reductive hydroxylation. Based on the principle of substrate structure similarity, the reported glucose dehydrogenases (BmGDH, BmGDH-M3, and EsGDH) from Bacillus subtilis and Exiguobacterium sibiricum, as well as the ketone reductase (SsCR) from Sporobolomyces salmonicolor, were screened. SDS-PAGE results indicated that all four enzymes exhibited good soluble expression, with BmGDH having the highest protein expression (Figure S3). The capabilities for cofactor regeneration and asymmetric reductive hydroxylation were evaluated by measuring the specific activity of regenerated NADPH and the titer of asymmetrically reduced hydroxylation 16a to 16c. The results revealed that BmGDH regenerated NADPH with the highest specific activity, while BmGDH^M3 catalyzed 16a to produce 16c with the highest titer (Figure c). Subsequently, the purified dual-enzyme cascade system of the four enzymes mentioned above and PvDAPDH was constructed to determine the production of 16b. The results indicated that the system containing BmGDH achieved the highest titer of 16b, which was 0.2 ± 0.02 mM (Figure d). Therefore, BmGDH was selected for a subsequent study.

Mechanism Analysis of PvDAPDH

Given that the accumulation of intermediate 16c and the catalytic efficiency of PvDAPDH for intermediate 16c are lower than those of BmGDH for substrate 16a, PvDAPDH is the rate-limiting enzyme in this synthetic pathway (Figure b and Table ). To elucidate the reason for the low catalytic efficiency of PvDAPDH toward intermediate 16c, we compared the kinetic parameters of PvDAPDH for intermediate 16c and its natural substrate. The K _m value of PvDAPDH for intermediate 16c (K _m = 9.42 ± 0.30 mM) was 23 times higher than that of StDAPDH for its natural substrate meso-DAP (K _m = 0.41 ± 0.07 mM). The k _cat value of PvDAPDH for intermediate 16c (k _cat = 0.24 s^–1) was approximately 1/169 that of StDAPDH for the natural substrate (k _cat = 40.47 ± 2.95 s^–1). Therefore, the low catalytic efficiency of PvDAPDH for intermediate 16c is primarily attributed to the low turnover number.

1. Kinetic Parameters and Specific Activities of BmGDH, PvDAPDH, and Its Mutant.

enzyme	substrate	Km (mM)	kcat (s–1)	kcat/Km (mM–1·s–1)	specific activity (U/mg)
BmGDH	16a	50.03 ± 7.70	34.80 ± 0.50	0.70	6.72 ± 0.10
PvDAPDH	16c	9.42 ± 3.00	0.24 ± 0.03	0.03 (1)	0.05 ± 0.01 (1)
PvDAPDHM4	16c	1.47 ± 0.14	0.50 ± 0.01	0.34 (11.3)	0.21 ± 0.02 (4.2)

^aThe kinetic parameters of BmGDH, PvDAPDH, and its mutant were determined in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C, which contained 200 mM NH₄Cl, 5 mM NADPH with various concentrations of 16a or 16c (0–50 mM), and purified enzymes. All experiments were conducted in triplicate. K _m values were obtained by fitting the Michaelis–Menten equation.

Our previous studies have demonstrated that the reductive amination of PvDAPDH can be divided into four steps: the activation of the ammonium ion (RC → [TS1]), the formation of an alcoholamine intermediate via nucleophilic attack (Int1 → [TS2]), the dehydration of an alcoholamine intermediate to form an iminium ion intermediate (Int2 → [TS3]), and the reduction of an iminium ion intermediate to generate a chiral amine (Int3 → [TS4]). Among these steps, the activation of ammonium ions, the nucleophilic attack, and the dehydration of the alcoholamine intermediate represent high-energy barrier steps. Therefore, we docked 16c and intermediate 16c_Int2 into the active center of the PvDAPDH-NADP⁺ binary complex and performed molecular dynamics (MD) simulations to obtain a dynamically stable ternary complex structure.

For the activation of the ammonium ion and the nucleophilic attack (RC → [TS1] and Int1 → [TS2]) reaction step, a two-dimensional descriptor was employed to characterize the catalytic conformation of 16c (Figure b). The first dimension, distance 1 [d1­(N@NH₄ ⁺-NE1@H154)], denotes the distance between the nitrogen atom of NH₄ ⁺ and the NE1 atom on the side chain imidazolium ring of residue H154. This distance was utilized to depict NH₄ ⁺ activation by residue H154. The second dimension, distance 2 [d2­(N@NH₄ ⁺-C1@16c)], refers to the distance between the nitrogen atom of NH₄ ⁺ and the C1 atom of 16c. This distance was employed to represent the activated NH₄ ⁺ nucleophilic attack on the C1 atom of 16c. The catalytic conformation needs to satisfy both d1 ≤ 3.3 and d2 ≤ 3.5 Å. MD simulations revealed that the catalytic conformation ratio of 16c is extremely low (Figure c, 0.01%), suggesting that the probability of activation and nucleophilic attack of ammonium ion is low. The nitrogen atom of the ammonium ion forms hydrogen bonds with the side chain carboxyl oxygen atom of residue D92 and the main chain carbonyl oxygen atom of residue D122, which may result in the ammonium ion being far away from residue H154. The carbonyl oxygen on C1 of 16c forms hydrogen bonds with the side chain hydroxyl groups of residues S151 and T171. Additionally, the CH-π interaction between the alicyclic ring of 16c and the imidazole ring of residue H227 may lead to a long distance between the ammonium ion and 16c. Therefore, altering the charged properties of the residues surrounding the ammonium ion shortens the activation distance of the ammonium ion. Additionally, strengthening the interaction between 16c and the surrounding residues could decrease the distance for the amine nucleophilic attack.

4.

Catalytic mechanism of PvDAPDH. (a) Proposed molecular mechanism of PvDAPDH. (b) The possible binding mode of 16c in PvDAPDH (16c in cyan stick, NH₄ ⁺ in blue stick, and hydrogen bonds in yellow dashed lines). (c) MD plots for the distances of the ammonium ion activation (d1) and the formation of the N–C bond (d2) in PvDAPDH-16c-NADPH. (d) Possible binding mode of 16c_Int2 in PvDAPDH (16c_Int2 in purple sticks and hydrogen bonds in yellow dashed lines). (e) MD plots for proton-transfer distance d3 in PvDAPDH-16c_int2-NADPH.

For the alcoholamine intermediate dehydration Int2 → [TS3] reaction step, the key distance of the catalytic conformation of 16c_Int2 is d3 (HD1@H154-O@16c_Int2), denoting the distance between the hydrogen atom HD1 on the side chain imidazole ring of residue H154 and the oxygen atom within the hydroxyl group on 16c_Int2. This distance represents the proton transfer from residue H154 to 16c_Int2 (Figure d). MD simulations showed that the conformational proportion of d3 ≤ 3.5 Å was only 2.4% (Figure e), indicating a low probability of proton transfer. Hydrogen bonds are formed between the amino hydrogen atom on C1 of 16c_Int2 and the main chain carbonyl oxygen atom of residue D122, as well as between the hydroxyl oxygen atom on C1 of 16c_Int2 and the side chain hydroxyl hydrogen atom of residue S151. Additionally, the hydroxyl group on C4 of 16c_Int2 forms hydrogen bonds with the side chain amide oxygen atom of residue N253 and the main chain amino hydrogen atom of residue M152. The aforementioned forces may cause 16c_Int2 to deviate from the active center. Consequently, by altering the interaction between 16c_Int2 and the surrounding residues, either 16c_Int2 is pushed toward residue H154 or the α-helix of residue H154 is brought closer to 16c_Int2, thereby reducing the distance for proton transfer. Overall, we identified nine nonconserved amino acid residues (S90, D92, M150, S151, M152, T171, R181, H227, and N253) near the ammonium ion, 16c, and 16c_Int2 as potential mutation sites.

Protein Engineering of PvDAPDH

Based on the chromogenic principle of 2,4-dinitrophenylhydrazine (DNPs), a strong linear relationship was observed between the absorbance at 520 nm and the concentration of 16c (R ² = 0.9974), which was used as a high-throughput screening method (Figure S6). A saturated single mutant library on the nine potential mutation sites was constructed using NNK degenerate codons (Figure a and Table S3).

5.

Protein engineering of PvDAPDH. (a) Potential mutation sites (light orange sphere). (b) Mutation process of PvDAPDH. The assay system consisted of 10 mM 16c, 100 mM NH₄Cl, 2 mM NADPH, and crude enzyme in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C. All experiments were conducted in triplicate. (c) The possible binding mode of 16c in mutant M4 (16c in cyan stick, NH₄ ⁺ in blue stick, and hydrogen bonds in yellow dashed lines). (d) The possible binding mode of 16c_Int2 in mutant M4 (16c_Int2 in purple stick and hydrogen bonds in yellow dashed lines). (e) MD plots for the distances of the ammonium ion activation (d1) and the forming of the N–C bond (d2) in M4-16c-NADPH. (f) The probability distribution of proton transfer distance d3 between 16c_Int2 and H154 in mutant M4.

In the first round of mutation, 27 potential positive mutants (Table S5) were obtained through DNPs screening, and 6 beneficial single mutants (S90Y, M150L, S151Y, T171M, R181S, and H227A) were identified by HPLC determination. Among these, the titer of the optimal single mutant T171 M increased by 155.6% compared to the wild type (WT), which was recorded as M1 (Figure b and Table S6). In the second round of mutations, M1 was used as a template for iterative combinatorial mutations with the other five beneficial single mutants, resulting in the construction of five double mutants (Figure b and Table S6). The results indicated that the positive double mutants were T171M/H227A and T171M/M150L, with the optimal double mutant T171M/H227A showing a 21.5% increase in titer over M1, which was recorded as M2. In the third round of mutations, M2 served as a template for iterative combinatorial mutations with the remaining four beneficial single mutants to construct a four triple mutants’ library (Figure b and Table S6). The results showed that the positive triple mutants were T171M/H227A/M150L and T171M/H227A/S151Y, and the optimal triple mutant T171M/H227A/M150L was 45.8% higher than M2, denoted as M3. In the fourth round of mutations, M3 was used as a template, and the remaining three beneficial single mutants were iteratively combined to construct three quadruple mutants (Figure b and Table S6). The positive quadruple mutants were T171M/H227A/M150L/R181S, and the titer was 13.2% higher than that of M3, which was recorded as M4. In the fifth round of mutation, M4 served as the template for further iterative combinatorial mutations with the remaining two beneficial single mutants, resulting in the construction of two quintuple mutants (Figure b and Table S6). However, no positive quintuple mutants were identified. Therefore, mutant M4 was chosen for further study (Figure b and Table S6).

To characterize the catalytic performance of mutant M4, we determined the specific activity and kinetic parameters of mutant M4 for 16c. Compared to that of the WT, the specific activity of mutant M4 for 16c increased by 3.2 times (Table ). The K _m value of mutant M4 for 16c was 5.4 times lower than that of the WT, indicating an enhanced affinity of mutant M4 for 16c. Additionally, the k _cat value of mutant M4 for 16c improved by 1.1 times, suggesting an increased reaction rate. Consequently, the catalytic efficiency (k _cat/K _m) of mutant M4 for 16c was 10.3 times higher than that of the WT (Table ).

To elucidate the molecular mechanism underlying the enhanced catalytic activity of mutant M4, we analyzed the proportion of catalytic conformations in three high-energy barrier steps (RC → [TS1], Int1 → [TS2], and Int2 → TS3). 16c and 16c_Int2 were docked into the active center of mutant M4, and MD simulations were subsequently performed. The results revealed that the proportion of catalytic conformations (d1 and d2) increased from 0.01% in WT to 0.20% in mutant M4 (Figures c and e). The T171 M mutation disrupted the hydrogen bond interaction with the carbonyl oxygen at the C1 position of 16c, leading to an upward deflection of the 16c conformation. This alteration resulted in the formation of new hydrogen bond interactions with residues D92, D122, G153, and NADPH, thereby reducing the distance for the nucleophilic attack (d2: 4.5 → 3.3 Å; Figures b and c). Additionally, a new hydrogen bond interaction between the ammonium ion and the phenolic hydroxyl group of residue Y205 draws the ammonium ion closer to residue H154, reducing the activation distance of the ammonium ion (d1: 6.4 → 4.5 Å) (Figures b and c). Furthermore, the proportion of the catalytic conformation (d3) in mutant M4 increased from 2.4% in the WT to 9.0% (Figures d and f). The H227A mutation disrupted the CH-π interaction with 16c_Int2 and established a new hydrogen bond with D92, resulting in an upward shift of 16c_Int2. Additionally, the R181S mutation may eliminate the electrostatic repulsion between residue H154 and facilitate the free rotation of residue H154 toward 16c_Int2, thereby shortening the proton transfer distance between 16c_Int2 and residue H154 (d3: 4.1 → 3.1 Å) (Figures d and d). These results indicated that mutants M150S, T171M, R181S, and H227A altered the electric field environment within the active site, shortened the proton transfer distance, increased the proportion of the catalytic conformation, and consequently enhanced the catalytic efficiency.

The substrate scope of the mutant M4 was evaluated, and the results indicated that compared with WT, the mutant M4 exhibited better catalytic activity for alicyclic ketones (1a–16a), including 9a, which WT could not convert. The yield of mutant M4 to various alicyclic ketones increased by 1.1–37 times, with mutant M4 exhibiting the highest enhancement of 37-fold for 14a. Unfortunately, no activity was observed for aliphatic and aromatic ketones (Figure ).

Whole Cell Synthesis of High-Value Bichiral Amino Alcohols

To achieve the biosynthesis of high-value bichiral amino alcohols, we inserted the genes encoding PvDAPDH^M4 and BmGDH into the high-copy plasmid pRSFDuet-1 (Figure S7). This plasmid was then transformed into E. coli BL21 (DE3) to create the recombinant engineering strain E. coli 01. SDS-PAGE results indicated that both enzymes were effectively expressed in E. coli 01 (Figure S8). Compounds 14b and 16b, which serve as active pharmaceutical scaffolds, were selected for scale preparation under optimized transformation conditions (Figure ). The whole-cell E. coli 01 converted the diketone substrates 14a and 16a into bichiral amino alcohols 14b and 16b. The diastereomeric ratio (dr) of 14b (precursor of vernakalant (channel blocker) and anxiolytic medications) was 98:2, with a titer of 6.5 ± 0.05 g L^–1. The dr value of 16b (precursor of ambroxol (mucus dissolving agent) and lomibuvir (hepatitis C protease inhibitor)) was 99:1, with a titer of 12.5 ± 0.15 g L^–1 (Figure and Figures S28 and S29).

6.

One-pot biosynthesis of high-value bichiral amino alcohols. The transformation reaction system consisted of a 20 mL Tris-HCl buffer (100 mM, pH 8.5) containing 50 g L^–1 cyclohexanedione substrate, 30 g L^–1 NH₄Cl, 100 g L^–1 d-glucose, 1.0 mM NADP⁺, and whole cell E. coli 01 biocatalysts. The products were determined via chiral HPLC after reaction for 30 h at 30 °C. All experiments were conducted in triplicate.

Conclusions

In conclusion, we developed a substrate-coupled, cofactor-regenerating dual-enzyme cascade system for the one-pot synthesis of high-value bichiral amino alcohols from inexpensive diketones. It is reported for the first time that PvDAPDH can utilize ketone substrates for the asymmetric reductive amination in the synthesis of (R)-chiral amines. Furthermore, natural BmGDH was found to have dual functions, in which it not only had the function of cofactor recycling but also reductively hydroxylated ketones to generate chiral alcohols. This work provides a potential biosynthetic pathway for the green production of amino alcohols with multiple chiral centers. However, its catalytic activity still remains low, and it is expected to reform by artificial intelligence to enable the rate-liming enzyme. −

Methods

Materials

The expression plasmid pET28a (+) and pRSFDuet-1 and the host strain E. coli BL21 (DE3) were purchased from Novagen. Diaminopimelate dehydrogenase (NCBI: WP_075674287.1) was cloned from Proteus vulgaris, which was obtained by a prelaboratory screen. l -Amino acid deaminase from Proteus mirabilis and glucose dehydrogenase from Bacillus megaterium (UniProtKB: A7XZE6_BACME) were preserved by our lab. Commercial reagents, standards, and solvents were purchased from Sigma-Aldrich (Shanghai, China), Meryer Chemicals (Shanghai, China), Aladdin Chemicals (Shanghai, China), Macklin Chemicals (Shanghai, China), and Bide Chemicals (Shanghai, China) and used without further purification.

Expression and Purification of PvDAPDH and GDH

The single colony of recombinant E. coli strains PvDAPDH and GDH was cultivated overnight (12–14 h, 37 °C, 180 rpm) in an LB medium containing kanamycin (50 μg mL^–1) and used as the inoculum (1%). When the culture’s optical density (OD₆₀₀) reached 0.6–0.8, isopropyl β-d-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.4 mM to induce gene expression at 25 °C for a further 16 h. The cells were collected by centrifugation (6000g, 10 min) and resuspended in buffer A (500 mM NaCl, 20 mM imidazole, 20 mM phosphate buffer, pH 8.5; 10 mL g^–1 of wet weight). The cell suspensions were lysed by sonication and centrifuged at 14,000g for 0.5 h. The subsequent experiments were performed on an ÄKTA pure system (GE Healthcare) with a HisTrap HP column (5 mL, GE Healthcare). The protein was then concentrated for activity assay or catalytic reaction. The purity of proteins was determined by SDS-PAGE. All purification operations were conducted at 4 °C.

Construction, Expression, and Screening of the PvDAPDH Mutation Library

The PvDAPDH mutant libraries of nine amino acid residues including S90, D92, M150, S151, M152, T171, R181, H227, and N253 were generated by applying a saturated mutation with NNK degenerate codons. The mutation libraries were performed with the template plasmid pET28a (+) containing the PvDAPDH gene, and the primers used are listed in Table S3. The colonies were incubated 12 h at 37 °C in 96-well plates containing 200 mL of a fresh lysogeny broth (LB) medium per well. The individual mutants were then transferred to a new 96-deep-well plate (40 mL seed per well containing 800 mL TBA medium) and cultured to an OD₆₀₀ of 0.6 to 0.8, and then the induced cells were cultured overnight at 25 °C. The cell pellets were harvested by centrifugation at 4000g for 30 min. The pellets were resuspended in 200 μL of Tris-HCl buffer with 2 mg mL^–1 lysozyme, and the mixture was then incubated at 37 °C for 2 h with shaking at 400 rpm. The cell-free lysis supernatant was obtained by centrifugation at 4000g for 30 min. The reductive amination reaction system (200 μL) consisted of 100 μL of the lysis supernatant, 100 mM NH₄Cl, 10 mM 16c, and 1.0 mM NADPH in a Tris-HCl buffer (100 mM, pH 8.5). The mixture was reacted in a shaker at 30 °C and 400 rpm for 24 h. To efficiently screen for positive mutants from a 96-well plate, 30 μL of the reaction solution was added to a 96-well plate containing 20 μL of 20 mM 2,4-dinitrophenylhydrazine solution and 70 μL of H₂O. Then the 96-well plates were mixed thoroughly and allowed to react at room temperature for 15 min. Following this, 80 μL of 4 M NaOH solution was added to facilitate color development. After standing in the dark for 15 min, the absorbance of the solution was measured at 520 nm. Next, the combinatorial mutation libraries were obtained via directed evolution methods. All site-directed mutagenesis was performed with the template plasmid pET28a (+) containing the PvDAPDH gene or its mutants, and the primers used are listed in Table S4. All of the mutant libraries were conducted by whole plasmid PCR using KOD-Plus-Neo (TOYOBO). The resultant PCR products were digested with DpnI to eliminate the template plasmid. Furthermore, 10.0 μL of digested products was transformed into E. coli BL21­(DE3) cells for the following screening or DNA sequencing (GENEWIZ, China).

Construction of the Coexpressed Strain of PvDAPDH and BmGDH

To construct a highly efficient conversion system, coexpression strains were constructed by single enzyme digestion experiments combined with one-step homologous recombination. The main primers used for constructing coexpressed strains are summarized in Table S7. PvDAPDH^M4 gene was first inserted into pRSFDuet-1 after the first T7 promoter using restriction sites BamHI and EcoRI. Then BmGDH gene was inserted after the second promoter using the restriction sites KpnI and XhoI.

Activity Assays and Kinetic Parameters

The activities of PvDAPDH wild type and its mutant were assayed by measuring the amount of product 16b that was determined by FDAA derivatization. The activity of GDH was assayed by measuring the initial rate of change in absorbance at 340 nm (corresponding to the amount of NADH (ε = 6220 M^–1 cm^–1)) using a microplate reader. One unit (1 U) of activity is defined as the amount of purified enzyme required to produce 1 μM of product. The kinetic parameters of the PvDAPDH wild type and its mutant for reductive amination were determined in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C, which contained 200 mM NH₄Cl and 5 mM NADPH with various concentrations of 16c (0–50 mM), and the purified enzyme was added to start the reaction. The kinetic parameter of BmGDH was determined in a Tris-HCl buffer (100 mM, pH 8.5) at 30 °C, which contained 1 mM NADP⁺, 50 mM d-glucose with various concentrations of 16a (0–50 mM), and purified BmGDH. All of the experiments were conducted with three replicates. The K _m and k _cat values are calculated by fitting the initial rate to the Michaelis–Menten equation with the software Origin.

HPLC Analysis

The concentration and stereoselectivity of amino alcohols were measured with Marfey’s reagent (1-fluoro-2,4-dinitrophenyl-5-l-alanine, FDAA) for precolumn derivatization. The mixture consisted of 50 μL of the reaction sample, 100 μL of Marfey’s reagent (14 mM) in acetone solution, 80 μL of NaHCO₃ (1 M), and 100 μL of ddH₂O. Subsequently, the above mixture was reacted at 80 °C for 10 min. Finally, 40 μL HCl (2 M) was added to stop the reaction. The obtained mixture samples were filtered through a 0.22 μm organic membrane and were then prepared for HPLC analysis. HPLC was performed on an Agilent 1260 system equipped with an Zorbax SB-C18 column (4.6 × 150 mm, 5 μm; Agilent Technologies, USA). The mobile phase was a mixture of ddH₂O (0.1% trifluoroacetic acid; buffer A) and methanol (0.1% trifluoroacetic acid; buffer B). The flow rate was 1.0 mL min^–1, the temperature was 25 °C, and detection wavelength was 340 nm. The gradient program was as follows: 40% B, hold for 3 min, increase B to 60% in 4 min, increase B to 80% in 3 min, hold for 3 min, and finally decrease B to 60% in 3 min.

GC Analysis

The reaction samples were repeatedly extracted with an equal volume of ethyl acetate three times, and then the obtained organic phase was dried by anhydrous Na₂SO₄ and filtered through a 0.22 μm organic membrane. The concentration of alcohol products was detected using GC. GC was performed on a SHIMADZU Nexis GC-2030AF system, which is equipped with a DB-5 chromatographic column (30 m × 0.25 mm × 0.25 μm; Agilent Technologies, lnc.) and nitrogen as the carrier gas. GC conditions were as follows: carrier gas flow rate: 1.0 mL min^–1, split ratio: 10:1, injector temperature: 250 °C, and FID detector: 270 °C. The temperature program was as follows: the initial column temperature was set at 60 °C and hold for 1 min; then increased at 10 °C min^–1 up to 90 °C; finally, increased at 20 °C min^–1 up to 250 °C , maintained for 2 min.

The concentration and stereoselectivity of alcohol products were detected using chiral GC. Chiral GC was performed on a SHIMADZU Nexis GC-2030AF system, which is equipped with a CP-Chiralsil-Dex CB chromatographic column (25 m × 0.25 mm × 0.25 μm; Agilent Technologies, lnc.) and nitrogen as the carrier gas. Chiral GC conditions were as follows: injector: 250 °C, FID detector: 270 °C, carrier gas flow rate: 1.0 mL min^–1, and split ratio: 10:1. The temperature program was as follows: the initial column temperature was set at 80 °C; then increased at 20 °C min^–1 up to 100 °C; next, increased at 10 °C min^–1 up to 130 °C hold for 1 min, then increased at 5 °C min^–1 up to 160 °C hold for 1 min;finally, increased at 5 °C min^–1 up to 200 °C, maintained for 2 min.

Molecular Docking

Using the PvDAPDH-NADPH binary complex as a model, 16c and NH₄ ⁺ were docked to the active center, and then the dynamically stable PvDAPDH-16c_NH₄ ⁺_NADPH complex structures were obtained through MD simulation. Similarly, 16c_Int2 was docked to the active center of the PvDAPDH-NADPH binary complex, and then the dynamically stable PvDAPDH-16c_Int2-NADPH complex structure was obtained through MD simulation. Next, the starting structure coordinates of mutant M4 were created by changing multiple mutated residues. Then, substrate 16c and intermediate 16c_Int2 were docked into the active center of mutant M4 using AutoDock Vina followed by MD simulations. The docking parameters used for the molecular docking of the ligand into the active site of PvDAPDH via AutoDock Vina were as follows: The grid box was centered on the active site of PvDAPDH with dimensions of 30 × 30 × 30 Å to ensure sufficient coverage of the binding pocket, and the exhaustiveness parameter was set to 8 to balance computational efficiency and docking accuracy. The top 20 docking poses were generated and analyzed using the default scoring function of AutoDock Vina.

Molecular Dynamic Simulations

The substrate-bound complex conformation was selected for the molecular dynamics (MD) simulation. The titratable residues (His, Glu, Asp, Lys, and Arg) were assigned based on their pK _a data derived from the PROPKA program. Moreover, the protonation states and local hydrogen bond networks were further carefully checked using the VMD software. Histidine residues His111, His179, His241, and His227 were set as HID, while His94, His154, His212, His230, and His233 were set as HIE. All glutamic acid and aspartic acid residues were deprotonated. All of the lysine and arginine residues were protonated. For the further MD simulations, the typical AMBER force field (GAFF) was implemented for the ligands, whereas the partial atomic charges were received from the RESP method utilizing the B3LYP-D3/6-31G (d, p) level of theory. The Amber ff14SB force field was used for protein residues. Each simulation system was solvated in a rectangular box filled with TIP3P water and neutralized by Na⁺ counterions. Next, each system was equilibrated with a series of minimizations interspersed by short MD simulations during which restraints on the protein backbone heavy atoms were gradually released (with a force constant of 10, 2, 0.1, and 0 kcal/(mol·Å²)) and heated slowly from 0 to 310 K for 50 ps in which we applied a 10 kcal/(mol·Å²) restraint on the protein backbone heavy atoms. Finally, an extensive MD simulation of 100 ns was performed at a constant temperature and pressure. All MD simulations were performed using the Amber 18 package.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c01864.

• DNA sequence, amino acid sequence, and functional verification of the BmGDH/PvDAPDH data; mutant primers; mutation library screening data; substrate scope determination data; kinetic parameter data; SDS-PAGE; construction of gene coexpression plasmid; substrate concentration ratio optimization data; and HPLC, GC, and MS analysis data of the production (PDF)

T.W. and L.L. designed this project. T.W. performed this project. J.W., J.W., and W.S. analyzed this project. T.W. and W.W. wrote this manuscript.

The authors declare no competing financial interest.