A Whole-Cell Catalytic System for Equol Production Based on Daidzein Reductase Engineering

Abstract

As an isoflavone metabolite with diverse physiological activities, the development of efficient and sustainable manufacturing technologies for (S)-equol holds significant importance. This study focuses on the semi-rational design of daidzein reductase (DZNR), the first key enzyme in the (S)-equol biotransformation pathway. Through multiple sequence alignment and three-dimensional structural analysis, two critical residues, Gly30 and Ala105, were identified in DZNR. A library of single and combinatorial mutants was constructed and screened, yielding the double variant DZNR30S+105S with substantially enhanced catalytic performance. In a whole-cell biocatalytic system, the recombinant E. coli (Escherichia coli) strain harboring this combinatorial mutant achieved a yield of 238.3 mg/L (S)-equol at a substrate concentration of 1 mM daidzein, demonstrating markedly improved catalytic efficiency. Upon increasing the daidzein concentration to 2 mM, the reaction reached equilibrium within 5 h, producing 384.6 mg/L (S)-equol, which highlights the mutant’s excellent potential for high-substrate-concentration applications. This study not only provides novel mechanistic insights into DZNR catalysis but also successfully establishes a DZNR variant with enhanced activity, offering an efficient biocatalytic component for the industrial-scale biomanufacturing of (S)-equol and thereby advancing the development of green biosynthesis technologies for this valuable compound.

1. Introduction

(S)-Equol is a natural metabolite produced by gut microbiota through the metabolism of soy isoflavones. It functions as a highly selective estrogen receptor modulator, exhibiting potent agonist activity at estrogen receptor β (ERβ) while showing low affinity for estrogen receptor α (ERα) and minimal activation thereof, thereby offering distinctive health benefits and an excellent safety profile [1,2,3]. Accumulating evidence indicates that (S)-equol is a bioactive compound that targets multiple pathways and modulates diverse signaling cascades in vivo, displaying antioxidant, estrogen-modulating, anti-inflammatory, anti-obesity, cardioprotective, neuroprotective, and bone-protective activities [4,5,6,7,8,9,10].

Although (S)-equol exhibits broad application potential, its natural production is constrained by a critical bottleneck: only 30–50% of the global population can metabolize daidzein into (S)-equol [11,12]. Currently, biosynthesis represents a highly promising approach for the in vitro production of (S)-equol. This method relies on the cascade catalysis of four key enzymes—daidzein reductase (DZNR), dihydrodaidzein racemase (DDRC), dihydrodaidzein reductase (DHDR), and tetrahydrodaidzein reductase (THDR)—to accomplish the biotransformation of daidzein to (S)-equol [13]. DZNR serves as the initiating enzyme in this metabolic pathway. Understanding of this enzyme originated from the seminal work of Shimada et al. in 2010, who first cloned and identified its encoding gene from the intestinal lactic acid bacterium Lactococcus Strain 20-92 [14]. Subsequent studies have confirmed that recombinant DZNR heterologously expressed in E. coli can specifically reduce the C2=C3 double bond to produce dihydrodaidzein (DHD) (Note: later studies corrected the product from the initially reported (S)-DHD to (R)-DHD) [15]. As a member of the old yellow enzyme (OYE) family (EC 1.6.99.1), DZNR possesses the remarkable capability to catalyze the asymmetric reduction of carbon–carbon double bonds (C=C). Such enzymes can generate chiral molecules with specific substituents at the α- or β-positions of the chiral center with high enantioselectivity (typically >99% ee) [16].

Enzyme engineering represents an essential strategy and critical breakthrough for achieving efficient and cost-effective industrial biosynthesis of (S)-equol [17]. DZNR belongs to the OYE family. Despite extensive protein engineering efforts targeting OYEs in general, DZNR has received surprisingly limited attention. Several inherent characteristics contribute to this research gap and pose substantial challenges for enzyme engineering. Unlike typical OYE family members, DZNR possesses an extended primary amino acid sequence and a more complex three-dimensional structure. Furthermore, most characterized DZNRs originate from intestinal microorganisms, where they have evolved under strictly anaerobic conditions. Consequently, these enzymes exhibit pronounced oxygen sensitivity, substantially complicating their heterologous expression, purification, and subsequent engineering efforts. Collectively, these factors—structural complexity, sequence divergence, and strict anaerobic requirements—have hindered systematic protein engineering studies on DZNR, leaving its catalytic potential largely unexplored relative to other well-characterized OYE enzymes.

This study employed a semi-rational design approach to engineer DZNR for enhanced (S)-equol biosynthesis efficiency. Through sequence alignment and structural analysis, two critical residues, Gly30 and Ala105, were identified as targets for mutagenesis to obtain variants with improved catalytic performance and to evaluate their utility in whole-cell biocatalytic systems. These findings provide novel insights into the structure–activity relationships of DZNR and establish the foundational components and theoretical framework for developing an efficient biomanufacturing process for (S)-equol (Figure 1).

2. Results

2.1. Sequence and Structural Analysis of DZNR

DZNR belongs to the OYE family. Previous studies have identified four conserved “hot-spot” regions in OYE family enzymes that are closely associated with catalytic properties. Site-directed or combinatorial mutations within these hotspots can alter enzymatic characteristics, including activity and substrate specificity [16,18]. Accordingly, the present study aimed to elucidate the roles of corresponding residues in DZNR. Primary sequence alignment analysis revealed that DZNR contains two residues that align with hotspot positions III and IV of the OYE family (Figure 2). Furthermore, homology modeling was employed to predict the three-dimensional structure of DZNR, which was subsequently compared with those of the characterized OYE family enzymes PETNR (PDB: 5LGX) (Figure 3a) and YqjM (PDB: 1Z41) (Figure 3b). Structural superposition demonstrated that these two residues occupy spatial positions consistent with the identified hotspots. These findings suggest that these sites may be intimately involved in modulating the catalytic properties of DZNR.

2.2. Site-Directed Mutations of DZNR

The DZNR sequence deposited in NCBI (GenBank: AFV15453.1) served as the parental template for codon optimization according to the preference of E. coli. The optimized gene was chemically synthesized and cloned into the pETDuet-1 vector to generate the recombinant expression plasmid pETDuet-1-dznr. Whole-plasmid amplification was employed to perform semi-saturation mutagenesis at positions 30 and 105 of DZNR, respectively.

Whole-cell enzymatic activities were determined for all variants, as DZNR is potentially oxygen-sensitive and its activity could not be detected following purification under standard laboratory conditions. Among the position-30 variants, mutant DZNR30S exhibited 1.39-fold higher activity (17.10 × 10⁻³ U/g^wcw) than the wild-type DZNR (12.30 × 10⁻³ U/g^wcw) (Figure 4a). At position 105, mutants DZNR105C (18.82 × 10⁻³ U/g^wcw) and DZNR105S (17.71 × 10⁻³ U/g^wcw) showed 1.53- and 1.44-fold increases in activity, respectively, relative to the wild-type enzyme (Figure 4b). Furthermore, combinatorial mutagenesis of both sites revealed that the activities of DZNR30S+105S and DZNR30S+105C were increased by 3.23- and 2.44-fold, respectively, compared with the wild-type DZNR (

Table 1. Whole-cell activity and relative activity of DZNR and its mutants.

Enzyme Name	Cell Activity (U/gwcw) (×10−3)	Relative Activity (%)
DZNR	12.30 (±1.12)	100
DZNR30S+105S	39.67 (±0.97)	322.58
DZNR30S+105C	30.01 (±1.76)	243.99

).

2.3. Characterization of Catalytic Properties of Mutants

Recombinant strains were induced under identical conditions, and SDS-PAGE was subsequently employed to assess the expression levels of DZNR and its variants. The results demonstrated that mutation did not significantly alter the expression levels of recombinant DZNR relative to the wild-type enzyme (Figure S1). To characterize the enzymatic properties of DZNR and its mutants, enzyme-catalyzed reactions were conducted across a range of temperatures and pH values. As shown in Figure 5, the optimal temperature for DZNR, DZNR30S+105S, and DZNR30S+105C against daidzein was 38 °C (Figure 5a). The suitable pH range for DZNR and both mutants was 4.0–6.0, with an optimum at pH 7.5 (Figure 5b).

Previous studies have demonstrated that (S)-equol exerts feedback inhibition on DZNR activity during the whole-cell conversion of daidzein to (S)-equol [20]. Accordingly, 0–2 mM (S)-equol was supplemented into the reaction mixture to evaluate the inhibitory effects on wild-type DZNR and its variants. As shown in Figure 6, compared with the control, wild-type DZNR activity was markedly inhibited in the presence of 1 mM (S)-equol, decreasing to 54% of its original value, and was essentially abolished at 2 mM (S)-equol. The mutant DZNR30S+105C exhibited a trend similar to that of wild-type DZNR: its activity was reduced to 57% at 1 mM (S)-equol and was completely abolished at 2 mM (S)-equol, paralleling the behavior observed for the wild-type enzyme. In contrast, mutant DZNR30S+105S displayed substantially enhanced tolerance to (S)-equol: at 1 mM (S)-equol, its activity remained essentially unchanged (90.2% of the original value), and even at 2 mM (S)-equol, 21.9% residual activity was retained, indicating significantly attenuated feedback inhibition relative to the wild-type enzyme.

The docking results provide a putative mechanistic explanation for the enhanced catalytic activity and attenuated product inhibition observed in the DZNR variants. The structural model predicts that Gly30 resides within a highly flexible loop region lining the interior of the active site (Figure 7a). Given that glycine lacks a side chain, this segment likely possesses exceptional conformational flexibility. The G30S mutation potentially stabilizes the loop conformation through the formation of a novel hydrogen-bonding network involving the introduced hydroxyl group and surrounding residues. Position 105, located at the entrance to the substrate-access tunnel, appears to become more hydrophilic upon mutation to serine or cysteine (A105S/A105C), presumably facilitating product release (Figure 7b,c). The enhanced tolerance of the double mutant toward (S)-equol may reflect reduced enzyme affinity for this isoflavonoid, attributable to the replacement of a hydrophobic residue with a polar serine at position 105. However, although molecular docking provides valuable structural insights, these computational predictions regarding loop stability, hydrogen-bonding network alterations, and quantitative binding affinity changes require further experimental validation to fully elucidate the underlying mechanisms.

2.4. Whole-Cell Performance of the Mutant

Based on the above results, the DZNR30S+105S variant, which exhibited the greatest improvement in catalytic activity, was employed in the whole-cell biocatalytic system M7 for the biotransformation of daidzein to (S)-equol. As shown in Figure 8, at a substrate concentration of 1 mM, the whole-cell catalyst harboring this variant achieved a 1.22-fold higher substrate conversion within 1 h compared to the control, reaching equilibrium within 5 h. Under these conditions, daidzein conversion reached 98.4%, with a product concentration of 238.3 mg/L (S)-equol (Figure 8a). Upon increasing the substrate concentration to 2 mM, the daidzein conversion at equilibrium (5 h) was 79.4% (Figure 8b). Although this conversion was lower than that observed at 1 mM, the final (S)-equol concentration reached 384.6 mg/L.

3. Discussion

Although enzyme engineering technologies have advanced considerably, systematic engineering of DZNR—particularly efforts to decipher and manipulate its distinct structural determinants governing catalysis and inhibition—remains scarce. In this study, we demonstrate for the first time that the catalytic performance of DZNR can be substantially enhanced through synergistic mutations at two functionally distinct sites: a flexible loop residue (Gly30) and a substrate-channel residue (Ala105). The resulting double mutant DZNR30S+105S exhibits not only a 3.2-fold increase in activity but, more importantly, a remarkable alleviation of product feedback inhibition—a critical bottleneck that has remained largely unresolved in previous DZNR studies. These findings provide unprecedented mechanistic insights and establish a dual-functional engineering paradigm that opens new avenues for DZNR optimization.

Semi-rational design strategies have been widely employed to engineer OYE family enzymes, targeting improvements in stability, catalytic activity, and substrate specificity [21]. These enzymes catalyze the highly stereoselective trans-hydrogenation of α,β-unsaturated aldehydes, ketones, carboxylates, nitriles, and nitroalkenes, rendering them valuable biocatalysts for asymmetric synthesis. However, DZNR exhibits distinct structural features that differentiate it from most OYE homologs, including a larger molecular mass (Figure 2) and pronounced oxygen sensitivity. These characteristics have historically impeded systematic engineering efforts on this enzyme. Consequently, the present study addresses not only the practical challenge of enhancing DZNR catalytic performance but also provides an opportunity to gain deeper insights into the catalytic mechanisms of this atypical OYE member. The dual mutants (DZNR30S+105S and DZNR30S+105C) exhibited significantly enhanced catalytic activity compared with the corresponding single mutants, suggesting that these two sites act in a cooperative and coordinated manner. The G30S mutation optimizes substrate binding through conformational stabilization of the active site, whereas the A105S/A105C mutation facilitates product release by altering the substrate-access channel. These modifications function synergistically to address the two major bottlenecks of substrate input and product output in the catalytic cycle, thereby maximizing overall catalytic efficiency.

Within the microbial metabolic network, the conversion of daidzein to (S)-equol proceeds through a multi-step reduction process. The reaction catalyzed by DZNR constitutes a critical step in this pathway and represents a recognized rate-limiting bottleneck. As a member of the OYE family, the catalytic efficiency of DZNR directly governs the metabolic flux toward (S)-equol. Notably, the catalytic behavior of DZNR can be highly context-dependent and even reversible, underscoring the complexity of its functional regulation. Hu et al. (2022) identified a novel DZNR (K-07020) from Clostridium sp. ZJ6, a chicken intestinal bacterium [20]. This enzyme exhibited significantly reduced catalytic efficiency for DHD production compared with characterized DZNRs. However, the distinctive feature of K-07020 lies in its bidirectional catalytic capability: it functions as a reductase to convert daidzein to DHD under NADH-dependent conditions, yet can also operate as an oxidase to reverse-catalyze DHD to daidzein under aerobic, cofactor-independent conditions [20]. This discovery reveals the catalytic plasticity of DZNR and highlights a potential challenge for biocatalytic applications. In contrast, the engineering strategy employed in the present study aimed to decouple catalysis from such conditional constraints. Deng et al. (2022) established an important benchmark for DZNR performance evaluation [22]. Through phylogenetic analysis of microbial libraries, they identified Ac-DZNR from Asaccharobacter celatus as the most active variant, achieving a product concentration of 180.1 mg/L (S)-equol in shake flask culture—substantially outperforming Si-DZNR from Slackia isoflavoniconvertens (56.7 mg/L) [22]. Relative to these benchmarks, the whole-cell conversion efficiency of the engineered DZNR30S+105S variant described herein surpasses that of naturally occurring DZNRs. The conversion yield increased from 182.6 mg/L to 238.3 mg/L within a 5 h reaction, representing a relative improvement of approximately 30%. More importantly, Deng et al. ultimately achieved a titer of 3418.5 mg/L through bioreactor optimization [22], demonstrating the substantial industrial potential of this biosynthetic pathway.

Future application of enzyme engineering technologies will further advance the synthesis efficiency of (S)-equol. Optimization and modification of the four enzymes in the (S)-equol pathway will enhance their mutual compatibility. For example, introduction of the T169A single mutation and the S118G/T169A double mutation in DHDR strengthens substrate–enzyme interactions, thereby improving substrate binding affinity and catalytic efficiency [23]. Additionally, the P212A mutation in DHDR enhances selectivity toward (S)-DHD under specific pH conditions, providing theoretical support for optimizing the (S)-equol biosynthetic pathway [24]. Furthermore, engineering of THDR, such as the P464A mutation, significantly enhances the productivity of 5-hydroxyequol, a key derivative of (S)-equol. In whole-cell biocatalytic systems, this mutation increases (S)-equol yield and effectively addresses the issue of THDR becoming rate-limiting during high-concentration substrate conversion, which leads to intermediate accumulation [25], thereby providing an essential catalytic tool for high-substrate-concentration biosynthesis of (S)-equol. Moreover, elucidation of the reversible catalytic mechanism of THDR has enabled rational design and engineering of its substrate channels, catalytic pockets, and water channels, yielding THDR variants with efficient unidirectional catalysis. This breakthrough resolves a critical bottleneck in (S)-equol biosynthesis, achieving highly efficient synthesis of this compound [26]. Ultimately, integration of pathway engineering, solvent optimization, and complementary strategies will enable efficient large-scale production of (S)-equol and its derivatives in the near future.

4. Materials and Methods

4.1. Conventional Materials and Strains

Genetic information was obtained from Li [27]. All genes were synthesized by Ruibo Biotech (Beijing, China) according to the codon preference of E. coli. Plasmid construction and proliferation were performed in E. coli DH5α, while E. coli BL21 (DE3) (Servicebio, Wuhan, China) served as the host for recombinant protein expression and metabolic engineering applications. All plasmids were constructed and maintained in E. coli DH5α, and then transferred to the host for expression. Seed cultures and routine cultivation were conducted using Luria–Bertani (LB) medium. To ensure plasmid stability, ampicillin sodium was supplemented to the culture medium at an appropriate concentration. Isopropyl-β-D-1-thiogalactopyranoside (IPTG), obtained from Solarbio Biotech (Beijing, China), was employed as the inducer for recombinant protein expression. The substrate daidzein and products dihydrodaidzein and (S)-equol were purchased from J&K Scientific Ltd. (Beijing, China). All restriction endonucleases, T4 DNA ligases and the corresponding reaction buffer kits were purchased from New England Biolabs (Ipswich, MA, USA). All other chemicals and molecular biology reagents were of analytical grade unless otherwise specified.

4.2. Construction of DZNR Mutants

The dznr gene encoding daidzein reductase (GenBank: AFV15453.1) was retrieved from the NCBI database and optimized according to the codon preference of E. coli. The codon-optimized sequence was chemically synthesized and subsequently cloned into the pETDuet-1 vector using BamH I and Sac I restriction sites to generate the recombinant expression plasmid pETDuet-1-dznr. This plasmid was transformed into E. coli BL21(DE3), yielding the recombinant strain E. coli BL21(DE3)/pETDuet-1-dznr.

In this study, we constructed various mutants at specific amino acid positions in DZNR to investigate site-specific functions. Here, 30Ala denotes the substitution of the amino acid at position 30 with alanine via site-directed mutagenesis. To further explore the function of this residue, we generated a semi-saturation mutagenesis library at position 30 by introducing the NDT degenerate codon (where N represents A/T/G/C, and D = A/T/G) at the corresponding codon in the gene. Similarly, a semi-saturation mutagenesis library was constructed at position 105 by introducing the NDT degenerate codon at the codon corresponding to this amino acid position. Site-directed mutations were introduced using the recombinant plasmid pETDuet-dznr as a template. The forward primer (F: ATAGCATGGCAACATATCTGAA) and the reverse primer (R: ATATGTTGCCATGCTATTACGAATC) were used to amplify the 30Ala gene at the mutation site. The 30 NDT gene was amplified using forward primers (F: ATAGCATGNDTACATATCTGAA) and reverse primers (R: ATATGTAHNCATGCTATTACGAATC). The 105NDT gene was amplified using forward primers (F: ATGCAGCTGNDTCATCCGGGTCG) and reverse primers (R: CCCGGATGAHNCAGCTGCATACCTG). The amplification of alanine series mutant or semi-saturated mutant gene sequences was carried out respectively by using whole plasmid PCR technology with Phusion High-Fidelity DNA Polymerase (NEB) (see Supplementary Information S1 for the screening method of the mutant library).

4.3. Induction Culture of Recombinant Strains

A single colony was inoculated into 1 mL of LB liquid medium supplemented with ampicilin (100 µg/mL) and cultured at 37 °C with shaking at 150 rpm for 12 h to prepare the seed culture. The overnight seed culture was transferred to 25 mL fresh medium at an inoculation ratio of 1% (v/v), with ampicillin added to a final concentration of 100 µg/mL. The culture was carried out at 37 °C and 220 rpm for approximately 3 h until OD₆₀₀ reached 0.8. Subsequently, 0.8 mM IPTG was added and the expression was induced at 24 °C and 220 rpm for 14 h. After induction, the culture was centrifuged at 8000× g for 3 min, and the bacterial sediment was washed twice with double-distilled water. The resulting cell pellet was resuspended and subjected to ultrasonic disruption. Soluble expression of recombinant enzymes was analyzed by SDS-PAGE, followed by Coomassie Brilliant Blue staining and destaining [28].

4.4. Whole-Cell Catalytic Activity Analysis of DZNR and Its Mutants

To investigate the variation in transformation efficiency over time, the induced expression of recombinant E. coli was resuspended in a 100 μL (OD₆₀₀ = 20) reaction system containing 200 mM sodium phosphate buffer (pH = 8.0), 1 mM daidzein, and 4% glucose. The reaction was performed at 30 °C, and samples were withdrawn at 1, 2, 3, and 5 h. The reaction was terminated by adding acetonitrile, and the products were analyzed by HPLC. Whole-cell activity was calculated based on product formation within the initial 2 h of the reaction. Cell biomass was determined by measuring wet cell weight (WCW). Cell suspensions were centrifuged (8000× g, 10 min), washed twice with distilled water, and weighed immediately to obtain the wet weight. The whole-cell enzyme activity unit (U) is defined as the amount of cells required to produce 1 μmoL of dihydrodaidzein per minute under conditions of 30 °C, pH 8.0, using daidzein as the substrate.

Induced recombinant E. coli cells (OD₆₀₀ = 20) were resuspended in reaction mixtures containing 200 mM sodium phosphate buffer (pH specified below), 1 mM daidzein, and 4% glucose. The optimal pH was determined using 100 μL reactions at pH 6.0–8.0 (30 °C, 5 h). The optimal temperature was assessed in 100 μL reactions (pH 8.0) incubated at 28–40 °C for 5 h. Product inhibition by (S)-equol was evaluated in 100 μL reactions (pH 8.0, 30 °C, 5 h) supplemented with 0–2 mM (S)-equol. Time-course analysis was conducted using 500 μL reactions (pH 8.0, 30 °C) with sampling at 1–5 h intervals. All reactions were terminated by adding an equal volume of acetonitrile and analyzed by HPLC.

4.5. High-Performance Liquid Chromatography Analysis

Analyses were performed on an Agilent 1260 Infinity II system equipped with a diode array detector (DAD). Samples were prepared by vortex mixing with an equal volume of acetonitrile, followed by centrifugation at 12,000× g for 2 min. The supernatant was then filtered through a 0.22 μm membrane and stored at −40 °C pending analysis. Chromatographic separation was performed using a Cosmosil-C18 column (250 mm × 4.6 mm, 5 μm) maintained at 30 °C. The mobile phase consisted of acetonitrile (B) and water (C). For the mutant transformation assay, the gradient elution program was as follows: 0–30 min, 95% B; 30–45 min, 50% B; 45–75 min, 35% B, at a flow rate of 1.0 mL/min. After the program, the flow rate was reduced to 0.6 mL/min to equilibrate the system for 15 min. For the whole-cell transformation assay, the gradient program was: 0–30 min, 95% B; 30–45 min, 70% B; 45–75 min, 60% B, at a flow rate of 1.0 mL/min, followed by a 15 min equilibration. The injection volume was 15 μL, and the detection wavelength was set to 280 nm. The conversion rate was calculated using the formula (product concentration/substrate concentration × 100%), where both concentrations were quantified from their respective standard curves. All data are presented as mean ± standard deviation (mean ± SD, n = 3).

4.6. Homology Modeling and Molecular Docking

The three-dimensional model of DZNR was built by SWISS-MODEL Server “http://swissmodel.expasy.org/,(accessed on 12 January 2025)” using the AlphaFold DB model of M9NZ71_9ACTN (gene: dzr, organism: Slackia isoflavoniconvertens) as a template. Predicted structure and the substrate daidzein were selected for the docking experiments with AutoDockTools 4.2. All the structure figures were prepared using Chimera (https://www.cgl.ucsf.edu/chimera/ (accessed on 12 January 2025)).

5. Conclusions

(S)-Equol is an isoflavone metabolite with diverse physiological activities. In this study, DZNR, the first committed enzyme in the (S)-equol biosynthetic pathway, was successfully engineered through a semi-rational design approach, resulting in substantially enhanced catalytic efficiency. Residues Gly30 and Ala105, located within the active site, were targeted for mutagenesis. Structural analysis suggests that modification of these sites likely optimizes the microenvironment of the substrate-binding pocket or modulates substrate orientation to enhance catalytic performance. Through this strategy, the DZNR30S+105S variant was successfully identified, representing a novel synergistic mutant that simultaneously alleviates the dual bottlenecks of limited catalytic efficiency and product inhibition. The performance of this variant was further validated in a whole-cell biocatalytic system. At a substrate concentration of 1 mM, recombinant E. coli expressing this variant achieved near-complete substrate conversion, with an (S)-equol yield of 238.3 mg/L, demonstrating excellent catalytic capability. More importantly, upon increasing the substrate concentration to 2 mM, the yield remained high at 384.6 mg/L. These results indicate that the variant possesses enhanced substrate tolerance and highlight its considerable potential for industrial-scale production of (S)-equol. Collectively, this work provides both an efficient biocatalyst and a strategic framework for future engineering of DZNR.