Engineering diaryl alcohol dehydrogenase KpADH reveals importance of retaining hydration shell in organic solvent tolerance

Abstract

Alcohol dehydrogenases (ADHs) are synthetically important biocatalysts for the asymmetric synthesis of chiral alcohols. The catalytic performance of ADHs in the presence of organic solvents is often important since most prochiral ketones are highly hydrophobic. Here, the organic solvent tolerance of KpADH from Kluyveromyces polyspora was semi‐rationally evolved. Using tolerant variants obtained, meticulous experiments and computational studies were conducted to explore properties including stability, activity and kinetics in the presence of various organic solvents. Compared with WT, variant V231D exhibited 1.9‐fold improvement in ethanol tolerance, while S237G showed a 6‐fold increase in catalytic efficiency, a higher ${\mathrm{T}}_{50}^{15}$, as well as 15% higher tolerance in 7.5% (v/v) ethanol. Based on 3 × 100 ns MD simulations, the increased tolerance of V231D and S237G against ethanol may be ascribed to their enhanced ability in retaining water molecules and repelling ethanol molecules. Moreover, 6.3‐fold decreased K _M value of V231D toward hydrophilic ketone substrate confirmed its capability of retaining hydration shell. Our results suggest that retaining hydration shell surrounding KpADH is critical for its tolerance to organic solvents, as well as catalytic performance. This study provides useful guidance for engineering organic solvent tolerance of KpADH and other ADHs.

1. INTRODUCTION

Biocatalysis in non‐conventional systems, such as organic solvents and ionic liquids, has attracted increasingly attentions. Biocatalytic reactions in aqueous‐organic biphasic systems or pure organic systems allow higher substrate loading, minimized spontaneous hydrolysis of water‐sensitive compounds, as well as favorable thermodynamics (Abdelraheem et al., 2019; Yang et al., ²⁰⁰⁴). Interestingly, some enzymes exhibited novel properties in organic solvents, such as significantly increased thermostability or super‐activity (Carrea & Riva, 2000; Wangikar et al., ¹⁹⁹⁷).

However, organic solvents could also influence structure and/or catalytic performance of enzymes. As widely accepted, organic solvents may influence the properties of enzymes via a variety of mechanisms. (I) Solvents may affect the activity by occupying substrate channels and active sites of enzymes, and result in competitive inhibition (Graber et al., 2007; Valivety et al., ¹⁹⁹³). Previous study reported that the bulky phenyl side chain could protect the active center from being occupied by DMSO molecules, which explains the higher tolerance of Bacillus subtilis Lipase A (BSLA) toward organic solvents (Kuper et al., 2012). (II) Organic solvents can influence the organic solvent resistance of enzymes through strengthening or weakening the structural flexibility (Dodson & Verma, 2006; Yenenler et al., ²⁰¹⁸). Structural features, including hydrophobicity of protein surface, tightness of loop regions, more salt bridges at interface, and increased hydrogen bonds and electrostatic interactions, could increase the “rigidity” of protein. These structural features are considered to help improve the stability of enzymes in organic solvents (Bezsudnova et al., 2012; Karabec et al., ²⁰¹⁰). (III) The hydrophobicity of organic solvents often has a massive effect on enzyme activity, and several studies have proved that lipases exhibited higher activity in solvents with higher logP values (Liu, Tan, et al., 2010; Liu, Zhang, et al., ²⁰¹⁰; Su & Wei, ²⁰⁰⁸). Enzymes usually require a certain amount of water molecules in their structures as bounded waters, to maintain their natural conformation and flexibility required for catalytic functionality (Rezaei et al., 2007; Zhao et al., ²⁰¹³). Hydrophilic solvents have a greater tendency to “strip” tightly bound water from the surface of enzymes, leading to changes in scaffold structure. Previous findings support a hypothesis that water molecules can significantly influence conformational dynamics of proteins and facilitate transition toward catalytic conformations. Consequently, essential waters are vital for the physiological and catalytic functions of enzymes (Zhao et al., 2013). The flexibility of proteins is considered to be most related with local interfacial viscosity, or kinetic ratio of fluidic organic solvents to water in regional solvation layer (Bellissent‐Funel et al., 2016; Dahanayake & Mitchell‐Koch, ²⁰¹⁸). In reaction system lacking appropriate amount of essential water, enzymes will lose flexibility and become inactivated. As the water content increases, enzymes become more flexible and display increased activity (Tsuzuki et al., 2003; Wang et al., ²⁰¹⁶). Water molecules only interact with charged functional groups and certain polar amino acid clusters. As a result, surface electrification engineering strategy by substituting amino acids with polar or charged amino acids will enhance hydration degree and change the equilibrium of protein motions which is beneficial to organic solvent resistance (Cui et al., 2020; Oroguchi & Nakasako, ²⁰¹⁶).

Striking catalytic performance of enzymes can be achieved through combined solvent and protein engineering (Dordick, 1992; Martin‐Diaz et al., ²⁰²¹). The former involves changing the polarity, hydrophobicity, and water content of an organic environment (Stepankova et al., 2013), while the latter applies site‐directed mutagenesis and immobilization to change the physical and chemical properties of enzymes (Kartal et al., 2011; Zumarraga et al., ²⁰⁰⁷). In 1990s, Arnold and co‐workers reported the protein engineering of Bacillus subtilis protease E to improve enzymatic activity and stability in organic solvents (Arnold, 1993; Arnold et al., ¹⁹⁹³; Martinez & Arnold, ¹⁹⁹¹; Moore & Arnold, ¹⁹⁹⁶; You & Arnold, ¹⁹⁹⁶). Tian et al. (2017) performed directed enzyme evolution to enhance methanol tolerance by targeting high B‐factor residues for iterative saturation mutagenesis (ISM). Highly organic‐solvent‐tolerant enzymes were obtained through recombination of compatible substitutions (Cui, Jaeger, et al., 2021; Cui, Pramanik, et al., ²⁰²¹). Libraries of epPCR and sequence saturation mutagenesis (SeSaM‐Tv) were developed by Ulrich and coworkers for amino acid substitutions that increase resistance toward 1,4‐dioxane (DOX), 2,2,2‐trifluoroethanol (TFE), and dimethyl sulfoxide (DMSO) of BSLA. The results showed that 5%–11% of all possible single substitutions in site‐saturation mutagenesis library contribute to the improved solvent resistance, and introducing charged residues on protein surface is important for the organic solvent resistance (Markel et al., 2017). Therefore, surface charge engineering is regarded as a basic principle in the engineering solvent tolerance of enzymes (Pedersen et al., 2019). Cui, Eltoukhy, et al. (2021) reported a smart salt bridge design strategy for simultaneously improving solvent resistance and thermostability of BSLA. And structure‐based strategy through rational surface engineering was explored to stabilize BSLA in water–ethanol system (Min et al., 2021).

However, there has been few reports on the engineering and mechanism of organic solvent tolerance of alcohol dehydrogenases (ADHs). ADHs is typically the first choice for catalyzing the asymmetric reduction of carbonyl compounds to their corresponding alcohols due to its high evolvability, easy operation and 100% theoretical yield (Zhou et al., 2020). Among them, asymmetric reduction of “difficult‐to‐reduce” diaryl ketones to produce enantiopure diaryl alcohols is of fundamental importance in organic synthesis and medicinal chemistry (Gladiali & Alberico, 2006; Morris, ²⁰⁰⁹; Sun et al., ²⁰¹⁶). As a typical diaryl ketone, (4‐chlorophenyl)(pyridine‐2‐yl) ketone (CPMK), its asymmetric reduction product (S)‐(4‐chlorophenyl)‐(pyridin‐2‐yl) methanol [(S)‐CPMA] is a key intermediate for synthesizing antiallergy drug bepotastine (Zhou et al., 2018). In our previous study, KpADH from Kluyveromyces polyspora was discovered by genome mining and showed excellent substrate tolerance and moderate ee value of 82% (R) toward CPMK (Zhou et al., 2016). To improve the stereoselectivity of KpADH, hydroclassified combinatorial saturation mutagenesis and polarity scanning have been proposed and validated to be efficient in engineering variants with elevated and inversed enantioselectivity (Xu et al., 2018).

However, organic solvents are required as co‐solvents to improve the solubility of hydrophobic substrates (such as CPMK) in reactions. In this work, rational engineering was attempted to improve the organic solvent tolerance of KpADH. Catalytic kinetic parameters of the beneficial variants toward both hydrophobic and hydrophilic substrates in the presence of ethanol were characterized. MD simulation was also conducted to understand the mechanisms of improved solvent tolerance of KpADH.

2. MATERIALS AND METHODS

2.1. Chemical reagents

(4‐Chlorophenyl)‐pyridin‐2‐yl‐ketone (CPMK) and methyl N‐Boc‐4‐hydroxypiperidine‐2‐carboxylate (NBPO) were purchased from Heowns Co. Ltd (China). Methanol and ethanol with a purity of 99.7% were purchased from Shanghai Titan Scientific Co. Ltd. (China). TFE, DMF, DOX, DMSO of analytical purity were obtained from Shanghai Macklin Biochemical Co. Ltd. (China).

2.2. Site‐directed mutagenesis and expression and purification of KpADH

Variants were generated by PCR procedure using plasmid pET28a‐KpADH as the template, and the corresponding primers are listed in Table S1. The resultant PCR products were digested with DpnI and further transformed into E. coli BL21 (DE3). Recombinant expression and purification process were as reported in our previous work (Xu et al., 2018).

2.3. Standard enzyme activity assay protocol and organic solvent tolerance determination

Activity of KpADH was calculated based on the consumption of coenzyme NADPH as determined by optical changes at 340 nm. The 200‐μL reaction system consisted of 10 μL substrate CPMK, 10 μL NADPH, 10 μL purified enzyme of appropriate concentration, and 170 μL sodium phosphate buffer (PBS) (pH 6.0, 100 mM). For activity assay, PBS buffer was preheated in a 30°C water bath for 5 min, followed by addition of all other components, then OD_340nm was monitored. The enzyme activity (U) is defined as the amount of enzyme needed to catalyze the oxidation of 1 μmol NADPH per min in the above systems.

Relative activity of KpADH (wild type [WT] or variant) was calculated as the activity in higher concentrations of ethanol (or methanol, TFE, DMF, DOX, DMSO) divided by activity in 5% (v/v) ethanol. The following Equation (1) was used:

$$\text{Relative activity}=\begin{array}{l}\frac{\text{Activity}\left(\mathrm{WT}/\text{variant}\right)\text{in higher concentrations of solvents}}{\text{Activity}\left(\mathrm{WT}/\text{variant}\right)\text{in}5\%\text{of ethanol}}\times 100\%.\end{array}$$ (1)

Organic solvents tolerance fold was assessed as Equation (2):

$$\text{Solvent tolerance fold}=\frac{\text{Relative activity(variant)}\text{in certain concentration of solvents}}{\text{Relative activity}\left(\mathrm{WT}\right)\text{in certain concentration of solvents}}.$$ (2)

2.4. Determination of kinetic parameters

Kinetic parameters were calculated based on initial reaction rates of purified WT and its variants at different concentrations of CPMK and NBPO ranging from 0.005 to 0.25 mM. Kinetic parameters were obtained by fitting the data with the Michaelis–Menten equation using software Origin 9.0. All assays were performed in triplicate.

2.5. Molecular dynamics simulations

The crystal structure of KpADH‐WT (PDB ID: 5Z2X) was used as the template to generate homology models of V231D and S237G using Discovery Studio. Structure model with the lowest total energy was selected for further experiments. The relative folding free energies (ΔΔG_fold = ΔG_fold,mut − ΔG_fold,WT) were calculated using FoldX (Cui, Jaeger, et al., 2021; Guerois et al., ²⁰⁰²). Molecular dynamics (MD) simulations of KpADH at 303 K were performed using GROMACS 2016. Detailed procedures of MD simulation were as following: (Abdelraheem et al., 2019) generate topology files of protein using pdb2gmx module and all‐atomic force field (OPLS‐AA); (Yang et al., 2004) define a 1000 nm³ cubic box using editconf module and place the protein in the center of the box; (Carrea & Riva, 2000) an approximate number (~1549 ethanol molecules) of ethanol molecules were added to the box and then the system was filled with SPCE water model (~25,563 water molecules); (Wangikar et al., 1997) the Genion module was used to add counter‐ions to neutralize the total net charge of the simulation systems (Graber et al., 2007) before simulations, energy minimization was performed using the steepest descent method in order to ensure that the structure of the system is normal, the geometry was reasonable, and no atoms was collided; (Valivety et al., 1993) the solvents and ions around the protein were pre‐balanced by 100 ps isothermal isovolumetric ensemble (NVT) and 100 ps isothermal isobaric ensemble (NPT) at 30°C; (Kuper et al., 2012) after pre‐equilibrium phase, the system was under appropriate temperature and pressure, the positions restriction of protein were released, and then 100 ns MD simulations were carried out; (Dodson & Verma, 2006) GROMACS simulation package tools were used for the subsequent analysis. Different replicates of 3 × 100 ns MD simulations were starting from different initial velocities or coordinates.

2.6. Asymmetric reduction of CPMK using KpADH and variants

A 20‐mL reaction mixture contained 0.3 U/mL of KpADH or variants cell‐free extract, 100 mM CPMK, 1 U/mL of GDH lyophilized enzyme, 150 mM glucose, 1 mM NADP⁺, and 5% ethanol in PBS buffer (pH 6.0, 100 mM). The reaction was magnetically stirred at 300 rpm and 30°C, and maintained at pH 6.0 by titration with 1 M NaHCO₃. At time interval, 100 μL sample was retrieved and mixed with 900 μL PBS and 1000 μL ethyl acetate. The mixture was thoroughly extracted, and then 400 μL upper organic phase was withdrawn to evaporate the solvent, and further dissolved using 400 μL ethanol. The conversion ratio was analyzed using HPLC equipped with OB‐H column at 254 nm. The conversion ratio was calculated as Equation (3):

$$\text{Conversion ratio}\left(\%\right)=\frac{\text{Product CPMA}\left(\mathrm{mM}\right)}{\text{Initial substrate CPMK}\left(\mathrm{mM}\right)}\times 100\%.$$ (3)

3. RESULTS AND DISCUSSION

3.1. Effect of organic solvents on activity of KpADH

KpADH is a promising biocatalyst for the synthesis of chiral diaryl alcohols through asymmetric reduction of bulky–bulky ketones. However, ketones such as CPMK, are highly hydrophobic and sparingly soluble in water, and are regarded as hard‐to‐be‐reduced due to their steric hindrance. To increase the collision probability between the enzyme and CPMK, co‐solvents need to be added in the reactions. Initially, the influence of various organic solvents as co‐solvent on KpADH was evaluated, including methanol (logP: −0.76), ethanol (logP: −0.24), TFE (logP: 0.54), DMF (N, N‐dimethylformamide, logP: −1.00), DOX (dioxane, logP: −0.42), and DMSO (dimethyl sulfoxide, logP: −1.35). As shown in Figure 1a, the highest specific activity of 26 U mg⁻¹ was determined with methanol, which features a relatively higher polarity and MW of 32.04 g/mol. The lowest specific activity of 0.3 U mg⁻¹ was observed in DMSO, which has the highest polarity and relatively higher MW of 78.13 g/mol. It has been reported that organic solvents with S=O group (such as DMSO) could have quite different effects on enzyme activity and structure compared with aliphatic alcohols (Roy et al., 2012). Figure 1b illustrates the relative activities of WT in the presence of 0%–20% (v/v) of above solvents. Notably, the relative activity decreased in a concentration‐dependent manner, in which higher solvent concentrations resulted in lower activities. Here, the solvent concentration corresponding to 50% relative activity of KpADH was defined as semi‐inactivation concentration (SIC). The highest SIC of 11.5% was determined with DOX. The SICs of TFE and methanol were about 7.5%, while ethanol was about 2.5% lower than that of TFE and methanol. The lowest SIC of 2% was obtained with DMF and DMSO. The SIC values of various solvents were ranked as follows: DOX > TFE ≈ methanol > ethanol > DMF ≈ DMSO. It is common knowledge that DMF and DMSO are aprotic while methanol and ethanol are protic solvents. More subdivided, DMF and DMSO are dipolar aprotic solvents. Therefore, dipolar aprotic solvents are more destructive to the activity of KpADH than protic solvents. Although the relative activities under DOX and TFE were higher than those under methanol and ethanol, the specific activities of KpADH in the presence of methanol and ethanol were much higher (Figure 1a). Therefore, methanol and ethanol are more biocompatible for the asymmetric reduction of CPMK catalyzed by KpADH. Considering economic efficiency and the environmental benignity, ethanol was chosen as the model solvent for subsequent experiments.

FIGURE 1.

(a) Specific activity of KpADH toward CPMK in various solvents of 5% (v/v); (b) relative activity of KpADH toward CPMK in solvents of different concentrations. To maintain consistency with other solvents, 100% initial activity was defined as the enzyme activity in 5% (v/v) ethanol due to the insolubility of CPMK in water. All data shown are average values from measurements in triplicates or more.

3.2. Mutagenesis of KpADH for enhanced organic solvent tolerance

In order to achieve better understanding of the mechanism of solvent tolerance, KpADH was engineered for improved organic solvents tolerance. Based on the overall crystal structure of KpADH (PDB: 5Z2X), loop region of 228–235 was identified, which locates above the active center (Figure 2). This region is consisted of polar or charged amino acids except for V231, which is a highly hydrophobic residue. Therefore, saturation mutagenesis was performed at 231. Variant V231D (polar amino acid mutation to retain water molecules) was identified with improved solvent tolerance (Figure S1). Furthermore, the downward extending hydrophobic loop 236–239 was also analyzed and a polar residue S237 was observed. It was presumed that the opposite characteristic of S237 with other residues on loop 236–239 might influence the stability of the microenvironment. Hence, nonpolar amino acids were introduced at site 237 to repel polar solvents molecules. Interestingly, variant S237G was obtained with improved solvent tolerance.

FIGURE 2.

3D structures of catalytic triad Ser 126‐Tyr 164‐Lys 168 (blue) from different viewing angles. Residue 231 is shown in red stick, and 237 in magenta stick.

Furthermore, influence of different organic solvents on the relative activity of WT and variants were investigated and illustrated in Figure 3. Overall, increased organic solvents caused a continuous decreasing in relative activity of variants. It appears that effect of various organic solvents on enzyme is primarily related to their hydrophobicity. Although both methanol and ethanol are alcoholic solvents, the relative activities of WT and S237G are similar at varied concentrations of methanol (Figure 3a). For ethanol, however, the relative activities of S237G and WT are 69% and 84% at 7.5% (v/v) ethanol, respectively (Figure 3b ). In order to highlight the differences, the concentration range of 0%–5% is omitted in Figure 3b. To demonstrate the improved ethanol tolerance of the variants, relative activity at below 5% ethanol was evaluated in Figure 3g. Remarkably, V231D displayed significantly improved tolerance against methanol and ethanol than WT. The relative activity of V231D was about 20% higher than that of WT at 7.5% (v/v) methanol and ethanol.

FIGURE 3.

(a–f) Relative activity toward CPMK of KpADH and its variants in (a) methanol, (b) ethanol, (c) DMSO, (d) DOX, (e) DMF, and (f) TFE; (g) Relative activity toward NBPO of KpADH and its variants in ethanol. Due to the insolubility of CPMK in water, 100% initial activity was defined as the enzyme activity in 5% (v/v) ethanol. All data shown are average values from measurements in triplicates or more.

With respect to DMSO system, S237G and V231D displayed better performance compared with methanol and ethanol (Figure 3c). At 5% (v/v) DMSO, the relative activity of V231D was 34% higher than WT. This result indicates that V231D possesses better solvent tolerance in DMSO system compared with methanol and ethanol systems. Additionally, among three organic solvents, the relative activities of WT and variants were similar at enhanced solvent concentrations up to 10% (v/v), suggesting a general damaging effect at higher concentrations. Compared with WT, variants S237G and V231D have improved organic solvent tolerance toward all three solvents. Nevertheless, due to distinct chemical and physical properties of organic solvents (methanol, ethanol, and DMSO), different improvements in solvent tolerance were observed for S237G and V231D. However, in DOX and TFE systems, the relative activity of WT is higher than that of two variants, where S237G exhibited 20%–30% higher relative activity than V231D (Figure 3d,f). In 5% (v/v) DMF system, WT, S237G, and V231D have similar relative activities of around 25% (Figure 3e). However, the relative activity of S237G at the second point is greater than 100% in Figure 3e,f. This may due to the higher thermostability of S237G (Table 1), and therefore better conformational stability. At lower concentrations of DMF and TFE, the conformational flexibility of S237G could be enhanced, leading to certain activation effect on enzyme activity.

TABLE 1.

Solvent stability properties of KpADH and variants.

Enzyme	PBS buffer (pH 6.0, 100 mM)		10% (v/v) ethanol		Relative activity a (%)
	${\mathrm{T}}_{50}^{15}$ (°C)	E d (kJ mol−1)	${\mathrm{T}}_{50}^{15}$ (°C)	E d (kJ mol−1)
WT	42.1 ± 0.3	769.7	37.5 ± 0.4	573.5	69.3 ± 3.1
V231D	32.0 ± 0.5	402.5	27.0 ± 0.6	191.3	93.1 ± 2.7
S237G	45.3 ± 0.2	854.5	40.0 ± 0.5	631.7	84.0 ± 2.5

^a Relative activity (%) = specific activity in 7.5% (v/v) ethanol/specific activity in 5% (v/v) ethanol.

Compared with other solvents, ethanol has better biocompatibility and is more environmentally friendly. Therefore, further study was conducted using ethanol as co‐solvent. As shown in Table 1, S237G has a ${\mathrm{T}}_{50}^{15}$ value of 2.5°C higher than that of WT, and a slightly higher relative activity in 7.5% (v/v) ethanol system than that of WT. Site 237 is adjacent to the active center, and could affect the enzymatic activity and spatial conformation of catalytic center. Although relative activity of V231D was 24% higher in 7.5% (v/v) ethanol than that of WT, V231D has a ${\mathrm{T}}_{50}^{15}$ value of 10.1°C lower. The energy of deactivation (E _d) values calculated for WT (769.7 kJ mol⁻¹), V231D (402.5 kJ mol⁻¹) and S237G (854.5 kJ mol⁻¹), representing their differential thermostability in PBS buffer. Compared with PBS buffer system, both ${\mathrm{T}}_{50}^{15}$ and E _d values decreased in 10% (v/v) ethanol system, regardless of WT or variants. This result indicates that variant S237G with enhanced thermostability is not robust enough to resist the disruption of enzyme stability caused by organic solvents.

Kinetic parameters toward CPMK in the presence of ethanol were determined. WT and variants displayed similar organic solvent tolerance patterns. Compared with 5% (v/v) ethanol, WT and variants exhibited up to 2.0‐fold elevated K _M value while around 1.5‐fold reduced k _cat value in 10% (v/v) ethanol. The increased K _M and decreased k _cat resulted in 2.1–3.0 fold decrease in catalytic efficiency (k _cat/K _M) in 10% ethanol system, indicating higher (10%) ethanol could affect the catalytic efficiency of KpADH. This result suggests a dramatically decreased substrate affinity of V231D toward hydrophobic CPMK. The substitution of Val with Asp at 231 could result in a greatly enhanced polarity of loop 236–239 above the active center, thus affecting the entry of hydrophobic CPMK into active center (Table 2).

TABLE 2.

Kinetic parameters of KpADH and variants in the presence of ethanol.

Substrate	Enzyme	Ethanol (v/v)	K M (mM)	k cat (s−1)	k cat/K M (s−1 mM−1)
CPMK	WT	5%	2.29 ± 0.80	32.05 ± 6.27	14.00
	V231D		22.74 ± 5.20	12.81 ± 1.29	0.56
	S237G		1.07 ± 0.40	97.70 ± 20.36	91.56
	WT	10%	3.13 ± 2.45	20.82 ± 3.83	6.65
	V231D		44.23 ± 1.95	8.68 ± 1.36	0.196
	S237G		2.15 ± 1.59	72.05 ± 13.12	33.51
NBPO	WT	5%	24.50 ± 7.36	33.78 ± 8.73	1.38
	V231D		3.62 ± 0.72	1.71 ± 0.18	0.47
	S237G		14.70 ± 4.22	265.79 ± 60.35	18.08

Compared with WT, variant V231D exhibited improved organic solvent tolerance fold, specifically 1.5‐fold in 10% (v/v) methanol, 1.9‐fold in 15% (v/v) ethanol, and 2.3‐fold in 5% (v/v) DMSO. In addition, S237G features improved organic solvent tolerance, as well as better thermostability than WT. Hence, V231D and S237G are two interesting mutants for understanding the mechanism of organic solvent tolerance of KpADH.

3.3. MD simulations of KpADH and variants

MD simulation provides insights into molecular mechanism at nanosecond scale with improved accuracy (Dodson & Verma, 2006; Radhakrishnan et al., ²⁰²²; Rezaei et al., ²⁰⁰⁷). To explore the mechanism of solvent‐tolerance variants, 100 ns MD simulations of KpADH and its variants (V231D and S237G) were performed three times (Figures S2–S5). At 15% (v/v) ethanol, WT and the variants could retain 25%–40% relative activity, while at higher ethanol concentrations, they were almost inactivated. Furthermore, at lower ethanol concentrations, the proportion of ethanol molecules in MD simulation system will become lower, leading to compromised effect of ethanol on the enzyme, which is not conducive to MD analysis. Therefore, 15% (v/v) ethanol concentration was adopted in MD simulation.

Radius of gyration (R _g) reflects the overall volume and density of the protein, the higher the R _g value, the more bulky the protein is. For all the MD simulation results, the maximum differences of R _g values are within 0.4 Å (Figure 4a). The results showed that substitutes at sites 231 and 237 did not change the overall structural compactness of WT and variants. In addition, the root mean square deviation (RMSD) value represents the overall structural flexibility of the protein (Figure 4b). Consistent with R _g values, the fluctuation of RMSD values of WT and variants are within the acceptable standard deviation of 5 Å (the same is true for MD simulations with varied filling solvents). R _g and RMSD results show that the obtained variants (V231D and S237G) could maintain the overall integrity and conformation in the presence of ethanol as co‐solvent. In MD simulation, no significant changes were observed in R _g and RMSD values in the presence of pure water or 15% (v/v) ethanol. This further supports that small amount ethanol has minimal impact on stability of KpADH.

FIGURE 4.

(a) R_g and (b) RMSD of protein backbone atoms; (c) RMSF of Cα atoms of each residue; (d) SASA for WT, V231D, and S237G in pure water and 15% (v/v) ethanol system. MD simulations were performed at 303 K. RMSF was calculated over the last 50 ns. All MD data shown are average values from three independent runs.

Moreover, the flexibility of each residue in the protein can be characterized by the root mean square fluctuation (RMSF) values. As shown in Figure 4c, due to the change of hydrophobicity/hydrophilicity in microenvironment, the RMSF value of certain amino acid residue in 15% (v/v) ethanol system might be different compared with that in pure aqueous system. However, compared with WT, the RMSF value of variants V231D and S237G did not change much around the mutational regions. Therefore, the obtained solvent‐tolerant variants could maintain the overall and localized conformation of the protein, such as flexibility, rigidity etc. In addition, ΔΔG_fold of V231D and S237G were calculated using FoldX. Compared with WT, V231D is highly destabilized with ΔΔG_fold value of 4.67 kcal mol⁻¹ which is consistent with its decreased thermostability. For S237G, there is a neutral effect on the stability with ΔΔG_fold value of −0.31 kcal mol⁻¹. Consistent with computational result of Computer‐Assisted Recombination (CompassR), combinatorial variant V231D/S237G did not show better performance (Table S2) (Cui, Jaeger, et al., 2021).

Solvent accessible surface areas (SASAs) play an important role in understanding how ethanol affects enzyme activity. It was observed that decrease of solvent polarity results in the exposure of hydrophobic residues of the protein. The increased solvent accessible surface area will affect the conformation of the entire protein. For WT and variants, SASAs were slightly increased for about 5 nm² in 15% (v/v) ethanol system compared with those in pure water system (Figure 4d). And the enhanced SASAs in ethanol system suggest that the change of solvent molecules around the mutation site may be responsible for the enhanced ethanol tolerance of the variants.

3.4. Importance of hydration shell on ethanol tolerance of KpADH

The mechanisms of surface charged substitutions have been reported in a broad range of enzymes (Cui et al., 2020; Jakob et al., ²⁰¹³; Nordwald et al., ²⁰¹⁴; Warden et al., ²⁰¹⁵). Based on systematic computational studies, surface hydration shell leads to retention of essential water molecules, and thus improves the organic solvent tolerance of enzymes (Cui et al., 2020; Pramanik et al., ²⁰²²). To understand the importance of surface hydration shell, distribution of water molecules around WT and variants (V231D and S237G) was statistically counted. The number of water molecules whose O atom is within 3.5 Å of any non‐hydrogen atom of residues was determined as the hydration shell level (Cui et al., 2022; Wedberg et al., ²⁰¹²). Spatial distribution of water occupancy at the surface of KpADH and variants is visualized in Figure S6. Compared with MD simulation in pure water, addition of 15% (v/v) ethanol obviously destroyed the formation of surface hydration layer by occupying the original region of water molecules. However, the water stripping behavior by ethanol is dynamic according to analysis of water molecules distribution in different MD frames in 15% (v/v) ethanol (Figure S6). As shown in Table 3, 10.3 water molecules around site 231 were observed in V231D, which is significantly higher than 8.9 of WT. In order to verify whether the increased number of water molecules is caused by the ionization of Asp, the number of water molecules around V231E, in which Asp is replaced by acidic Glu, was also calculated. It was interesting to find that V231E, which exhibited the same ethanol tolerance as WT (Table S2), has a higher water molecule number of 13.2 than that of V231D (10.3). However, the water molecular retention rate (number of water molecules around certain residue in 15% [v/v] ethanol/in pure water) of V231D is 99%, while that of WT and V231E is 74% and 75%, respectively (Table 3). These results indicate that beneficial variant V231D has enhanced capability to retain surface water molecules in 15% (v/v) ethanol than WT and V231E.

TABLE 3.

Total number of water molecules within 3.5 Å distance of residues 231 and 237 of KpADH and variants in 15% ethanol and without ethanol.

	Without ethanol	15% (v/v) ethanol	Water molecular retention rate (%)
WT‐231	8.9 ± 0.3	6.6 ± 0.2	74
V231D‐231	10.3 ± 0.2	10.2 ± 0.3	99
WT‐237	6.9 ± 0.4	3.7 ± 0.3	54
S237G‐237	5.1 ± 0.2	3.3 ± 0.2	65

To further confirm the importance of increased surface water retention rate for the ethanol tolerance of V231D, kinetic parameters of two different substrates (hydrophobic CPMK and hydrophilic NBPO) catalyzed by KpADH and variants were analyzed (Table 2). Since NBPO has excellent water solubility, it was assumed that NBPO can easily penetrate the hydrated shell around site 231 above the catalytic center (Figure 2). Opposite to NBPO, CPMK could not penetrate the hydrated shell easily. Thus, the ability of substrate penetration through hydration layer could largely represent the substrate affinity of the enzyme.

As shown in Table 2, the K _M value of V231D toward CPMK is 22.74 mM in 5% (v/v) ethanol, representing about 11 times higher than that of WT (2.29 mM). Therefore, V231D has a much lower substrate affinity for CPMK than WT. In contrast, the K _M value of V231D toward NBPO is 3.62 mM in 5% (v/v) ethanol, which is only one seventh of that of WT (24.50 mM). However, compared with WT, V231D exhibited 19.8‐fold decrease in k _cat toward NBPO, leading to 2.9‐fold decreased catalytic efficiency. It was assumed that mutation at 231, located within a loop region proximal to the active site of KpADH, may alter the loop conformation and affect its catalytic performance.

Compared with hydrophobic CPMK, a greater affinity toward hydrophilic NBPO was observed with V231D than WT. Therefore, it is presumably the stronger hydration layer formed by V231D blocks CPMK from accessing the active center. In addition, compared with 5% (v/v) ethanol system, K _M values of both WT and variants toward CPMK showed a further 1.4–2.0 folds increase in the presence of 10% (v/v) ethanol, indicating that higher ethanol content resulted in reduced enzyme catalytic efficiency by disrupting substrate affinity (Table 2).

For variant S237G with enhanced hydrophobicity in active center, the number of water molecules around site 237 was also calculated (Table 3). In pure water, the number of water molecules is 5.1 in S237G, which is lower than 6.9 of WT. In 15% (v/v) ethanol, however, the number of water molecules is 3.3 in S237G, almost equal to that of WT (3.7). The results suggest that the substitution of hydrophobic Gly at 237 could not only block the entry of ethanol molecules, but also water molecules. However, variant S23G also exhibited higher water molecular retention rate of 65% than that of WT (54%) around site 237. In addition, S237G has improved solvent tolerance, as well as 6.5‐fold improved catalytic efficiency (k _cat/K _M) toward CPMK in 5% (v/v) ethanol than WT. For NBPO, S237G exhibited an 8.0‐fold increased k _cat and a 1.7‐fold decrease in K _M, leading to 13.0‐fold improved catalytic efficiency than WT (Table 2).

Consequently, as validated by MD simulation and kinetic parameters, retention water molecules could be enhanced by improving hydration around the substituted residues, and increased hydrophobicity of catalytic center is conducive to repel the entry of solvent molecules.

3.5. Asymmetric preparation of CPMA employing KpADH variants

Preparation of chiral CPMA catalyzed by KpADH and variants was carried out at 30°C, along with glucose dehydrogenase from Bacillus megaterium (BmGDH) for coenzyme regeneration. Initially, the reactions were conducted in 20‐mL systems containing 100 mM CPMK and 5% (v/v) ethanol. As shown in Figure 5a, the reaction rate of S237G is obviously higher than that of WT, although a complete conversion was reached by both WT and S237G. Variant S237G catalyzed the complete conversion within 1 h, when about 75% was achieved by WT. Due to the relatively poor thermostability of V231D, the enzymatic activity was seriously damaged after 1 h, resulting an incomplete conversion. For KpADH, the increased solvent tolerance does not correspond to improved thermostability. Thus, reactions at reduced temperature of 16°C were conducted to evaluate the catalytic ability of V231D. In Figure 5b, the final conversion ratios of WT and both variants were about 90% at 12 h. Whereas, the overall reaction rate of V231D is faster than that of WT. S237G also showed a better reaction rate than V231D at 16°C.

FIGURE 5.

Biocatalytic preparation of chiral CPMA employing KpADH and variants at (a) 30°C and (b) 16°C.

Since V231D was characterized with a higher ΔΔG_fold of 4.67 kcal mol⁻¹ compared with WT (0 kcal mol⁻¹), the substitution of hydrophobic Val with acidic Asp at 231 may increase the flexibility of loops 236–239, and have negative effect on thermostability. It further confirms that the thermostability of the enzyme is not necessarily related to its solvent tolerance. In addition, the space–time yields of WT, V231D and S237G were calculated to be 0.7, 1.0, and 1.3 g L⁻¹ h⁻¹ at 16°C, respectively. Consequently, solvent‐tolerant variants V231D and S237G have better catalytic performance than that of WT. Improved hydration shell by semi‐rational engineering may represent an alternative approach for improving solvent tolerance of KpADH and other ADHs.

4. CONCLUSIONS

Organic solvent‐tolerant variants S237G and V231D of KpADH were developed by semi‐rational engineering. Remarkably, V231D showed 20%–34% improved tolerance in different organic solvents. Interestingly, S237G possesses enhanced solvent tolerance, as well as improved thermostability along with 6.5‐fold higher catalytic efficiency toward CPMK compared with WT. It was noted that ethanol affect the activity of KpADH via replacing the surface essential water. MD simulation reveals that enhanced ability of retaining hydration shell is critical to the improved solvent tolerance. Additionally, increased hydrophobicity of catalytic center in S237G plays a key role in its enhanced stability.

Water molecules account for a large proportion of the crystal structures of most ADHs. Also, water molecules facilitate hydride transfer during the oxidative‐reductive reactions of alcohol dehydrogenases/reductases (Li et al., 2020). Thus, it is necessary to form stronger water barrier to prevent interference of solvent molecules surrounding the catalytic pocket. Our study allows a better understanding of mechanisms of solvent tolerance, and reveals retaining hydration shell is crucial to the organic solvent tolerance of KpADH and other ADHs.

AUTHOR CONTRIBUTIONS

Lu Zhang: Methodology; data curation; investigation; validation; formal analysis; writing – original draft. Wei Dai: Methodology. Shuo Rong: Data curation. Ulrich Schwaneberg: Methodology; writing – review and editing. Guochao Xu: Conceptualization; methodology; supervision; funding acquisition; writing – review and editing Ye Ni: Conceptualization; methodology; funding acquisition; writing ‐ review and editing; project administration; supervision; resources..

FUNDING INFORMATION

This work was financially supported by the National Key Research and Development Program (2019YFA0906400), and the National Natural Science Foundation of China (22077054, 22078127).

CONFLICT OF INTEREST STATEMENT

The authors declare no competing financial interest.

Supporting information

Data S1. Supporting Information.

Zhang L, Dai W, Rong S, Schwaneberg U, Xu G, Ni Y. Engineering diaryl alcohol dehydrogenase KpADH reveals importance of retaining hydration shell in organic solvent tolerance. Protein Science. 2024;33(4):e4933. 10.1002/pro.4933

DATA AVAILABILITY STATEMENT

All data generated and analyzed during this study were included in this manuscript and its additional files.