Facilitating the Evolution of Esterase Activity from a Promiscuous Enzyme (Mhg) with Catalytic Functions of Amide Hydrolysis and Carboxylic Acid Perhydrolysis by Engineering the Substrate Entrance Tunnel

ABSTRACT

Promiscuous enzymes are generally considered to be starting points in the evolution of offspring enzymes with more specific or even novel catalytic activities, which is the molecular basis of producing new biological functions. Mhg, a typical α/β fold hydrolase, was previously reported to have both γ-lactamase and perhydrolase activities. However, despite having high structural similarity to and sharing an identical catalytic triad with an extensively studied esterase from Pseudomonas fluorescens, this enzyme did not show any esterase activity. Molecular docking and sequence analysis suggested a possible role for the entry of the binding pocket in blocking the entrance tunnel, preventing the ester compounds from entering into the pocket. By engineering the entrance tunnel with only one or two amino acid substitutions, we successfully obtained five esterase variants of Mhg. The variants exhibited a very broad substrate acceptance, hydrolyzing not only the classical p-nitrophenol esters but also various types of chiral esters, which are widely used as drug intermediates. Site 233 at the entrance tunnel of Mhg was found to play a pivotal role in modulating the three catalytic activities by adjusting the size and shape of the tunnel, with different amino acid substitutions at this site facilitating different activities. Remarkably, the variant with the L233G mutation was a very specific esterase without any γ-lactamase and perhydrolase activities. Considering the amino acid conservation and differentiation, this site could be a key target for future protein engineering. In addition, we demonstrate that engineering the entrance tunnel is an efficient strategy to regulate enzyme catalytic capabilities.

IMPORTANCE Promiscuous enzymes can act as starting points in the evolution of novel catalytic activities, thus providing a molecular basis for the production of new biological functions. In this study, we identified a critical amino acid residue (Leu233) at the entry of the substrate tunnel of a promiscuous enzyme, Mhg. We found that substitution of this residue with smaller amino acids such as Gly, Ala, Ser, or Pro endowed the enzyme with novel esterase activity. Different amino acids at this site can facilitate different catalytic activities. These findings exhibited universal significance in this subset of α/β fold hydrolases, including Mhg. Furthermore, we demonstrate that engineering the entrance tunnel is an efficient strategy to evolve new enzyme catalytic capabilities. Our study has important implications for the regulation of enzyme catalytic promiscuity and development of protein engineering methodologies.

INTRODUCTION

Catalytic promiscuity refers to the ability of an enzyme to catalyze different substrates and different types of chemical reactions in the same active center. Although high substrate specificity has been well known, increasing evidence indicates that enzyme catalytic promiscuity exists widely in nature (1, 2, 3, 4). Recently, enzyme promiscuity has captured much attention because it is related to the diversity of metabolic pathways and natural products, thereby contributing to the source of compounds for new drug screening (5). More importantly, promiscuous enzymes can act as starting points in the evolution of more specific and even novel activities. This serves as a molecular basis for producing new biological functions in living organisms (6). According to Darwin's theory of evolution, ancestral enzymes present in the primeval environment exhibited broad substrate scope and reaction types, but the level of their catalytic activities was low, indicating that they were highly promiscuous (7). After several million years of evolution involving gene duplications, mutations, and natural selection, an ancestral enzyme gradually turned into multiple highly differentiated progeny enzymes with diverse functions (6). Thus, promiscuous enzymes are found at key branch points in the evolutionary process. In this sense, rather than being completely specific, promiscuous enzymes are ideal models of continuous evolution of novel catalytic activities.

Sometimes, wild-type enzymes may not exhibit promiscuous activities, but will gain those activities by “fine-tuning” of certain amino acid residues through protein engineering. Several methods are used to achieve this in practical research. The first method of choice is to engineer the active center, which is the main site for binding and catalyzing substrates. For example, wild-type Pseudomonas fluorescens esterase (PFE) does not catalyze perhydrolysis of acetic acid or does so with a very low catalytic rate, while its variants PFE-L29P and PFE-L29T show marked perhydrolase activity (8). The second method is to induce allosteric effects. The enzyme activity is modified by altering residues at a site away from the binding pocket (even more than 10 Å). These residues can remotely influence the binding states between substrate and active center and then influence the enzyme catalytic activity (9, 10). A third and very different approach is to engineer the substrate entrance tunnel. Entry into the binding pocket is the prerequisite to catalysis, and the residues at the entrance tunnel can influence its size and polarity, thereby determining substrate specificity and hence catalytic activity. For example, Chaloupková et al. found a crucial site, Leu177, which is positioned at the tunnel opening of the haloalkane dehalogenase. Introducing a small and nonpolar amino acid at position 177 resulted in an increase in catalytic activity (11). Kotik et al. found that saturation mutagenesis at site 217 of the entrance tunnel of an epoxide hydrolase caused profound differences in activity and enantioselectivity toward various epoxides (12).

Mhg, a (−)-γ-lactamase with high stability and enantioselectivity, was first isolated and purified from Microbacterium hydrocarbonoxydans (13). Mhg can be utilized to prepare chiral γ-lactam (14), which is an important chiral intermediate for the synthesis of a series of antiviral drugs, such as abacavir (targeting HIV) and peramivir (targeting hepatitis and pandemic influenza viruses). Recently, by performing structure similarity analysis combined with classical enzyme assays, we have found that Mhg has perhydrolase activity (Fig. 1) and that its catalytic efficiency as a perhydrolase was 10-fold higher than as a γ-lactamase. Active site mutagenesis confirmed that the two reactions occur in the same active center (15). Structure analysis revealed that Mhg shares high structural similarity with an aryl esterase from Pseudomonas fluorescens and that both have the same Ser-His-Asp catalytic triad. Despite its potential as an esterase, Mhg cannot catalyze the hydrolysis of ester compounds. In this study, we describe the evolution of the esterase with high activity and wide substrate acceptance from promiscuous Mhg by engineering of the substrate entrance tunnel.

FIG 1.

Reaction schemes of (−)-γ-lactamase (A) and perhydrolase (B).

MATERIALS AND METHODS

Chemicals, reagents, bacterial strains, and culture medium.

4-Nitrophenyl butyrate (pNPB), (±)-2-azabicyclo[2.2.1]hept-5-en-3-one [(±)-γ-lactam], phenol red, and bromophenol blue were purchased from Sigma-Aldrich (Shanghai, People's Republic of China). Methyl acetate, methyl 2-tetrahydrofuroate, 3-cyclohexene-1-carboxylic acid, methyl 2-methylbutyrate, and methyl cyclopentanone-2-carboxylate were purchased from Adamas (Shanghai, People's Republic of China). Methyl phenyl glycidate, methyl-4-methoxyphenyl-oxiranecarboxylate, and 1-naphthyl acetate were purchased from Aladdin (Shanghai, People's Republic of China).

Methanol, acetonitrile, n-hexane, and isopropanol were of high-performance liquid chromatography (HPLC) grade and were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Other reagents were purchased from Sinopharm Chemical Reagent (Shanghai, People's Republic of China). Unless otherwise stated, all reagents were of analytical grade and used without any further purification.

Escherichia coli Top10 cells were used for both cloning studies and protein expression. The E. coli strains were routinely grown in LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl) and on LB agar plates (1% tryptone, 0.5% yeast extract, 1% NaCl, 1.5% agar) at 37°C. To isolate bacterial strains carrying the appropriate recombinant plasmids, 50 μg/ml ampicillin was added to the medium.

Gene, plasmids, and primers.

The gene mhg (GenBank accession number GU116480) was amplified from genomic DNA isolated from Microbacterium hydrocarbonoxydans (NBRC 103074). The plasmid pBAD/M, which derived from the commercial plasmid pBAD/Myc-HisA (Invitrogen, Carlsbad, CA, USA) with an NdeI restriction site introduced in it, was used for standard cloning and expression in E. coli. The gene mhg was inserted into pBAD/M between the NdeI and HindIII restriction sites, resulting in plasmid pBAD-mhg. All primers used in this study were synthesized by BGI (Shenzhen, China) and are listed in Table 1.

TABLE 1.

Primers used in this study

^a Mutation sites are indicated in bold.

Alanine-scanning mutagenesis of the substrate entrance tunnel.

All alanine-scanning mutants were constructed by PCR using primers containing the corresponding mutations and the QuikChange site-directed mutagenesis method (16). PCR was performed using KOD-Plus DNA polymerase (Toyobo, Osaka, Japan). The PCR mix (25-μl final volume) contained 10× KOD buffer (2.5 μl), MgCl₂ (1 μl, 25 mM), deoxynucleoside triphosphate (dNTP) (2.5 μl, 2 mM each), primers (0.5 μl, 2.5 μM each), template plasmid (1 μl, 10 ng/μl), and 1 unit of KOD-Plus DNA polymerase. The PCR consisted of two rounds of cycles. Predenaturation was performed at 94°C for 5 min, followed by the first round for 5 cycles with the following conditions: 94°C for 35 s, 55°C for 35 s, and 68°C for 30 s. The second round was for 30 cycles with the following conditions: 94°C for 35 s, 55°C for 35 s, and 68°C for 5 min. The final extension was at 68°C for 10 min. The PCR products were then digested by DpnI (Thermo Fisher Scientific) to remove the methylated template DNA. The digestion mix (12 μl final volume) contained the PCR product (10 μl), 10× buffer Tango (1.2 μl), and DpnI (0.8 μl, 10 U/μl). After 3 h of digestion at 37°C, the mix was transformed into E. coli Top10 cells and selected on LB agar with 50 μg/ml ampicillin. The transformants obtained were verified to contain the mutation of interest by gene sequencing.

Creation and screening of the mutant libraries.

Saturation mutagenesis at the Leu233 site was also performed by the QuikChange PCR method, using NNK codon degeneracy (N, adenine/cytosine/guanine/thymine; K, guanine/thymine) with pBAD-mhg as the template. Saturation mutagenesis of other sites at the entrance tunnel was performed using mhg-L233A as the template. All reactions were carried out according to the above-mentioned PCR procedures. For screening the mutant libraries, 10 μl of the PCR product was digested, transformed into E. coli Top10 cells, and selected on LB agar with ampicillin. The library clones were then transferred to 96-well plates containing 200 μl of LB medium with 50 μg/ml ampicillin and l-arabinose as inducers to a final concentration of 0.01%. Assuming that when the library contains 200 samples, the coverage rate is greater than 95% (17), we used two 96-well plates for each site. After inoculation, the 96-well plates were incubated at 28°C for 16 h. The cells in each well were then harvested by centrifugation at 2,800 × g for 30 min and resuspended in 200 μl of the NaH₂PO₄-Na₂HPO₄ buffer (50 mM, pH 7.5) containing pNPB (0.6 mM final concentration) on a Thermo shaker at 30°C and 1,000 rpm for 15 min. To remove the cells after reaction, the 96-well plates were centrifuged again, and the supernatants were transferred onto fresh 96-well plates. Subsequently, the product, p-nitrophenol, was detected at 405 nm by a Bio-Rad 680 microplate reader (Bio-Rad, Shanghai, People's Republic of China).

Gene expression and enzyme purification.

The recombinant protein Mhg with a C-terminal His₆ tag was expressed in E. coli Top10 cells. The constructed plasmid pBAD-mhg was transformed into E. coli Top10 competent cells, which were grown overnight at 37°C. Then 5 ml of the overnight culture was transferred into 100 ml of LB medium with 50 μg/ml ampicillin. When the culture reached an optical density at 580 nm (OD_580|nm) of 0.8, l-arabinose was added to a final concentration of 0.01%, and the cells were then induced for another 20 h at 28°C.

Purification was conducted as follows. Briefly, 100 ml of induced cells was harvested by centrifugation at 4,000 × g for 10 min and resuspended in 30 ml of binding buffer (50 mM Tris, 100 mM NaCl, 10 mM imidazole [pH 8.0]). The cells were then disrupted by ultrasonication in an ice bath for 20 min (work for 1 s and stay for 2 s) at 400 W. To remove the cell debris, the lysate was centrifuged at 15,000 × g for 30 min. Subsequently, the supernatant was filtered through a 0.22-μm membrane and then collected and loaded onto a 5-ml HisTrap FF crude column (GE Healthcare, Wauwatosa, WI, USA) equilibrated with binding buffer. Impurities were removed with washing buffer (50 mM Tris, 100 mM NaCl, 20 mM imidazole [pH 8.0]). Finally, the target protein was eluted by elution buffer (50 mM Tris, 100 mM NaCl, 500 mM imidazole [pH 8.0]). An Amicon Ultra-15 10K centrifugal filter device (Merck KGaA, Darmstadt, Germany) was utilized to remove the imidazole present in the elution buffer. In the end, the target protein was dissolved in storage buffer (20 mM Tris, 5 mM EDTA, 5% glycerol [pH 8.0]) and stored at −80°C to prevent loss of activity.

Quantification and SDS-PAGE analysis.

The protein was quantified using a standard BCA protein assay kit (Pierce, Rockford, IL, USA). SDS-PAGE with a 6% polyacrylamide stacking gel and a 12% polyacrylamide separating gel was performed to confirm the purity.

Structure similarity analysis and molecular docking.

Homology modeling was performed using Swiss-Model. The Dali server (http://ekhidna.biocenter.helsinki.fi/dali_server) was used for a protein structure similarity search in the Protein Data Bank (PDB). The Dali server is a structure alignment tool based on comparing the similarities of intramolecular distance matrices. The result was displayed with a Z-score, the value of which represents the similarity between two proteins (18, 19). The molecular docking of pNPB onto Mhg was performed by the Autodock plug-in (20) of PyMOL software (21), which performs an exhaustive search (based on a combination of simulated annealing and genetic algorithm optimizations) of orientation, position, and conformation of rotatable bonds in the binding site (22).

Activity assay.

Aryl esterase activity was detected according to the protocol developed by Krebsfänger et al. (23). The substrate pNPB was first dissolved in isopropanol (12 mM), and the reaction system contained 200 μl of NaH₂PO₄-Na₂HPO₄ buffer (50 mM [pH 7.5]), 500 μl of induced cells (or 5 μg of purified enzyme), and 10 μl of dissolved substrate to a final concentration of 0.6 mM. The reaction system was then incubated at 30°C for 20 min on a Thermo shaker. Finally, the amount of product (p-nitrophenol) was estimated at 405 nm using a DU800 spectrophotometer (Beckman Coulter, Brea, CA, USA). As for other ester compounds, we used a final substrate concentration of 5 to 10 mM, and the reaction times ranged from 30 min to 16 h for different substrates. Decreases in substrate and increases in product were determined by utilizing an Agilent gas chromatography (GC) 7890A system equipped with a HP-5 (30 m by 250 μm by 0.25 μm) column. The GC oven temperature was programmed from 60°C (held for 3 min) to 280°C at 15°C/min and then held for another 10 min. Split injection was used with a split ratio of 10:1, and the injector temperature was set at 280°C.

We used a bromophenol blue assay for detection of perhydrolase activity (24). Each reaction was performed in the following reaction system: 200 μl of buffer (1 M NaAc and 1 M NaBr in H₂O [pH 5.5]), phenol red (final concentration of 0.056 mM), hydrogen peroxide substrate (final concentration of 50 mM), and purified enzyme (5 μg). The reaction mixtures were incubated at 30°C for 15 min on a Thermo shaker. Finally, we determined the amount of bromophenol blue produced by perhydrolysis activity at 595 nm using a DU800 spectrophotometer.

(−)-γ-Lactamase activity was analyzed in a reaction assay in 300 μl of NaH₂PO₄-Na₂HPO₄ buffer (50 mM [pH 7.5]) containing (±)-γ-lactam (100 mM) and purified enzyme (5 μg). The reaction solution was incubated at 30°C for 10 min on a Thermo shaker and was then extracted with 200 μl of ethyl acetate. The decreases in (−)-γ-lactam were determined via an Acquity H-class ultraperformance liquid chromatograph (UPLC) (Waters, Milford, MA, USA) equipped with a Daicel Chiralpak AS-H column (250 by 4 mm; Daicel Chemical Industries, Tokyo, Japan). A mixture of acetonitrile and isopropanol with a volume ratio of 85:15 was used as the mobile phase at a flow rate of 0.8 ml/min. In addition, the UV absorbance of γ-lactam was measured at 230 nm. (+)-γ-Lactam and (−)-γ-lactam had retention times of 8.55 and 11.05 min, respectively.

Chiral analysis methods.

The decreases in (−)-γ-lactam were determined via an Acquity H-class UPLC equipped with a Daicel Chiralpak AS-H column, as described above.

The conversion rate and enantiomeric excess (ee) of the methyl phenyl glycidate (MPG) and methyl-(−)-3-(4-methoxyphenyl) oxiranecarboxylate were determined by an HPLC (Waters, MA, USA) equipped with a Daicel Chiralpak OD-H column (250 by 4 mm; Daicel Chemical Industries). A mixture of hexane and isopropyl alcohol with a volume ratio of 90:10 was used as the mobile phase at a flow rate of 1.0 ml/min. The remaining substrate was detected by measuring the absorbance at 254 nm using a UV-visible (UV-Vis) detector (486 series; Waters).

The conversion rate and ee of methyl 2-tetrahydrofuroate, 3-cyclohexene-1-carboxylic acid, and methyl 2-methylbutyrate were determined by an Agilent GC 7890A system equipped with a β-Dex 120 column (0.25-mm diameter, 30-m long; Supelco). Injection and detection were done at 220°C, with the helium flow rate at 1.0 ml/min. The conditions were different for the three substrates: for methyl 2-tetrahydrofuroate, the column temperature was kept at 85°C for 5 min and then increased to 95°C at a rate of 1°C per min and maintained for 3 min at 95°C; for 3-cyclohexene-1-carboxylic acid, the column temperature was kept at 30°C for 5 min and then raised to 90°C at 1°C per min and held for another 5 min; for methyl 2-methylbutyrate, the column temperature was kept at 40°C for 20 min.

Determination of kinetic constants.

Kinetic constants for the hydrolysis of pNPB were determined using the DU800 spectrophotometer as described above. The substrate concentrations were altered from 10 to 400 mM to give valid data points. Kinetic constants were then calculated by fitting the data to the Michaelis-Menten equation utilizing Origin 8.0 software (OriginLab, Northampton, MA, USA).

Software and online services.

The software Vector NTI 8.0 (Informax) was used for sequence alignments. Three-dimensional (3D) structure analysis and generation of computer-aided simulation figures were performed by using PyMOL software (21).

RESULTS

Structure similarity search and activity testing.

The Mhg homology model was constructed by Swiss-Model using (−)-γ-lactamase from Aureobacterium sp. (PDB code 1HKH) as the template. Subsequently, a structure similarity search in PDB database was performed using the Dali server. The search results indicated that, apart from γ-lactamase and perhydrolase, the enzyme most similar to Mhg was an aryl esterase from Pseudomonas fluorescens (PDB code 1VA4; Z-score of 43). Superimposition of the 1VA4 and Mhg models showed that they had the same catalytic triad (Ser-His-Asp) in the same spatial location. Protein sequence analysis showed that the two enzymes shared 52% similarity and 36% identity. Highly similar structures and the presence of the same catalytic triad between Mhg and esterase 1VA4 hinted that Mhg may have potential esterase activity. Subsequently, we used several classical esterase substrates, 4-nitrophenyl butyrate (pNPB), 4-nitrophenylcaproate (pNPC), and 1-naphthyl acetate, to test the possible aryl esterase activity of Mhg. Unfortunately, Mhg did not exhibit any detectable activity with these compounds.

Alanine-scanning mutagenesis at the substrate entrance tunnel.

Since wild-type Mhg exhibited no esterase activity, we reasoned that it could be due to some crucial amino acid residues near to or distal from the active site influencing the binding or catalysis of ester compounds. In order to find out how these compounds interacted with the residues of the binding pocket, we chose pNPB as the target molecule to be docked on the Mhg homology model. As suggested by the docking pose (Fig. 2A), pNPB forms two hydrogen bonds with NH of Met99 and Tyr32, and its carbonyl carbon is in close proximity to HO of Ser98. The catalytic mechanism of an esterase was reported to be similar to that of an amidase (25). In an amidase, the carbonyl oxygen atom of the substrate initially binds to the active site through the oxyanion hole, which is formed by nitrogen atoms of Tyr32 and Met99. Subsequently, the active site Ser98, deprotonated by His259 of the catalytic triad, attacks the activated carbonyl group to form the tetrahedral intermediate, following which the reaction takes place (26). As shown by the docking result, pNPB binds directly in the binding pocket with relatively proper positioning. The position of serine on the ester carbonyl carbon and fixing of carbonyl oxygen by the oxyanion hole make it possible for the catalytic activity to occur. Thus, there was an indication of catalytic activity of Mhg as an esterase. We then focused on the composition of the entrance tunnel to the binding pocket. There were six amino acid residues: Tyr32, Phe144, Trp204, Phe162, Ile232, and Leu233. We reasoned that large amino acid side chains lie in the entrance and may partially block the entrance tunnel of Mhg, thus making it difficult for the substrate to enter (Fig. 2B). To find out the roles of these amino acid residues, alanine-scanning mutagenesis was performed at the entrance tunnel. Six mutants (with mutations Y32A, F144A, F162A, W204A, I232A, and L233A) were generated by the QuikChange method. Initially, we used whole cells of the six mutants to catalyze the hydrolysis of pNPB and found that only the L233A mutant exhibited esterase activity. We then extracted the purified enzymes and repeated the activity assays, which again demonstrated that the L233A variant exhibited obvious esterase activity toward the classical substrate pNPB.

FIG 2.

Illustration of the substrate binding state of Mhg by docking of the pNPB molecule. (A) Detailed view of the interaction of pNPB with an oxyanion hole and active site. (B) Surface representation of the entry of the binding pocket bound to pNPB.

Saturation mutagenesis.

Since the substitution of Leu233 with alanine conferred esterase activity, this residue might offer functional plasticity. In order to determine the effects of different residues on the esterase activity, saturation mutagenesis was then performed at site 233 of Mhg. Nineteen mutants were constructed, and the purified enzymes were used to perform esterase activity assays. Four variants, including the L233A variant and three others (L233G, L233S, and L233P variants), showed various levels of esterase activity. The specific activities for hydrolyzing pNPB by these variants are listed in Table 2. The fact that the side chains of each of these four residues are relatively shorter than leucine possibly highlights the effect of steric hindrance in blocking the entrance tunnel, thereby preventing the substrates from entering the binding pocket. These results were consistent with our prediction. The smaller the side chain at site 233, the higher was the catalytic activity. Notably, substitution with glycine exhibited the highest activity, while substitution with proline with a pyrrole ring at site 233 was the least active among the four variants.

TABLE 2.

Kinetic constants for hydrolysis of pNPB

Catalyst (mutation)	Km (mM)	kcat (s−1)	kcat/Km (M−1·s−1)	Specific activity (U/mg)
L233A	14.7 ± 1.2	36.0 ± 0.2	2,444	696 ± 25
L233S	18.5 ± 1.9	19.2 ± 0.5	1,037	372 ± 18
L233P	13.3 ± 0.3	10.6 ± 0.3	799	206 ± 37
L233G	27.4 ± 2.0	137 ± 1.7	5,017	2,665 ± 58
L233A-F144R	23.7 ± 1.7	153 ± 1.2	6,439	2,953 ± 39

Since engineering the substrate entrance tunnel successfully endowed Mhg with ester hydrolysis activity, to further improve the catalytic efficiency, we performed saturation mutagenesis at other sites (Y32, F144, F162, W204, and I232) lying at the entrance tunnel with the L233A protein as the template. Five libraries were constructed by the QuikChange method, and 200 mutants for each library were screened for the catalytic capability of hydrolyzing pNPB. After screening about 1,000 mutants, the L233A-F144R variant was found to have the highest specific activity to pNPB at 2,953 U/mg, a value 4.2-fold higher than that of the L233A protein. To test whether mutations need to occur in any order for improvement of activity, an F144R mutant was constructed. Activity assays showed that the F144R variant, like the F144A variant, could not hydrolyze pNPB to produce p-nitrophenol. This deconvolution experiment argued for the existence of an order of mutagenesis for optimization of activity. Subsequently, we coupled F144R with L233S, L233P, or L233G, but unlike the case with L233A-F144R, none of the three variants exhibited an obvious increase in activity beyond that of their counterparts with single mutations.

Measurement of kinetics constants.

In order to precisely evaluate the catalytic capabilities of the five variants in hydrolyzing ester compounds, their kinetic constants toward pNPB were measured. The k_cat and K_m values were determined using purified enzymes with an excess of substrate. The data are listed in Table 2. Consistent with whole-cell and purified enzyme activity assays, the L233G variant displayed the highest esterase activity among the four variants owing to its high turnover number (k_cat), although the smallest-size side chain of glycine led to the weakest substrate affinity. The specific activities of the L233P and L233S variants with pNPB were 206 U/mg and 372 U/mg, respectively, and kinetic data showed that the L233S variant had higher catalytic efficiency because of its higher k_cat value. The L233A-F144R variant exhibited the highest catalytic capability among the five active variants, and its catalytic efficiency (k_cat/K_m) was 6,439 M⁻¹·s⁻¹, which indicated that this variant possessed excellent esterase activity. Compared to L233A, the newly introduced mutation F144R increased the values of k_cat and k_cat/K_m 4.3- and 2.6-fold, respectively.

Activity of the variants toward a series of ester compounds.

To test the substrate acceptance of the active variants, a series of diverse structural ester compounds was selected and examined by the esterase assay (Fig. 3). Different from the classical substrate pNPB (1a), many compounds were important intermediates in the synthesis of chiral drugs. For example, the chiral enantiomer of methyl phenyl glycidate (rac-5a) is the precursor of the widely used anticancer drug paclitaxel. The chiral enantiomer of methyl 2-tetrahydrofuroate (rac-10a), a pharmaceutical intermediate, can be used for synthesis of cephalosporin antibiotics. The catalytic activities of the five active variants toward ester compounds are shown in Fig. 4. The five variants exhibited different activities toward the 10 different esters tested, and the activity profile for each variant was substrate dependent. The classical substrates 1a, 2a (pNPC), and rac-7a could be hydrolyzed by all five variants to produce corresponding hydroxyl compounds. The specific activities of the five variants to these three substrates ranged from 45 U/mg to 2,953 U/mg, with the L233A-F144R and L233G variants exhibiting much higher activities than the other three variants. However, for the compounds rac-5a and rac-9a, the specific activities ranged from 1 U/mg to 4.5 U/mg, with the L233S variant showing the highest activity. The other esters, rac-4a, rac-8a, and rac-10a, could be catalyzed by all five variants, but the specific activities were lower than 1 U/mg, while rac-3a and rac-6a could not be hydrolyzed by the L233G or L233A-F144R variant. The L233P variant exhibited highest activity toward the latter five esters. To summarize, among all five variants, the L233A, L233S, and L233P variants could catalyze the hydrolysis of each of the 10 esters tested, while the L233G and L233A-F144R variants could catalyze the hydrolysis of eight. In addition, between the classical esterase substrates 1a and 2a, all five variants showed a 10-fold higher activity for 1a than for 2a, which indicated that the active variants we created were aryl esterases rather than lipases.

FIG 3.

Ester compounds chosen to assay the hydrolytic activities of the Mhg variants.

FIG 4.

Overview of the performance of the Mhg variants in hydrolysis of various esters. Data are shown as logarithmic values in the radar map, and the unit is units per milligram. Activities ≤0.01 U/mg are depicted as 0.01 U/mg. The original data are provided in the supplemental material.

The enantiomeric excesses (ee) of the racemic substrates after the reactions were determined by chiral HPLC and GC as described in Materials and Methods. For each racemic substrate, only the data for the highest activities and ee values among the five variants are listed in Table 3. Unfortunately, the four esterase variants did not exhibit excellent enantioselectivity. The highest ee value was 46% when the L233S variant catalyzed rac-5a. In this study, we did not make any further efforts to improve their enantioselectivities. In this regard, several strategies, such as iterative saturation mutagenesis (ISM) and inducing allostery, etc., could potentially be employed (9, 27).

TABLE 3.

Enantioselectivity of racemic esters catalyzed by the Mhg variants

Substrate	Variant (mutation)	Reaction time (h)	Conversion (%)a	ee (%)
rac-3a	L233P	16	53	25
rac-4a	L233P	16	67	21
rac-5a	L233S	2	51	46
rac-6a	L233P	16	71	NDb
rac-7a	L233G	0.25	70	ND
rac-8a	L233P	16	80	ND
rac-9a	L233S	2	45	27
rac-10a	L233P	16	72	25

^aConditions: 10 mM substrate, 100 μg of enzyme in phosphate buffer (50 mM, pH 7.5), 30°C for varying time intervals ranging from 15 min to 16 h. Conversion rates were adjusted for proper determination of ee values.

^bND, not determined.

γ-Lactamase and perhydrolase activities of the variants.

In view of the multiple effects of mutations on enzyme properties, the original γ-lactamase and perhydrolase activities of the five variants were analyzed (Fig. 5). The results showed that none of the five variants retained their perhydrolase activity. The L233A, L233S, and L233P variants retained their γ-lactam-hydrolyzing capability but showed an obvious decrease in activity. The L233G and L233A-F144R variants lost γ-lactamase as well as perhydrolase activities and exhibited esterase activity exclusively. These observations indicated that we successfully evolved a promiscuous Mhg to a novel specific enzyme, an esterase.

FIG 5.

Relative activities of wild-type (WT) Mhg and its esterase variants for perhydrolase, (−)-γ-lactamase, and esterase activities. Data are shown as logarithmic values in the histogram, and reactions were performed under standard enzyme assay conditions. Hydrogen peroxide, (±)-γ-lactam, and 4-nitrophenyl butyrate (pNPB) were used to determine the specific activities of perhydrolase, (−)-γ-lactamase, and esterase, respectively. The specific perhydrolase activity of the wild type (6,040 U/mg) was taken as 100%. Relative activity ≤0.01 U/mg is shown as 0.012 U/mg for legibility. Error bars represent standard deviations from three independent experiments.

DISCUSSION

Increasing evidence suggests that promiscuous enzymes are not rare exceptions but are rather widespread (28). Promiscuous enzymes are considered to be the starting points for the evolution of new activities. In this study, we chose the promiscuous enzyme Mhg as a model, which showed both γ-lactamase and perhydrolase activities but did not possess any esterase activity. Molecular docking and sequence analysis suggested a possible role for the substrate entrance tunnel in preventing the esterase activity of Mhg. In this study, we validated this hypothesis and successfully constructed esterase activity in Mhg by engineering the entrance tunnel. Through alanine-scanning mutagenesis and saturation site-directed mutagenesis, we successfully obtained five variants that exhibited esterase activity, illustrating the effect of steric hindrance in blocking the entrance tunnel of wild-type Mhg. The five esterases could catalyze not only the classical substrates but also a variety of diversely structural chiral ester substrates.

The L233 site is located at the rim of the entrance tunnel (Fig. 2B) and can significantly influence the size and shape of the tunnel. In Mhg, substitution of leucine at this site with amino acids (glycine, alanine, serine, and proline) that have relatively shorter side chains successfully conferred esterase activity since the obstacle at the entrance was removed. Among the four active variants, the L233G variant exhibited the highest esterase activity with pNPB as the substrate, although its affinity was the lowest. This fact may due to a “slipping off” effect of the substrate. Glycine has the shortest carbon chain, leading to a cavity in which substrate can not only enter but also easily slip off. Similarly, the five-membered pyrrole ring in proline and the relatively complex structure may cause the substrate to fix to the binding pocket, which could account for its highest affinity. The L233A-F144R variant exhibited the highest activity, which may result from the synergistic effects of the two sites (9).

Esterase is predicted to be the ancestor of a number of different types of enzymes (25). In the research of evolutionary analysis, Devamani and his coworkers found that the predicted ancestral enzymes of hydroxyl nitrile lyases shared the Ser-His-Asp catalytic triad and that they had a conserved function of hydrolysis of esters (29). We found that the members of the γ-lactamase family always have characteristics that are more or less similar to those of the esterase family. Almost all of the γ-lactamases discovered so far belong to the α/β hydrolase family, which also includes the esterase family (30). Therefore, we presumed that the ancestral enzymes from which Mhg evolved probably possessed esterase activity and that γ-lactamase and perhydrolase activities may be modern activities that gradually evolved from the esterase activity though divergent evolution.

Previous studies indicate that in many promiscuous enzymes, there is a trade-off between acquired and original activities (11, 31). In this study, the trade-off relationship between the three activities was also observed. Mutations have a great impact on the enzyme, leading to a sharp decline in original activities. In our research, the L233 site is crucial because a single point mutation at this site turns Mhg into an esterase. Combined with previous research by our team, we developed a histogram that displayed the relationship of the different L233 site variants with their respective enzyme activities (Fig. 6). Consistent with the trade-off mentioned above, we did not recover any variant that has all three enzyme activities simultaneously. We believe that the new activity is obtained at the expense of the original activity. Additionally, three variants only retain one kind of activity: the L233M variant has only perhydrolase activity; the L233T variant has only γ-lactamase activity; and the L233G variant has only esterase activity. Thus, we obtained variants with single activity from promiscuous Mhg through saturation mutagenesis at the L233 site.

FIG 6.

Relative activities of all Mhg variants at the L233 site for perhydrolase, (−)-γ-lactamase, and esterase. Data are shown as logarithmic values in the histogram, and reactions were performed under standard enzyme assay conditions. Hydrogen peroxide, (±)-γ-lactam, and pNPB were used to determine the specific activities of perhydrolase, (−)-γ-lactamase, and esterase, respectively. The other 12 variants lost all three activities or totally formed inclusion bodies. The specific perhydrolase activity of wild-type Mhg (6,040 U/mg) was taken as 100%. Relative activity ≤0.01 U/mg is shown as 0.012 U/mg for legibility. Error bars represent standard deviations from three independent experiments.

The L233 site is crucial for Mhg because it differentiates between the three possible activities: perhydrolase, (−)-γ-lactamase, and esterase. Subsequently, in the process of computational analysis of a series of promiscuous enzymes that share high structural similarities with Mhg, such as PDB entries 1BRO, 1VA4, and 1M33, we found that the corresponding site was a conserved leucine or valine. Our results indicate that the Leu/Val site has a universal significance for determining the catalytic activities in this correlative α/β hydrolase family. As proven by our study, single activities are expected to be easily separated through saturation mutagenesis at this site.

In the present study, we demonstrate that engineering of the substrate entrance tunnel is an effective strategy to regulate the catalytic activities and substrate specificities. Although not commonly used, this method is important for engineering enzymes in their binding pocket. Moreover, enzyme promiscuity plays an important role in evolving new catalytic activities, and fine-tuning of protein engineering might influence enzyme performance. For example, mutagenesis at one single point could dramatically regulate several activities of a promiscuous enzyme.