process-Engineering the optimum pH of β-galactosidase from Aspergillus oryzae for efficient hydrolysis of lactose

β-Galactosidase (lacA) from Aspergillus oryzae is widely used in the dairy industry. Its acidic pH optimum and severe product inhibition limit its application for lactose hydrolysis in milk. In the present study, structure-based sequence alignment was conducted to determine the candidate mutations to shift the pH optimum of lacA to the neutral range. The Y138F and Y364F mutants shifted the pH optimum of lacA from 4.5 to 5.5 and 6.0, respectively. The acid dissociation constant (p K a) values of catalytic acid/base residues with upwards shift were consistent with the increased pH optimum. All variants in the present study also alleviated galactose inhibition to various extents. Molecular dynamics demonstrated that the less rigid tertiary structures and lower galactose-binding free energy of Y138F and Y364F might facilitate the release of the end product. Both Y138F and Y364F mutants exhibited better hydrolytic ability than lacA in milk lactose hydrolysis. The higher pH optimum and lower galactose inhibition of Y138F and Y364F may explain their superiority over lacA. The Y138F and Y364F mutants in the present study showed potential in producing low-lactose milk, and our studies provide a novel strategy for engineering the pH optimum of glycoside hydrolase.


INTRODUCTION
Lactose is the principal carbohydrate in milk and makes up 7.2% of human milk, and 4.7% of cow milk (He et al., 2008).In mammals, lactose is hydrolyzed by enterocytes via β-galactosidase (EC 3.2.1.23)into its monosaccharide constituents, glucose and galactose, in the small intestine (Misselwitz et al., 2019).Commonly, newborns express sufficient β-galactosidase to digest the lactose in mammalian milk.However, the expression level of the LCT gene coding for β-galactosidase in the human body declines gradually with age after weaning, resulting in lactose maldigestion (Deng et al., 2015).Approximately 70% of the world population suffers from lactose malabsorption.Symptoms of intestinal discomfort such as bloating, diarrhea, and gas may arise when lactose in dairy products is consumed by these individuals (Storhaug et al., 2017).Complete avoidance of milk and dairy products could cause nutrient shortcomings such as low calcium intake (Black et al., 2002).In the dairy industry, β-galactosidase is widely used to produce lactose-free milk for people with lactose intolerance.
Based on sequence homology, β-galactosidases can be mainly classified into 4 different glycoside hydrolase (GH) families (GH 1,GH 2,GH 35,and GH 42) in the CAZy database (http: / / www .cazy.org/).All these families belong to a superfamily termed GH-A.The GH-A enzymes possess a classical triose-phosphate isomerase (TIM)-barrel fold structure acting as the catalytic domain and cleave glycosidic bonds via a double displacement mechanism.Two Glu residues serving as proton donor and nucleophile, respectively, are pervasive in retaining glycoside hydrolases (Vuong and Wilson, 2010).
The commercially available β-galactosidases used in the dairy industry have been mainly obtained from fungi.The most common fungal sources are Kluyveromyces lactis, Kluyveromyces fragilis, and Aspergillus species, which are in GRAS (generally recognized as safe) status (Xavier et al., 2018).Aspergillus oryzae is one of the most important fungal sources of β-galactosidase.The β-galactosidase (lacA) produced by A. oryzae has a pH optimum of 4.5 (Maksimainen et al., 2013), which is appropriate for lactose hydrolysis of acid whey.Compared with the intracellular, neutral β-galactosidases produced by Kluyveromyces sp., lacA showed great advantages of extracellular expression with higher culture yield, less contaminated proteins, and easier downstream process-ing.However, the acidic pH optimum of lacA limits its application in the hydrolysis of lactose in milk with a pH of around 6.5 to 6.7.Thus, as a secreted, extracellular enzyme, shifting the pH optimum of lacA to the neutral range could decrease the cost of downstream purification of neutral β-galactosidase in industry.
Nowadays, protein engineering is a commonly used tool in modern biotechnology to modify various characteristics of enzymes.Only limited papers reported successful cases of engineering the pH optimum of enzymes.The strategies of these cases focused on modifying the surface charges of enzymes or changing the acid dissociation constant (pKa) values of the catalytic residues.Li et al. (2019) replaced specific Glu residues on the surface of β-glucuronidase from A. oryzae by Arg, which contributed to an increased overall amount of positive charges on the surface and shifted the pH optimum from 4.5 to 6.5.Similarly, Wang et al. (2020b) introduced negatively charged residues to the surface of aspartase from Bacillus sp.YM55-1 and shifted its pH optimum from 9.0 to 8.0.Commonly, the pKa values of the catalytic residues decide the pH profile of enzymes (Dey and Roy, 2018).Since the ionization states of catalytic residues are vital in terms of the enzymatic catalyzation of a specific reaction, mutation design is usually based on comparative enzyme engineering.The mutations selected have been determined by comparison to homologous enzymes with the desired pH optimum range.According to the alignment of the primary sequences of xylanases with different pH optimum, an Asp near the acid/base Glu residue was experimentally proved to play a critical role in the pH optimum of these xylanases (Joshi et al., 2000).
In the present study, we employed a structure-based rational design to the β-galactosidase from A. oryzae by replacing amino acids close to the catalytic residues to shift its pH optimum for enhancement of its performance in lactose-hydrolyzed milk.The enzymatic properties of lacA and its mutants were characterized, including pH optimum, pH stability, substrate specificity, and end product inhibition.Then, hydrolysis of lactose in milk by lacA and its mutants was evaluated.Finally, prediction of the pKa values of the catalytic acid/base residue and molecular dynamics simulations were employed to analyze the underlying mechanisms of the mutants.

Strains, Plasmids, Cultures, and Chemicals
Escherichia coli strain JM109 and Pichia pastoris X33 were used as gene cloning and expression hosts, respectively.The vector pPICZαA was used as the expression vector.The lacA gene [GenBank (https: / / www .ncbi.nlm.nih.gov/genbank/ ) number: XP_001727461.1]was optimized in accordance with the codon preference of P. pastoris and synthesized by Sangon.The yeast extract peptone dextrose medium (YPD), buffered glycerolcomplex medium (BMGY), and buffered methanolcomplex medium (BMMY) were prepared according to the Multi-Copy Pichia Expression Kit (Invitrogen) for yeast growth and protein production.o-Nitrophenylβ-d-galactopyranoside (oNPG) and o-nitrophenol were obtained from Sangon.The Glucose Assay Kit was purchased from Megazyme.The UHT milk was purchased from Mengniu Dairy Co. Ltd.Other chemicals and reagents not mentioned were of reagent grade, purchased from Sinopharm Chemical Reagent Co. Ltd.The commercial lactase used for milk lactose hydrolysis was kindly provided from Sunsonenzymes Co. Ltd.
No animals were used in this study, and ethical approval for the use of animals was thus deemed unnecessary.

Gene Cloning and Site-Directed Mutagenesis
The lacA gene synthesized was digested with KpnI and NotI and cloned into KpnI/NotI sites at the multiple cloning site of the pPICZαA plasmid.The resulting plasmid was termed as pPICZαA-lacA.Five different single-point mutations (Y138F, Y138V, Y138A, Y138L, and Y364F) and one double-point mutation (Y138F/Y364F) were introduced into pPICZαA-lacA by whole plasmid PCR.The template plasmids were digested with DpnI, and the mutated plasmids were transformed into E. coli JM109 competent cells for nick repair.The site-directed mutagenesis of the lacA gene from Aspergillus oryzae was identified by sequence analysis.All primers are described in Table 1.

Transformant Screening, Expression, and Enzyme Purification
The mutant plasmids were linearized with BglII restrict enzyme and transformed into P. pastoris X33 competent cells by electroporation.Transformants were randomly picked up on YPD plates containing 100 μg/ mL zeocin.The positive variants were identified by colony PCR with the primer of 5′AOX/3′AOX according to the Multi-Copy Pichia Expression Kit instructions.
The positive variants were cultured in 25 mL of BMGY medium at 30°C for 16 h.The cells were harvested by centrifugation at a cell density of optical density at 600 nm (4.0-6.0) and resuspended in 100 mL of BMMY medium.Methanol was supplemented to a final concentration of 1% (vol/vol) at an interval of 24 h for 5 d.

Shi et al.: β-GALACTOSIDASE pH OPTIMUM
Finally, the medium was centrifuged at 6,300 × g for 10 min at 4°C.The isoelectric points of the wild-type enzyme and its mutants were predicted to be 5.34 by Compute pI tools of Expasy (https: / / web .expasy.org/).These enzymes were negatively charged at pH 7.0.Thus, anion exchange chromatography was used for the purification of these enzymes.After dialysis of the supernatant with a 10-kDa dialysis membrane against phosphate buffer (20 mM pH 7.0) at 4°C for 24 h, the β-galactosidase in the supernatant was purified by passing through an anion exchange column HiTrap Q HP (GE Healthcare Life Sciences) and eluted with an eluent solution (20 mM phosphate buffer containing 1 M NaCl, pH 7.0) in an AKTA purifier chromatographic system (GE Healthcare Life Sciences) by gradient elution.The purified protein was analyzed by SDS-PAGE.The concentration of the purified protein was determined by NanoDrop 2000c Spectrophotometer (Thermo Fisher).

Assay for ONPG Hydrolysis
The reaction was allowed to proceed at 37°C for 15 min containing 0.8 mL of 12 mM oNPG in sodium acetate buffer (50 mM, pH 5.0) and 0.2 mL of the suitably diluted enzyme.The reaction was terminated by adding 1 mL of 1 M Na 2 CO 3 and diluted to 10 mL by adding water.The absorbance was determined at optical density of 420 nm.One unit of oNPG-hydrolyzing activity (OHA) was defined as the quantity of enzyme required to catalyze the liberation of 1 μmol of o-nitrophenol per minute under the assay conditions.

Assay for Lactose Hydrolysis
The enzyme reaction mixture contained 0.9 mL 0.2 M lactose in sodium acetate buffer (50 mM, pH 5.0) and 0.1 mL of the suitably diluted enzyme.After incubation at 37°C for 15 min, the reaction was terminated by adding 0.2 mL of 1 M HClO 4 .Then, 0.1 mL of 2 M KOH was added to neutralize the reaction solution.The amount of glucose released was determined according to the Glucose Assay Kit manual.One unit of lactose-hydrolyzing activity (LHA) was defined as the amount of enzyme required to release 1 μmol of glucose per minute.

Measurement of Optimal pH and pH Stability
The optimal pH of lacA and its variants were determined by measuring the enzyme activity at 37°C over the pH range of 4.0 to 9.0 at an interval of 0.5 pH unit.The pH stability of enzymes was studied by pre-incubating the enzymes in buffer solutions with different pH values (3.0-8.0) for 1 h at 37°C, followed by adjusting pH to 5.0.The residual hydrolysis activities were assayed under standard conditions.The unincubated enzyme was used as a control.

Galactose Inhibition of LacA and Variants
To study the galactose inhibition of lacA and mutants, the reactions were performed at pH 6.5, and other conditions were the same as enzyme assay conditions with various concentrations of galactose (from 0 to 20 mg/mL).The relative activity was calculated as the residual activity divided by the activity measured without galactose individually.

Enzymatic Hydrolysis of Milk Lactose
The hydrolysis process was started by adding the purified enzyme to a 250-mL Erlenmeyer flask containing 50 mL of UHT milk.The final enzyme concentration was 200 μg/mL, corresponding to 21.76, 10.12, 6.42, 4.32, 4.21, 6.39, 1.86, and 22.37 U/mL (LHA at pH 6.5, 37°C) of lacA, Y138F, Y138V, Y138A, Y138L, Y364F, Y138F/Y364F, and the commercial lactase, respectively.The flasks were incubated on an orbital rotation shaker at 200 rpm, 37°C.The samples were collected after 0, 1, 2, 4, 6, 10, 12, 14, 16, 18, and 20 h of the hydrolysis process and incubated at 100°C for 10 min to inactivate the enzymes.The concentration of glucose released was determined according to the Glucose Assay Kit instructions.In addition, lacA, Y138F, Y364F, and the commercial lactase were added at a final concentration of 10 U/mL (LHA at pH 6.5, 37°C), respectively, for 3 h to evaluate the lactose hydrolysis ability.The purified enzymes were sterilized by passing through 0.22-μm filters.All the operations in the experiments including enzyme sterilization, adding enzymes into milk, and sampling were performed in a sterile environment.

Homology Modeling and pKa Prediction
Homology modeling of Y138F and Y364F was carried out using AlphaFold2 (Jumper et al., 2021).The pKa values of titratable groups in enzymes were calculated through the PDB2PQR web server (https: / / server .poissonboltzmann.org/).The protonation states of titratable groups were determined by empirical PROPKA at pH 7. The pKa values predicted are listed in Table 3.

Conventional Molecular Dynamics Simulations of LacA, Y138F, and Y364F
The molecular dynamics simulations of lacA, Y138F, and Y364F were carried out by using GROMACS 2019.3 (http: / / www .gromacs.org/).The ligand molecule, galactose, was optimized with Gaussian 16 (Frisch et al., 2016) at the calculation level of B3LYP (D3)/6-31G(d) (Grimme et al., 2010).AmberTools (Pearlman et al., 1995) was used to generate the parameter files of the ligand.The protein and ligand were parametrized by Amber ff14SB force field and the general Amber force field respectively.The protein-ligand complex was solvated with 1.2 nm of transferable-intermolecular potential-with-3-points (TIP3P) water along each dimension.Sodium ions were added to neutralize the system.The steepest descent method was used for the energy minimization of the system for 5,000 steps.Subsequently, 2 ns of canonical ensemble (NVT) equilibration was performed at 310 K.The LinCS algorithm was used to apply bond length constraints to H-bonds (Hess et al., 1997).Long-range electrostatic interactions were calculated through the particle mesh Ewald method (PME; Essmann et al., 1995).After NVT equilibration, isothermal-isobaric ensemble (NPT) equilibration was performed at 310 K for 2 ns.The molecular dynamics simulations were carried out for 100 ns.The 10 snapshots isolated from the last 1 ns at an interval of 100 ps of the 100 ns molecular dynamics trajectory were used for galactose-binding free energy calculation via the Poisson-Boltzmann surface area (MMPBSA) method (Kumari et al., 2014).

Rational Design of Mutation Points
The retaining β-galactosidase lacA belongs to the GH35 family, which employs a double displacement mechanism (Figure 1).E200 and E298 serve as the proton donor and nucleophile, respectively.The protonation state of the acid/base residue is vital for the proton transfer to the glycosidic oxygen in the first step of the hydrolysis reaction.Mutations of lacA were introduced close to the acid/base residue, E200, to enhance its pKa value so that E200 can maintain the protonation state in more alkaline pH conditions.In the study of β-galactosidase (lacB) from Aspergillus candidus, Zhang et al. (2018) performed saturation mutagenesis to the first-shell residues, which make hydrogen bonds to the galactose.However, the enzyme activity of most mutants declined dramatically except the Y364F mutant.As lacA and lacB shared amino acid sequence similarity of 99.7%, Y364 in lacA was chosen to be replaced by F to study the pH effect of substituting residue in the first shell.Because the mutations of the residues in the second shell usually exhibit a less adverse effect on enzymatic activity compared with the mutations in the first shell, other target residues selected for mutagenesis were based on the following principles: (1) being located in the second-shell of lacA, which make hydrogen bonds to the residues in the first shell except for E200 and E298, which are listed in Table 2, and (2) being not highly conserved from structure-based comparison to the β-galactosidases with higher pH optimum.
PyMOL (https: / / pymol .org/ ) was used to align the catalytic domains of lacA [Protein Data Bank (http: / / www .rcsb.org/ ) entry: 4iug] and the other 3 β-galactosidases (Protein Data Bank entry: 3thc, 4mad, 4e8c) from GH35 family (Figure 2).These have an optimum pH of 4.5, 4.0, 6.0, and 6.5, respectively (Maksimainen et al., 2013;Asp, 1971;Bernal et al., 2012;Jeong et al., 2009).According to the alignment of the 4 β-galactosidases, 4 groups of residues at homologous locations met the first principle (Table 2).Only the group containing Y138 of lacA met all the criteria among these groups.As shown in Figure 2, in the structural alignment, the corresponding spatial location of Y138 in lacA was not conserved, which was F in the β-galactosidases with higher pH optimum.According to the design principles, the mutation Y138F in lacA was chosen.In addition, other 3 hydrophobic amino acid (V, A, L) mutations were also introduced to Y138 of lacA according to the size of their side chains.A double mutant Y138F/Y364F was also constructed.

pH Dependency and pH Stability of the Recombinant LacA and Variants
To evaluate the effect of site-directed mutations on the pH profile of lacA, the pH dependency of lacA and its mutants were performed over the pH range of 3.0 to 8.0.
When oNPG was used as the substrate, the Y138F mutant shifted the pH optimum of lacA from 4.5 to 5.5 and showed increased specific activity under alkaline conditions (Figure 3a).At pH 7.0, the OHA of the Y138F mutant was 185.52 U/mg, 2.04-fold higher than that of lacA.The Y138A and Y138L mutants both shifted the pH optimum of lacA from 4.5 to 5.5 and exhibited no loss in OHA under alkaline conditions.The Y364F and Y138F/Y364F mutant enzymes both exhibited a drastic decline of hydrolysis activity, though they shifted the pH optimum from 4.5 to 6.0.
When lactose was used as the substrate, the specific activity of all mutant enzymes exhibited declines to different extents (Figure 3c).The pH optimum of the variants also shifted to a more alkaline pH, which showed no apparent change compared with the pH profile when oNPG was used as the substrate.
All enzymes were stable over the pH range of 4.0 to 7.5 with activities retained above 90%.Notably, the Y364F variant was more stable under extremely acidic conditions than other variants.

Lactose Hydrolysis in Milk
Lactose hydrolysis in milk by the commercial lactase, lacA, and the mutant enzymes was tested with 2 conditions, with the addition of the same amount or the same activity units (LHA) of the enzyme.As illustrated in Figure 4, under both conditions, Y138F, Y364F, and the commercial lactase exhibited higher hydrolysis efficiency than lacA.When added the same amount of the enzyme, lacA, Y138F, Y364F, and the commercial lactase hydrolyzed 88.51 ± 0.25% (SD), 92 ± 0.02%, 96.89 ± 1.13%, and 93 ± 0.82%, respectively, of the initial lactose in milk after 20 h of reaction, whereas other mutants exhibited lower hydrolytic ability than lacA.With the addition of the same activity units (LHA) of the enzyme, the Y138F, Y364F, and commercial lactase presented higher hydrolytic rates than lacA.After 3 h of reaction, lacA, Y138F, Y364F, and the commercial lactase hydrolyzed 43.48 ± 1.23%, 57.33% ± 0.41%, 82.67 ± 1.15%, and 53.83 ± 0.92%, respectively, of the initial lactose in milk (Figure 4b).Both Y138F and Y364F exhibited a better hydrolytic ability of lactose in milk than lacA.However, their specific activity using lactose as the substrate under neutral pH was lower  The residues in the table represent the residues in second-shell contacting the residues in the first-shell via hydrogen bonds, and the residues in the first-shell make hydrogen bonds with the ligand galactose.
2 PDB entry represents the accessory number of Protein Data Bank database of 4 β-galactosidases from the GH35 family. 3 The pH optimum is the pH value in which the enzyme can exhibit maximum activity.
than lacA.We speculated that end product inhibition of the enzyme might have an important effect on the efficiency of lactose hydrolysis.Thus, the end product inhibition of β-galactosidase needs to be studied.

Galactose Inhibition of LacA and Variants
The galactose inhibition of lacA and its mutants was studied by determining the hydrolytic activities using lactose and oNPG as the substrate with varying galactose concentrations.As shown in Figure 5, the hydrolysis activity of the parent lacA declined sharply with the addition of increased galactose concentration using lactose and oNPG as the substrate, respectively.
When lactose was used as the substrate, the galactose inhibiting effect was very noticeable (Figure 5a).The lacA, Y138F, and Y364F exhibited 10.91 ± 1.28%, 26.43 ± 1.12%, and 71.86 ± 3.82%, respectively, remaining activity when the galactose concentration was 20 mg/mL.The double-point mutant Y138F/Y364F showed the greatest resistance to galactose inhibition.When the galactose concentration reached 20 mg/mL, the Y138F/Y364F remained at 89.31 ± 0.23% of its initial activity.However, the Y138F/Y364F showed no advantages over other mutants in hydrolyzing lactose from milk due to the lowest specific activity (Figure 3c).
When oNPG was used as the substrate, the galactose showed weak inhibition against all mutants (Figure 5b).When the galactose concentration reached 20 mg/ mL, the OHA of lacA fell to 40% of its initial activity, whereas all mutants exhibited more than 90% of their initial OHA.These results indicated that these variants were less sensitive to galactose inhibition using oNPG as the substrate.

pKa Value Changes of the Acid/Base Residues
The pKa values of the acid/base residue, E200, in lacA, Y138F, and Y364F were predicted, which were  (PDB entries: 3thc,4mad,4e8c).The side chains of important residues are shown as sticks.The pH opt represent the pH optimum of each β-galactosidase.The Glu residues acting as nucleophile and acid/base are strictly conserved among these enzymes.PDB: Protein Data Bank database (http: / / www .rcsb.org/).The structural alignment was produced using Pymol and the rendering performed in Blender (https: / / www .blender.org/).6.10, 8.06, and 8.14, respectively (Table 3).The result correlated well with the increased pH optimum of Y138F and Y364F mutants.With higher pKa values, the acid/base residue of the mutants can maintain the correct protonation state in a more alkaline pH range to make the hydrolysis reaction occur more efficiently.

Dynamics of Enzyme-Galactose Complexes
For quantification of the enzyme-galactose-binding abilities of lacA, Y138F, and Y364F, molecular dynamics simulations were carried out.According to the root mean square deviation (RMSD) of backbone atoms, all 3 systems (lacA-galactose complex, Y138F-galactose complex, and Y364F-galactose complex) showed obvious fluctuations in the first 20 ns (Figure 6).After 100 ns simulation, the RMSD values of lacA, Y138F, and Y364F were stabilized at 1.7 Å, 1.9 Å, and 2.0 Å, re-spectively.The probability density distribution curve of RMSD of lacA was steeper than that of Y138F and Y364F, which indicated that lacA reached a steady state more easily.These results indicated the less stable structures of the 2 mutants.The interactions between the galactose and β-galactosidase at the −1 subsites of the active site tunnel contributed to the end product inhibition.The galactose-binding free energy of lacA, Y138F, and Y364F was −51.049, −33.203, and −32.039 kcal/mol, respectively.The lower affinity for galactose of Y138F and Y364F may increase the speed of the galactose expulsion.

DISCUSSION
An ideal β-galactosidase for milk process application should possess high hydrolysis activity in neutral pH conditions and low end product inhibition.In our study, 4 β-galactosidases with different pH optimum from the GH35 family were compared (Table 2).The active domains of these β-galactosidases were aligned to determine the candidate mutations (Figure 2).Y138 and Y364 of lacA were chosen to be mutated, which might have a potential influence on the pH optimum.All mutations shifted the pH optimum of lacA to the alkaline side, and the Y138F mutants showed increased OHA at alkaline pH (Figure 3a).In the milk lactose hydrolysis process, the single mutant, Y138F and Y364F, both exhibited better hydrolytic ability of lactose than lacA and commercial lactase (Figure 4b), whereas the double mutant Y138F/Y364F did not reflect the positive superimposed effects, whose lactose-hydrolyzed rate was the lowest (Figure 4).
For retaining GH, the proton state of the proton donor and nucleophile residues are important for the bell-shaped pH profile (Vuong and Wilson, 2010).Typically, the nucleophile residues should be deprotonated and negatively charged to attack the anomeric carbon of the substrates (Kirby, 1984).Thus, nucleophile residues usually have a pKa below 5 (Wan et al., 2015).The acid/base residues should be protonated at the first step of the hydrolysis reaction.The protonation states are determined by their pKa values.Pokhrel et al. (2013) introduced Arg residues close to the acid/ base residue, E172 of xylanase from Bacillus circulans.The pKa value of E172 decreased from 7.27 to 5.33 in the V37R mutant, shifting the pH optimum from 6.5 to 5. In our study, the pKa values of the acid/base residue,  E200, showed an upwards shift in variants, which may result in the increased pH optimum.
Lactose is the natural substrate hydrolyzed by β-galactosidase in the production of lactose-free milk.However, the oNPG is the most commonly used substrate to quantify the hydrolytic activity of β-galactosidase in the dairy industry.In our study, a significant difference between OHA and LHA was found by lacA and its mutants (P < 0.01).Among these mutations, the specific activity of Y138F was 2.04-fold higher than that of lacA toward the chromogenic substrate (oNPG) at pH 7.0.However, the LHA of Y138F was 2.48-fold lower than that of lacA at pH 7.0.In addition, when lactose was used as the substrate, the pH profile of Y138F was much steeper than that of lacA.Various studies have been undertaken to probe the substrate specificity of β-galactosidase (Kim et al., 1997;Jeong et al., 2009).
However, few reports have stressed that there is no certain relationship between OHA and LHA.The incorrect estimation of lactase activity could be caused when solely using the oNPG assay method.Thus, the OHA should not be estimated as LHA in actual application.
The effect of galactose concentration on OHA and LHA also differs.The end product inhibition is noticeable when lactose is used as the substrate.Site-directed mutagenesis introduced into the active site of lacA might impair the interactions between galactose and β-galactosidase, which results in lower galactose inhibition.However, these interactions may also promote the hydrolysis reaction by stabilizing the reactant, lactose.Thus, a trade-off between hydrolytic activity and end product inhibition could be found (Figure 5a).In our study, the Y138F and Y364F exhibited better performance than lacA in the milk lactose process due to the low galactose inhibition.Other researchers also reported similar phenomena.Atreya et al. (2016) performed Ala substitution to the residues of cellulase making hydrogen bonds with the end product, cellobiose.All mutations decreased the cellobiose inhibition at the cost of decreased activity.Liu et al. (2021) performed Ala scanning to the first-shell residues that make hydrogen bonds to the galactose of β-galactosidases from Bacillus coagulans, and the N148D variants alleviated galactose inhibition with the activity retained in only 13.49 ± 0.36% of the wild-type enzyme.The pKa values were predicted by the PDB2PQR web server.The protonation states of titratable groups in protein were determined by PROPKA at pH 7.
2 lacA represents the wild-type enzyme.Molecular dynamics simulations were carried out to quantify the galactose-binding free energy of lacA, Y138F, and Y364F.The backbone RMSD analysis revealed the larger structure fluctuations of Y138F and Y364F than that of lacA (Figure 6).Furthermore, the galactose-binding free energy of Y138F and Y364F was higher than that of lacA.These results indicated that the mutants were less favorable for the binding of galactose, which was correlated well with the relieved end product inhibition by galactose.A similar result was also reported by other researchers.Wang et al. (2020a) performed directed evolution to improve the catalytic efficiency of β-glucosidase (16BGL) from Penicillium oxalicum.The experimental data showed that the M280T/V484L/D589E mutant had reduced binding affinity to p-nitrophenyl-β-d-glucopyranoside (pNPG) compared with 16BGL.The binding free energy of 16BGL and M280T/V484L/D589E to pNPG calculated by MMPBSA were −35.52 ± 8.08 and −17.50 ± 4.60 kcal/mol, respectively, which was consistent with the experimental data.
In the present study, the heterologous expression of the eukaryotic β-galactosidase (lacA) and its mutants was performed in P. pastoris utilizing methanol-induced alcohol oxidase 1 promoter P AOX1 , which is commonly used for high-level expression of recombinant proteins (Karbalaei et al., 2020).Pichia pastoris is widely used for the production of heterologous proteins.Various pharmaceutical proteins and industrial enzymes have been produced by fed-batch high-cell-density fermentation.High cell density can be easily achieved and large amounts of recombinant proteins can be obtained at low costs.In addition to the high production of β-galactosidase in industry, the safety of the mutant β-galactosidases produced must be considered in future studies.Methanol can pose a serious safety risk for use in the food industry due to its toxicity when the promoter P AOX1 is used for lactase production.Methanolfree promoters need to be evaluated by engineering the P AOX1 promoter or using strong constitutive promoters.Shen et al. (2016) constructed methanol-free P. pastoris expression systems induced in the presence of glycerol and dihydroxyacetone by engineering the kinase involved in P AOX1 activation.This novel methanol-free expression system could reach 50 to 60% of the traditional methanol-induced system.In addition, the safety of the expression host and the mutant β-galactosidases need to be evaluated in a future study.
In addition to the safe production of the extracellular mutant lactases, the off-flavor formation in the UHT milk treated with lactase is another important problem making the product unattractive for consumption after long storage.In the present study, sensory analysis of the lactose-free milk was not conducted because of safety concerns.Commonly, UHT milk has a shelf-life of several months at room temperature, making these products prone to off-flavor formation.Even a very low off-flavor formation rate can lead to considerable off-flavor formation due to the extended shelf-life.In the research of neutral lactase produced from Kluyveromyces lactis, studies have revealed that the high levels of protease in the lactase preparation may cause rapid development of off-flavors in UHT milk treated with lactase (Troise et al., 2016).Arylsulfatase is also found as contaminating side activity in enzyme preparations, leading to a strong development of off-flavor even at very low levels (Stressler et al., 2016).Thus, the purification strategy of extracellular lactase with a high degree of purity needs to be developed to make the final lactase product substantially free from contaminating side activity in future research.
In summary, our studies provide a new strategy for engineering the pH optimum of β-galactosidase from A. oryzae, along with the alleviation of galactose inhibition.The Y138F and Y364F mutants in the present study performed better than the wild-type β-galactosidase, which showed potential application in milk lactose hydrolysis.

Figure 1 .
Figure1.The double displacement mechanism followed by β-galactosidase from Aspergillus oryzae using lactose as a substrate and water as a glycosyl acceptor.Acid/base and nucleophile: catalytic Glu residues.

Figure 2 .
Figure 2. Structural alignment of β-galactosidases.(a) The overall tertiary structure of lacA (PDB entry: 4iug); (b-d) alignment of the catalytic domains of lacA with other 3 β-galactosidases (PDB entries: 3thc, 4mad, 4e8c).The side chains of important residues are shown as sticks.The pH opt represent the pH optimum of each β-galactosidase.The Glu residues acting as nucleophile and acid/base are strictly conserved among these enzymes.PDB: Protein Data Bank database (http: / / www .rcsb.org/).The structural alignment was produced using Pymol and the rendering performed in Blender (https: / / www .blender.org/).

Figure 3 .
Figure 3.Effect of pH on the activity and stability of the purified β-galactosidases.(a) pH profiles of lacA and its mutants with oNPG as the substrate; (b) pH stability of lacA and its mutants with oNPG as the substrate; (c) pH profiles of lacA and its mutants with lactose as the substrate; (d) pH stability of lacA and its mutants with lactose as the substrate.The pH profiles and pH stability were determined in different buffers by varying pH values from 3.0 to 8.0 at 37°C.The mean and standard deviation of experiments conducted with 3 individual replicates are presented.oNPG: o-nitrophenyl-β-d-galactopyranoside; lacA: wild-type enzyme.

Figure 4 .
Figure 4. Time courses of lactose hydrolysis in milk by the commercial lactase, lacA, and the mutant enzymes.(a) Addition of the same amount of enzyme in each reaction.Reaction condition: 50 mL of milk, 10 mg of enzymes, 37°C, 200 rpm.(b) Addition of the same activity units of enzyme in each reaction.Reaction condition: 50 mL of milk, 500 U (LHA, pH 6.5, 37°C) of enzymes, 37°C, 200 rpm.The mean and standard deviation of experiments conducted with 3 individual replicates are presented.lacA: wild-type enzyme; LHA: lactose-hydrolyzing activity.

Figure 6 .
Figure 6.Backbone root mean square deviation (RMSD) and probability density distribution of the RMSD of lacA, Y138F, and Y364F.Backbone atoms: C, Cα, and N atoms of each residue; lacA: wild-type enzyme.

Table 3 .
The acid dissociation constant (pKa) values of the catalytic Glu residues of lacA, Y138F, and Y364F 1