The E. coli strains, BL21 (DE3), and expression vector, pET28a(+), were purchased from Takara (Dalian, China). O-Nitrophenol (ONP) and o-nitrophenyl-β-D-galactopyranoside (ONPG) were purchased from Sangon Biotech (Shanghai, China). Lactose, glucose, and galactose used for this study were supplied by Solarbio (Beijing, China). The thin-layer chromatography (TLC) silica gel plates were purchased from Merck (Darmstadt, Germany). All other materials used were of the analytical degree.

The seawater samples were isolated and obtained from the surface of Yellow Sea sediment (depth 20 m, 120.16° E 35.01° N). They were diluted to 10 times then spread and cultured on Zobell 2216E medium (5.0 g of peptone, 1.0 g of yeast extract, 0.01 g of iron phosphate, 16.0 g of agar, and 1,000 ml of natural seawater were placed in the dark for several weeks, and the pH was adjusted to 7.6–7.8). A few strains that were positive were isolated after incubation at 25°C for 7 days. In order to identify which one can produce β-galactosidase, the selective medium [1% (w/v) peptone, 1% (w/v) yeast extract, 0.5% (w/v) lactose, 2% (w/v) NaCl, 0.004% (w/v) X-gal, and 1.5% (w/v) agar, pH 7.0] was used to culture and screen the strain. Among them, strain BY02 with high activity was collected and used for a further study. According to the 16S rRNA gene sequence analysis, strain BY02 was identified as a member of genus Bacillus. In order to evaluate the influence of lactose and glucose on the expression of β-galactosidase, strain BY02 was cultured on three kinds of medium (with 2% glucose, 2% lactose, 2% glucose, and lactose) for 3 days at 25°C.

The gene for β-galactosidase was amplified from the genome of Bacillus sp. BY02. The open reading frame (ORF) finder in the National Center for Biotechnology Information (NCBI)² was used to identify the ORFs. SignalP-5.0 server was used to analyze the signal peptide of the amino acid sequence.³ ExPASy was used to calculate the theoretical isoelectric point (pI) and molecular weight (Mw) of Gal42.⁴ Conserved Domain Database (CDD) was used to improve the phylogenetic analysis. In addition, a BLAST algorithm program was performed to search similar sequences for Gal42. The ESPript⁵ and ClustalX program were used for multiple comparison analysis of Gal42 and BCA-β-Gal (PDB ID: 3TTS). MEGA 7.0 software was used to construct the evolutionary tree.

The three-dimensional (3D) structure of Gal42 was built by homology modeling using the SWISS MODEL.⁶ The crystalline structure of β-galactosidase from Bacillus circulans sp. alkalophilus (PDB ID: 3TTS) with sequence identity of 64.36% was chosen as the template. The Global Model Quality Estimation (GMQE) and Qualitative Model Energy Analysis (QMEAN) values for the homology model were calculated to evaluate the model quality and reliability. Pymol (version 2.4.1) was used to visualize the 3D structure, build organization graphics, and draw illustrations.

gal42 gene was synthesized by Synbio Technologies (Suzhou, China) and inserted into corresponding sites of plasmid pET-28a(+) between the recognition sites NcoI and XhoI. After the recombination, the plasmid was expressed in E. coli BL21 (DE3), which was cultured in Terrific Broth (TB) medium containing 30 μg/ml of kanamycin at 37°C until the optical density at 600 nm (OD₍₆₀₀₎) reached 0.6–0.8. Afterward, the addition of 0.1 mM of isopropyl β-D-thiogalactoside (IPTG) was performed to induce the expression of the target protein at 20°C and 200 rpm for 36 h. The target β-galactosidase was purified and harvested in AKTA150 FPLC system using Ni-NTA Sepharose affinity column (5 ml His-Trap^TM High Performance, GE Healthcare, Madison, WI, United States). The supernatant that was obtained by the crude enzyme incubated in 50 mM of phosphate buffer (pH 7.6) was centrifuged for 10 min at 12,000 rpm. After sonication at ice condition (150 W, 2-s burst, and 2-s stop, 30 min), the obtained supernatant was loaded into the previously equilibrated affinity column. Then, the washing buffer (20 mM of imidazole, 20 mM of phosphate buffer, and 500 mM of NaCl, pH 7.6) was used to deplete proteins other than target proteins. The target protein was eluted by elution buffer (150 mM of imidazole, 20 mM of phosphate buffer, and 500 mM of NaCl, pH 7.6). Mw and purity of the enzymes were assayed by 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), and the protein concentration was determined by the bicinchoninic acid (BCA) protein assay kit (Beyotime Biotechnology, Shanghai, China). In brief, protein samples (2 μg) were resolved by 10% SDS-PAGE (the voltage of 75 V for concentrated glue and 120 V for separating glue), and Coomassie brilliant blue was used for staining.

The activity of β-galactosidase can be reflected by the production of ONP using ONPG as substrate. Briefly, 50 μl of diluted enzyme sample was mixed with 450 μl of 10 mM ONPG solution (20 mM of phosphate buffer, pH 7.0). Then, the reaction system was incubated at 40°C for 10 min and stopped by adding 500 μl of Na₂CO₃ (1 M). Furthermore, the released ONP was measured at 420 nm. Under the experimental conditions, one unit (U) of β-galactosidase activity was defined as the amount of enzyme required to release 1 μmol of ONP per minute (Li et al., 2019).

In order to determine the optimal temperature of the Gal42, substrate reactions for Gal42 and ONPG were carried out at different temperature ranging from 0 to 70°C. To determine the thermal stability of Gal42, the residual activity was measured after incubation at 0–70°C for 60 min. To examine the optimal pH, substrates for Gal42 and ONPG were carried out in Britton–Robinson buffers at pH 5.11–10.42. To determine the pH stability of Gal42, the residual enzyme activity was measured after Gal42 was cultured in buffers with different pH values at 4°C for 12 h. By monitoring the enzyme activity in the presence of different cations [using 1 mM of BaCl₂, FeSO₄, CaCl₂, CuSO₄, FeCl₃, KCl, MnCl₂, Al₂(SO₄)₃, CoCl₂, NaCl, (NH₄)₂SO₄, NiCl, MgSO₄, Li₂SO₄, and ZnSO₄ in 20 mM of phosphate buffer] or chelating agents (1 mM of ethylenediamine tetraacetic acid (EDTA) and SDS), the effects of metal ions and chelating agents on Gal42 activity were determined as previously. Meanwhile, the effects of different concentrations of zinc ions (1, 5, and 10 mM) on optimal temperature and thermal stability were also detected according to the above conditions.

The reaction product of Gal42 in lactose and milk was determined by TLC. Briefly, 900 μl of lactose (5 mg/ml) and 100 μl of Gal42 were mixed and incubated under pH 8.0 at 40°C for 480 min. The reaction products were collected and boiled for 10 min at different time points (0, 1, 10, 30, 60, 360, and 480 min) to inactivate the enzyme. After that, 4 μl of sample from each time point was spotted in a horizon line on the silica gel TLC plate, and the plate was developed with a 3:2:5 (by vol.) mixture of glacial acetic acid, water, and 1-butanol. Later, after drying, the plate was treated with a 9:1 mixture of atomized ethanol and sulfuric acid. Heat the plate for 30 min at a temperature of 80°C, and the hydrolyzed products could be observed. The hydrolyzed products of lactose in drinking milk (4.2–5.0% lactose) (Inner Mongolia Mengniu Dairy Group Co., Ltd., Hohhot, China) was also observed by using the method described above. The diluted milk was mixed with enzymes, sampled at the same point, and then analyzed by TLC method, with galactose as the standard. Negative-ion electrospray ionization–mass spectrometry (ESI-MS) system (Thermo Fisher Scientific^TM Q Exactive^TM Hybrid Quadrupole-Orbitrap^TM, Waltham, MA, United States) was employed to further investigate the composition of the reaction products. The instrument conditions were syringe pump injection-flow rate 50 μl/min, sheath gas 15 L/min, auxiliary gas 5 L/min, auxiliary gas temperature 150°C, capillary temperature 300°C, and S-lens voltage 50 V.

The β-galactosidase gene (gal42) of strain Bacillus sp. BY02 was deposited in GenBank under accession number MW246571.

Thus far, several β-galactosidases have been purified and identified from bacteria (Mavromatis et al., 2010; Tian et al., 2013) and fungi (Rico-Díaz et al., 2017), including genus of Bacillus (Boon et al., 2000), Bifidobacterium (Hsu et al., 2007), and Aspergillus (Saqib et al., 2017). However, the productivity of β-galactosidases from natural microbial cells hardly meets the needs of industrial application due to the low yield (Park and Oh, 2010; Oliveira et al., 2011). Boosting the reaction efficiency of β-galactosidases is a precondition for achieving its industrial application. In a previous study, Becerra et al. (2001) has improved the partial secretion percentage of β-galactosidases in the culture medium through directing mutations at the N-terminus of the protein. Gutshall et al. (1995) reported that the β-galactosidase LacG isolated from Arthrobacter sp. B7 has a high specific activity. Thus far, the main industrial sources of β-galactosidase are Aspergillus sp. and Kluyveromyces (de Freitas et al., 2020).

In this study, the marine bacterium Bacillus sp. BY02 showed high intracellular β-galactosidase activity (19 U/ml) when grown in the 2216E medium containing 2% lactose. As shown in Supplementary Figure 1, it can produce β-galactosidase in the presence of lactose on 2216E X-Gal agar. The genome sequence analysis of Bacillus sp. BY02 showed that it contains a putative galactosidase-encoding gene, gal42, which consists of an ORF of 2,010 bp. The prepared β-galactosidase Gal42 contains 669 amino acid residues. The predicted Mw and theoretical pI of the expressed β-galactosidase were 78.38 kDa and 4.46, respectively.

A phylogenetic tree was constructed containing Gal42 and other reported β-galactosidases in GH families 1, 2, 35, and 42 (Figure 1). In this neighbor-joining tree, Gal42 formed a monophyletic cluster with the β-galactosidase from GH42. In order to further explore the conserved and catalytic domains of Gal42, a multiple sequence alignment was established among Gal42 and other three β-galactosidase, Bca-β-gal from Bacillus (GenBank code: QCG73672), BI Gal42A from Bifidobacterium animalis subsp. (GenBank code: ACS45863), Bbg II from Bifidobacterium bifidum S17 (GenBank code: ADO53518), β-gal II from Bifidobacterium adolescentis (GenBank code: AAR2413), and A4-β-Gal from Thermus thermophilus A4 (GenBank code: BAA28362) (Supplementary Figure 2). According to multiple sequence alignment analysis, the acid/base catalyst site (Glu148) and the nucleophilic site (Glu305) conserved in the GH42 β-galactosidases were also present in the amino acid sequence of Gal42. The further BLAST analysis and comparison with the NCBI CDD clarified Gal42 as a GH42 enzyme, which showed the highest homology with the β-galactosidase from Neobacillus bataviensis (GenBank code: TWE08672, 80.45% identity) and the β-galactosidase from Mesobacillus foraminis (GenBank code: TCN25424, 79.10% identity). These results indicated that Gal42 is a new member of the GH42 β-galactosidase.

The structural characterization indicated that the encoding β-galactosidase, Gal42, is a homotrimer in solution. As shown in Figure 2A, the 3D structure of Gal42 was constructed using SWISS-MODEL on the basis of homologs of known structure (Bca-β-gal, PDB ID: 3TTS). The GMQE and QMEAN values for the homology model were 0.87 and −3.05, respectively, indicating good model quality and high reliability. The homologous modeling result showed that Gal42 preserved the catalytic sites (Asn147, Glu148, Met304, and Glu305) and zinc binding sites of Cys cluster (Cys11, Cys153, Cys155, and Cys158). According to the sequence logo analysis (Figure 2B), two regions, “NEY” (Asn147-Glu148-Tyr149) and “ME” (Met304-Glu305), were conserved in the catalytic cave (Maksimainen et al., 2012). The catalytic residues Glu148 and Glu305 are highly conserved in all GH42 β-galactosidases, which play a role as an acid/base catalyst and a nucleophilic site, respectively (Barrangou et al., 2009). In addition, the Asn147 and Met304 preceding the two sites are also highly conserved. Tyr275, which is invariant in GH42, forms a hydrogen bond with the catalytic nucleophile Glu305, consistent with its proposed role as regulator. Trp313 provides aromatic stacking in GH42 and may modulate specificity by adopting different conformations (Hinz et al., 2004). The Zn²⁺ binding sites of Gal42 maintain its zinc coordination function in the Gal42 structure (Figure 2A). Meanwhile, as shown in Figure 2C, Cys cluster was not conserved in GH42 β-galactosidases, which illustrated that possession of metal binding sites was not a commonality in β-galactosidases.

gal42 gene was cloned and overexpressed in E. coli BL21(DE3) through the pET28a(+) plasmid. The formed E. coli BL21-pET28a-gal42 strain was cultured in Luria Bertani (LB) broth for proliferation and then in TB medium for enzyme production with IPTG. As shown in Figure 3A, the β-galactosidases activity of purified Gal42 was up to 217 U/ml, with biomass of 27 g/L at 60 h. The Mw of the purified Gal42 was analyzed by SDS-PAGE and showed to be about 74 kDa with a dominating band (Figure 3B), which is consistent with its theoretical Mw (78.38 kDa).

As shown in Figure 4A, the optimal reaction temperature of the purified Gal42 was 40°C. In the thermal stability assays, Gal42 was kept stable at 0–30°C (Figure 4B). The enzyme remained approximately 50% activity even after 40°C incubation. The optimal pH of Gal42 in Britton–Robinson buffers was determined to be 7.44 (Supplementary Figure 3A), which is similar to that of other GH42 β-galactosidases (Hu et al., 2007; Hildebrandt et al., 2009). Additionally, after a 12-h pretreatment in Britton–Robinson buffers at pH 5.11–10.42, the enzyme remained stable at pH 6.0–8.5, with an activity exceeding 60% (Supplementary Figure 3B). Especially, the residual activity of Gal42 was nearly 100% of its maximum activity at pH range 7.0–8.0. Compared with a pH-stable β-galactosidase BgaL from Paracoccus sp. 32d (Wierzbicka-Woś et al., 2011), Gal42 has shown wider pH-stability range. Considering the neutral pH of milk, β-galactosidase Gal42 was easily utilized and suitable in producing lactose-free milk with its properties mentioned above.

β-Galactosidase Gal42 contained a zinc binding site of the Cys cluster. Herein, the effect of zinc ions on biochemical characterization of Gal42 was determined (Figure 4). In the presence of Zn²⁺, the enzyme activity of Gal42 was significantly increased at the temperature of 10–60°C. At 40°C, the enzyme activity of Gal42 increased to 119 and 126% with 5 and 10 mM of zinc ions, respectively. However, the optimal temperature of the Gal42 was not affected by Zn²⁺ (Figure 4A). In the presence of 1 mM of Zn²⁺, the thermo-stability of Gal42 was slightly increased; and in the presence of 5 mM of Zn²⁺, the thermo-stability of Gal42 was significantly increased on 40 and 50°C (p < 0.001). Meanwhile, by increasing the concentration of Zn²⁺ to 10 mM, the activity of Gal42 was not significantly increased anymore (Figure 4B). As shown in Figure 4C, Cu²⁺, EDTA, and SDS showed significant inactivation effects on the activity of Gal42. The activity of Gal42 can be activated by Li⁺, Ba²⁺, Fe²⁺, Ca²⁺, K^+, NH₄⁺, Ni⁺, and Na⁺; while it was slightly inhibited by Fe³⁺ and Al³⁺. More interestingly, Gal42 can be activated not only by Zn²⁺ but also by divalent metal ions (Co²⁺, Mn²⁺, and Mg²⁺). As shown in previous studies, the metal ions of Mn²⁺, Mg²⁺, Zn²⁺, and Ni⁺ could greatly inhibit the activity of a β-galactosidase from Alteromonas sp. ML52 over 32.4% (Sun et al., 2018). The results further showed that zinc ion plays an important role in Gal42, and it is a novel enzyme with different characteristics from other enzymes belonging to GH42.

The presence of ions often affects the activity and stability of β-galactosidase. The β-galactosidase obtained from bacterial strain Erwinia sp. E602 (Xia et al., 2018) was also obviously inactivated by Cu²⁺ with about 5% activity. Mg²⁺ and Na⁺ are both required to achieve the maximal activity of the β-galactosidase (Juers et al., 2001). The active site of Mg²⁺ can be substituted by Mn²⁺ without many differences in activity. In this study, Gal42 showed obviously enhancing activity when Mn²⁺ and Mg²⁺ presented, and the presence of Na ⁺ also increases the activity of Gal42. Mn²⁺ and Mg²⁺ are beneficial to many reported β-galactosidase activities, such as enzymes from Pediococcus pentosaceus and Pediococcus acidilactici (Lee et al., 2017; Chanalia et al., 2018). In the aspect of structural analysis, Mg²⁺ binds to each subunit and has the largest influence on the nearest substrate binding site. It also can modulate the chemistry of active site components (Lo et al., 2010). Another important role of Mg²⁺ is that it can stabilize a mobile loop at the active site through interacting with Glu residue (Sutendra et al., 2007). What is more interesting is, in the presence of Zn²⁺, the activity and thermo-stability of Gal42 were significantly increased. There are three types Zn²⁺ binding sites in proteins: catalytic, co-catalytic, and structural. Among them, the Zn²⁺ binding is coordinated by four amino acid side chains, including either four cysteines or two histidines in combination with two cysteines (Shumilina et al., 2014). Maksimainen et al. (2012) have demonstrated that Zn²⁺ does not play a role in the catalytic cycle in Bca-β-gal; however, the binding of Zn²⁺ stabilized the structure of the catalytic domain. In the same sense, our experimental results showed that the activity and thermal stability of Gal42 were improved in the presence of zinc ion (Figure 4). The improved thermo-stability of Gal42 with the presence of Zn²⁺ may be due to the greatly increased stability of Zn²⁺ binding site (Cys113, Cys153, Cys155, and Cys158) (Figure 2).

In this study, lactose and milk were used as the substrates, and the reaction was carried out for a certain time under the catalysis of Gal42 to detect and analyze the decomposition of lactose. TLC was used to analyze its reaction product. As shown in Figure 5A, monosaccharides are formed by hydrolysis of lactose during the reaction. After 360 min, lactose had been fully hydrolyzed into monosaccharides. The reaction products were further identified by negative ESI-MS. As shown in Figure 5C, the main peak at 341.1 and 377.1 m/z corresponded to lactose, and 179.1 and 223.0 m/z corresponded to galactose. These results indicated that the lactose almost fully hydrolyzed into galactose within 360 min.

According to the properties of Gal42, the enzyme activity can be relatively stable in a certain range of pH (covers the pH range of milk). Therefore, it was completely feasible to use Gal42 to hydrolyze lactose in milk in industrial application. In this study, it could be observed that the lactose content in milk can decreased with the enzyme reaction of Gal42. The reaction was completed within 60 min, and all lactose was decomposed into monomers (Figure 5B). In conclusion, this characteristic indicates that Gal42 may have a good application prospect in dairy industry.