The GH family has potential

The GH family has potential applications in various industries. A putative ORF encoding α-galactosidase, namely GalR, with a low identical match and e-value and high query coverage in the metagenome was selected because it demonstrated considerable activity. GalR shared 65% identity with the GH 27 protein of M. polysiphoniae, which was not experimentally characterized. Much research has characterized GH 27 α-galactosidase in Eukaryota, such as Hericium erinaceus and Fusarium oxysporum fungi, which usually have a small molecular weight and are monomeric proteins 32, 33. According to the Naumoff's study (34), GH 27 can be classified into three major subfamilies (GH 27a, GH 27b, and GH 27c). GH 27a enzymes containing four cysteine residues in the N-terminus are highly conserved to form two disulfide bridges for the maintenance of catalytic structure (33). However, only three cysteine residues were observed in GalR, and no disulfide bridge could be determined according to the DISULFIND prediction (35). In addition, on the basis of the in silico prediction of SignalP 4.1 Server, the first 25 (+)-Apogossypol of GalR with the high hydrophobic amino acids were assumed to be the signal secretion sequence (36). The C-terminal domain of GH 27 containing a β-sandwich structure was less conserved and its function was still unknown (37). In this study, based on the prediction of protein structure, GalR did not have a C-terminal β-sandwich domain (Fig. S2). Nevertheless, the hydrolysis activity of α-1,6-linked galactoside moieties from galacto-oligosaccharides can be identified in GalR. This suggested that a β-sandwich domain did not involve in the enzymatic function of GalR. Many studies have been performed regarding the heterologous overexpression of α-galactosidase in E. coli. However, incompatibility of codon usage bias can cause a decrease in recombinant protein productivity or protein misfolding, resulting in the formation of an inclusion body (38). In this study, to improve the protein productivity, a gene optimization program using synonymous codon substitution by OptimumGene was performed for the GalR design. Our result revealed that the optimal GalR accounted for >1.7 times the original GalR protein content with high translational efficiency, which might result from the codon usage bias in E. coli (Fig. 3) (38). According to our review of relevant literature, five reports have been published on the characterization and overexpression of GH 27 α-galactosidases in E. coli 33, 39, 40, 41, 42. The optimal temperature and pH of the six reported α-galactosidases were detected at 35°C-50°C and pH 4–8.2. Four of these α-galactosidases from Bacteroides fragilis, F. oxysporum, and rice were acidophilic between pH 4 and 5. By contrast, our study indicated that GalR was alkaliphilic at the optimal pH 9 (Fig. 4). The decrease of lysine-aspartate ion pairs replaced by the increase of arginine-glutamate ion pairs and high hydrophobic contacts were important for stabilizing proteins in alkaline media (43). Comparing the amino acid composition between these α-galactosidases in GH 27, the lysine frequency of GalR was lower than that of B. fragilis, F. oxysporum and rice (α-Gal III). Furthermore, GalR showed high arginine and glutamate residues. The 36.4% frequency of hydrophobic residues, including Ala, Ile, Leu, Phe, Trp, and Val in GalR, was higher than that of the other reported α-galactosidases. These factors may reinforce the interactions of protein residues and contribute to enzymatic stabilization in alkaline conditions (43). The 50% relative activity of GalR was attained for 1.7 and 0.7 h preincubation at 40°C and 50°C. The result showed higher thermostability than those of previous studies in GH 27 α-galactosidases 39, 42. This may result from the thermostability characterizations containing nonpolar residues (Val, Ala, Leu, and Ile) and aromatic residues (Phe and Trp) for enhancing hydrophobicity (44). In addition, determination of the enzymatic sensitivity to different reagents, and metal ions indicated that GalR was gradually inhibited with the increase of EDTA, implying the involvement of metal ions in the enzymatic activity (Fig. 5). This was consistent with a GH 27 α-galactosidase from B. fragilis, whereas the enzymatic activity of several GH 36 α-galactosidases from Bacillus coagulans, Bacillus megaterium, and Sulfolobus solfataricus was not altered in the presence of EDTA 40, 45, 46, 47. It suggested that the different GH families indeed display different enzymatic characteristics. Moreover, Ca2+ had no significant effect on GalR, whereas Na+ and K+ slightly promoted enzyme activity. This was similar to the property of α-galactosidase from F. oxysporum (33). GalR had higher tolerance to organic solvents of methanol and ethanol than to Mg2+, H2O2, and SDS. Furthermore, GalR was also active against raffinose (Fig. 6), but its activity was lower than that of pNPGal utilization and was similar to that of other GH 27 enzymes 33, 39, 41.