Sequence analysis and identification of PcMan113
A novel endo-1.4-β-mannanases named as PcMan113 was found in a proprietary database derived from a Paenibacillus cineris. The PcMan113 sequence was 987 bp, encoding a protein of 329 amino acid residues with a pI and molecular weights predicted to be 4.54 and 39.6 kDa, respectively. The protein showed 65% similarity with a well characterized endo-1,4-β-mannanase from Bacillus sp. N16-5 (Liu et al. 2021). BLAST searches against the NCBI nonredundant and CAZy databases (Lombard et al. 2014), combined with conserved domain analysis, implied that PcMan113 belongs to the GH113 family. A phylogenetic tree was constructed based on the amino acid sequences of several similar proteins and representative β-mannanases form the GH 5, GH 26, and GH 113 families using the neighbor-joining method, which further revealed that PcMan113 clustered with the GH113 like β-mannanase BaMan113A (Fig. S1).
PcMan113 was therefore identified as a novel member of GH family 113 with low sequence identity. The PcMan113 was predicted to be an intracellular enzyme according to SignalP analysis and it had no signal peptide. Sequence alignment with other identified GH113 family mannanases showed that several conserved residues (Fig. 1), and two catalytic residues (proton donor, E152; nucleophile, E232) and ten chemical substrate binding site residues (N98, W104, R105, E152, D183, Y204, E232, W282, D283, Y300) were strictly conserved among the GH113 family members.
Expression, purification and identification of PcMan113
PcMan113 was successfully overexpressed in E. coli BL21(DE3) and purified using Ni-NTA affinity chromatography (Fig. 2a). On size-exclusion chromatography, PcMan113 eluted as a single peak between the standard protein markers rabbit muscle actin (15.7 kDa) and conalbumin (43.0 kDa) (Fig. 2b), which indicated that purified PcMan113 was a monomer in solution. Based on these results and the SDS-PAGE, which revealed a single protein band, the mass of the purified protein was in good agreement with the predicted molecular weight of 39.6 kDa (Fig. 2).
Characterization of purified PcMan113
Using LBG as the substrate, the PcMan113 exhibited an optimal activity in Na2HPO4-citrate buffer at pH 5.0 (Fig. 3a) and maintained more than 80% of its maximal activity at pH values between 4.5 and 7.0. Moreover, the enzyme also showed considerable stability over an extensive pH range from 3.0 to 8.0, retaining over 60% of its initial enzyme activity after incubation without substrate at 37°C for 1h (Fig. 3b).
The temperature-activity profile indicated that PcMan113 was a typical mesophilic hydrolase, exhibiting the highest activity at 55°C (Fig. 3c) and thermostability at 45°C in Na2HPO4-citrate buffer. It retained more than 60% relative activity at temperatures ranging from 40 to 65°C (Fig. 3d). However, when incubated at 70°C, the enzyme activity declined rapidly.
The effects of metal ions and the chelator EDTA on purified PcMan113 were explored by incubating the enzyme at 55°C for 1.0 h. The results showed that 5 mM Zn2+ could slightly enhance the enzyme activity to 115%, while Cu2+ and Ag+ greatly inhibited the enzyme activity. Moreover, Mn2+ and Fe2+ reduced the enzyme activity by approximately 10%, and EDTA also moderately inhibited its activity (Fig. S2).
Substrate selectivity and kinetic properties of PcMan113
To determine the substrate selectivity of PcMan113, we incubated purified PcMan113 with various substrates including LBG, KG, GG, sodium carboxymethyl cellulose (SCC), beechwood xylan (BX), mannobiose (M2), mannotriose (M3), mannotetraose (M4), mannopentaose (M5) and mannohexaose (M6). The purified PcMan113 showed maximal activity towards M5 (8.2×103 U/mg), followed by M4 (7.5×103 U/mg), M6 (6.9×103 U/mg), M3 (3.2×103 U/mg) and M2 (2.6×102 U/mg). The purified PcMan113 showed much lower activity towards KG (68.2×102 U/mg), LBG (31.7 U/mg) and GG (2.1 U/mg) orderly (Table1), and no activity was observed with SCC or BX. The specific activity of PcMan113 toward KG was higher than toward LBG or GG, which was consistent with other enzymes from the GH 113 family, such as BaMan113A from Bacillus sp. N16-5 (Liu et al. 2021), as well as AaManA (David et al. 2018) and Man113A from Alicyclobacillus sp. strain A4 (You et al., 2018). By contrast, mannanases from GH families of 134, 26 and 5, such as AoMan134A (Hogg et al. 2003), CrMan26 (Mandelli et al. 2020), and RmMan5A (Song et al. 2018), exhibit a high preference for LBG. PcMan113 and other GH family mannanases have different preferences toward LBG and KG, which are related to structural differences in space for accommodating the substrate with galactose side chains in a non-catalytic binding mode (Jin et al. 2016; Kumagai et al. 2015; Le et al. 2005).
The kinetic properties of purified PcMan113 were analyzed using mannooligosaccharides as substrates, including mannobiose (M2), mannotriose (M3), mannotetraose (M4), mannopentaose (M5) and mannohexaose (M6). The kinetic parameters of PcMan113 for these substrates were calculated by nonlinear fitting to the Michaelis-Menten equation (Table 2) (Kumagai et al. 2015) The Km values for M2, M3, M4, M5 and M6 were 66.1, 35.4, 5.1, 4.7 and 8.2 mM, respectively, indicating that PcMan113 has the highest affinity for M5, followed by M4 and M6. Moreover, PcMan113 showed higher catalytic efficiency with M5 (6.55 mM−1s−1) and M4 (5.19 mM−1s−1) than with M6, and the lowest catalytic efficiency with M2 (0.035 mM−1s−1).
It was reported that GH113 family members showed considerable transglycosylation activity and could hydrolyze mannooligosaccharides with degrees of polymerization (DP) above 3 (Zhang et al. 2008; Xia et al. 2016). Our results indicated that the smallest substrate of PcMan113 was M3 (Table 1 and 2), which was consistent with the catalytic properties of other GH113 family members. Compared to other endo-β-mannanases, PcMan113 showed poor catalytic efficiency on M2 (0.035 mM−1 s−1) and M3 (0.20 mM−1s−1), and the catalytic efficiency toward M3 much lower than that of Man113A (6.54 mM−1s−1) (Xia et al. 2016) and PaMan5A (26.67 mM−1s−1) (Couturier et al. 2013). The enzyme hydrolyzed M2 slowly (data not shown), suggesting that the hydrolysis products of PcMan113 would be mainly M2 and mannose, which makes it suitable to degrade mannans.
Structural homology analysis and active site screening of PcMan113 protein
A structural homology model of PcMan113 was generated using BaMan113A (PDB ID: 7DV7) as the template. The Ramachandran plot indicated that 97% of the residues were in the favored and allowed regions, indicating that the model of PcMan113 is plausible. PcMan113 possesses two polypeptide chains in an asymmetric unit, each folding into a typical (β/α) 8TIM-barrel architecture (Fig. 4a and Fig. S3), which is consistent with the typical features of GH113 family enzymes (You et al., 2018). The catalytic residues E152 and E232 are located at the center of the β-barrel, which forms a hydrophobic area near which a deep cavity is observed. The PcMan113 may have a semi-enclosed substrate-binding pocket cleft, which was occupied by amino acids residues such as N98, W104, R105, F110, D183, Y204, W282, D283, Y300, which formed the substrate binding site of PcMan113 (Figs. 4b and S4).
Thus, this cavity accommodates the substrate and these residues interact with the substrate. The inter-molecular interface involves four regions, T58-H60, F110-E118, V154-Q155, and K184-Q186, which are located in loops (Fig. S5). Among them, loop F110-E118 might act as a lid that seals the active site, which has an effect on the entrance of substrates and release of products.
Rational engineering of PcMan113 enzyme
Based on the discussed structural features of PcMan113 enzyme, a single point mutant library containing 23 single mutants at 9 residues was constructed and evaluated for activity towards the industrial substrates KG and LBG (Fig. 5a). Mutation of residues W20, W104, E118, P119, K184, W248 and W284 reduced the enzyme activity, indicating these residues played important roles in substrate binding. By contrast, mutation of residues F110 and N246 enhanced activity towards substrates of KG and LBG, and these two residues were selected as targets for site-directed mutagenesis.
On the basis of substrate-binding pocket analysis, the six mutants F110A, F110E, N246A, N246F, N246W and N246Y were constructed and evaluated. Among them, the four mutants F110E and N246F/W/Y showed higher catalytic activity towards KG and LBG than wild-type PcMan113 (WT) (Fig. 5a). The mutant F110E (PcMT1) showed respective 2.0- and 2.2-fold increases of catalytic activity toward KG and LBG compared with the WT. PcMT1 showed high substrate affinity and catalytic efficiency with a Km of 1.40 mM and a kcat of 35.20 s−1, respectively (Table S1). The biochemical data revealed that the F110-E118 loop may affects enzymatic hydrolysis by interacting with F110 and E118 from the adjacent monomers. In addition, the loop F110-E118 was proposed to interact with the substrate and active sites, and F110 was mutated to the acidic residue glutamate or nonpolar residues with small side chains, which may affect substrate recognition and enzymatic hydrolysis.
Mutating N246 to aromatic residues led to a slight reduction in enzyme activity (Fig. 5a), and some studies reported that Trp-mediated binding plays an important role in β-mannanses of GH113 (Kumagai et al. 2015; Xia et al. 2016; Liu et al. 2021), suggesting that the stacking interaction at this position is conducive to improving enzyme activity. Therefore, N246 was replaced with Phe, Trp and Tyr in three different mutants. The most remarkable results were observed for N236Y (PcMT2), which exhibited 233% and 246% relative activity with KG and LBG, respectively (Fig. 5a). The enzyme kinetic parameters of PcMT2 were significantly enhanced, which much higher substrate affinity and catalytic efficiency on M5 (Km, 1.8 mM; kcat/Km, 19.30) compared with the WT (Table S1). These results were in agreement with previous reports that improving stacking interactions between the substrate and protein at the position of N246 could significantly enhance the catalytic activity (Liu et al. 2021; You et al. 2018; Xia et al. 2016).
Furthermore, the mutations with improved activity on KG and LBG were combined, resulting in three additional double mutants of PcMan113 (F110E/N246F, F110E/N246Y and F110E/N246W). The enzyme activity measurements reveled that all double mutants had much higher hydrolysis activity towards LBG and KG than the WT and the corresponding single point mutants (Fig. 5b). The F110E/N246Y mutant (PcMT3) not only exhibited a great improvement in hydrolytic activity (4.6- and 3.5-fold respective increases with KG and LBG), but also showed higher substrate affinity and catalytic efficiency (Km, 0.9 mM; kcat/Km, 37.44) with M5 than the WT enzyme (Fig. 5b and Table S1). However, these results were different from similar mutants of BaMan113A, which exhibited reduced substrate affinity and catalytic efficiency [35]. Thus, PcMT1, PcMT2 and PcMT3 had clearly increased activity in the hydrolysis of mannans and manno-oligosaccharides, which indicated that the substrate preference of the enzyme has not been altered. At the same time, these data also demonstrated that PcMan113 hydrolyzed substrates in a standard endo-acting mode, indicating that PcMan113 can degrade mannans more completely, which was similar to the reported enzyme Man113A (Xia et al. 2016).
MD simulations and structural analysis of the PcMT3 variant
To further explore the differences between the WT enzyme and improved variant PcMT3 in catalytic activity and protein-substrate interactions, mannobiose was docked into the binding sites of WT and PcMT3 in 30 ns MD simulations. The results indicated that the binding energy of PcMT3 with mannobiose was dramatically increased compared with the WT (−7.85 kcal/mol vs. −37.83 kcal/mol) (Table S2). The root mean-square deviations (RMSD) over 30 ns were calculated to investigate the stability of the WT/PcMT3-mannobiose complexes (Fig. S6a). PcMan113-WT reached the equilibrium state after 15 ns of simulation, while Pc-MT3 was stable state after 10 ns of simulation. The average RMSD values of variants PcMT1/2/3 were 0.26±0.06 nm, 0.30±0.06 nm and 0.25±0.04 nm, respectively, all of which were lower than that of the Pc-WT complex (0.32±0.08 nm), indicating that all variants were more stable.
Moreover, the root mean square fluctuations (RMSF) were calculated to assess the mobility of protein residues, and the results showed that both the WT and the variants had low fluctuations ranging from 0.2 to 0.3 nm (Fig. S6b). Compared with the WT, some residues in loops that are part of the substrate binding site (such as F110 and N246) of PcMT3 had higher RMSF values, reaching >0.4 nm in loop R243-N255, and >0.3 nm in loop F110-E118, indicating greater flexibility in these loops than in the WT (Fig. S6b). The radius of gyration (Rg) values are an indicator of the reliability of protein homology modeling, and the Rg values of WT-PcMan113 and variants all reached between1.9 and 1.95 nm (Fig. S7), suggesting that the homology modeling was reliable.
In the WT-mannobiose complex, a narrow substrate access channel to the active site was observed (Fig. 6a). By contrast, the mutations in PcMT3 led to a somewhat bigger space in the substrate binding channel and the substrate could easily enter the catalytic center (Figs. 6b, 6d and 7). Based on the previous structural analysis of the homology model and the N246Y-mediated structural effects on the substrate-binding tunnel, we hypothesized that N246 may play a pivotal role in blocking the access of the substrate into the active site, while the newly introduced Y246 side chain adds a stacking interaction for substrate binding. The N246Y mutation possibly removed steric hindrance to widen the available substrate-binding space inside the binding pocket (Fig. 6a−d). Indeed, the N246Y mutant exhibited a 46% activity improvement with LBG compared with the WT, which was also in agreement with the MD simulations (Fig. S6 and Table S2). When F110 was mutated to Glu in PcMT3, its enzyme activity was further enhanced, and its RMSD indicated that the F110-E118 loop had greater flexibility (Fig. S6b). We therefore hypothesized that loop F110-E118 possibly interacted with the substrate and active site (Figs. S5 and 7). In addition, when site of F110 was mutated to E, E110 was found has a small side chain, which may affect substrate recognition and allow easy entrance into the binding pocket.
Furthermore, we also observed an interesting conformational change in residue W248, which was altered upon ligand binding (Fig. 7). This resulted in a more conducive conformation of the mannobiose binding channel, bringing the residue W248 that was proposed to act as a lid of the substrate-binding channel in close proximity to the substrate in the active site.