Three major databases, AlphaFoldDB, UniProt and Swiss-Model, were searched for proteins with high similarity to MlrB from Sphingopyxis sp. USTB-05. Physicochemical properties of enzymes are available from the ExPAsy ProtParam online server. Signal peptide and transmembrane region structures were predicted using SignalP 5.0 and TMHMM 2.0, respectively. The Protein Structure Prediction Server (PSIPRED) and SOPMA were used for online secondary structure prediction. In the normal mode of Phyre2 website (Kelley et al., 2015), the full-length amino acid sequence was input for tertiary structure prediction and model construction, and the simulation results from the AlphaFold Protein Structure Database was compared. All software packages build the MlrB model based on the same template, and the final model evaluation is presented in a Ramachandran plot.

The molecular structures of linearized MC-LR, linearized MC-RR and the docking ligands of linearized microcystinase were drawn using software. Autodock Vina was used to calculate the binding modes and the affinities of the linearized MCs to MlrB (Trott and Olson, 2010). The ligand was hydrogenated and the receptor protein was subjected to pre-processing such as adding polar hydrogen and calculating charge, and then the docking box was built (complete wrapping of the receptor protein). For each initial conformation, up to 9 generated binding conformations can be recorded. The docking results of the optimal binding model between receptor and ligand were visually displayed using Pymol 2.3 (DeLano, 2002).

Based on our previous studies cloning and expressing MlrB from Sphingopyxis sp. USTB-05, derived from freshwater lake substrate, for linearized MC-RR biodegradation (Wang et al., 2015). Initially, the linearized microcystinase gene mlrB from Sphingopyxis sp. USTB-05 was amplified via polymerase chain reaction (PCR). In the second step, an overlap-extension PCR was used to construct the fixed point mutation of the mlrB gene. The oligonucleotide primers are listed in Table 1, and a detailed oligonucleotide primer scheme is provided by Sambrook and Russell (2016). The PCR products and the pET30a (+) vector plasmid were digested with BamH I and Xho I restriction enzymes, respectively, and ligated with T4 DNA ligase. The constructed recombinant plasmid was transformed into E. coli DH5α cloning vector and positive clones were screened in antibiotic LB medium and sequenced by Sangon Biotech (Shanghai) Co., Ltd. The plasmid was extracted from E. coli DH5α/pET30a (+)/mlrB using the TIANprep mini plasmid kit (Tiangen, Beijing China) and transferred into the E. coli BL21 (DE3) expression vector, and positive clones were selected and further verified by gene sequencing. Consistent with previous work, the recombinant MlrB mutants (MlrB S77A, MlrB K80A, MlrB Y171A, MlrB N173A, MlrB D245A, MlrB H307A) were constructed and expressed successful (Wang et al., 2015). In the case of MlrB S77A, for example, the mutation is the replacement of the Ser at position 77 with an Ala, and so on. The bacteria cells were collected and washed three times using phosphate buffer solution (PBS, pH 7.4) and subsequently frozen in −20°C.

Buffers were precooled to 4°C and all subsequent steps were performed on ice. Afterward, the cell pellet was resuspended in PBS (pH 7.4). Cells were disrupted by sonication and cell debris was removed by centrifugation (20 min, 22,351 ×g, 4°C). After filtration, the supernatant was mixed for 1 h with a Ni-NTA agar resin (Sangon Biotech, Shanghai, China). As a result, MlrB enzyme was firmly attached to the resin. The resin, which contained bound target proteins, was washed with a gradient of 5–10 column volumes of imidazole-containing PBS (pH 8), and the mobile fraction’s absorbance at 280 nm was monitored using Nanodrop (Thermo Fisher Scientific, Waltham, MA, USA) until it approached baseline level. Following that, the target proteins were isolated using PBS (125 mM imidazole, pH 8.0). The target proteins were ultimately removed imidazole and concentrated using ultrafiltration. It was necessary to assess the expression level of active MlrB by quantifying total soluble protein concentration according to Bradford (1976). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to check purified protein quality.

The MlrA, prepared according to previous methods, was used to produce the first biodegradation products (linearized MCs) of 12.5 mg/L MC-LR and MC-RR (Xu et al., 2020). MC-LR and MC-RR were each mixed with 53.8 mg/L of MlrA and allowed to fully react at 30°C, 200 rpm for 3 h to produce linearized MCs with a final concentration of 12.5 mg/L. In order to verify the biodegradation activity of the recombinant MlrB enzyme toward the linearized products, two experimental groups and two control groups were formed by adding linearized MC-LR and MC-RR to solutions containing the MlrB enzyme (at a final concentration of approximately 1 mg/mL) and without the enzyme (control). Incubate the reaction system [30°C, 200 revolutions per minute (rpm)] and remove 200 μL of sample to a centrifuge tube at 0, 5, 15, 30, 45, and 60 min (min), then immediately add 2 μL of concentrated hydrochloric acid to stop the enzymatic reaction. All samples were centrifuged at 14,000 rpm for 20 min. Following the procedure described above, the activities of the six MlrB enzyme mutants were determined. Processed samples were stored at 4°C for subsequent monitoring of linearized MCs concentration by high-performance liquid chromatography (HPLC). The HPLC system (Shimadzu LC-20AT, Shimadzu Co., Ltd., Tokyo, Japan) was equipped with an ultraviolet Diode Array Detector at 238 nm using an Agilent TC-C18 column (4.6 mm × 250 mm; Agilent, 1200 series, Wilmington, DE, United States), in which the mobile phases for linearized MC-LR and MC-RR were 36% and 40% (v/v) acetonitrile-water solution containing 0.05% (v/v) trifluoroacetic acid, with a flow rate of 1.0 mL/min and an effective injection volume of 20 μL. All other reagents are chromatographic grade.

To comprehensively elucidate the physicochemical properties of the MlrB (also known as USTB-05-B enzyme in Sphingopyxis sp. USTB-05) enzyme and make systematic structural predictions, a range of bioinformatic analysis software were used, including ProtParam, ProtScale, SignalP 5.0, TMHMM 2.0, PSIPRED, SOPMA, Phyre2, SWISS-MODEL, AlphaFoldDB and SAVES. ProtParam analysis revealed that the MlrB enzyme consists of 541 amino acid residues, of which alanine (11.5%) is the most abundant and cysteine (0.4%) the least abundant. There are 65 positively charged amino acid residues and 66 negatively charged amino acid residues. The theoretical isoelectric point (pI) of MlrB is 6.78, which is clearly different from most of the proteins in Table 2 and suggests that it is a biased acidic protein. The molecular formula is C₂₆₃₅H₄₁₆₀N₇₄₆O₇₇₉S₁₀. The half-life in E. coli was more than 10 h, the in vitro instability coefficient was 24.85 (less than 40), and the aliphatic index was 89.48 (greater than 80), indicating that the protein was stable and beneficial for in vitro experiments (John Flensburg, 2005).

The amino acid sequence of USTB-05-B was analyzed using SignalP 5.0 signal peptide software. There is a signal peptide sequence of MlrB, which consists of amino acids at positions 1 to 19, with a break point between serine at position 19 and histidine at position 20 of the sequence, and then the amino acids at positions 20 to 541 to form a mature chain. The TMHMM2.0 server predicted no transmembrane sequence in MlrB, and MlrB was assumed to be a secreted protein. The hydrophilicity of MlrB was predicted with ProtScale software. The number of hydrophilic amino acid residues in MlrB’s polypeptide chain was significantly higher than that of hydrophobic residues, suggesting MlrB is a hydrophilic protein. Based on SOMPA prediction analysis, MlrB contains four secondary structures, alpha helix (30.87%), random coil (42.88%), extended strand (19.04%), beta turn (7.21%), indicating that the major component of the secondary structure is the random coil, followed by the alpha helix. PROSITE analysis indicates that MlrB belongs to the class A beta-lactamases and possesses one active site situated at amino acid positions 73–88.

Phyre2 was used for the MlrB similarity search and the coverage of the first high score template with MlrB was 96%. MlrB shows 100% homology with the template and 91% of protein residues are modeled with 100% confidence. The highest-scoring template (Protein Data Bank database ID: 1ei5; chain A) used for modeling was the D-aminopeptidase (DAP) from Brucella anthropi, which belongs to the new member of the “penicillin-recognizing enzyme” family (Bompard-Gilles et al., 2000). The sequence identity of the template to MlrB is 25%, which is below the lower limit of expected accuracy (typically 30%; Fiser, 2010), but the high confidence level observed (>91%) suggests that the target MlrB sequence may be a true homolog of the template, and therefore ensures the reliability of the modeling. The model was evaluated using the Ramachandran plot, revealing that 88.1% (378/429) of amino acid residues were located in the most favored regions and 98.1% (421/429) of amino acid sites were located in permissive regions in the model constructed by phyre2 (Figure 1A). Correspondingly, alphafold model (accessions:M4NG93) shows that 89.6% (412/460) of the amino acid residues are in the most favorable region, with the percentage increases to 100% in the permissive region (Figure 1B). The results demonstrate that the model constructed by the latter is more accurate and reliable, making it suitable for further docking analyses.

Multiple sequence comparison of the template enzyme DAP (1ei5, chain A) with MlrB (AGG86526.1) is shown in Figure 2. The alpha helix is represented by the green helix, the β-folding by the blue arrow and the conserved active sites of DAP by the five red boxes. The six red arrows point to possible key sites for MlrB, serine at position 77 (S77), lysine at position 80 (K80), tyrosine at position 171 (Y171), asparagine at position 173 (N173), aspartic acid at position 245 (D245) and histidine at position 307 (H307).

The template’s high homology and coverage allow us to speculate about the six potential key sites of MlrB (S77, K80, Y171, N173, D245 and H307) based on DAP’s mechanism of action. The results of molecular docking further substantiated our hypothesis regarding the enzyme catalytic process through computer simulation. The docking scores of the linearized microcystinase model with linearized MC-LR and MC-RR were -8.8 kcal/mol and -8.3 kcal/mol, respectively, and the binding energies were much less than -5 kcal/mol, indicating that the enzyme binds to the substrate with high affinity (Ge et al., 2023). As depicted in Figure 3, the substrate molecules were fully accommodated within the active pocket. Interestingly, the spatial positions of the six active sites in MlrB enzyme coincide with the positions of the catalytically active residues in the template enzyme DAP. It was evident that Y171 occupied an optimal position with respect to the substrate, while the other five amino acids surrounded the substrate molecules. D245 is positioned close to the N terminus of the substrate, whereas S77, K80 and H307 are situated on one side of the substrate, with Y171 and N173 located on the opposite side, thereby encapsulating the substrate within. Analysis of receptor-ligand interactions revealed multiple direct hydrogen bonds between Y181 and the substrate molecules, which could potentially be crucial for catalysis. However, it should be noted that molecular docking is a computational method that often lacks receptor flexibility and thus introduces uncertainty into protein-ligand complex reliability (Bompard-Gilles et al., 2000). Therefore, it is necessary to supplement experimental evidence to ensure the theoretical conjecture. Using genetic engineering techniques, mutant enzymes can be constructed, purified and subsequently mixed with linearized MCs to verify each site’s role in degradation experiments, thereby revealing the mechanism of linearized MCs biodegradation by enzyme MlrB.

Based on the analysis of the molecular docking binding pattern results, six mutants of MlrB were constructed using previous method (Teng et al., 2023). After inducing the expression of the target protein, we conducted chromatography purification using affinity, concentrated the sample by ultrafiltration, and analyzed by SDS-PAGE. Successful purification of the six mutants and a positive control was confirmed, as shown in Figure 4. Lane 1 represents the control group (wild-type MlrB), while lanes 2 to 7 represent the mutant groups. The molecular weight of the MlrB is theoretically 59 kDa. It should be noted that the expected band, ranging from 50 kDa to 65 kDa, has been obtained and underlined, MlrB and its mutations were purified successfully.

The biodegradation kinetics of MCs shown for linearized MC-LR, point mutants of S77A, K80A, Y171A, N173A and H307A resulted in a complete loss of enzyme activity (Figure 5A). The D245A mutation greatly reduces MlrB activity, requiring 1 h to fully biodegrade linearized MC-LR. Interestingly, aspartate also is not a major site for DAP (Bompard-Gilles et al., 2000). For linearized MC-RR, S77A, K80A, Y171A, N173A, H307A point mutants also resulted in complete loss of enzyme activity (Figure 5B). D245A mutations significantly reduced MlrB activity.