Identification and sequence analysis of CrmE10 and AlinE4
To compare the enzymatic activities of different esterases, the open reading frames of CrmE10 (Accession number: ARU15426.1) and AlinE4 (Accession number: WP_160739227) were identified from the whole-genomes of strains C. marinus E4A9T and A. indicus DSM 18604T, respectively. CrmE10 and AlinE4 have similar amino acid sequences (59.66% identity, 85% coverage). To reveal the relationship between CrmE10 and AlinE4, we performed a phylogenetic analysis with other known lipolytic enzymes using MEGA software. The results showed that both CrmE10 and AlinE4 belong to the SGNH-hydrolase superfamily, and the bacterial lipolytic enzyme GDSL family (Fig. S1). The presence of strictly conserved Ser (residues Ser29 in CrmE10 and Ser13 in AlinE4), Gly (residues Gly66 in CrmE10 and Gly55 in AlinE4), Asn (residues Asn97 in CrmE10 and Asn81 in AlinE4), and His (residues His181 in CrmE10 and His165 in AlinE4) in blocks I, II, III, and V, respectively, confirmed that these two enzymes were new members of the SGNH-hydrolase family  (Fig. S2).
Biochemical characterizations of CrmE10 and AlinE4
To better understand the catalytic mechanism for CrmE10 and AlinE4, both enzymes were expressed and purified in E. coli. CrmE10 exhibited the highest activity (approximately 29.4 U/mg) toward p-nitrophenyl (p-NP) hexanoate (C6) at pH 7.5 and 20 °C (Km, Vmax and kcat of 0.16 mM, 33.5 µmol/mg/min and 29.4 s-1, respectively) (Fig. 1A, B, and C). AlinE4 displayed the highest activity (approximately 25.8 U/mg) toward p-NP butyrate (C4) at pH 7.5 and 40 °C (Km, Vmax and kcat of 0.10 mM, 26.9 µmol/mg/min and 25.8 s-1, respectively) (Fig. 1A, B and C). CrmE10 had about 30% activity at 4 °C and showed no activity above 40 °C, which indicated that CrmE10 was a cold-active enzyme. In contrast, AlinE4 was a mesophilic enzyme with over 32% activity above 60 °C. Interestingly, AlinE4 showed a relatively high thermostability, evidenced by that the enzyme activity was hardly affected after heat treatment at 70 °C for 1 h, and retained 20%-40% activity after heat treatment at 100 °C for 1 h (Fig. 1D and E).
CrmE10 showed tolerance to low NaCl concentrations and retained over 83% and 41% of its initial activity at 1 M and 2 M NaCl, respectively (Fig. 1F). The activity of CrmE10 was abolished when NaCl concentrations increased to 5 M. AlinE4 exhibited higher NaCl tolerance, retained over 61% activity at concentrations up to 5 M (Fig. 1F). Metal ions Ba2+, Ca2+, Mg2+, and Sr2+ had minimal effects on the activities of the two enzymes (Fig. 1G). Cd2+, Cu2+, and Mn2+ completely abolished the enzymatic activity of CrmE10, but did not work for AlinE4 as these metal ions retained over 20%, 5% and 70% activities, respectively (Fig. 1G). The addition of Co2+, Ni2+, and Zn2+ decreased the enzymatic activity of CrmE10 over 10%, 35% and 30%, as well as for the activity of AlinE4 over 51%, 85% and 79%, respectively (Fig. 1G). The chelating agent EDTA did not decrease the activity of CrmE10 and AlinE4, which indicated that CrmE10 and AlinE4 were not metalloenzyme (Fig. 1G). Besides, compared with CrmE10, AlinE4 showed higher tolerance toward organic solvents, including acetone, ethanol, DMF, DMSO, glycerol, isopropanol, and methanol (15%, v/v) (Fig. 1H).
Overall Structures of CrmE10 and AlinE4
CrmE10 and AlinE4 were expressed in Escherichia coli BL21 (DE3) cells as described previously. They were further purified by metal ion affinity chromatography and gel-filtration, finally came out at peak positions of 15 ml (CrmE10) on a Superdex 200 10/300 column and 74 ml (AlinE4) on a Superdex 75 16/600 column, which corresponded to masses of around 22 kDa and 23 kDa, respectively (Fig. S3A and B). The theoretical molecular weights of CrmE10 and AlinE4 are 22.36 kDa and 20.59 kDa, respectively. The multi-angle light scattering (MALS) results indicated that these two enzymes were monomeric in solution (Fig. S3C and D).
To reveal the molecular basis of CrmE10 and AlinE4, we solved the crystal structures of the two proteins with high resolutions of 1.90 Å for CrmE10 and 1.18 Å for AlinE4, respectively (Table S1). The CrmE10 structure was determined using the molecular replacement method with esterase TesA (PDB code: 4jgg) as an initial model. In the crystal structure of CrmE10, two identical molecules were found in one asymmetric unit. The CrmE10 structure was composed of twelve α-helices (including three 310-helices) and five β-strands. The N-terminal region had a compact architecture that was mainly composed of five predominant β-strands (β1-β5) surrounded by eight α-helices (α1-α8). The C-terminal region was a long α-helix, α9 (Ala184-Asp203), that extended away from the subunit core and wrapped in the other chain. Thus, CrmE10 formed an intertwined dimer though the swapped C-terminal domain (Fig. 2A).
AlinE4 structure was determined by using esterase TesA (PDB code: 4jgg) as an initial search model, which revealed similar topology with CrmE10 (Fig. 2B). AlinE4 had one molecule in an asymmetric unit and was consisted of eleven α-helices (including two 310-helices) and five predominant β-strands. AlinE4 had a similar topology with CrmE10, as well as the α9 (Ala168-Ala185) wrapped in other chains and formed a symmetric dimer through the swapped C-terminal domain (Fig. 2C and D). The cap-domains and nucleophilic elbows were extensively important components of lipolytic enzymes in most other SGNH families; however, they were not presented CrmE10 and AlinE4 [31-33]. Furthermore, the PISA analysis for CrmE10 and AlinE4 on the PDBePISA server showed that the multimeric state was 2 for the two enzymes, and the buried areas between two subunits were 8150 Å2 (CrmE10) and 8010 Å2 (AlinE4), respectively. This suggested CrmE10 and AlinE4 were a dimer in crystal structure.
Structural comparison of CrmE10 and AlinE4 with other homologs
CrmE10 shared 28.42% (89% coverage), 32.04% (88% coverage) and 34.76% (80% coverage) sequence identities with its homologous EstA (PDB code: 3HP4), TAP (PDB code: 1IVN) , and TesA (PDB code: 4JGG) , respectively. AlinE4 shared 31.33% (87% coverage), 39.51% (89% coverage) and 39.26% (85% coverage) identities with these proteins, respectively. However, the overall structures among CrmE10, AlinE4, EstA, TAP, and TesA were very similar, evidenced by the low RMSD values of Cα atoms. For CrmE10 with EstA, TAP, and TesA, the values were 1.22 Å, 1.02 Å and 1.25 Å, respectively; For AlinE4 with EstA, TAP, and TesA, the values were 1.136 Å, 1.082 Å, and 0.979 Å, respectively. The obvious difference was at the loop between α8 and α9 and an α-helix (α9) in the C-terminal region (Fig. 3A). In CrmE10 and AlinE4, the loop and the α9-helix extended away and were wrapped with the other chain in the dimeric structure; whereas in EstA, TAP, and TesA, which have been proved that they can catalyze substrates by one molecule, this region was embedded inward and surrounded by predominant β-strands with other helices. However, in CrmE10 and AlinE4, the α8-α9 loop and α9-helix of the other chain were highly conserved with the same regions, suggesting that the catalytic reaction of CrmE10 and AlinE4 required coordination of two molecules, which might be a new mechanism of SGNH-hydrolase family esterases (Fig. 3B).
Dimerization contributed to the catalytic activities of CrmE10 and AlinE4
According to sequence alignment, the catalytic triad of CrmE10 consisted of Ser29, Asp178, and His181, and it was composed of Ser13, Asp162, and His165 in AlnE4 (Fig. S2). Mutants CrmE10-S29A, CrmE10-D178A, and CrmE10-H181A had no enzymatic activities, which confirmed that these residues were crucial for the activity (Fig. S4A). In CrmE10 and AlinE4 structures, catalytic residue Ser was located on helix α1, and residues Asp and His were located on the loop between helix η3 (CrmE10) or helix η2 (AlinE4) and helix α9, which were typically different from other esterases. Interestingly, the catalytic triads of CrmE10 and AlinE4 were not composed of Ser, Asp, and His from the same chain, which was common in other esterases (Fig. 4). For AlinE4, residues Ser13 and Asp162 on chain A and His165 on chain B were in a reasonable position of the catalytic triad, of which hydrogen bonds within the catalytic triad could be formed from Ser13-Oγ to His165-Nε2 and from His165-Nδ1 to Asp162-Oδ1 (Fig. 4B and 4D). In mutants AlinE4-S13A, AlinE4-D162A, and AlinE4-H165A, catalytic activities toward p-NP esters were almost abolished (Fig. S4B). Moreover, substrates p-NP butyrate and p-NP hexanoate could be successfully docked into the active sites of AlinE4 and CrmE10 by using AutoDock software, respectively (Fig. S4C and S4D). In CrmE10, the active sites Ser and His also formed hydrogen bonds between Ser29-Oγ and His181-Nε2 (3.4 Å /3.9 Å) (Fig. 4A and 4C).
As the catalytic triad was composed of the swapped dimeric structure, to further identify dimerization contributes to the catalytic activities of CrmE10 and AlinE4, mutagenesis analysis was performed for residues Asp178 and Ser29. The polycistronic plasmid CrmE10-W1 was composed of His-sumo-tagged WT CrmE10 and untagged mutant CrmE10-D178A, whereas the plasmid CrmE10-W2 was composed of His-sumo-tagged CrmE10 and untagged mutant CrmE10-D178A-S29A. The heated proteins did not have enzymatic activities and were used as a control. After expression and purification from E. coli, CrmE10-W1 had higher enzymatic activity than CrmE10-W2 (Fig. S4A). As mutants CrmE10-S29A and CrmE10-D178A had no activity, the results showed that residues Ser29 and Asp178 influenced the enzymatic activity of CrmE10 by domain swapping. Thus, dimerization participated in the active sites of CrmE10, and this mechanism might also fit AlinE4 due to the similar architecture, which was different from other homologs.
Metal ion Cd2+ affected the catalytic activity of AlinE4
The mechanism of SGNH-hydrolase family esterase activity involves a two-step-reaction (acylation and deacylation) similar to those proposed for lipolytic enzymes and serine proteases. In this reaction, Ser is a nucleophile residue, and His is the proton donor/acceptor. Many heavy metal ions have impacts on enzymatic activity, but the mechanism was not clear[12, 35-37]. There was one density map around the catalytic sites in the crystal structure of AlinE4, which turned out to be one Cd2+ evidenced by electrochemistry analysis (Table S2). According to the characterization of AlinE4, Cd2+ had negative impacts on the enzymatic activity of AlinE4, evidenced by only retaining 20% activity at 10 mM CdCl2 (Fig. 1G). In the crystal structure of AlinE4, Cd2+ interacted with residues Ser13 and His165, which were components of the catalytic triad (Fig. 4E). The results suggested that Cd2+ might act on activity through (i) blocking proton transfer and (ii) protecting substrates from nucleophile attack.
Structure-based mutation dramatically increased the enzymatic activity
CrmE10 and AlinE4 shared similar atomic architectures (RMSD value of Cα was 0.570 Å, Fig. 2C); however, the enzymatic properties exhibited significant difference, including substrate specificity, alkaline adaptability, temperature adaptability, metal ion tolerance, and organic solvent tolerance (Fig. 1A-1H). To further investigate the possible mechanism, we analyzed the sequences based on the 3D-structures and found five specific residues might contribute to these enzymatic property difference. The residues were acidic in CrmE10, including Asp77, Glu86, Asp123, Glu159, and Asp200, whereas the corresponding residues were basic in AlinE4, including Lys61, Lys70, Lys107, Lys143, and Lys184 (Fig. S2). The region of these five sites in CrmE10 formed a negative potential surface; however, the corresponding region was full of positive potential in AlinE4.
AlinE4 exhibited high alkaline adaptability evidenced by retaining about 60% activity at pH 10.5, whereas CrmE10 only retained about 10% activity at pH 8.5 (Fig. 1B). The enzymatic assay results showed that mutants AlinE4-K61D, AlinE4-K107D, and AlinE4-K143E retained about 45% or lower activity when pH was equal to or higher than 9.0 (Fig. 5C). The engineered enzyme CrmE10-mut5 (D77K/E86K/D123K/E159K/D200K) significantly increased to about 20% activity at pH 10, whereas CrmE10-E159K/D200K increased a little enzymatic activity at pH 9.0-9.5 and CrmE10-mut3 (D77K/E86K/D123K) had no difference as compare to wild-type CrmE10 (Fig. 5B), which meant these five sites synergistically participated in alkaline adaptability. The mutants of these five residues might change the surface charge of CrmE10 and increase its stability in an alkaline environment, therefore increasing its activity. AlinE4 had higher enzymatic activity towards long-chain substrates (C8, C10, C12 and C14) than CrmE10 (Fig. 1A). Compared with wild-type CrmE10, CrmE10-mut5 exhibited higher activity towards long-chain substrates (Fig. S5A), which suggested these five sites might contribute to substrate binding. Furthermore, the temperature preferences of mutants CrmE10-mut5 and CrmE10-E159K/D200K were different from wild-type CrmE10 (Fig. S5B). Mutants CrmE10-mut5 and CrmE10-E159K/D200K exhibited more than 50% activity with the addition of 10 mM Cd2+ or Mn2+, which completely inhibited wild-type CrmE10 activity. Moreover, these two engineered enzymes had lower activities with the addition of 10 mM Ni2+ than wild-type CrmE10, which was similar to AlinE4 (Fig. S5C). Besides, compared with wild-type CrmE10, CrmE10-mut5 exhibited higher activity with the addition of DMF or DMSO (Fig. S5D). Therefore, changing the charge properties of the esterases would dramatically affect the enzymatic properties.