Computer-aided rational design strategy based on protein surface charge to improve the thermal stability of a novel esterase from Geobacillus jurassicus

Abstract


Introduction
Green process is a strategic choice in the bio-economic world (Ferreira-Leitão et al. 2017).Enzymes are the most promising green catalysts among the different green methods available, giving modi cations that lessen hazardous waste and assist the chemical sector in moving toward a sustainable and eco-friendly future (Kate et al. 2022).Enzymes enable the manufacture of enzymes in a gentle, ecologically friendly production approach, as opposed to the synthesis of chemical catalysts.Lipolytic enzymes are extensively employed in the baking, pharmaceutical, beverage, paper, cosmetics and organic synthesis industries (Singh et al. 2016).Esterases (EC 3.1.1.1)favor short-chain (less than 10 carbon atoms) water-soluble fatty acid esters as opposed to lipases (EC 3.1.1.3)(Bornscheuer 2002).Esterases are α/β hydrolases and the catalytic triad consists of the catalytic triad consisting of Ser-Asp-His.Esterase enzymes are widely distributed in plants, animals, and microorganisms.The most commercially useful sources of esterase come from microorganisms such as bacteria, yeasts, and fungi (Li et al. 2010; Li et al. 2014).Numerous reaction circumstances call for high-temperature resistant enzymes, especially thermophilic esterases, which are mainly produced from thermophilic microorganisms (D'Auria et al. 2000; Khaswal et al. 2022).Natural thermophilic esterases typically lack the requisite thermal stability and are not thermally stable enough to sustain demanding industrial conditions.Protein engineering techniques have been developed to bypass industrial barriers and create thermally stable enzymes using directed evolution, rational design, and semi-rational design (Ribeiro et al. 2019;Lin et al. 2018).Directed evolution is a method of enhancing protein characteristics by replicating the natural selection process to produce large mutant libraries (Qiu et al. 2021).Wang et al (Wang et al. 2018) created a random mutagenesis library of 11,000 colonies and obtained mutant K47E, which outperformed wild-type alkaline pectate lyase PEL168 in speci c activity and T 1/2 at 50°C by 1.8-and 2.0-fold, respectively.However, directed evolution is arduous in enzyme engineering and produces and screens for a huge number of mutants.Semi-rational design is a strategy for generating smaller and smarter mutant libraries based on sequence comparison and structural analysis, Nakatani et al (Nakatani et al. 2018) obtained mutant S92E by site-saturated xylanase mutagenesis from Bacillus sp.strain TAR-1, which showed a 4.3-fold increase in residual activity compared to wild-type after 15 min heat treatment at 80°C.Rational design offers smaller mutation libraries and better prediction accuracy when compared to directed evolution and semi-rational design, with the important bene ts of high e ciency and cost savings.Liao et al (Liao et al. 2019) developed a mutant K573V-E631F of the α-L-rhamnosidase from Aspergillus niger JMU-TS528 via a rational design approach, resulting in an increase in the T 1/2 of 2.3 h at 60°C.In addition, an integrated rational design strategy to improve the enzymatic thermal stability would be bene cia (Tong et al. 2021).Tan et al (Tan et al. 2023) coupled B-factor and Gibbs unfolding free energy change (ΔΔG) for rational design of β-mannanase from Aspergillus niger to obtain mutants with increased thermal stability.The development of enzymes through rational engineering is capable of enhancing their e cacy, accuracy, e ciency, and productivity, thereby facilitating the evolution process at a faster rate.
Surface charge engineering is a potent method based on protein structure for changing protein properties, including activity, thermal stability, and tolerance to ionic liquids to enrich mutant libraries (Zhou et al. 2019).According to numerous studies, surface charged residues are critical for protein stability.Therefore, by optimizing the interactions between protein surface charges, the thermal stability of proteins can be effectively improved (Strickler et al. 2006).This approach focuses on protein surface banding amino acids, which may potentially boost the thermal stability of the enzyme without altering its activity because it is farther from the catalytic center.It also minimizes the diversity of potential alternative amino acid residues in the mutant library (Gribenko et al. 2009).Zhang et al (Zhang et al. 2023) enhanced the thermal stability and activity of β-1,3 − 1,4-glucanase by mutating the negatively charged aspartic amino acid into an uncharged alanine by mutation, thereby improving the thermal stability and activity of glucanase.Cao et al (Cao et al. 2021) obtained two mutants with improved thermal stability by mutating the negatively charged glutamate to the uncharged glutamine and alanine in the transaminase from Aspergillus terreus.However, the lack of an e cient selection approach has prevented the widespread deployment of surface charge engineering techniques.The Tanford-Kirkwood (TK) model was originally developed to determine the contribution of single charged residue to the total energy of protein molecules by Tanford and Kirkwood (Tanford et al. 1957), and was subsequently enhanced by the addition of solvent accessibility (Richmond 1984), Gibbs free energy (Bashford et al. 1991), and electrostatic interactions (Havranek et al. 1999).The ETSS system (Zhang et al. 2014), recently introduced for protein redesign, includes the capability to compute surface charge interactions and optimize them for enzyme redesign.Tu et al (Tu et al. 2015) utilized the ETSS approach to enhance the thermal stability of PG8fn by optimizing charge-charge interactions, resulting in the development of a double mutant D244A/D299R.This mutant displayed a 1.4-fold increase in T 1/2 at 55°C, indicating improved thermal stability.Although the ETSS model has demonstrated success in enhancing the thermal stability of enzymes, it still requires further re nement as it fails to take into account the impact of mutations on enzyme activity.
In this study, an esterase gene from Geobacillus jurassicus was cloned and expressed in Escherichia coli.Subsequently, a computer-aided rational design based on surface charge protein engineering was applied to enhance its thermostability.The T 1/2 of the most thermostable mutant E253Q increased 1.8-fold comparing with the wild-type.Intermolecular forces, surface potential changes and B-factor analysis were utilized to reveal the mechanism of thermal stability enhancement.In conclusion, we identi ed an esterase from Geobacillus sp. with low homologous similarity to esterases with reported properties derived from the same species.The nding advances research on thermophilic bacteria and the identi cation of the thermophilic esterases.Additionally, the computer-aided rational strategy based on surface charge engineering provides a favorable impact on the following use of thermophilic esterases and serves as an example for the rational design of other esterases.
All experiments were performed in triplicate and the results were presented as the average of these values and the standard deviation.

Gene cloning, expression and puri cation of gju768
The gju768 gene was cloned from G. jurassicus DSMZ 15726.Primers gju768-F/gju768-R (gju768-F:5'-cgcggatccATGGTGATCATTGAACAGGAACTAT-3', gju768-R:5'-ccgctcgagCACATGCTCGCGAAACCAG-3') were used to amplify the origin gju768 with restriction endonucleases (BamH I/Xho I).Primers gju768-U/gju768-D (gju768-U:5'-atgggtcgcggatccgaattcATGGTGATCATTGAACAGGAACTATT-3', gju768-D:5' ttgtcgacggagctcgaattcTTACACATGCTCGCGAAACCA-3') were used to amplify the mutated gju768 with restriction endonuclease (EcoR I).After further ampli cation of gju768 using the assembled genome as a template via polymerase chain reaction (PCR), the result was con rmed and puri ed via agarose gel electrophoresis.E. coli (DH5) carrying pET-28a (+) was grown at the same time to obtain the vector pET-28a (+).Subsequently, gju768 was linked to the vector digested with BamH I and Xho I using the T4 ligase to construct BL21-pET-28a-gju768, and the mutant gju768 linked vector digested with EcoR I via one-step cloning (Vazyme, China) to produce strains that harbored mutants.Finally, to create heterologous expression strains, the linkage products were individually transformed into E. coli BL21 (DE3).The constructed positive strains were transferred to Luria-Bertani medium (LB medium) for enrichment culture.Following an overnight incubation at 37℃ in LB medium (10 mL) supplemented with kanamycin (50 µg/mL), the colonies were transferred to 250 mL shake asks for further ampli cation.When the optical density at 600 nm (OD 600 ) reached 0.6, IPTG (24 µg/mL nal concentration) was added for protein induction.And subsequently, the introduced bacteria were incubated at 20℃ for 20 h.The harvested bacteria were puri ed by double distilled water and resuspended in 50 mM Tris/HCl buffer (pH 8.0).Cell pellets were sonicated by sonication (work 3 s, stop 5 s, 200 W) on ice for 15 min, and supernatant including soluble recombinant esterase Gju768 was obtained (8000 × g at 4°C for 10 min).A Ni-NTA Super ow Column (1 mL, Qiagen, Hilden, Germany) was utilized to collect the target protein.The protein that was attached to the column was eluted with a linear gradient of imidazole (from NPI 20 to NPI 250).The crude and puri ed proteins were examined by SDS-PAGE.

Selection of the mutation sites
A computer-aided rational design strategy based on the protein surface charge was proposed and applied to screen mutants.The Alphafold2 was used to predict the 3D structure of the protein.GetArea (https://curie.utmb.edu/getarea.html)(Fraczkiewicz et al. 1998) was used to search for amino acids located on the surface of a protein.The Mutation Energy section of Discovery Studio 3.5 was used to predict the amino acid positions at which stable mutations can be obtained.PoPMuSiC (https://soft.dezyme.com/)(Dehouck et al. 2009) and FoldX (https://foldxsuite.crg.eu/)(Guerois et al. 2002) were used to calculate free energy changes induced by amino acid substitutions.PyMOL was used to measure the distance between negatively charged surface amino acid residues and catalytic triplet residues.When implementing amino acid site mutagenesis, the following ve principles must be respected simultaneously: (1) Three-dimensional model structures of the protein were used as a starting point for computer-aided rational design; (2) Negative charged amino acid residues with above 50% average solvent accessibility surface area ratio were selected by mutating Aspartate (D) to Asparagine (N) and Glutamate (E) to Glutamine (Q) to eliminate the negative surface charge; (3) Mutation stability was tested for mutants D to N and E to Q by using the Predictive Stable Mutations module which in Discovery Studio 3.5, and mutants anticipated to be stabilizing were screened; (4) In both PoPMuSiC and FoldX, mutants should show ∆∆G < 0; (5) In order to minimize the damage to the enzyme activity, preferred mutation sites at a distance of more than 20 Å away from the core region of the catalytic triads were selected as the ultimate criterion for the selection of mutation sites.Mutations that matched all the aforementioned ve criteria were used in subsequent experiments.

Site-directed mutagenesis
Site-directed gju768 mutagenesis using plasmid pET-28a-gju768 as the template was performed directly by the quick-change site-directed mutagenesis technique (Stratagene, CA, USA).The reaction procedure setup was as follows: initial denaturation at 98℃ for 3 min, 30 cycles at 98℃ for 10 s, 68℃ for 3.5 min, and a nal extension at 72℃ for 5 min.PCR products were digested using Dpn at 37 ℃ for 1.5 h and subsequently transformed into E. coli DH5α for expression analysis.Veri ed plasmids were then transformed into E. coli BL21 (DE3) for expression analysis.The mutant structures were predicted by Alphafold2 and all generated models were evaluated and veri ed by the Ramachandran plot.

Enzymatic properties analysis
The enzyme was determined spectrophotometrically by measuring the conversion of p-nitrophenyl acetate to p-nitrophenol at a certain temperature for 5 min at 405 nm.The reactions were then terminated by adding 1 mL of 95% ethanol.One unit of enzyme activity (U) was de ned as the amount of enzyme required to release 1 µM p-nitrophenol per min under the conditions of the assay.Substrate speci city, optimum temperature, and optimum pH of the enzyme were varied and determined by controlling a single variable.Substrate speci city was identi ed by applying p-nitrophenyl ester (6.25 mM) in the C2 to C16 range.Enzyme optimum temperature was determined in the range of 45°C to 75°C under optimum substrate control.The optimal pH of the enzyme was determined in the range of pH 7.0-10.5 within the optimum substrate and temperature conditions for 5 min of reactivity.
Thermal stability was determined by measuring residual enzyme activity after exposing the enzyme to different temperatures for a controlled period of time, respectively.The T 1/2 of the esterase was measured by incubating the enzyme at different temperatures for various time intervals.The activation with non-incubated enzyme was assumed to be 100%.Determination and analysis of residual esterase activities after incubation at different time intervals.

Enzymatic kinetic parameters assay
The Michaelis-Menten constant (K m ) and maximum reaction velocity (V max ) of Gju768 and its mutants were determined by the p-nitrophenol method (pH 9.0, 60°C or 65°C) using a series of concentrations of p-NP-ester (0.03-0.30mM) as substrates.V max and K m values were obtained by non-linear regression based on Michaelis-Menten.Kinetic values were calculated from Lineweaver Burk diagrams and the Michaelis-Menten equation.

Sequence alignment and modelling analysis of Gju768
By PCR ampli cation, a 768 bp long (named gju768) DNA fragment from G. jurassicus DSMZ 15726 encoding a polypeptide of 255 amino acid residues was cloned and sequenced.The protein expressed by gju768 was named Gju768.Basic bioinformatics investigation of gju768 demonstrated a G + C content of 55.9%.The theoretical pI value of Gju768 was calculated by the Compute pI/Mw program of ExPASy to have a value of 6.37 and a molecular weight of 29.12 kDa.
Gju768 has been indicated to show low sequence homology with esterases from Geobacillus sp. that have been reported properties (Table 1).Homology analysis indicated that Gju768 has the highest identity of 15.20% to the esterase EstGSU753 from Geobacillus subterraneus DSM13552 (Cai et al. 2020), which was previously investigated and reported in our lab, indicating its novelty and high research value.Multiple sequence alignment (Fig. 1a) revealed that Ser116 (S), Asp180 (D), and His237 (H) constituted the catalytic triad.The Alphafold2 was used to predict the three-dimensional structure of Gju768.The 3D structure of Gju768 was visualized and modi ed in PyMOL.The catalytic triad (Asp180 (D), His237 (H), and Ser116 (S)) was shown in Fig. 1b as stick form.Ramachandran plot analysis of the model proteins modeled in 3D was conducted to verify the structural reliability of the deduced model.As shown in Fig. S1, more than 91.9% of the residues from the modeled proteins were found to be located in the most favorable regions and 7.7% in the additional allowed regions, which validated the integrity of the modeled structure.Mutation site selection and site-directed mutagenesis of Gju768 In this study, a computer-aided rational design strategy based on the protein surface charge, as described in the Methods section, incorporates the use of GetArea, Discovery Studio, PoPMuSiC, FoldX, and PyMOL to improve the thermal stability of esterase Gju768 (Fig. 2).Protein stability depends on surface charge interactions, and improving these connections makes proteins more thermally stable (Strickler et al. 2006).Rational design of the protein surface charge also allows optimizing only the interactions away from the catalytically active site, enabling improved thermal stability without affecting protein activity (Schweiker et al. 2009).GetArea was used to determine whether an amino acid was present inside a protein or on protein surface, and an amino acid with Ratio > 50% were thought to be on the protein surface.
Amino acids predicted to be located on the protein surface were listed in Table S2.The focus was on Glutamic acids and Aspartic acids, which were found on the protein surface.These amino acids were switched to side-chain amide analogs (D to N and E to Q), which effectively eliminated negative surface charge and achieved charge neutralization while causing minimal side-chain structure change (Akke et al. 1990).
Mutation stability was tested for D to N and E to Q mutations in Discovery Studio 3.5 by using the Predictive Stable Mutations module.The mutation results predicted by Discovery Studio 3.5 were shown in Table S3.Of the 17 mutation results, 6 mutation results were expected to have a "Neutral" effect and 11 mutation results were predicted to have a "Stabilizing" effect.In accordance with the second criterion, the 11 stable mutations were identi ed to be positive mutants and were then used in the followings.The effects of non-synonymous variants on the protein stability were quanti ed in terms of the Gibbs free energy of unfolding ( ΔG).PoPMuSiC is a web server for predicting the fold free energy change (ΔΔG) caused by a single point mutation in a protein.ΔΔG represented a linear combination of statistical potentials, with regard to the amino acid type, torsion angles de ning the backbone conformation and the solvent accessibility (Dehouk et al. 2011).ΔΔG < 0 indicates that the mutant is more stable than wild-type (WT).Studies have revealed that the strategy of screening mutants for higher positive by integrating several computer design methods has been widely employed (Chen et al. 2012).FoldX is another rational design software that can quantify the change in free energy before and after mutation.Screening mutants by combining the free energy results calculated by PoPMuSiC and FoldX can effectively narrow the mutation scope.The results of simulation of single mutations by both PoPMuSiC and FoldX were provided in Table S4.According to the third criterion, ve out of the eleven mutations ful ll both ΔΔG < 0 predicted by both PoPMuSiC and FoldX, and were thus selected.The distances of Glutamic acids and Aspartic acids located on the protein surface from the catalytic triad were measured by PyMOL and amino acids with Ds Cα−Cα > 20 Å were further screened since they were far from the catalytic active center according to the nal criteria, of which the mutations had no effect on the catalytic activity.As shown in Table S5, the distance between the single point mutation and the catalytic triplet was evaluated and 4 single point mutants met the lter criterion of Ds Cα−Cα > 20 Å.
On the basis of the selection criteria, the results of the mutants that satis ed all the screening criteria were listed in Table S6, in total four potential mutant sites were obtained, which were selected and investigated (D24N, E221Q, D225N, and E253Q).The locations of the four potential mutants in the three-dimensional structure were shown in Fig. S2.

Expression, puri cation and enzymatic properties of Gju768 and its mutants
The recombinant plasmid pET-28a-gju768 was constructed as a template and expressed in E. coli B21 (DE3).Molecular mass of recombinant Gju768 was predicted to be 29.12 kDa, which was in accordance with the SDS-PAGE estimate (Fig. S3).The primers utilized to constrain the mutants were presented in Table S7.The mutants were constructed on the pET-28a (+) vector by the same method and expressed in E. coli.The protein expression was veri ed by SDS-PAGE (Fig. S4).Upon comparison with the pET-28a (+) empty vector, all of these variants had a distinct band of approximately 29 kDa, indicating that all variants were successfully expressed.
A total of four mutants were obtained based on the rational design strategy of eliminating the negative charge on the protein surface, and their enzymatic activity and thermal stability were subsequently initially assayed.The results were shown in Fig. 3.All mutants showed increased enzyme activities compared to the WT.To determine the thermal stability of the mutants, the WT and mutants were incubated at 65°C for 20 min and then analyzed for their residual activity, which was retained by 48.8%, 81.0%, and 84.8% for D24N, E221Q, and E253Q, respectively.The thermal stability of D24N, E221Q, and E253Q was improved by 25.8%, 108.8%, and 118.6% compared with the 38.8% residual activity of the WT.Thus, the enzymatic properties of WT, D24N, E221Q, and E253Q were thus considered to be selective for the characterization of their properties.The thermal stability of D225N was not notably improved compared to the WT and thus it was not considered for subsequent investigations.
To investigate the substrate speci city of the WT and mutants, the relative activities (C2-C16) of the WT and three single point mutants were measured for short-chain substrates and long-chain substrates at pH 9.0, 60°C.The results, as shown in Fig. 4a, showed that the hydrolytic preference of the mutants for the best substrate of nitrobenzenes with that of the WT, which were all C8.Compared to the other two mutants, D24N preferred hydrolysis of short-chain substrates over hydrolysis of long-chain substrates, whereas the E221Q and E253Q mutants had reduced hydrolase activity for C2 compared to the WT, whereas the hydrolase activity for C4-C16 substrates was signi cantly increased.
Subsequently, the optimal temperature of the WT and the mutants was determined within the temperature range of 45°C to 70°C at pH 9.0.
Results showed in Fig. 4b indicated that the optimal temperature of D24N and E221Q were still 60°C, which was the same as the WT.The optimal temperature of mutant E253Q was 65°C, an increase of 5°C over the WT (60°C).It was found that although the optimum temperatures of D24N and E221Q were the same as compared to the WT, their speci c enzyme activities were in general higher than that of the WT in the range of 45°C to 70°C.When the temperature was increased to 70°C, all mutants retained over 70% of the relative enzyme activity, which was much higher than that of the WT (45%), suggesting that the three mutants were more resistant to high temperatures than the WT.In addition, the optimal pH of the WT and four active mutants was also measured within the pH range from 7.5 to 10.0 at the optimal temperature (Fig. 4c).Changes in residue E253Q did not affect its optimal pH (pH 9.0).Mutants D24N and E221Q had an optimal pH of 9.5, which allowed them to adapt to more alkaline reaction conditions than the WT optimal pH of 9.0.At pH 10.0, the maximum enzyme activity of the WT was only 45%, whereas all three mutants retained more than 90% of the maximum enzyme activity, suggesting that the pH adaptive range of the mutants shifted to alkaline.
Thermostabilities of the WT and its mutants were estimated by their T 1/2 values and the thermostability analysis was performed at 65°C, from 0 to 50 min incubation time with 10 min time intervals (Fig. 4d).After 50 min of incubation at 65°C, all four mutants were capable of retaining higher residual activity than the WT, with mutant E253Q being the most thermally stable, retaining 15% of its maximum enzyme activity after 50 min of incubation, whereas the WT was nearly deactivated.At 65°C, the T 1/2 of mutants E221Q and E253Q were 31.7 and 32.4 min, respectively, which were 77.1% and 81.0% higher than those of the WT (17.9 min).In addition, the T 1/2 of mutant D24N was 20 min, which was close to but higher than the T 1/2 of the WT.While the residual enzyme activities after 20 min of incubation was 48.8%, which was superior than that of WT with 38.8% residual enzyme activity.To conclude, all three mutants designed based on the rational design strategy of eliminating the negative charge on the protein surface exhibited with a better thermal stability than the WT.
Kinetic parameters for Gju768 and its mutants K m and V max were the core properties of the enzyme which were determined by the Michaelis-Menten model.By using the Lineweaver-Burk double reciprocal mapping approach, K m and V max for the WT and its mutants were calculated.Kinetic parameters for the WT and three mutants were determined at their respective optimum conditions.Under the conditions of C8 and their respective optimal temperatures and optimal pH (WT: 60°C, pH 9.0, D24N and E221Q: 60°C, pH 9.5, E253Q: 65°C, pH 9.5), the K m values of the WT, D24N, E221Q, and E253Q were 0.118 mM, 0.081 mM, 0.031 mM, and 0.027 mM, respectively (Table 2).It is clear that all three mutants had smaller K m values than the WT, which implied that the mutants had a stronger a nity binding to the substrate.Meanwhile, as shown in Table 2, the D24N mutant exhibited the highest speci c enzyme activity, 11251.743U/mg, whereas the speci c enzyme activities of the other mutants did not differ much from those of the WT.

Thermostability mechanism of Gju768 and its mutants
As mentioned previously, we proposed a computer-aided surface charge engineering rational design strategy to improve the thermal stability of the enzyme, and the improved thermal stability of the mutants D24N, E221Q, and E253Q obtained from the selecting veri ed the effectiveness of our strategy.Thereafter, we explored the mechanism of the improved thermal stability.Researchers have observed that increasing intrinsic forces signi cantly increases the thermal stability of enzymes (Ruslan et al. 2012;Yu et al. 2022), and recent studies have demonstrated that the formation of new hydrogen bonds and Cation-π interactions contributes signi cantly to protein stability (Xie et al. 2020).In order to investigate the mechanisms of the increased thermal stability of the mutant, we evaluated the intramolecular interactions of hydrogen bonds and Cation-π interactions involved in the mutant using a three-dimensional structure, and the results were shown in Fig. 5.
The replacement D24N was located in the LOOP region on the protein surface there, as shown in Fig. 5a, b, and the O atom of Asn24 formed a new hydrogen bond with the N atom of Arg23 at a distance of 3.5 Å, which was not observed in Asp24.Mutant E221Q similarly lied in the surface loop region of the protein, as shown in Fig. 5c, d, the N atoms of both Glu221 and Gln221 formed hydrogen bonds with the O atom of Pro218 in the both periods before and after the mutation, with bond lengths of 3.5 Å and 3.4 Å, respectively, whereas in the partial conformation of the mutant, it could be observed that the N atom of Gly222 and the OE1 and OE2 atoms of Glu224 formed two hydrogen bonds, whereas only one hydrogen bond was observed in the partial conformation of the WT.The binding of a cation to the π-plane of an aromatic structure produces a surprisingly non-covalent force that has been termed the Cation-π interaction (Dougherty 1996).Mutant E253Q was located in the alpha helix region of the protein surface, away from the center of catalytic activity, and intermolecular forces in the partial conformation of mutant E253Q were also characterized, and the results were shown in Fig. 5e, f.In the WT partial conformation, the N atom of Glu253 formed two hydrogen bonds with the O atoms of Arg249 and Trp250 at distances of 3.2 Å and 3.3 Å, respectively.In addition, the His254 and Trp250 formed a π-π stacking interactions with a distant of 4.4 Å.In the mutant E253Q partial conformation, the N atom of Gln253 formed two hydrogen bonds with the O atoms of Arg249, and Trp250 as well, all with bond lengths of 3.2 Å.The change from the negatively charged Glutamic Acid at position 253 to the uncharged Glutamine was what caused a Cation-π interaction to occur between Trp250 and His254.The Cation-π interaction has been proven to carry larger energies than π-π stacking and to be as important as hydrogen bonds, ion pairing, and hydrophobic effects (Liao et al. 2013).
Site-directed mutagenesis enhances thermal stability by altering the amino acid sequence and the structure of the affected proteins (Taylor et al. 2010).Hydrogen bonds and increased Cation-π interactions have been recognized as the main determinants of enhanced thermal stability (Guo et al. 2015;Liu et al. 2013).In this study, the general structural stability of the mutant was increased by increasing hydrogen bonds and Cation-π interactions, which ultimately improved the thermal stability.
Besides intermolecular forces, optimizing the surface charge of proteins also made a signi cant contribution to the thermal stability of proteins (Chan et al. 2012;Strickler et al. 2006).The surface potential distributions of the WT and mutants are shown in Fig. 6, with the WT Asp24, Glu221 and Glu253 located in the localized negative potential region on the protein surface (Fig. 6a, c, e).When the negatively charged Asp 24, Glu 221, and Glu253 were mutated to the uncharged Asn24, Gln221, and Gln253 (Fig. 6b, d, f), respectively, the potential of the mutation site became neutral, and an increase in the distribution of positive potential around the mutation site was also observed.The symmetry of the electrostatic potential on the protein surface was also a primary factor in thermal stability, and greater symmetrical positive potentials may lead to superior thermal stability (Chi et al. 2023).As a result, the mutation created a favorable change in the electrostatic potential on the protein surface facilitating the mutant to maintain a stable conformation with high stability and thus improved thermal stability.
Early studies indicated that B-factors could be used to recognize the exibility of proteins, and it has been proposed that high B-factors indicate that proteins were more exible than average, whereas low B-factors were assumed to occur in more rigid positions (Liang et

Conclusions
In this study, a novel esterase from strain G. jurassicus DSMZ 15726 was obtained (named Gju768) and a computer-aided rational design strategy based on the protein surface charge was proposed and applied to select single mutants.Subsequently, three thermal stable mutants, D24N, E221Q, and E253Q, was obtained according to the selecting strategy.These three mutants improved the thermal stability of the enzyme without decreasing its activity.The most thermostable mutant, E253Q, increased its T 1/2 from 17.9 min to 32.4 min, a 1.8-fold longer than the WT (incubation at 65°C, pH 9.0).Meanwhile, we have explored the reasons for the improved thermal stability of the mutants compared with the WT in terms of intermolecular forces, surface potential distribution and B-factor values.The computer-aided surface charged engineering rational design strategy proposed in this study provided a feasible strategy to improve the thermal stability of esterase and was expected to facilitate its application in industrial manufacturing. Protein al. 2009; Parthasarathy et al. 2000;Yuan et al. 2003).The B-factors of the mutation sites of the WT and mutants were measured using ResQ(Yang et al. 2016) to measure uctuations in residues relative to the average position.As shown in Fig.7, we found that the B-factor values of D24N, E221Q, and E253Q all decreased after mutation, indicating that the mutants all increased in rigidity and became more structurally stable.It has been demonstrated that structural exibility decreased when the thermal stability of proteins increased(Vihinen 1987), and our research results supported this conclusion.
sequence and structure of Gju768.(a) Amino acid sequence of Gju768 compared to the sequences of homology searched by BLAST on NCBI.The catalytic triad of Gju768 (Ser116, Asp180, and His237) were labeled with black triangles.(b) The 3D model of Gju768 was predicted by Alphafold2.The model was visualized by PyMOL and the catalytic triad were shown in green stick forms.

Figure 4 The
Figure 4

Table 1
Pairwise similarity between the esterase amino acid sequences from Geobacillus sp.

Table 2
Kinetic parameters of Gju768 and its mutants on p-NP-esters.