Alteration of Industrial Lipases by Protein Engineering

全文

(1)Alteration of Industrial Lipases by Protein Engineering. March 2021. Kazunori Yoshida. The Graduate School of Natural Science and Technology (Doctor Course) OKAYAMA UNIVERSITY.

(2)

(3) Contents. Page. Chapter 1.. General Introduction. Chapter 2.. Synthetically Useful Variants of Industrial lipases from. 1. 29. Burkholderia cepacia and Pseudomonas fluorescens.. Chapter 3.. Improvement of Thermostability of Burkholderia cepacia Lipase. 63. by the Loop-Walking Method.. Chapter 4.. Prediction of Optimal Mutants by Multivariate Analysis:. 95. Further Improvement of Thermostability.. List of Publication and Oral Presentation. 119. Acknowledgement. 123.

(4)

(5) Chapter 1. Chapter 1. General Introduction. 1. Industrial use of lipases. Lipases catalyze not only the hydrolysis of triacylglycerol but also esterification, transesterification, and aminolysis. In addition, lipases are characterized by wide substrate applicability and high stereoselectivity. Therefore, lipases are used for the industrial production of detergents, fat and oils, flavors, digestive agents, drugs, fine chemicals, cosmetics, perfumes, and biofuels.1 For example, in fat and oil processing, the high regioselectivity and acyl chain length specificity are utilized. When triacylglyceride is used to produce free fatty acid, the reactive position of triglyceride (sn-1, 2, 3 position) depends on the identity of the lipase. The Rhizopus oryzae lipase2 and Rhizomucor miehei lipase3 hydrolyze the ester bonds of sn-1,3 positions regioselectively to produce free fatty acids . The Burkholderia cepacia lipase4 and Candida cylindracea lipase5 hydrolyze the ester bonds of sn-1, 2, 3 positions to produce free fatty acids. The Penicillium camemberti lipase6 shows no activity for. triacylglycerides. but. shows. good activity for. diacylglycerides. and. monoacylglycerides. In a hydrophobic phase, these lipases can also catalyze the regioselective esterification and transesterification. Lipase also has alkyl chain length specificity. Utilizing these properties, 1,3-dioleoyl-2-palmitoylglycerol (OPO) for cocoa butter equivalent (CBE) production is produced by the transesterification of palm oil and oleic acid with a Rhizopus oryzae lipase2. Lipases are also widely used as biocatalysts for the production of pharmaceutical and agrochemical intermediates. The Burkholderia cepacia lipase4 and Candida antarctica lipase7 are often used in these industries. It is necessary to select a lipase having the performance most suitable for the stereoselectivity of the desired compound.. 1.

(6) Chapter 1. 2. Protein engineering methods for enzyme modifications. There are two ways for obtaining an enzyme with a desired enzyme function. One is the screening of microorganisms from nature, and the other is protein engineering. The latter method has been growing in the last three decades. The methods of protein engineering are roughly divided into three types: directed evolution, semi-rational design, and rational design (Figure 1).8. Protein engineering method. 2-1) Directed evolution Random mutagenesis. 2-2) Semi-rational design Site-saturation mutagenesis. 2-3) Rational design Site-directed mutagenesis. 2-2-1) Sequence-based design 2-2-2) Structure-based design 2-1-1） Random mutagenesis 2-1-2） SeSaM 2-1-3) DNA shuffling 2-1-4) Random mutagenesis & High-throughput screening 2-1-5) Random mutagenesis & Artificial intelligence (Machine learning). 2-2-1-1) Hot spot wizard 2-2-1-2) Ancestral mutagenesis 2-2-1-3) PROSS, INTMSAlign. 2-2-2-1) CAST 2-2-2-2) ISM, FRISM 2-2-2-3) Circular permutation (CP) 2-2-2-4) S-S bond formation 2-2-2-5) B-factor. 2-3-1) MD simuration 2-3-2) QM/MM 2-3-3) Protein design (Rosetta). Figure 1. Protein engineering methods for the improvement of enzymatic properties.. 2-1. Directed evolution. Directed evolution is a method of artificially promoting the evolution of enzymes by mimicking the process of natural selection in nature. The directed evolution consists of two processes. The first is the creation of a mutant library, and the second is the selection of useful mutants by activity evaluation. By introducing mutations into an obtained useful mutant, it is possible to obtain another mutant having a further enhanced enzyme function (Figure 2).. 2.

(7) Chapter 1. Creation of mutant gene library by Error-prone PCR. E. coli expressing mutant gene library. Isoration of disired mutant gene. Evaluation of mutant gene library. Figure 2. Scheme of directed evolution by random mutagenesis. 2-1-1. Random mutagenesis. The most common directed evolution is achieved by random mutagenesis using error-prone PCR. The advantage of this method is that it requires no comprehensive information such as the three-dimensional structure and the catalytic mechanism of the enzyme, and it is possible to construct a mutant library using only the primary structure information. In this method, mutations are introduced into the entire target gene to obtain a mutant with excellent characteristics.9 On the other hand, the disadvantage of this method is the complexity of selecting a mutant from a huge mutant library. When the enzyme is composed of 300 amino acid residues, there are 20300 variations of the mutant. Therefore, the screening method is a key to the success of this method.10 The most famous report on this method is the creation of an organic solvent-resistant mutant of subtilisin B by Arnold and co-workers.11 The halo assay of milk casein has been used to screen for promising strains from a mutant library. The improvement of the organic-solvent resistance of subtilisin E for dimethylformamide (DMF) was 256-fold in 60% DMF and 131-fold in 85% DMF as compared to the wild-type enzyme, respectively.. 3.

(8) Chapter 1. 2-1-2. Sequence saturation mutagenesis (SeSaM). Screening of random mutation libraries constructed by error-prone PCR may overlook effective mutants because the variation of mutants is limited by the following two factors: base substitution bias and mutation site bias. The base substitution bias is brought about by the difference in the frequency of nucleic acid transversions and transitions. The nucleic acid transversions have been reported to be much less likely than transitions.12 It also occurs due to the difference in the frequency of codons when translating a base sequence into an amino acid. For example, tryptophan has one codon (TGG), while leucine has six codons (TTA, TTG, CTT, CTC, CTA and CTG). The mutation site bias is brought about by the secondary structure of the base sequence, as there are portions where the frequency of mutation induction is high or low. The sequence saturation mutagenesis (SeSaM), which is a chemotherapeutic random mutagenesis method, has been reported as a solution to the bias of the mutation library occurred by these factors.13 This method consists of four steps: Step 1: creation of a pool of DNA fragments with random size distribution, Step 2: enzymatic elongation of DNA fragments with universal base, Step 3: full-length gene synthesis, Step 4: universal base replacement by standard nucleotides. The most important points of this technique are the binding of DNA fragments of universal bases to 3’ ends using terminal transferase (Step 2), and the replacement of universal bases with four types of standard bases (Step 4). In these steps, both biases are solved. As an example of this method, the surfactant resistance of subtilisin E has been improved. The half-life of the mutant (SeSaM1-5_S62I/ A153V/G166S/I205V) was improved from < 2 min to 4.7 hour in 1 M guanidinium chloride and from < 2 min to 2.7 hour in 0.5% sodium dodecyl sulfate.14 In addition, thermostability of phytase has also been improved. The residual activity of the mutant after heat treatment at 58 °C for 20 minutes was improved by approximately 1.5-fold (mutant: 58%, wild-type: 38%).15 4.

(9) Chapter 1. 2-1-3. DNA shuffling. DNA shuffling is one of the gene recombination technologies developed for directed evolution. In this method, a library of chimeric genes is obtained by fragmenting multiple homologous DNA template sequences into DNA fragments using a DNase and reconstructing the DNA fragments into single-stranded DNA by a polymerase reaction. By this method, a β-lactamase mutant showing 32,000-fold antibacterial activity has been obtained.16 It is ideal to obtain DNA template sequences with about 50 to 100 bp, which is difficult to achieve by controlling the reaction conditions. To solve this problem, the nucleotide exchange and excision technology (NExT) has been developed.17 This method uses uracil-DNA glycosylase which specifically recognizes dUTP, to simplify the DNA fragmentation of DNA template sequences. This method has been applied to chloramphenicol acetyltransferase I (CAT). The staggered extension process (StEP) is a method of constructing a mutation library containing many mutational variations by repeating a short PCR cycle with many types of template DNA and one PCR primer.18 By this method, short DNA strands synthesized during the short extension reaction time of PCR are annealed with another template DNA in the next PCR cycle. As a result, chimeric genes in which various types of template DNA sequences are combined are created. By using this method, a mutant strain of subtilisin E with a 50-fold thermostability was obtained. The above methods of creating chimeric gene libraries using homologous DNA template sequences are limited. This is because a homologous region is required to reconstitute a gene sequence in a single-stranded DNA. The incremental truncation for the creation of hybrid enzyme (ITCHY) has been developed as a method for creating chimeric gene libraries that does not depend on the homology of DNA template sequences.19 In this method, an insert gene in which different species of DNA sequences are bound is cloned into an expression vector, and the insert gene is DNA-fragmented by an exonuclease. The DNA fragment is 5.

(10) Chapter 1. reconstituted with DNA ligase to generate a chimeric sequence that is ligated into an expression vector. By this method, the catalytic residue of glycinamide ribonucleotide formyltransferase (PurN) derived from Escherichia coli has been elucidated. SCRATCHY, which combines ITCHY with DNA shuffling, has been reported as a method for further introducing variations in the mutation library.20 A SCRATCHY library was constructed by two types of glycinamide ribonucleotide formyltransferase (PurN, GART), and the functional elucidation of PurN was studied in detail.. 2-1-4. Random mutagenesis and high-throughput screening. The probability of obtaining a beneficial mutant by random mutagenesis is related to the number of mutants whose characteristics are evaluated; the possibility of obtaining beneficial mutants increases with the increase in the number of mutants evaluated.21 As an approach from this aspect, a high-throughput screening system that combines microfluidics and a fluorescence-activated cell sorting (FACS) has been developed.22, 23, 24 By using a fluorescent substrate specifically designed for an enzyme, this method can evaluate mutants with a processing capacity of 1.0 × 106 CFU/min by sorting with fluorescent activity. If the number of strains in a culture broth is 107–108, candidate strains can be screened in 2 to 3 hours. Therefore, it is possible to repeat the preparation and evaluation of the mutant many times in a short period. In fact, the functional improvement of enzymes such as β-galactosidase25, cellulase26, glycosidase27, glucose oxidase28, horseradish peroxidase29, and laccase30 has been achieved. Although many functional improvements of enzymes have been reported by this method, most of them are saccharification enzymes and oxidases. This is because the fluorescent substrates that can be used with this method are primarily limited to water-soluble substrates. Therefore, there are no reports on lipases, which accept lipophilic substrates. A method with Raman spectroscopy has appeared as a new method for overcoming the limitation of a fluorescent substrate. This method uses the wavelength and scattering intensity 6.

(11) Chapter 1. of Raman rays of a compound in the droplet, and it is expected that applications for protein engineering will expand. For example, sorting by using the intracellular CO2 accumulation of yeast31 or pollen sorting32 have been reported.. 2-1-5. Random mutagenesis and artificial intelligence (machine learning). Another method for efficiently preparing and evaluating multiple mutants in random mutagenesis is the prediction of amino acid substitution in mutant using artificial intelligence (machine learning). The machine learning can search for an optimal amino acid substitution from teacher data which combines experimental and sequence data of the obtained mutants. By this method, a mutant having a desired enzymatic function can be obtained from a smaller library. This method mainly consists of repeating the following four steps: (i) obtaining experimental data of mutants; (ii) construction of a functional value prediction model from experimental data; (iii) prediction for the suitable amino acid substitution in mutant using a prediction model; (iv) experimental evaluation of the predicted mutant. Husiman and co-workers have reported the enzymatic functional improvement of halohydrin dehalogenase (HHDH) using protein sequence activity relationships (ProSAR)33. The ProSAR34 is an data mining algorism that creates protein sequence–activity relationship models from experimental data sets of enzyme mutant libraries. Mutagenesis, activity evaluation, and activity prediction by partial least-squares regression were repeated for 18 cycles, and an HHDH mutant showing approximately 4,000-fold activity in the conversion of ethyl (S)-4-chloro-3-hydroxybutyrate into ethyl (R)-4-cyano-3-hydroxybutyrate was obtained. Umetsu and co-workers have changed green fluorescent protein (GFP) from green fluorescence to yellow fluorescence.35 155 GFP mutants were prepared by changing four amino acid residues identified from the alignment of reported GFP and yellow fluorescent protein (YFP), and the fluorescence wavelength and intensity were measured. Finally, suitable mutants were predicted from all mutants (204) by machine learning (Bayesian optimization). 7.

(12) Chapter 1. A mutant showing a longer wavelength (yellow) and stronger fluorescence intensity than the reported YFP was obtained from the upper-ranking of the predicted mutants. Arnold and co-workers have reported the improved stereoselectivity of putative nitric oxide dioxygenase (NOD) from Rhodothermus marinus.36 The NOD-catalyzed stereoselective reaction of phenyldimethylsilane and ethyl 2-diazopropanoate was selected as a model. By repeating machine learning for 2 cycles, an NOD mutant showing 93% ee was obtained (wild-type: 73% ee). About 100 mutants were used in each cycle for machine learning teacher data, and multiple scikit-learn models have been tested as learning models.. 2-2. Semi-rational design. The semi-rational design narrows down the promising target sites and amino acid residues on the basis of protein sequences, structures, and functions, avoiding complex experimental operations. This allows for the production of small-scale and high-quality mutant libraries, and greatly eliminating the need for high-throughput screening in library evaluation.37 The disadvantage is that this method cannot be applied to enzymes that do not have sufficient information. The semi-rational design is classified into sequence-based design and structure-based design, depending on information used for the mutant design. Arnold and co-workers have compared the effectiveness of the semi-rational design method with that of directed evolution in improving the function of P450-BM3.38 A method that combines combinatorial site-saturation mutagenesis (CSSM)39 and structure-based computer design programs such as Corbit40 and CRAM was examined, and it is suggested that the semi-rational design can reduce the number of mutants to be evaluated in directed evolution.. 2-2-1. Sequence-based design. The sequence-based design is a method based on the evolutionary information of proteins. In this method, protein sequences are compared with one another by multiple alignment41 or. 8.

(13) Chapter 1. phylogenetic analyzes. The following three methods are described below.. 2-2-1-1. Hot spot wizard. Damborsky and co-worked have developed the hot spot wizard method to identify the mutagenesis hot spot.42 This method is a web-based tool that identifies mutagenesis hot spots on the basis of information combining protein sequences and structures. It is possible to estimate the candidate residues involved in catalytic function and stability. This method has been effective in estimating active site access tunnels and catalytic residues to improve the enzymatic function of haloalkane dehalogenase (DhaA) from Rhodococcus rhodochrous.43 This tool is open to the public (https://loschmidt.chemi.muni.cz/hotspotwizard/) on the web site as hot spot wizard v3.0.44. 2-2-1-2. Ancestral mutagenesis. Ancestral mutagenesis is a method of identifying mutational hot spots by combining multiple alignments of protein sequences and a phylogenetic tree. This method traces the ancestors of the constructed phylogenetic tree to estimate mutations that occur at species bifurcations as mutation candidates. This method is mainly constructed by following three steps: (i) preparation of multiple alignments of homologous amino acids; (ii) creation of evolutionary tree and estimation of ancestral sequence; (iii) mutation of the estimated amino acid sequence into the target protein. Yamagishi and co-workers have developed the thermostable mutant of 3-isopropylmalate dehydrogenase (IPMDH) from Caldococcus noboribetus using this method.45a Yamagishi and co-workers have also improved the thermostability of IPMDH from Thermus thermophiles. 45b. , glycyl-tRNA synthetase (GlyRS). from Thermus thermophiles45c, and the activity of β-amylase from Bacillus circulans45d. Akanuma and co-workers have obtained a mutant of IPMDH from Thermus thermophiles with improved catalytic activity at low temperature by developing an ancestral sequence 9.

(14) Chapter 1. reconstruction (ASR).46. 2-2-1-3. PROSS, INTMSAlign. The multiple sequence alignment of protein sequences can be used not only to improve enzymatic stability and activity but also to identify mutational hot spots for soluble protein expression in heterologous expression systems. The formation of inclusion bodies (IBs) in heterologous expression systems of target proteins was suppressed to improve the protein yield. Various computer software such as AGGRESCAN47, Solubis48, and CamSol49 have been developed for this method. Fleishman and co-workers have developed the computer software called PROSS, which can identify the mutational hot spots to improve protein stability and solubility in heterologous expression systems50. This software estimates hot spots based on both multiple sequence alignment and structural information. Plasmodium falciparum reticulocyte-binding protein homolog 5 (PfRH5) was selected as model protein, and the soluble expression and thermostability of PfRH5 variants were improved. Asano and co-workers have developed the computer software called INTMSAlign, which focuses on the polarity of amino acid residue of α-helix construction.51 This software estimates mutational hot spots by selecting mutational candidates based on multiple sequence alignments of protein sequences and evaluating these candidates with the α-helix rule and hydropathy contradiction rule. In the heterologous expression of luciferase from Metidia pacifica (MpLUC), a mutant in which the soluble expression is about two-folds that compared with wild-type was obtained.51c. 2-2-2. Structure-based design. Structure-based design is a method for designing mutation points based on protein structure information. Mutation sites are selected by observing the structure around the active site or substrate pocket or by comparing the structures with high/low protein sequence homology. As 10.

(15) Chapter 1. the number of protein structures registered in public data base such as protein data bank (PDBs) has increased rapidly to promote homology modeling studies, many methods for improving the enzyme function by this method have been developed. Among several reported methods, three methods are described below.. 2-2-2-1. Combinatorial active-site saturation test (CAST). The combinatorial active-site saturation test (CAST) is a method for preparing a mutation library by site-saturation mutagenesis (SSM), targeting amino acids that compose the surface of the substrate-binding pocket of an enzyme. By screening a desired substrate using this mutation library, it is possible to identify amino acid residues improving the catalytic function, such as substrate specificity, stereoselectivity, and specific activity. This method requires structural data or a homology model of the structure and cannot be applied to an unknown enzyme. Vogel and co-workers have improved substrate specificity for lipase from Pseudomonas aeruginosa.52 Scott and co-workers have improved the catalytic activity of phosphotriesterases from Agrobcaterium radiobacter (PETAr) on organophosphorus insecticides such as malathion, parathion, demeton, diazinon and chlorpyrifos.53 A mutant showing approximately 5,000-fold catalytic activity for malathion was created.. 2-2-2-2. Iterative site-specific mutagenesis (ISM), focused rational iterative site-specific mutagenesis (FRISM). The iterative saturation mutagenesis (ISM) is a semi-rational method of directed evolution that mimics the natural process of Darwinian evolution. This method is characterized by the systematic improvement of the enzyme phenotype. By preparing and evaluating a mutation library using a mutant having an effective phenotype in the first round as a template for the next round, and accumulating effective mutation points in candidate mutant, a mutant having a desired enzyme function can be efficiently obtained. Bocola and co-workers have reported 11.

(16) Chapter 1. the efficacy of ISM by comparing the previous report52 on the mutants of lipase from Pseudomonas aeruginosa.54 By this method, a mutant with a significantly improved enantiomer ratio (E value = 582) was obtained from a systematically constructed mutation library (10,000 mutants). It is suggested that ISM is more effective than the conventional mutation method such as error-prone PCR, DNA shuffling, and saturation mutagenesis at hot spots. Recently, Reetz and co-workers have developed focused rational iterative site-specific mutagenesis (FRISM), which combines rational design and ISM54 to further reduce the mutation library.55 This method consists of the following five steps: (i) preparation of structural data and ligand model necessary for rational design; (ii) selecting the hot spots of the mutational points by the docking simulation of the ligand to the substrate-binding pocket; (iii) selecting a few substituted amino acid residues for individual hot spots; (iv) preparation and evaluation of designed mutants; (v) systematic creation of multiple mutants by ISM. In step iii, the substituted amino acids are selected using computational design, consensus data, and B-factor56. Mutants of lipase B from Candida antarctica (CalB) with improved stereoselectivity for the transacylation between p-nitrophenyl rac-2-phenylpropionate and rac-1-phenylethanol have been created.. 2-2-2-3. Circular permutation (CP). It is known that in directed evolution in nature, proteins rarely experience larger mutations than amino acid substitutions.57 The circular permutation is one of these mutations. Although they have almost the same protein sequence, the positions of the protein terminals (C-terminal and N-terminal) are different. Therefore, a subtle change occurs in the protein structure, and a slight change occurs in the enzymatic properties such as substrate specificity. Lutz and co-workers have developed the circular permutation (CP) method.58 The N-terminal and C-terminal of the target enzyme are linked with an artificial linker, and then the loop structure is cleaved to create the N-terminal and C-terminal at a new position. Lipase B from Candida 12.

(17) Chapter 1. antarctica (CALB) was used as a model enzyme to prepare mutants, and kinetic analysis was done. Some mutants surpassed wild-type CALB activity toward standard substrates. Lutz and co-workers have also reported that substrate specificity can be changed by this method.59a The old yellow enzyme from Saccharomyces pastorianus was used as a model enzyme. Furthermore, it has been proposed by the structural analysis of these mutants that this change of substrate specificity is due to a slight change in the position of the protein main chain that constitutes the substrate pocket.59b. 2-2-2-4. S–S bond formation. The disulfide bond engineering, which introduces a new S–S bond into an enzyme, is one of the most evaluated methods for improving the thermostability. The formation of a disulfide bond has improved protein stability by 2.3 to 5.2 kcal/mol60, and the dramatic improvement of thermostability is expected. It is possible to predict the introduction site of disulfide bond by an algorithm such as SSBOND61a, MODIP61b, Bridge D61c and disulfide by Design 261d. There are some reports that the thermostability has been improved by the formation of disulfide bond.62 On the other hand, the introduction of disulfide bonds affects the flexibility of protein structures, and it is necessary to pay attention to the effect on substrate specificity of the enzyme.63. 2-2-2-5. B-factor. The B-factor is also called the Debye-Waller factor or the temperature factor, which is used as an index for estimating the destabilizing sites in a protein structure. B-FIT is known as a method that combines B-factor and directional evolution, and the thermostability improvement of LipA from Bacillus subtilis has been reported.64 Although the B-factor can be an effective index for the prediction of mutation points, a three-dimensional structure is necessary for accurate prediction. To solve this problem, methods for B-factor calculation 13.

(18) Chapter 1. from the amino acid sequence, such as PROFbval65, MoRFpred66, and ResQ67, have been devised.. 2-3. Rational design. The rational design is a method of designing a mutant enzyme based on the three-dimensional structure of the enzyme. It is also called the computational design, in silico design, or de novo design. The advantage of this method is that the target enzyme can be obtained from a small mutation library. On the other hand, the disadvantage is that it requires three-dimensional structure data. Molecular dynamics (MD) simulations, QM/MM (quantum mechanics/molecular mechanics) method, and protein design method are known. Among the various computer softwares used for the protein design method, Rosetta is the most famous.. 2-3-1. Molecular dynamics (MD) simulation method. MD simulations use the interaction between atoms to calculate the energy of a molecule. MD simulations can give quantitative information of the number, type and duration of different interactions (e.g. hydrogen bonds, electrostatic, polar interactions), and changes on properties (e.g. size, polar vs non polar) of solvent-accessible surface. MD simulations can simulate various molecular events such as catalysis, ligand recognition, folding in hundred nanoseconds to milliseconds. In the protein engineering of enzymes, these simulations have been applied to estimate amino acid residues that affect the enzyme properties such as substrate specificity, enantioselectivity, and stability.68 Dalbya and co-workers have reported the Escherichia coli transketolase (TK) variant which counteracts the enzyme activity– stability trade-off by exploiting correlated molecular-dynamics networks.69 The mutational hot spots affecting the thermostability of TK-M3 mutant were predicted by MD simulations. The best TK-M3 variant had a 10.8-fold improved half-life at 55 °C, and specific activity was increased 3-fold toward aromatic substrates compared to the wild-type enzyme. Chunstivirot 14.

(19) Chapter 1. and co-workers have reported the insight into the substrate specificity change caused by the Y227H mutation of α-glucosidases III (HBGase III) from Apis mellifera through MD simulations.70 The wild-type enzyme mainly hydrolyzes sucrose, while HBGase III_Y227H mainly hydrolyzes maltose. Comparison of relative free binding energies calculated by the MM/GBAS method71 suggested that the difference in substrate specificity is caused by that of free binding energies between the substrate and 227th amino acid residue.. 2-3-2. QM/MM (quantum mechanics/molecular mechanics) method. The QM/MM (quantum mechanics/molecular mechanics) method is a computational chemistry method that combines the advantages of the QM method and the MM method.72a The QM method can be applied to chemical reactions. In the QM/MM method, the regions where chemical reactions occur (e.g. active sites, cofactors, and substrates) are calculated by the QM method, and the remaining regions and solvents are calculated by the MM method.72b These calculations enable the modeling of enzyme catalysis. Warshel and Frushicheva have reported the QM/MM calculations based on the empirical valence bond (EVB) method for the enantioselective reaction of 4-nitrophenyl 2-methylheptanoate with lipase A from Candida antarctica (CALA).73. 2-3-3. Protein design method (Rosetta). The computer software called Rosetta, developed by Baker, is well known as software used for. de. novo. protein. design.. This. software. is. published. on. the. web. site. (https://www.rosettacommons. org/softwar) as RosettaCommons and was continuously developed to improve functionality.74 As a performance of de novo protein design with Rosetta, Baker and co-workers have developed the novel protein such as a large unnatural protein library75a and a mini-protein that binds to botulinum toxin B with high affinity.75b It is considered that Rosetta has reached a level applicable to the de novo design of mini-proteins 15.

(20) Chapter 1. (4 to 12 kDa). The control of the structure of the ligand-binding pocket is the most important point in the catalytic function of the enzyme. However, Rosetta still has a problem of systematically designing the shape and size of ligand-binding pockets. In a recent study, Baker and co-workers have developed an enumerated algorithm to support de novo designs for diverse pocket structures of protein.76 In this study, using the NTF2-like structural superfamily as a model protein, five artificially designed proteins with structures similar to NTF2 were obtained. It is expected that this algorithm will be applied to the catalytic function design of enzymes in the future.. 3. Summary of this thesis. 3-1. Purpose of this study. Lipases are useful enzymes used in academia and industry.1 They have limitations in substrate specificity, enantioselectivity, pH stability, thermostability, and organic solvent tolerance. Because natural lipases do not necessarily meet industrial criteria, enzymatic properties have been strengthened by various methods such as immobilization77, medium engineering78, and protein engineering79. Among them, protein engineering is one of the most powerful methods. In this study, we utilized protein engineering to improve substrate scope, enantioselectivity, and thermostability of industrial lipases called lipase PS (LPS, Amano Enzyme Inc.) and lipase AK (LAK, Amano Enzyme Inc.), both of which are widely used in organic synthesis.. 3-2. Expansion of the substrate scope of LPS and LAK (Chapter 2). Comparison of amino acid sequences of homologous lipases has revealed that the amino acid residues that are distant from the ligand-binding pocket have a great impact on substrate specificity of lipases.80 On the other hand, CAST has been used for improving the substrate application scope of lipases.52,53 Ema and co-workers have developed a BCL14595 variant 16.

(21) Chapter 1. (BCL14595_I287F/I290A) showing expanded substrate scope on the basis of the transition-state model proposed from kinetic and thermodynamic analysis and X-ray crystal structure of a lipase from Burkholderia cepacia.81 Although LPS and LAK are widely used in organic synthesis, they show poor activity for bulky substrates such as 1-phenyl-1-hexanol. We therefore decided to employ site-directed mutagenesis to create excellent mutants of LPS and LAK. Compared to the amino acid sequence of BCL14595, LPS and LAK have 96% homology and 89% homology, respectively. LPS and LAK are therefore considered to have three-dimensional structures quite similar to that of BCL14595. We expected that the substrate scope of LPS and LAK will be expanded by applying the finding of BCL14595_I287F/I290A to LPS and LAK. In addition, the overexpression of the enzyme is important for the commercial production of lipase. Bacterial lipases such as LPS and LAK require an activator (chaperone) to form a folded structure with enzymatic activity. An Escherichia coli expression system has been constructed82 for the heterologous expression of a bacterial lipase although overexpression has not been successful. On the other hand, the extracellular overexpression of LPS and LAK by the Burkholderia cepacia expression system has been established.83 In the heterologous expression of proteins, the folding structure of the enzyme (e.g. disulfide bond formation) may be changed because of the difference in the expression system. Therefore, even if the LPS variant and LAK variant designed from the knowledge of BCL14595_I287F/I290A evaluated in the E. coli expression system is overexpressed in the B. cepacia expression system, the expected catalytic property may not be exhibited. In Chapter 2, for the commercial production of LPS and LAK variants with extended substrate scope, LPS and LAK mutants were designed by applying the knowledge of BCL14595_I287F/I290A to LPS and LAK. To evaluate the enzyme properties, each designed mutant enzyme was expressed in the B. cepacia expression system, and then the catalytic performance for bulky substrates such as 1-phenyl-1-hexanol was compared.. 17.

(22) Chapter 1. 3-2. Improvement of LPS thermostability by the loop-walking method (Chapter 3). The enzyme stability such as thermostability, pH stability, and organic solvent tolerance is an important factor in expanding the industrial use of lipases. Although LPS has relatively good stability, it may not be applied to a manufacturing process because of the problem of stability. Various methods have been reported for the improvement of lipase stability by protein engineering, such as random mutation, structure-based design, and sequence-based design. Random mutation is a method for evaluating a mutation library prepared by error-prone PCR. This method is not suitable for efficient mutant creation because it requires a large number of mutant evaluations in order to obtain a mutant having a target property. Structure-based design is a method for identifying hot spots that affect the enzyme stability from the three-dimensional structure information. It has been reported that this design method includes disulfide bond formation62,63, B-factor64, MD simulations84, and structural comparison with thermostable enzymes.85 In the development of a new mutation method, we focused on the loop structure of the enzyme. It has been reported that the alteration of the loop structure86 by protein engineering has an impact on stability.87 In addition, if the amino acid sequence of an enzyme is reported, the loop structure can be predicted using a computer software such as SWISS-MODEL.88 In Chapter 3, we have developed the loop-walking method (LWM). LMW is a mutation method that focuses on amino acid residues that consist of a loop structure, which may dramatically improve protein thermostability.. 3-4. Prediction of an optimal LPS thermostable variant by multivariate analysis (Chapter 4). In recent years, the enzyme property is further improved by multivariate analysis or machine learning. The thermostability of Bacillus subtilis lipase has been improved by using a sequence homology-based method,89 quantitative structure–thermostability relationship (QSTR) models, and nonlinear support vector machine (SVM),90 and that of Rhizomucor 18.

(23) Chapter 1. miehei lipase has been improved by using the convolution neural network-based (CNN-based) prediction model.91 In this study, we introduced machine learning to effectively narrow down the mutant-based protein engineering, inspired by Kato and co-worker’s study.92 In Chapter 4, in order to obtain a LPS variant with maximum thermostability improvement, we obtained the LPS-L7. variant. with. optimal. amino. acid. substitution. in. the. loop. region. (LPS-L7_P233/L234/V235). We tried to predict the most thermostable LPS-L7 variant with optimal amino acid substitution combination by multivariate analysis of 214 evaluation values of the LPS-L7 variants obtained in Chapter 3 as teaching values. In addition, we also elucidated the thermostability mechanism of the LPS-L7 variants by analyzing the amino acid indices of the structural model calculated by multivariate analysis.. 19.

(24) Chapter 1. References (1). (a) Hasan, F.; Shah, A. A.; Hameed, A. Enzyme Microb. Technol. 2006, 39, 235–251. (b) Andualema, B.; Gessesse, A. Biotechnol. 2012, 11, 100–118. (c) Chandra, P.; Ensepa; Shigh, R.; Arora, P. K. Microb. Cell Fact. 2020, 19, 169.. (2). Hiol, A.; Jonzo, M. D.; Rugani, N.; Druet, D.; Sarda, L.; Comeau, L. C. Enzyme Microb. Technol. 2000, 26, 421-430.. (3). Herrgard, S.; Gibas, C. J.; Subramaniam, S. Biochemistry 2000, 39, 2921–2930.. (4). Nakanishi, J.; Kurono, Y.; Kolde, Y.; Beppu, T. European Patent 1989, 0331376. (5). Toida, J.; Arikawa, Y.; Kondou, K.; Fukuzawa, M.; Sekiguchi, J. Biosci. Biotechnol. Biochem. 1998, 62, 759–763.. (6). Yamaguchi, S.; Mase, T. J. Ferment. Bioeng. 1991, 72, 162–167.. (7). Chen, S. J.; Cheng, C. Y.; Chen, T. L. J. Ferment. Bioeng. 1998, 86, 308–312.. (8). Dalby, P. A. Curr. Opin. Struct. Biol. 2011, 21, 473–480.. (9). Giger, L.; Caner, S.; Obexer, R.; Kast, P.; Baker, D.; Ban, N.; Hilvert, D. Nat. Chem. Biol. 2013, 9, 494–498.. (10) Leemhuis, H.; Kelly, R. M.; Dijkhuizen, L. IUBMB Life 2009, 61, 222–228. (11) Chen, K; Arnold, F. H. Proc. Natl. Acad. Sci. USA 1993, 90, 5618–5622. (12) Wong, T. S.; Zhurina, D.; Schwaneberg, U. Comb. Chem. High Throughput Screen. 2006, 9, 271–288. (13) Wong, T. S.; Tee, K. L.; Hauer, B.; Schwaneberg, U. Nucleic Acids Res. 2004, 32, e26. (14) Li, Z.; Roccatano, D.; Lorenz, M.; Schwaneberg, U. ChemBioChem 2012, 13, 691–699. (15) Shivange, A. V.; Roccatano, D.; Schwaneberg, U. Appl. Microbiol. Biotechnol. 2016, 100, 227–242. (16) Stemmer, W. P. C. Proc. Natl. Acad. Sci. USA 1994, 9, 10747–10751. (17) Müller, K. M.; Stebel, S. C.; Knall, S.; Zipf, G.; Bernauer, H. S.; Arndt, K. M. Nucleic Acids Res. 2005, 33, e117. 20.

(25) Chapter 1. (18) Zhao, H.; Giver, L.; Shao, Z.; Affholter, J. A.; Arnold, F. H. Nat. Biotechnol. 1998, 16, 258–261. (19) Ostermeier, M.; Nixon, A. E.; Shim, J. H.; Benkovic, S. J. Proc. Natl. Acad. Sci. USA, 1999, 96, 3562–3567. (20) Lutz, S.; Ostermeier, M.; Moore, G. L.; Maranas, C. D.; Benkovic, S. J. Proc. Natl. Acad. Sci. USA 2001, 98, 11248–11253. (21) Dalby, P. A. Curr. Opin. Struct. Biol. 2011, 21, 473–480. (22) Wójcik, M.; Telzerow, A.; Quax, W. J.; Boersma, Y. L. Int. J. Mol. Sci. 2015, 16, 24918 –24945. (23) Zhang, Y.; Minagawa, Y.; Kizoe, H.; Miyazaki, K.; Iino, R.; Ueno, H.; Tabata, K. V.; Shimane, Y.; Noji, H. Sci. Adv. 2019, 5, eaav8185. (24) Zeng, W.; Guo, L.; Xu, S.; Chen, J.; Zhou, J. Trends Biotechnol. 2020, 38, 888–906. (25) Baret, J. -C.; Miller, O. J.; Taly, V.; Ryckelynck, M.; El-Harrak, A.; Frenz, L.; Rick, C.; Samuels, M. L.; Hutchison, J. B.; Agresti, J. J.; Link, D. R.; Weitzc, D. A.; Griffiths, A. D. Lab. Chip. 2009, 9, 1850–1858. (26) a) Ostafe, R.; Prodanovic, R.; Commandeur, U.; Fischer, R. Anal. Biochem. 2013, 435, 93–98. b) Ostafe, R.; Prodanovic, R.; Ung, W. L.; Weitz, D. A.; Fischer, R. Biomicrofluidics 2014, 8, 41102. (27) Najah, M.; Mayot, E.; Mahendra-Wijaya, I. P.; Griffiths, A. D.; Ladame, S.; Drevelle, A. Anal. Chem. 2013, 85, 9807–9814. (28) Prodanovic, R.; Ostafe, R.; Blanusa, M.; Schwaneberg, U. Anal. Bioanal. Chem. 2012, 404, 1439–1447. (29) Agresti, J. J.; Antipov, E.; Abate, A. R.; Ahn, K.; Rowat, A. C.; Baret, J.-C.; Marquez, M.; Klibanov, A. M.; Griffiths, A. D.; Weitz, D. A. Proc. Natl. Acad. Sci. USA 2010, 107, 4004–4009. (30) Beneyton, T.; Coldren, F.; Baret, J.-C.; Griffiths, A. D.; Taly, V. Analyst 2014, 139, 21.

(26) Chapter 1. 3314–3323. (31) Nitta, N.; Sugimura, T.; Isozaki, A.; Mikami, H.; Hiraki, K.; Sakuma, S.; Iino, T.; Arai, F.; Endo, T.; Fujiwaki, Y.; Fukuzawa, H.; Hase, M.; Hayakawa, T.; Hiramatsu, K.; Hoshino, Y.; Inaba, M.; Ito, T.; Karakawa, H.; Kasai, Y.; Koizumi, K.; Lee, S. W.; Lei, C.; Li, M.; Maeno, T.; Matsusaka, S.; Murakami, D.; Nakagawa, A.; Oguchi, Y.; Oikawa, M.; Ota, T.; Shiba, K.; Shintaku, H.; Shirasaki, Y.; Suga, K.; Suzuki, Y.; Suzuki, N.; Tanaka, Y.; Tezuka, H.; Toyokawa, C.; Yalikun, Y.; Yamada, M.; Yamagishi, M.; Yamano, T.; Yasumoto, A.; Yatomi, Y.; Yazawa, M.; Carlo, D.D.; Hosokawa, Y.; Uemura, S.; Ozeki, Y.; Goda, K. Cell 2018, 175, 266–276. (32) Mondol, A. S.; Patel, M. D.; Rüger, J.; Stiebing, C.; Kleiber, A.; Henkel, T.; Popp, J.; Schie, I. W. Sensors 2019, 19, 4428 (33) Fox, R. J.; Davis, S. C.; Mundorff, E. C.; Newman, L. M.; Gavrilovic, V.; Ma, S. K.; Chung, L. M.; Ching, C.; Tam, S.; Sheela Muley, S.; Grate, J.; Gruber, J.; Whitman, J. C.; Sheldon, R. A.; Huisman, W. G. Nat. Biotechnol. 2007, 25, 338–344. (34) (a) Fox, R.; Roy, A.; Govindarajan, S.; Minshull, J.; Gustafsson, C.; Jones, J. T.; Emig, R. Protein Eng. 2003, 16, 589–597. (b) Fox, R. J. Theor. Biol. 2005, 234, 187–199. (35) Saito, Y.; Oikawa, M.; Nakazawa, H.; Niide, T.; Kameda, T.; Tsuda, K.; Umetsu, M. ACS Synth. Biol. 2018, 7, 2014–2022. (36) Wu, Z.; Kan, S. J.; Lewis, R. D.; Wittmann, B. J.; Arnold, F. H. Proc. Natl. Acad. Sci. USA 2019, 116, 8852-8858. (37) Lutz, S. Curr. Opin. Biotechnol. 2010, 21, 734–743. (38) Chen, M. M. Y.; Snow, C. D.; Vizcarra, C. L.; Mayo, S. L.; Arnold, F. H. Protein Eng. Des. Sel. 2012, 25, 171–178. (39) Geddie, M. L.; Matsumura, I. J. Mol. Biol. 2004, 279, 26462–26468. (40) Treynor, T. P.; Vizcarra, C. L.; Nedelcu, D.; Mayo, S. L. Proc. Natl. Acad. Sci. USA 2007, 104, 48–53. 22.

(27) Chapter 1. (41) Pei, J. Curr. Opin. Struct. Biol. 2008, 18, 382–386. (42) Pavelka, A.; Chovancova, E.; Damborsky, J. Nucleic Acids Res. 2009, 37, W376–383. (43) Pavlova, M.; Klvana, M.; Prokop, Z.; Chaloupkova, R.; Banas, P.; Otyepka, M.; Wade, R. C.; Tsuda, M.; Nagata, Y.; Damborsky, J. Nat. Chem. Biol. 2009, 5, 727–733. (44) Sumbalova, L.; Stourac, J.; Martinek, T.; Bednar, D.; Damborsky, J. Nucleic Acids Res. 2018, 46, W356–W362. (45) (a) Iwabata, H.; Watanabe, K.; Ohkuri, T.; Yokobori, S.; Yamagishi, A. FEMS Microbiol. Lett. 2005, 243, 393–398. (b) Watanabe K.; Ohkuri, T.; Yokobori, S.; Yamagishi, A. J. Mol. Biol. 2006, 355, 664–674. (c) Shimizu, H.; Yokobori, S.; Ohkuri, T.; Yokogawa, T.; Nishikawa, K.; Yamagishi, A. J. Mol. Biol. 2007, 369, 1060–1069. (d) Yamashiro, K.; Yokobori, S.; Koikeda, S.; Yamagishi, A. Protein Eng. Des. Sel. 2010, 7, 519–528. (46) Furukawa, R.; Toma, W.; Yamazaki, K.; Akanuma, S. Sci. Rep. 2020, 10, 15493. (47) Conchillo-Solé, O.; Groot, N. S.; Avilés, F. X.; Vendrell, J.; Daura, X.; Ventura, S. BMC Bioinformatics 2007, 8, 65. (48) Durme, J. V.; Baets, G. D.; Kant, R. V. D.; Ramakers, M.; Ganesan, A.; Wilkinson, H.; Gallardo, R.; Rousseau, F.; Schymkowitz, J. Protein Eng. Des. Sel. 2016, 29, 285–289. (49) Sormanni, P.; Aprile, F. A.; Vendruscolo, M. J. Mol. Biol. 2015, 427, 478–490. (50) Campeottoa, I.; Goldenzweig, A.; Daveya, J.; Barfodc, L.; Marshallc, J. M.; Silkc, S. E.; Wrighta, K. E.; Draperc, S. J.; Higginsa, M. K.; Fleishman, S. J. Proc. Natl. Acad. Sci. U S A 2017, 114, 998–1002. (51) (a) Nakano, S.; Asano Y. Sci. Rep. 2015, 5, 8193. (b) Matsui, D.; Asano, Y. Biosci. Biotechnol. Biochem. 2015, 79, 11473–1480. (c) Matsui, D.; Nakano, S.; Asano, Y. Sci. Rep. 2017, 7, 9558. (52) Reetz, M. T.; Bocola, M.; Carballeira, J. D.; Zha, D.; Vogel, A. Angew. Chem. Int. Ed. 2005, 44, 4192–4196. (53) Naqvi, T.; Warden, A. C.; French, N.; Sugrue, E.; Carr, P. D.; Jackson, C. J.; Scott, C. 23.

(28) Chapter 1. PLoS ONE 2014, 9, e94177. (54) Reetz, M. T.; Prasad, S.; Carballeira, J. D.; Gumulya, Y.; Bocola, M. J. Am. Chem. Soc. 2010, 132, 9144–9152. (55) Li, D.; Wu, Q.; Reetz, M. T. Methods Enzymol. 2020, 643, 225–242. (56) Sun, Z.; Liu, Q.; Qu, G.; Feng, Y.; Reetz, M. T. Chem. Rev. 2019, 119, 1626–1665. (57) Weiner, J.; Bornberg-Bauer, E. Mol. Biol. Evol. 2006, 23, 734. (58) Qian, Z.; Lutz, S. J. Am. Chem. Soc. 2005, 127, 13466. (59) (a) Daugherty, A. B.; Govindarajan, S.; Lutz, S. J. Am. Chem. Soc. 2013, 135, 14425. (b) Daugherty, A. B.; Horton, J. R.; Cheng, X.; Lutz, S. ACS Catal. 2015, 5, 892. (60) Tidor, B.; Karplus, M. Proteins 1993, 15, 71–79. (61) (a) Hazes, B.; Dijkstra, B. W. Protein Eng. 1988, 2, 119–125. (b) Dani, V. S.; Ramakrishnan, C.; Varadarajan, R. Protein Eng. 2003, 16, 187–193. (c) Pellequer, J. L.; Chen, S. W. Proteins 2006, 65, 192–202. (d) Craig, D. B.; Dombkowski, A. A. BMC Bioinform. 2013, 14, 346. (62) Le, Q. A. T.; Joo, J. C.; Yoo, Y. J.; Kim, Y. H. Biotechnol. Bioeng. 2012, 109, 867–876. (63) Yu, X.-W.; Tan, N.-J.; Xiao, R.; Xu, Y. PLoS One 2012, 7, e46388. (64) (a) Reetz, M. T.; Carballeira, J. D.; Vogel, A. Angew. Chem., Int. Ed. 2006, 45, 7745−7751 (b) Reetz, M. T.; Carballeira, J. D. Nat. Protoc. 2007, 2, 891–903. (65) Schlessinger, A.; Yachdav, G.; Rost, B. Bioinformatics 2006, 22, 891–893. (66) Disfani, F. M.; Hsu, W. L.; Mizianty, M. J.; Oldfield, C. J.; Xue, B.; Dunker, A. K.; Uversky, V. N.; Kurgan, L. Bioinformatics 2012, 28, i75–i83. (67) Yang, J.; Wang, Y.; Zhang, Y. J. Mol. Biol. 2016, 428, 693–701. (68) (a) Romero-Rivera, A.; Garcia-Borràs, M.; Osuna, S. Chem. Commun., 2017, 53, 284– 297. (b) Childers, M. C.; Daggett, V. Mol. Syst. Des. Eng. 2017, 2, 9–33. (69) Yua, H.; Dalbya, P. A. Proc. Natl. Acad. Sci. USA 2018, 115, E12192–E12200. (70) Ayutthaya, P. P. N.; Chanchao, C.; Chunsrivirot, S. PLoS One 2018, 13, e0198484. 24.

(29) Chapter 1. (71) (a) Swanson, J. M.; Henchman, R. H.; McCammon, J. A. Biophys. J. 2004, 86, 64–74. (b) Miller, B. R.; McGee, T. D.; Swails, J. M.; Homeyer, N.; Gohlke, H.; Roitberg, A. E. J. Chem. Theor. Comput. 2012, 8, 3314–3321. (72) (a) Warshel, A.; Levitt, M. J. Mol. Biol. 1976, 103, 227–249. (b) Mulholland, A. J. Biochem. Soc. Trans. 2008, 36, 22–26. (73) Frushicheva M. P.; Warshel, A. ChemBioChem 2012, 13, 215–223. (74) (a) Richter, F.; Leaver-Fay, A.; Khare, S. D.; Bjelic, S.; Baker, D. PLoS One 2011, 6, e19230. (b) Kuhlman, B. J. Biol. Chem. 2019, 294, 19436–19443. (c) Leman, J. K.; Weitzner, B. D.; Lewis, S. M.; Adolf-Bryfogle, J.; Alam, N.; Alford, R. F.; Aprahamian, M.; Baker, D.; Barlow, K. A.; Barth, P.; Basanta, B.; Bender, B. J.; Blacklock, K.; Bonet, J.; Boyken, S. E.; Bradley, P.; Bystroff, C.; Conway, P.; Cooper, S.; Correia, B. E.; Coventry, B.; Das, R.; Jong, R. M. D.; DiMaio, F.; Dsilva, L.; Dunbrack, R.; Ford, A. S.; Frenz, B.; Fu, D. Y.; Geniesse, C.; Goldschmidt, L.; Gowthaman, R.; Gray, J. J.; Gront, D.; Guffy, S.; Horowitz, S.; Huang, P. -S.; Huber, T.; Jacobs, T. M.; Jeliazkov, J. R.; Johnson, D. K.; Kappel, K.; Karanicolas, J.; Khakzad, H.; Khar, K. R.; Khare, S. D.; Khatib, F.; Khramushin, A.; King, I. C.; Kleffner, R.; Koepnick, B.; Kortemme, T.; Kuenze, G.; Kuhlman, B.; Kuroda, D.; Labonte, J. W.; Lai, J. K.; Lapidoth, G.; Leaver-Fay, A.; Lindert, S.; Linsky, T.; London, N.; Lubin, J. H.; Lyskov, S.; Maguire, J.; Malmström, L.; Marcos, E.; Marcu, O.; Marze, N. A.; Meiler, J.; Moretti, R.; Mulligan, V. K.; Nerli, S.; Norn, C.; Ó’Conchúir, S.; Ollikainen, N.; Ovchinnikov, S.; Pacella, M. P.; Pan, X.; Park, H.; Pavlovicz, R. E.; Pethe, M.; Pierce, B. G.; Pilla, K. B.; Raveh, B.; Renfrew, P. D.; Burman, S. S. R.; Rubenstein, A.; Sauer, M. F.; Scheck, A.; Schief, W.; Schueler-Furman, O.; Sedan, Y.; Sevy, A. M.; Sgourakis, N. G.; Shi, L.; Siegel, J. B.; Silva, D. -A.; Smith, S.; Song, Y.; Stein, A.; Szegedy, M.; Teets, F. D.; Thyme, S.B.; Wang, R.Y.-R.; Watkins, A.; Zimmerman, L.; Bonneau, R. Nat. Methods 2020, 17, 665–680. 25.

(30) Chapter 1. (75) (a) Rocklin, G. J.; Chidyausiku, T. M.; Goreshnik, I.; Ford, A.; Houliston, S.; Lemak, A.; Carter, L.; Ravichandran, R.; Mulligan, V. K.; Chevalier, A.; Arrowsmith, C. H.; Baker, D. Science 2017, 357, 168–175. (b) Chevalier, A.; Silva, D.-A.; Rocklin, G. J.; Hicks, D. R.; Vergara, R.; Murapa, P.; Bernard, S. M.; Zhang, L.; Lam, K.-H.; Yao, G.; Bahl, C. D.; Miyashita, S.-I.; Goreshnik, I.; Fuller, J. T.; Koday, M. T.; Jenkins, C. M.; Colvin, T.; Carter, L.; Bohn, A.; Bryan, C. M.; Fernández-Velasco, D. A.; Stewart, L.; Dong, M.; Huang, X.; Jin, R.; Wilson, I. A.; Fuller, D. H.; Baker, D. Nature 2017, 550, 74–79. (76) Basanta, B.; Bick, M. J.; Bera, A. K.; Norn, C.; Chow, C. M.; Carter, L. P.; Goreshnik, I.; Dimaio, F.; Baker, D. Proc. Natl. Acad. Sci. USA 2020, 117, 22135–22145. (77) (a) Takahashi, H.; Li, B.; Sasaki, T.; Miyazaki, C.; Kajino, T.; Inagaki, S. Chem. Mater. 2000, 12, 3301–3305. (b) Kato, K.; Seelan, S. J. Biosci. Bioeng. 2010, 109, 615–617. (c) Hernandez, K.; Fernandez-Lafuente, R. Enzyme Microb. Technol. 2011, 48, 107–122. (d) Royhaila, N.; Mohamad, N. R.; Marzuki, N. H. C.; Buang, N. A.; Huyop, F.; Wahab, R. A. Biotechnol. Biotechnol. Equip. 2015, 29, 205–220. (78) Stepankova, V.; Bidmanova, S.; Koudelakova, T.; Prokop, Z.; Chaloupkova, R.; Damborsky, J. ACS Catal. 2013, 3, 2823–2836. (79) (a) Svendsen, A. Biochim. Biophys. Acta 2000, 1543, 223–238. (b) Burcu, T. Y.; Ceren, A.; Petek, Z. C. Protein Engineering Methods and Applications, Pravin K. Ed., InTech, 2012, pp 33–58. (80) Hirose, Y.; Kariya, K.; Nakanishi, Y.; Kurono, Y.; Achiwa, K. Tetrahedron Lett. 1995, 36, 1063–1066. (81) (a) Ema, T.; Kobayashi, J.; Maeno, S.; Sakai, T.; Utaka, M. Bull. Chem. Soc. Jpn. 1998, 71, 443–453. (b) Ema, T.; Jittani, M.; Furuie, K.; Utaka, M.; Sakai, T. J. Org. Chem. 2002, 67, 2144–2151. (c) Ema, T.; Yamaguchi, K.; Wakasa, Y.; Yabe, A.; Okada, R.;. 26.

(31) Chapter 1. Fukumoto, M.; Yano, F.; Korenaga, T.; Utaka, M.; Sakai, T. J. Mol. Catal. B: Enz. 2003, 22, 181–192. (82) Xiangping, W.; Pengyong, Y.; Erzheng, S.; Jingjing, X.; Bei, G.; Dongzhi, W. BMC Biotechnol. 2012, 12, 58–66. (83) Nakanishi, Y.; Kurono, Y.; Hirose, S. Jpn. Kokai Tokkyo Koho, JP 06153965 A, 1994. (84) Singh, B.; Bulusu, G.; Mitra, A. J. Phys. Chem. B 2015, 119, 392–409. (85) Damnjanović, J.; Nakano, H.; Iwasaki, Y. Biotechnol. Bioeng. 2014, 111, 674–682. (86) (a) Ruslan, R.; Rahman, R. N. Z. R. A.; Leow, T. C.; Ali, M. S. M.; Basri, M.; Salleh, A. B. J. Mol. Sci. 2012, 13, 943–960. (b) Li, B.; Yang, G.; Wu, L.; Feng, Y. PLoS One 2012, 7, e46881. (c) Yedavalli, P.; Rao, N. M. Protein Eng. Des. Sel. 2013, 26, 317–324. (d) Wu, J.-P.; Li, M.; Zhou, Y.; Yang, L.-R.; Xu, G. Biotechnol. Lett. 2015, 37, 403–407. (87) Zhao, J.; Wang, Z.; Gao, F.; Lin, J.; Yang, L.; Wu, M. RSC Adv. 2018, 8, 41247. (88) Schwede, T.; Kopp, J.; Guex, N.; Peitsch, M. C. Nucleic Acids Res. 2003, 31, 3381– 3385. (89) Bommarius, A. S.; Broering, J. M.; Chaparro-Riggers, J. F.; Polizzi, K. M. Curr. Opin. Biotechnol. 2006, 17, 606–610. (90) Tian, F.; Yang, C.; Wang, C.; Guo, T.; Zhou, P. J. Mol. Model. 2014, 20, 2257. (91) Fang, X.; Huang, J.; Zhang, R.; Wang, F.; Zhang, Q.; Li, G.; Yan, J.; Zhang, H.; Yan, Y.; Xu, L. J. Chem. Inf. Model. 2019, 59, 4833–4843. (92) Kato, R.; Nakano, H.; Konishi, H.; Kato, K.; Koga, Y.; Yamane, T.; Kobayashi, T.; Honda, H. J. Mol. Biol. 2005, 351, 683–692.. 27.

(32) Chapter 1. 28.

(33) Chapter 2. Chapter 2. Synthetically Useful Variants of Industrial lipases from Burkholderia cepacia and Pseudomonas fluorescens. Abstract Industrial enzymes called lipase PS (LPS) and lipase AK (LAK), which originate from Burkholderia cepacia and Pseudomonas fluorescens, respectively, are synthetically useful biocatalysts. To strengthen their catalytic performances, we introduced two mutations into hot spots of the active sites (residues 287 and 290). The LPS_L287F/I290A double mutant showed high catalytic activity and enantioselectivity for poor substrates for which the wild-type enzyme showed very low activity. The LAK_V287F/I290A double mutant was also an excellent biocatalyst with expanded substrate scope, which was comparable to the LPS_L287F/I290A double mutant. Thermodynamic parameters were determined to address the origin of the high enantioselectivity of the double mutant. The ∆∆H‡ term, but not the ∆∆S‡ term, was predominant, which suggests that the enantioselectivity is driven by a differential energy associated with intermolecular interactions around Phe287 and Ala290. A remarkable solvent effect was observed, giving a bell-shaped profile between the E values and the log P or  values of solvents with the highest E value in i-Pr2O. This suggests that organic solvent with appropriate hydrophobicity and polarity provides the double mutant with some flexibility that is essential for the excellent catalytic performance.. 29.

(34) Chapter 2. Introduction Enzymes show high catalytic activity and stereoselectivity under mild conditions. Industrial enzymatic processes are widely accepted from the viewpoint of environmental harmony and sustainability.1 Because of the limited diversity of natural enzymes, new technologies have been developed to alter the structure and property of enzymes.2 The directed evolution method can evolve an enzyme stepwise using random mutagenesis and a high-throughput screening system. Although no information about the enzyme structure and the reaction mechanism is necessary for directed evolution, a large number of variants need to be screened. On the other hand, when the enzyme structure and the reaction mechanism are known, a rational design approach with site-directed mutagenesis is effective and efficient. Lipases are synthetically useful biocatalysts that show high catalytic activity and enantioselectivity for a broad range of unnatural substrates in both water and organic solvent.1a,3 In particular, they exert high enantioselectivity for various secondary alcohols (Scheme 1).. Scheme 1. Typical good substrates for lipases, where R and Ar designate the alkyl group and the aromatic/large substituent, respectively.. We performed mechanistic studies based on kinetic and thermodynamic analysis, X-ray crystal structures, and MO calculations and proposed a transition-state model (Figure 1a) to explain the origin of enantioselectivity of lipases for secondary alcohols.4 The transition-state model essentially represents a mechanism by which the slower-reacting (S)-enantiomer is 30.

(35) Chapter 2. disfavored. We also introduced point mutation(s) into the active site of a lipase to rationally control the enantioselectivity.5 We redesigned a Burkholderia cepacia NBRC14595 lipase, BCL14595, to create a more useful double mutant, which is herein called BCL14595_I287F/I290A.5b,c The wild-type BCL14595. showed. very. low. activity. for. 1-phenyl-1-hexanol,. while. the. BCL14595_I287F/I290A double mutant showed high activity and enantioselectivity for this secondary alcohol. We have proposed that the reaction is accelerated by the CH/ interactions between Phe287 and the alkyl chain of the (R)-enantiomer (Figure 1b).5b,c,6 On the other hand, steric repulsion takes place between Phe287 and the benzene ring of the (S)-enantiomer (not shown). Phe287 is thus considered to have dual mode interactions with the two enantiomers, improving both catalytic activity and enantioselectivity. In addition, the I290A mutation removes steric hindrance to accelerate the reaction of the (R)-enantiomer (Figure 1b). The wild-type BCL14595 showed very low activity with an E value of 5 for 1-phenyl-1-hexanol, while the BCL14595_I287F/I290A double mutant showed much higher activity with a high E value of >200. We employed E. coli for the heterologous expression of the BCL14595 gene and successfully converted a denatured protein (inclusion body) into an active enzyme by in vitro refolding with a separately overproduced activator (chaperon), which is however unsuitable for large-scale preparation.5 In contrast, an industrial enzyme called lipase PS (LPS, Amano Enzyme Inc.), which is a homologous protein of BCL14595 with 13 different amino acid residues (96% homology), is produced on a large scale with a B. cepacia expression system. Active LPS can be secreted into a culture broth.7 Here we prepared both the LPS_wild-type enzyme and the LPS_L287F/I290A double mutant using the B. cepacia expression system and compared their catalytic properties. We also investigated another industrial enzyme called lipase AK (LAK, Amano Enzyme Inc.), which originates from Pseudomonas fluorescens. LAK has 35 different amino acid residues (89% homology) as compared with BCL14595 or 31.

(36) Chapter 2. LPS. We compared the enzymatic characteristics of the LAK_wild-type enzyme and the LAK_V287F/I290A double mutant. Substrate mapping revealed excellent catalytic performances (expanded substrate scope) of the LPS_L287F/I290A and LAK_V287F/I290A double mutants. The temperature effect and solvent effect were investigated to address the origin of the high enantioselectivity of the double mutant.. 32.

(37) Chapter 2. Figure 1. (a) The transition-state model to explain the enantioselectivity of lipase toward secondary alcohols (residues 287 and 290 are added to the original version). (i) The C–O bond of the substrate takes the gauche conformation with respect to the breaking C–O bond, which is due to the stereoelectronic effect. (ii) The H atom attached to the stereocenter of the substrate is syn-oriented toward the carbonyl O atom to minimize the torsional strain. Enantioselectivity is explained by the conformational requirements and repulsive interactions and/or strains. Typically, the (R)-enantiomer reacts faster because, in this favorable conformation shown in blue, the larger substituent (R1) can be directed toward external solvent without severe strain and/or steric hindrance. (b) The catalytic activity of the BCL14595_I287F/I290A double mutant for (R)-1-phenyl-1-hexanol is enhanced by introducing attractive CH/ interactions and removing steric hindrance.. 33.

(38) Chapter 2. Result and Discussion The recombinant enzymes prepared and purified as described in the Experimental section were immobilized on Toyonite-200M according to the literature.5 A mixture of secondary alcohol 1, the immobilized enzyme, and molecular sieves 3A in i-Pr2O was stirred at 30 °C for 30 min, and vinyl acetate was added to start the reaction (Scheme 2). The progress of the reaction was monitored by TLC and NMR, and the reaction was stopped by filtration. Acetate 2 and alcohol 1 were separated. by. silica. gel. column. chromatography. The enantiomeric purity was determined by GC, HPLC, or NMR, and the E value was calculated according to the literature.8 The results are shown in Tables 1 and 2, where the reaction rates can be compared because the same amounts of enzyme and substrate were used in all cases.. Scheme 2. Lipase-catalyzed kinetic resolution of 1.. The results of kinetic resolution of 1 with the LPS_wild-type enzyme and the LPS_L287F/I290A double mutant are shown in Table 1. Alcohols 1a–e with a small substituent such as the methyl group were resolved well by the double mutant as well as the wild-type enzyme in most cases (entries 1–5). The double mutant exhibited superior activity for 1a and a comparable E value as compared with the wild-type enzyme (entry 1). Although the double mutant showed slightly lower activity for 1b than the wild-type enzyme, the enantioselectivity of the former was improved (entry 2).. 34.

(39) Chapter 2. Table 1. Substrate scope of the LPS_L287F/I290A double mutant and the LPS_wild-type enzyme.a entry 1 2 3 4 5 6 7 8 9 10 11 12 13. 1 1a 1b 1c 1d 1e 1f 1g 1h 1i 1j 1k 1l 1m. time (h) 1 4 2 1 10 15 2.5 96 24 2 3 2 1.5. L287F/I290A c (%)b Ec 50 >200 37 >200 50 90 51 >200 40 >200 45 >200 39 >200 40 105 16 31 50 >200 49 >200 49 >200 39 113. wild-type c (%)b Ec 40 >200 42 119 35 >200 53 >130 46 >200 36 >200 24 >200 19 43 49 117 d 9 – d 5 – d 5 – d 10 –. . a Reaction conditions: immobilized lipase (200 mg, 0.5% (w/w) enzyme/Toyonite-200M), 1 (0.50 mmol), vinyl acetate (1.0 mmol), molecular sieves 3A (three pieces), dry i-Pr2O (5 mL), 30 °C. b Conversion calculated from c = ee(1)/(ee(1) + ee(2)). c Calculated from E = ln[1 – c(1 + ee(2))]/ln[1 – c(1 – ee(2))]. d Conversion calculated from 1H NMR.. In the case of 1c, the catalytic activity of the double mutant was improved but with a drop of the E value (entry 3). We consider that the pocket comprising Phe287 and Ala290 (Figure 1b) attracts the methylene chain of (S)-1c to enhance the reactivity of (S)-1c, lowering the enantioselectivity. Alcohols 1f–i with a substituent that is slightly larger than the methyl group were also examined. The double mutant achieved higher conversions for 1f–h and higher enantioselectivity for 1h than the wild-type enzyme (entries 6–8). This outcome for 1h was unexpected because a fluorine-containing substrate has previously exhibited a dropped enantioselectivity because of the lack of CH/ interactions.5b,c The double mutant showed lower activity and enantioselectivity for 1i than the wild-type enzyme (entry 9). It is likely 35.

(40) Chapter 2. that the cleft comprising Phe287 and Ala290 (Figure 1b) cannot accommodate well the ethyl ester group of (R)-1i. To our delight, the double mutant showed much higher activity and enantioselectivity for 1j–m, for which the wild-type enzyme showed very low activity (entries 10–13). The trimethylsilyl group or the thiazole ring had a good influence on the outcome (entries 12,13). The results of kinetic resolution of 1 with the LAK_wild-type enzyme and the LAK_V287F/I290A double mutant are shown in Table 2. The double mutant and the wild-type enzyme showed comparable enantioselectivities for 1a (entry 1). The double mutant showed higher enantioselectivity for 1b than the wild-type enzyme (entry 2), whereas the former gave a lower E value for 1c than the latter (entry 3). The double mutant showed excellent enantioselectivity for 1d–e as the wild-type enzyme did (entries 4,5). The double mutant showed higher activity for 1f–g than the wild-type enzyme (entries 6–7). The E values of the double mutant for 1h–i were much improved (entries 8–9). Interestingly, the V287F/I290A double mutations in LAK enhanced enantioselectivity for 1i (Table 2, entry 9) although the L287F/I290A double mutations in LPS decreased enantioselectivity (Table 1, entry 9). The LAK_V287F/I290A double mutant is a useful biocatalyst because 1h–i are reported to be poor substrates for a wild-type enzyme.9,10 Furthermore, the double mutant exhibited high activity and enantioselectivity for 1j–m, for which the wild-type enzyme showed poor activity and enantioselectivity (entries 10–13).. 36.

(41) Chapter 2. Table 2. Substrate scope of the LAK_V287F/I290A double mutant and the LAK_wild-type enzyme.a entry 1 2 3 4 5 6 7 8 9 10 11 12 13 a. 1 1a 1b 1c 1d 1e 1f 1g 1h 1i 1j 1k 1l 1m. time (h) 1 4 2.5 0.25 10 24 3 60 12 1.5 4 3 1.5. V287F/I290A c (%)b Ec 50 >200 42 >200 49 55 43 >200 25 >200 43 >200 43 >200 48 90 41 >200 50 >200 46 >200 50 >200 48 134. wild-type c (%)b Ec 47 >200 50 78 47 >200 50 >200 49 >200 35 >200 26 >200 45 4 43 30 d 4 – d 6 – d 6 – 27 8. Reaction conditions: immobilized lipase (200 mg, 0.5% (w/w) enzyme/Toyonite-200M), 1. (0.50 mmol), vinyl acetate (1.0 mmol), molecular sieves 3A (three pieces), dry i-Pr2O (5 mL), 30 °C. b Conversion calculated from c = ee(1)/(ee(1) + ee(2)). c Calculated from E = ln[1 – c(1 + ee(2))]/ln[1 – c(1 – ee(2))]. d Conversion calculated from 1H NMR.. We performed molecular modeling (MOE, MOLSYS Inc.) to understand the catalytic behaviors of the wild-type enzymes and the double mutants of LPS and LAK. The structure of LPS was obtained by refining the X-ray crystal structure (PDB: 1OIL), and that of LAK was constructed by homology modeling using LPS as a template. The double mutants of LPS and LAK were then generated from the corresponding wild-type enzymes. The active sites of these lipases are shown in Figure 2. The LPS_wild-type enzyme with Leu287 has a narrow pocket as compared with the LAK_wild-type enzyme with Val287, which can account for a tendency that LPS is more enantioselective than LAK (Tables 1 and 2). The active-site pockets of the double mutants of LPS and LAK are deeper around residue 290 than those of the corresponding wild-type enzymes, the former of which can accommodate the substituents 37.

(42) Chapter 2. that are larger than the methyl group. Although it is reasonable that the double mutants of LPS and LAK with similar pockets in size and shape showed similar catalytic properties, the irregular behaviors of LPS and LAK toward 1i (entry 9 in Tables 1 and 2) may result from the different electrostatic potentials of their active sites (Figure 2).. (a). (b) Leu287 Ile290. (c). Leu287 Phe287. Phe287 Ala290 Asp264 His286 Ile290 Ala290 Ser87. (d). (e) Val287 Ile290. Val287 Phe287. (f) Phe287 Ala290. Asp264 His286 Ile290 Ala290 Ser87. Figure 2.. Electrostatic potential maps of the active sites of (a) the LPS_wild-type enzyme,. (b) the LPS_L287F/I290A double mutant, (d) the LAK_wild-type enzyme, and (e) the LAK_V287F/I290A double mutant. Superimposed views of the active sites of (c) the LPS_wild-type enzyme (blue) and the LPS_L287F/I290A double mutant (red) and (f) the LAK_wild-type enzyme (blue) and the LAK_V287F/I290A double mutant (red). Each of (a)– (c) and (d)–(f) is seen from the same direction.. 38.

(43) Chapter 2. The mechanism of enantioselectivity can be inspected by thermodynamic analysis.4c Plot of ln E against 1/T according to equation 1 gives the ∆∆H‡ and ∆∆S‡ values.11 ln E = –∆∆H‡/(RT) + ∆∆S‡/R. (eq 1). The ∆∆H‡ and ∆∆S‡ values represent the differences in activation enthalpy (∆H‡) and entropy. (∆S‡),. respectively,. between. the. faster-reacting. and. slower-reacting. enantiomers (equations 2,3). ∆∆H‡ = ∆H‡fast – ∆H‡slow. (eq 2). ∆∆S‡ = ∆S‡fast – ∆S‡slow. (eq 3). The ∆H‡ value involves a change of the energy associated with covalent bonds, strain, and intermolecular interactions, while the ∆S‡ value is associated with a change of the disorder of LPS T (°C) 1/T E of the lnEdouble mutant is enhanced by the additional the system. If the enantioselectivity L287F/I290A. 30. 0.0033. 57.4. 4.05. 4.5. attractive interaction and steric 45.6 repulsion L287F/I290A 35 0.003247 3.82 L287F/I290A. 40. 0.003195. 38.8. 3.66 3.58 L287F/I290A 50 0.003096 26.3 mutant should be negatively larger than 3.27 that wild-type 30 0.0033 4.92 1.59 0.003247 for wild-type the wild-type35 enzyme. We 4.09 selected1.41a wild-type 40 0.003195 3.92 1.37 wild-type of LPS 45 (wild-type 0.003145 enzyme 3.57 combination and 1.27 wild-type 50 0.003096 3.32 1.20. 4.0. ‡ (Figure 1), the ∆∆H value for the L287F/I290A 45 0.003145 35.7 double. the L287F/I290A double mutant) and 1m. 3.5. lnE. 3.0. y = 3531.6x - 7.6143 R² = 0.96812. 2.5 2.0. y = 1811.1x - 4.4212 R² = 0.95392. 1.5 1.0. because of the moderate to good E values and DDH. DDS. DDG. ‡. determined the thermodynamic values (∆∆H. 0.5 0.00308 0.00313 0.00318 0.00323 0.00328. 1/T. L287F/I290A -7.02 -15.1 -2.433 ‡ wild-type -3.60 -8.8 -0.9368 and ∆∆S ) from the E values at 30–50 °C. Figure 3. Temperature effect on the enantioselectivity in the kinetic resolution of. according to equation 1. The results are. 1m with the LPS_L287F/I290A double mutant (circle) and the LPS_wild-type enzyme (square) in i-Pr2O.. summarized in Tables 3 and 4 and Figure 3.. 39.

(44) Chapter 2. Table 3. Temperature effect in the kinetic resolution of 1m with the LPS_L287F/I290A double mutant and the LPS_wild-type enzyme.a. a. LPS. T (°C). time (h). c (%)b. Ec. L287F/I290A L287F/I290A L287F/I290A L287F/I290A. 30 35 40 45. 2 1 1 1. 50 33 41 44. 57 46 39 36. L287F/I290A. 50. 1. 50. 26. wild-type wild-type. 30 35. 7 5. 40 41. 4.9 4.1. wild-type wild-type wild-type. 40 45 50. 4 4 4. 39 44 53. 3.9 3.6 3.3. Reaction conditions: immobilized lipase (200 mg, 0.5% (w/w) enzyme/Toyonite-200M), 1m. (0.50 mmol), vinyl acetate (1.0 mmol), dry i-Pr2O (5 mL), molecular sieves 3A (three pieces). b. Conversion calculated from c = ee(1m)/(ee(1m) + ee(2m)). c Calculated from E = ln[1 – c(1. + ee(2m))]/ln[1 – c(1 – ee(2m))].. Table 4. Thermodynamic parameters for the kinetic resolution of 1m with the LPS_L287F/I290A double mutant and the LPS_wild-type enzyme in i-Pr2O.. a. ∆∆H‡ –1 (kcal·mol ). ‡ ∆∆S –1 (cal·K ·mol–1). ∆∆G‡ –1 a (kcal·mol ). L287F/I290A. –7.02. –15.1. –2.43. wild-type. –3.60. –8.8. –0.94. Calculated from ∆∆G‡ = ∆∆H‡ – 303∆∆S‡.. In both cases, the ∆∆H‡ value is a dominant factor in the ∆∆G‡ value, which indicates that enantioselectivity is driven by a differential energy associated with covalent bonds, strain, and intermolecular interactions (Table 4). The ∆∆H‡ value of the double mutant is two times negatively larger than that of the wild-type enzyme. The attractive interaction between 40.

(45) Chapter 2. Phe287 and the alkyl chain of (R)-1m would decrease the ∆H‡fast value, and steric repulsion between Phe287 and the thiazole ring of (S)-1m would increase the ∆H‡slow value, both of which give a negatively larger ∆∆H‡ value (equation 2). Table 4 also indicates a partial compensation effect; the ∆∆H‡ value, which becomes negatively larger, is counterbalanced by the ∆∆S‡ value, which also becomes negatively larger.4c Steric repulsion between Phe287 and (S)-1m favors the ∆∆H‡ term because the ∆H‡slow value becomes larger, whereas it disfavors the ∆∆S‡ term because the ∆S‡slow value increases with an increase in the disorder of the system. The CH/ interaction between Phe287 and (R)-1m as well as the removal of steric hindrance between Ala290 and (R)-1m also favor the ∆∆H‡ term because the ∆H‡fast value becomes smaller, whereas they disfavor the ∆∆S‡ term because the ∆S‡fast value decreases with a decrease in the disorder of the system. Therefore, the trends observed for the ∆∆H‡ and ∆∆S‡ values are consistent with the transition-state model (Figure 1). The solvent effect is often remarkable and even provides a valuable insight into the mechanism of biocatalysis.12 We therefore investigated the solvent effect on the kinetic resolution of 1m with the LPS_L287F/I290A double mutant (Table 5). The best solvent was found to be i-Pr2O. The log P value, which is the logarithm of a partition coefficient P of a solvent between 1-octanol and water, is a measure of hydrophobicity of the solvent.13 Table 5 indicates that the E value and the reaction rate sharply decreased with a decrease of the log P value. The relationships between the E value and the log P value or permittivity () are plotted in Figure 4.14 We speculate that hydrophilic solvent such as 1,4-dioxane deprives the lipase of the essential water, which lowers the protein flexibility that is essential for the catalytic activity.12a. 41.

(46) Chapter 2. Table 5. Solvent effect in the kinetic resolution of 1m with the LPS_L287F/I290A double mutant.a. a. solvent. log P. . time (h). c (%)b. Ec. 1,4-dioxane acetone THF Et2O. –1.1 –0.23 0.49 0.85. 2.2 21 7.5 4.3. 48 48 48 11. 38 11 19 37. 9 7 11 27. i-Pr2O toluene cyclohexane. 1.9 2.5 3.2. 3.4 2.4 2.0. 2 5 1. 50 50 42. 57 42 50. hexane. 3.5. 1.9. 0.5. 38. 24. Reaction conditions: LPS_L287F/I290A double mutant (200 mg, 0.5% (w/w). enzyme/Toyonite-200M), 1m (0.50 mmol), vinyl acetate (1.0 mmol), dry organic solvent (5 mL), molecular sieves 3A (three pieces), 30 °C.. b. Conversion calculated from c =. ee(1m)/(ee(1m) + ee(2m)). c Calculated from E = ln[1 – c(1 + ee(2m))]/ln[1 – c(1 – ee(2m))].. In contrast, hydrophobic solvent such as hexane enables the lipase to retain the essential water, which keeps the flexibility of the protein. The E value was the. highest. where. the. in. i-Pr2O,. lipase. is. considered to have the most appropriate. flexibility. Figure 4. The solvent effect in the LPS_L287F/ I290A-catalyzed kinetic resolution of 1m. (a) The correlation between the E value and the log P value of the solvent. (b) The correlation between the E value and the permittivity () of the solvent.. (Figure 4a). A bell-shaped profile with a peak at i-Pr2O is also seen when the E values are plotted against the  values (Figure 4b). It is likely that the polarity of the solvent also affects 42.

(47) Chapter 2. the protein flexibility and that the highest E value is achieved in the solvent giving the lipase appropriate flexibility.. Summary Industrial enzymes called lipase PS (LPS) and lipase AK (LAK), which originate from Burkholderia cepacia and Pseudomonas fluorescens, respectively, are synthetically useful biocatalysts. To strengthen their catalytic performances, we introduced two mutations into the hot spots of the active sites (residues 287 and 290). The LPS_L287F/I290A double mutant showed high catalytic activity and enantioselectivity for poor substrates for which the wild-type enzyme showed very low activity. This double mutant also exhibited high catalytic activity and enantioselectivity for good substrates of the wild-type enzyme. Clearly, the substrate scope of the double mutant has been broadened. It should be emphasized again that sterically demanding substrates possessing two bulky substituents on both sides are usually poor substrates.15 The LAK_V287F/I290A double mutant is also an excellent biocatalyst with expanded substrate scope, which was comparable to the LPS_L287F/I290A double mutant. Although the two double mutants were equally excellent on the whole, some differences were also observed between them. It is therefore recommended that the better one be selected on a case-by-case basis. The enantioselectivity of the LPS_L287F/I290A double mutant was driven by the differential activation enthalpy (∆∆H‡), and this ∆∆H‡ value for the double mutant was negatively larger than that for the wild-type enzyme, both of which suggest that attractive interactions and/or steric repulsion are used for chiral discrimination in the transition state. Bell-shaped profiles with a peak at i-Pr2O were obtained when the E values for the double mutant were plotted against the log P or.  values of organic solvents, which suggests that. appropriate protein flexibility is essential for the excellent catalytic performances. The LPS_L287F/I290A and LAK_V287F/I290A double mutants will find many applications in the kinetic resolution and dynamic kinetic resolution of various chiral alcohols.5c,16 43.