Enhanced Enzyme Kinetic Stability by Increasing Rigidity within the Active Site*

Background: Improving the kinetic stability of enzymes is a key issue for protein engineers. Results: Mutagenesis of residues with a high B factor located within 10 Å of the catalytic Ser105 residue enhances kinetic stability dramatically. Conclusion: Increasing the rigidity of the flexible segment within the active site improves enzymatic kinetic stability. Significance: Optimization of the active site may an alternative, efficient approach for enhancing protein stabilization. Enzyme stability is an important issue for protein engineers. Understanding how rigidity in the active site affects protein kinetic stability will provide new insight into enzyme stabilization. In this study, we demonstrated enhanced kinetic stability of Candida antarctica lipase B (CalB) by mutating the structurally flexible residues within the active site. Six residues within 10 Å of the catalytic Ser105 residue with a high B factor were selected for iterative saturation mutagenesis. After screening 2200 colonies, we obtained the D223G/L278M mutant, which exhibited a 13-fold increase in half-life at 48 °C and a 12 °C higher T5015, the temperature at which enzyme activity is reduced to 50% after a 15-min heat treatment. Further characterization showed that global unfolding resistance against both thermal and chemical denaturation also improved. Analysis of the crystal structures of wild-type CalB and the D223G/L278M mutant revealed that the latter formed an extra main chain hydrogen bond network with seven structurally coupled residues within the flexible α10 helix that are primarily involved in forming the active site. Further investigation of the relative B factor profile and molecular dynamics simulation confirmed that the enhanced rigidity decreased fluctuation of the active site residues at high temperature. These results indicate that enhancing the rigidity of the flexible segment within the active site may provide an efficient method for improving enzyme kinetic stability.


Enzyme stability is an important issue for protein engineers.
Understanding how rigidity in the active site affects protein kinetic stability will provide new insight into enzyme stabilization. In this study, we demonstrated enhanced kinetic stability of Candida antarctica lipase B (CalB) by mutating the structurally flexible residues within the active site. Six residues within 10 Å of the catalytic Ser 105 residue with a high B factor were selected for iterative saturation mutagenesis. After screening 2200 colonies, we obtained the D223G/L278M mutant, which exhibited a 13-fold increase in half-life at 48°C and a 12°C higher T 50 15 , the temperature at which enzyme activity is reduced to 50% after a 15-min heat treatment. Further characterization showed that global unfolding resistance against both thermal and chemical denaturation also improved. Analysis of the crystal structures of wild-type CalB and the D223G/L278M mutant revealed that the latter formed an extra main chain hydrogen bond network with seven structurally coupled residues within the flexible ␣10 helix that are primarily involved in forming the active site. Further investigation of the relative B factor profile and molecular dynamics simulation confirmed that the enhanced rigidity decreased fluctuation of the active site residues at high temperature. These results indicate that enhancing the rigidity of the flexible segment within the active site may provide an efficient method for improving enzyme kinetic stability.
Enzyme stability is a significant concern for protein engineers because of its great industrial importance. Numerous studies have revealed that enzymes maintain their structural stability through various kinds of non-covalent interactions, including hydrogen bonds, salt bridges, hydrophobic interactions, and van der Waals forces. Therefore, the most common strategy to improve enzymatic stability is to introduce new interactions that raise the transition state energy barrier for denaturation. In the past, stable biocatalysts were generated by amino acid substitution to increase structural rigidity (1-3), restrict conformational flexibility (4 -11), and increase interactions between unstable domains (12)(13)(14)(15). However, insufficient knowledge of the structure-function relationship hampered the identification of suitable targets for mutation. Exploring novel stabilization mechanisms and strategies for guiding enzyme design are therefore necessary.
Advances in structural biology and bioinformatics have provided new capabilities for evaluating the flexibility of segments/ residues in a protein that have subsequently helped in the selection of key mutation sites for protein stabilization. Tian et al. (16) used molecular dynamics (MD) 2 simulation to select residues with high root mean square fluctuation (RMSF) values for mutation to proline to improve protein thermostability. Moreover, Joo et al. (17) achieved thermostabilization of Bacillus circulans xylanase by designing a flexible surface cavity using computational tools. The most technological advance thus far involves improving the thermostability of the smallest lipase in nature, namely Bacillus subtilis lipase (composed of 181 residues that form the core ␣/␤-hydrolase fold), based on its B factor (18). This value represents the diffusion of atomic electron densities with respect to their equilibrium positions due to thermal motion and positional disorder. Thus, the B factor can be used to indicate the mobility of individual residues. Reetz et al. (18) chose 10 residues with a high B factor to construct iterative saturation mutagenesis libraries. After screening, the best mutant showed a half-life (t1 ⁄ 2 ) at 55°C of 980 min, which is ϳ490-fold higher than that of the wild-type protein. MD simulations have suggested that stepwise formation of an extensive hydrogen bond/salt bridge network of structurally coupled residues on the enzyme surface increases protein thermostability (19). Successful enhancement of B. subtilis lipase thermostability using this approach led to widespread focus on this strategy (20 -22). However, the analysis of enzymes larger and more complex than B. subtilis lipase yielded inconclusive results. For example, when Kim et al. (21) mutated the seven residues with the highest B factor on the surface of Candida antarctica lipase B (CalB; composed of 317 residues that constitute a lid domain and extra secondary structures around the core ␣/␤-hydrolase fold), they obtained a mutation that increased the t1 ⁄ 2 at 50°C (54 min) only 24%. This result implied that, because of the complexity of enzyme structures, only reducing the surface fluctuation may be insufficient for protecting the active site from heat-induced conformational changes that inactivate the enzyme.
Extensive research has resulted in two distinct definitions for in vitro protein stability, namely thermodynamic stability and kinetic stability. Thermodynamic stability involves the resistance of a folded protein conformation to denaturation, whereas kinetic stability measures its resistance to irreversible inactivation (23). Although enzyme unfolding and deactivation can sometimes be related, they are clearly the result of different processes. For most industrial enzymes, kinetic stability (i.e. loss of biological activity), not thermodynamic stability (i.e. protein unfolding), is the most important parameter used to describe overall stability (24). Comparative studies of the relationship between enzyme conformation and activity during denaturation suggest that the active site is more fragile than the enzyme as a whole (25,26). Because the active site plays a key role in enzymatic catalysis, maintaining its correct conformation is the key to engineering its kinetic stability.
Therefore, we developed a novel method for enhancing the stability of CalB that involves mutating residues within its active site. This method increased the rigidity of the active site to protect the enzyme against irreversible inactivation under harsh conditions. Lipases catalyze a wide range of reactions, such as ester hydrolysis, esterification, and transesterification, thereby rending them important biocatalysts in the production of detergents, food, flavors, energy, and fine chemicals (27)(28)(29)(30)(31). Until now, scientists have used computational modeling (21), inserted extra disulfide bridges (22), and used error-prone PCR (32) to obtain more stable mutants. To understand how local active site rigidity impacts enzyme kinetic stability, we mutated residues that demonstrated high mobility within the active site. Residues with a high B factor and located within 10 Å of the catalytic Ser 105 residue were chosen as candidates for constructing iterative saturation mutagenesis libraries. The kinetic stability and thermodynamic stability of the mutants were subjected to conditions that typically lead to heat-and chemicalinduced denaturation. Furthermore, crystallography and MD simulation were performed to elucidate the structural basis for improved stability. Results from this study have substantiated the concept of local rigidity as an efficient strategy for improving enzyme kinetic stabilization. It should be noted that insights regarding local rigidity emphasize the significance of local rigidity in thermal stabilization.

EXPERIMENTAL PROCEDURES
Media and Reagents-The CalB gene was synthesized at GenScript Crop (Nanjing, China). Restriction enzyme and T4 ligase were purchased from New England Biolabs (Ipswich, MA). PrimeSTAR polymerase was purchased from TaKaRa (Dalian, China). The QIAquick TM PCR purification kit was purchased from Qiagen (Hilden, Germany). Escherichia coli Rosetta (DE3) competent cells and pET-22b were purchased from Novagen (Madison, WI). E. coli was routinely cultured overnight at 37°C in 2ϫ YT broth containing Bacto tryptone (1.6%, w/v), Bacto yeast extract (1%, w/v), and sodium chloride (0.5%, w/v) or on 2ϫ YT agar plates with (in both cases) 100 g/ml ampicillin. All substrates were purchased from Sigma.
In Silico Design Procedure-We used the crystal structure of C. antarctica lipase B (Protein Data Bank code 1TCA) to design thermostable CalB variants. For analyzing the flexibility of the protein, the B factor profile of Protein Data Bank code 1TCA was used. The amino acids located within 10 Å of the catalytic Ser 105 residue, which showed a high B factor value, were chosen for saturation mutagenesis. Saturation mutagenesis libraries were created at amino acid positions Phe 71 , Asp 223 , Leu 277 , Leu 278 , Ala 281 , and Ile 285 .
In addition, regardless of the locations, the residues with the highest B factor values in CalB were also chosen as points for iterative saturation mutagenesis. Generally, the residues located on the protein surface have a higher B factor than residues in the protein core because few interactions are formed among surface residues. After deleting N-terminal and C-terminal amino acids, all sites were chosen on the surface of the protein (Arg 249 , Arg 309 , Arg 242 , Glu 269 , Leu 219 , Lys 13 , and Pro 218 ).
Recombinant Protein Expression and Purification-CalB gene (wild type) was amplified using primers containing the restriction sites for NcoI and XhoI (supplemental Table S1). Polymerase chain reaction (PCR) amplification was carried out with PrimeSTAR polymerase and a temperature program consisting of 98°C for 2 min; 30 cycles of 10 s at 98°C, 15 s at 55°C, and 1 min at 72°C; and a final 10-min extension at 72°C. The PCR product was digested with NcoI and XhoI and subsequently cloned into the vector pET-22b (pET22b-WT). pET22b-WT was transformed into the Rosetta (DE3) by electroporation. The expression and purification of recombinant protein followed the described method (33).
Library Creation and Thermostability Screening-The mutants were prepared by whole-plasmid PCR with the primer containing NNK (sense strand)/MNN (antisense strand) degeneracy at those target sites (supplemental Table S1). Here N represents any of the following A, T, G, or C; K represents G or T; M represents A or C. PCR was performed with PrimeSTAR polymerase and a temperature program consisting of 98°C for 2 min; 30 cycles of 10 s at 98°C, 15 s at 55°C, and 7 min at 72°C; and a final 10-min extension at 72°C. The PCR products were digested with DpnI to remove the parent plasmid and purified with a PCR purification kit. The created libraries were electroporated into Rosetta (DE3) cells and selected for ampicillin resistance. Cells were plated on 2ϫ YT agar plates containing 100 g/ml ampicillin and 0.5% (v/v) tributyrin emulsified in 0.4% (w/v) gum arabic followed by incubation at 37°C for 8 h and then incubation at 15°C for 24 h. The appearance of clear haloes around the colonies upon incubation indicates enzyme activity. All the clones that showed clear haloes were picked in 96-well plates containing 200 l of 2ϫ YT medium with 100 g/ml ampicillin. After growth at 37°C for 8 h, the cultures were used to inoculate fresh medium in an identical plate and incubated further for 3 h at 37°C. After 1 mM isopropyl 1-thio-␤-D-galactopyranoside was added, the cells were incubated for 24 h at 15°C. Each saturation mutagenesis library contained 200 colonies.
Cells were harvested by spinning the plates at 4000 rpm for 30 min at 4°C and lysed by a triple freeze-thaw from Ϫ80°C, and then 200 l of 50 mM sodium phosphate buffer (PBS, pH 7.5) was added in each well. The supernatant was divided into two identical 96-well PCR plates. One plate was incubated at high temperature for 15 min and cooled at 4°C for 20 min followed by equilibration at room temperature for 15 min. The other plate was incubated under identical conditions except for incubation at high temperature. Screening of the thermostability of the mutants used pNP caprylate as the substrate, and hydrolysis was monitored at 405 nm for 2 min. The ratio of activity of each clone incubated at higher temperature versus control without incubation was taken as residual activity, which was used to identify the positive clones. A limited number of iterative saturation mutagenesis (18) experiments were performed in the next round by choosing the genes of some of the "best hits" as templates for randomizing at another amino acid position.
Enzyme Activity Assays-pNP caprylate was used to compare the activity of CalB variants. The ability of the enzyme to hydrolyze pNP caprylate was determined by measuring the absorbance of p-nitrophenol liberated using a UV-2550 spectrophotometer with a thermal control unit (Shimadzu, Kyoto, Japan). The reaction mixture consisted of 0.02 ml of 10 mM pNP caprylate as a substrate in acetonitrile and 0.97 ml of 50 mM PBS, pH 7.5 containing an appropriate amount (10 l) of the enzyme solution. The enzyme reaction was performed for 1 min at 37°C. The activities were determined photometrically at 37°C, and the buffer was adjusted at 37°C unless otherwise stated. One lipase unit in this assay is defined as the amount of enzyme that liberates 1 mol of p-nitrophenol/min under these conditions. The effect of temperature on the activity of lipase (1 g/ml) was examined across the range 30 -70°C. The 50 mM PBS was adjusted to pH 7.5 at the respective temperature.
For the kinetic studies, the concentration of pNP caprylate increased from 50 to 1000 M. The enzymatic activity of CalB variants was determined at different temperatures (293, 298, 303, 308, 313, 318, 323, and 328 K). Kinetic parameters V max and K m were acquired by fitting enzymatic activities as a function of substrate concentrations to the Michaelis-Menten equation using non-linear regression of the software Origin 8.0. The parameter k cat was obtained by using the following equation: is the molar concentration of the enzymes. Values of the free energy (⌬G ‡ ), enthalpy (⌬H ‡ ), and entropy (⌬S ‡ ) of activation at 308 K were calculated using the following equation: ⌬G ‡ ϭ RT(ln k B T/h Ϫ ln k cat ), ⌬H ‡ ϭ E a Ϫ RT, and T⌬S ‡ ϭ ⌬G ‡ Ϫ ⌬H ‡ where k B is the Boltzmann constant, h is the Planck constant, and R is the universal gas constant.
Thermal Inactivation and Unfolding-The CalB variants (0.1 mg/ml) were incubated at different temperatures for different time intervals from 0 to 150 min and then cooled on ice for 10 min. Their residual enzyme activities were assayed at 37°C as described above. The data were fitted to first-order plots and analyzed with the first-order rate constants (k d ) determined by linear regression of ln(residual activity) versus the incubation time (t). The time required for the residual activity to be reduced to half (t1 ⁄ 2 ) of the CalB variants was calculated using the following equation: t1 ⁄ 2 ϭ ln 2/k d . The changes in transition state free energy (⌬⌬G) for inactivation between mutants and wild type were calculated using the following equation: ⌬⌬G ϭ ϪRT ln(k d mutant /k d wild type ) (34) where T and R are temperature and the gas constant (1.987 cal⅐K Ϫ1 ⅐mol Ϫ1 ), respectively. ⌬⌬G of wild type was used as a reference value. k d wild type and k d mutant were inactivation rate constants of wild-type and mutant CalB, respectively.
Heat treatment of purified protein was carried out by incubating the protein in 0.2-ml PCR tubes in a programmable thermal cycler for precise temperature control. Proteins (25 l of 0.1 mg/ml in 50 mM PBS, pH 7.5) were heated at different temperatures for 15 min and cooled at 4°C for 20 min followed by equilibration at room temperature for 15 min. Samples were centrifuged to remove any aggregated protein before assaying for enzymatic activity. The activity at 37°C was considered to be 100%; the residual activities were quantitatively measured after heating at different temperatures for 15 min. The T 50 15 value is the temperature at which enzyme activity is reduced to 50% after a 15-min heat treatment. The precise value was obtained by determination of the inflection point of a fit of the residual activities at certain temperatures to a sigmoidal plot (sigmoidal Boltzmann fit using Origin 8.0) (35).
Differential Scanning Calorimetry-All differential scanning calorimetry measurements were performed on a VP-Capillary differential scanning calorimeter (MicroCal, LLC) from GE Healthcare. Samples were degassed by stirring gently under vacuum prior to measurements. Protein samples were diluted to 2.5 mg/ml by 12.5 mM PBS, pH 7.5, and the corresponding buffer was used as a reference. Protein unfolding events were recorded between 25 and 85°C with a scan rate of 2.0 K/min with an initial 21 min of equilibration at 25°C. The scans were analyzed after subtraction of an instrument baseline recorded with buffer in both cells using the software package Origin provided by the manufacturer.
Urea-induced Unfolding-The enzyme inactivation was induced by incubating ϳ0.03 mg/ml proteins with a set of concentrations of urea for 24 h at room temperature. The residual activities of the enzymes were measured at 37°C using pNP caprylate as the substrate in 50 mM PBS, pH 7.5 containing the corresponding concentration of urea. The C 50 value, indicating the concentration of urea in which 50% of enzymatic activity was retained, was calculated as T 50 15 .
For fluorescence measurements, samples treated similarly were measured on an F-7000 fluorescence spectrophotometer (Hitachi, Japan) at 20°C. Samples were excited at 282 nm to monitor the tryptophan emission with a slit width of 5 nm for both excitation and emission. Spectra were recorded from 300 to 400 nm at a scanning speed of 1200 nm/min. The scanning experiments were performed three times, and the processed data were given as the means of three measurements. The shift in wavelength maximum was used to monitor the conformational change at various denaturant concentrations. All the sample spectra were corrected for background fluorescence. The data for wild type and mutants were analyzed by a twostate model using Origin 8.0 according to the equation: where S obs is the observed spectroscopic signal, S N and S U are the signals for the native and unfolded species, respectively, ⌬G is the free energy of unfolding in the absence of denaturant, m is the partial derivative of ⌬G with respect to denaturant, and d is the concentration of denaturant. S obs and d are the dependent and independent variables; S N , S U , ⌬G, and m are treated as fitting parameters (36). The C m , the urea concentration at the midpoint of urea-induced unfolding curve, was obtained using the following equation: C m ϭ ⌬G/m.
Crystallization of CalB Wild Type and Mutants-Crystals of CalB and its mutants were grown at 14°C by the hanging drop, vapor diffusion method. All crystals were grown in 25% PEG 3350, 0.2 M sodium acetate, 0.1 M Bis-Tris, pH 6.5 and cryoprotected by the addition of 20% ethylene glycol (v/v) to the crystallization conditions. 1.5-1.6-Å-resolution data sets were collected at the beamline BL17U1 at Shanghai Synchrotron Radiation Facility (China). All data sets were integrated, scaled, and reduced with the HKL2000 software package. The structures were solved by molecular replacement with the CCP4i program PHASER using the structure of C. antarctica lipase B (Protein Data Bank code 1TCA) as the model. Model building was performed by COOT, and all of the model qualities were checked with PROCHECK. All molecular structures were visualized by PyMOL software. Hydrophobic interactions, disulfide bridges, hydrogen bonds, ionic interactions, aromatic-aromatic interactions, aromatic-sulfur interactions, and cation-pi interactions in the whole protein were calculated using the Protein Interactions Calculator (37), and all parameters were given default values.
MD Simulation-The crystal structures of wild type (Protein Data Bank code 4K6G) and mutants (L278M, Protein Data Bank code 4K6H; D223G/L278M, Protein Data Bank code 4K5Q) were further optimized by energy minimization using the conjugate gradient algorithm supplement in Discovery Studio 3.0 (Accelrys, San Diego, CA). CHARMM force field and Momany-Rone partial charge were used to assign atom types. Maximum steps and root mean square gradient values were set to 10,000 and 0.001, respectively. An implicit solvent model was used, and the dielectric constants of protein and solvent were set to 1 and 80, respectively. Other parameters were set to default values.
The MD simulation was performed using GROMACS v4.5.5 (38) implementing the GROMACS 96.1 (53a6) force field. The initial structure was solvated with a simple point-charge model of water in a box. A sufficient number of Na ϩ ions were added to neutralize the negative charges in the system. The system was then subjected to a steepest descent energy minimization and the 30-ps MD simulation was performed at 300 K with the heavy atoms and C␣ atoms fixed. Finally, a 30-ns MD simulation was performed on the whole system at 300 and 330 K. All bond lengths were constrained using the Linear Constraint Solver (LINCS) algorithm. The cutoff value for van der Waals interactions was set at 1 nm, and electrostatic interactions were calculated using a particle mesh Ewald algorithm. The time step of the simulations was set at 2 fs, and the coordinates were saved for analysis every 2 ps. Postprocessing and analysis were performed using standard GROMACS tools and Visual Molecular Dynamics (VMD).

Construction of Iterative Saturation Mutagenesis Libraries-
CalB is a globular ␣/␤-hydrolase that is ϳ30 ϫ 40 ϫ 50 Å in size with a catalytic triad composed of Ser 105 , Asp 187 , and His 224 . Its substrate binding pocket is an elliptical, steep funnel 9.5 ϫ 4.5 Å in size. The catalytic Ser 105 residue is located within the tight turn between ␤4 and helix ␣4 at the bottom of the binding pocket. When bound, the substrate is oriented parallel to the long axis of the active site (39). To identify residues that play a role in enhancing the local stability of the CalB active site, we mutated amino acids based on two criteria: 1) location within a 10-Å radius around the Ser 105 residue and 2) a relatively high B factor. As shown in Fig. 1, six residues with an average B factor over 10 were selected using B-FITTER (40) based on the published x-ray crystallography structure of CalB (Protein Data Bank code 1TCA). These resides were Leu 278 , Ile 285 , Leu 277 , Ala 281 (all located within the ␣10 helix), Phe 71 (located in the loop connecting the ␤3 strand and ␣3 helix), and Asp 223 (located in the loop connecting the ␣8 and ␣9 helices) (Fig. 1). Saturation mutagenesis libraries of these residues were generated and screened for enhanced thermostability. To ensure reasonable coverage of the relevant protein sequence, 200 transformants were screened for each library (41). We found that amino acids 278, 281, and 285 were crucial for enzymatic activity because mutation led to a dramatic alteration in activity (data not shown). Heat treatment (55°C for 15 min) identified several positive mutants in the Leu 278 and Asp 223 libraries. In particular, sequencing revealed that the most thermostable mutants were L278M and D223G.
To further improve enzyme stability, the gene encoding L278M was used as a template for iterative saturation mutagenesis at the other five sites (Ile 285 , Leu 277 , Ala 281 , Phe 71 , and Asp 223 ). Mutagenesis of Phe 71 and Asp 223 did not significantly impact enzymatic activity; however, mutagenesis of Leu 277 , Ala 281 , and Ile 285 resulted in large fluctuations in enzyme activity. Heating these five libraries at 59°C for 15 min revealed that the best variant was L278M/D223G from the L278M/D223X library. Additional iterative saturation mutagenesis of the remaining four residues using L278M/D223G as a template did not further improve thermostability. More importantly, the enzyme lost most of its activity following multiple mutations within the active site. MARCH 14, 2014 • VOLUME 289 • NUMBER 11

JOURNAL OF BIOLOGICAL CHEMISTRY 7997
For comparison, residues located on the surface of CalB were also investigated using a similar method. Seven residues with high B factors were chosen for saturation mutagenesis, namely Arg 249 , Arg 309 , Arg 242 , Glu 269 , Leu 219 , Lys 13 , and Pro 218 (Fig. 1). The average B factors of these residues were much higher than those of residues within the active site. With the exception of those from the Pro 218 library, most of the mutants retained their catalytic activity (data not shown). No obvious thermostable variants were found in all the libraries examined under the same screening conditions.
Kinetic, Dynamic, and Chemical Stability of the CalB Variants-Irreversible thermal inactivation is an important parameter for describing the kinetic stability of enzymes. To assess this property, each variant was incubated at 48°C and then cooled on ice for 10 min before the residual activity was measured at 37°C with respect to the time of incubation. All traces of thermal inactivation were found to follow first-order kinetics. Calculation of the t1 ⁄ 2 and transition state free energy (⌬⌬G) of the CalB variants was based on the inactivation constant k d . As shown in Table 1, the t1 ⁄ 2 of D223G/L278M at 48°C was 49 min, which was about 13-fold higher than that of wild type. Thermodynamic analysis showed that the ⌬⌬G of D223G/L278M, L278M, and D223G increased by 6.3, 4.6, and 3.1 kJ⅐mol Ϫ1 , respectively, compared with wild type. The higher ⌬⌬G exhibited by D223G/L278M suggests that it possesses higher energy barriers against thermal inactivation, thus making it more stable than other single variants.
The temperature at which 50% of enzyme activity remains after incubation for 15 min (T 50 15 ) is another parameter used to evaluate kinetic stability. The activity of the CalB variants was assessed by incubating each enzyme from 37 to 65°C. No significant differences in residual activity were observed with treatment below 40°C ( Fig. 2A). However, incubation at temperatures above 45°C affected this activity. For example, at 50°C, the residual activity of wild-type CalB was only 14% after treatment; however, D223G and L278M retained ϳ45% activ-   (Table 1). Unexpectedly, T 50 15 of the best variant, D223G/L278M, increased to 58.5°C, which is ϳ12°C higher than that of wild type. This result demonstrated that a strong synergistic effect exists between D223G and L278M to enhance CalB kinetic stability, thereby leading D223G/L278M to greater kinetic stability than wild type.
To investigate the thermodynamic effect of the mutants, the melting temperature (T m ) of each CalB variant was measured by differential scanning calorimetry. Because a single peak appeared in the scans, we inferred that the wild type and three CalB variants underwent a single transition (Fig. 2B). As shown in Table 2, the T m values for D223G, L278M, and D223G/ L278M were 2.4, 2.3, and 3.6°C higher than that of wild type (56°C), respectively. Unlike the change in kinetic stability (T 50 15 ), no synergistic effect between D223G and L278M was observed on the thermodynamic stability (T m ) of CalB.
To further examine changes in thermodynamic stability, the enthalpy (⌬H u ) and entropy (⌬S u ) of the unfolding process were calculated by non-linear fitting ( Table 2). Although ⌬H u and ⌬S u of the single mutants decreased, both values increased in the double mutants. This result indicates that improving the stability of the double mutant was not due to a simple additive effect of the two single mutations.
In D223G/L278M, when the temperature was equal to T m , ⌬H u and T⌬S u canceled each other out. However, when the temperature was lower than T m , ⌬H u became larger than T⌬S u , resulting in a more positive ⌬G u value that reflects enhanced stability. Therefore, mutants exhibiting a higher ⌬H u value may be more stable when T Ͻ T m . This conclusion is consistent with the fact that a high ⌬H u value for D223G/L278M indicates that it has greater stability (e.g. longer t1 ⁄ 2 and higher T 50 15 ) than other variants when T Ͻ T m . Moreover, enhancement of ⌬H u contributed to increased thermodynamic stability of the double mutant.
To assess the chemical stability of the variants, urea-induced protein unfolding was performed. The maximum emission wavelength shift in intrinsic fluorescence was used as a measure of the global conformational change (Fig. 2C). The urea concentration at the unfolding curve midpoint (C m ) of wild-type CalB was 2.89 M. This value shifted 0.25, 0.43, and 0.55 M higher in concentration for D223G, L278M, and D223G/L278M,

FIGURE 2. Thermally and urea-induced inactivation/unfolding profiles of wild-type and mutant CalB. A, thermal inactivation profiles of CalB mutants.
Enzymes in 50 mM sodium phosphate buffer, pH 7.5 were incubated at various temperatures for 15 min and assayed for residual activity at 37°C. The activity measured at 37°C was considered to be 100%. B, results from differential scanning calorimetric analysis of the CalB mutants. The scans were analyzed after subtraction of an instrument-derived baseline recorded with buffer in both cells as calculated using the Origin software package. C, the fluorescence spectra were recorded at wavelengths between 300 and 400 nm with an excitation wavelength at 282 nm under a scanning speed of 1200 nm/min. D, urea-induced inactivation profiles of wild-type and mutant CalB. Enzymes were incubated with different concentrations of urea for 24 h and then assayed for residual activity at 37°C. Activity measured in the absence of urea was considered to be 100%. f, wild type; •, D223G; OE, L278M; , D223G/L278M. The fitted curves for protein unfolding are shown as black lines. Error bars represent S.D.

TABLE 2 Thermodynamic properties of wild-type and mutant CalB
Values represent the mean of three independent sets of experiments with S.D. Ͻ5%.  MARCH 14, 2014 • VOLUME 289 • NUMBER 11 respectively ( Table 2). The mutants also exhibited better chemical resistance than wild type in the urea-induced inactivation assay ( Fig. 2D and Table 1). After treatment with 2.4 M urea for 24 h, wild-type CalB retained only 33% of its original activity, whereas all the mutants retained ϳ62% activity. The C 50 value, which indicates the urea concentration that inhibits 50% of the enzymatic activity, increased 0.7, 0.8, and 0.8 M for D223G, L278M, and D223G/L278M, respectively. These results indicate that mutations within the active site could improve the thermal and chemical stability of CalB. Catalytic Properties of CalB Variants-The specific activity of wild-type CalB and D223G, L278M, and D223G/L278M mutants was determined at temperatures ranging from 25 to 70°C using pNP caprylate as the substrate. The optimal temperature for catalytic activity by the mutants increased to 57.5, 57.5, and 62.5°C for D223G, L278M, and D223G/L278M, respectively, which is 2.5-7.5°C higher than that of wild-type CalB (Fig. 3). Results from further analysis of the enzymatic activity of CalB variants at different temperatures are represented in the Arrhenius plot (ln k cat versus 1/T) shown in Fig. 3 (inset). The activation energy (E a ) of the reactions was calculated from the slope of the Arrhenius plot (Table 3). D223G displayed an E a similar to that of wild type, whereas the E a value of L278M was greater. In contrast, the E a of D223G/L278M (43.24 kJ⅐mol Ϫ1 ) was lower than that of wild type (45.23 kJ⅐mol Ϫ1 ), suggesting that the double mutation results in a weaker temperature dependence for enzymatic activity and a broader temperature range for catalytic activity (Fig. 3).

Local Active Site Rigidity Impacts Enzyme Kinetic Stability
Using pNP caprylate as the substrate, we determined the kinetic parameters of wild-type CalB and its mutants. All mutants except D223G, which demonstrated a minor decrease in k cat , showed an increase in both k cat and K m compared with wild type (Table 3). L278M displayed the highest catalytic efficiency (k cat /K m ) because its k cat value was nearly 2.0-fold higher than that of wild type. The other two mutants, however, showed catalytic efficiency similar to that of wild type. Taken together, these results showed that the mutants possessed enhanced kinetic stability without sacrificing catalytic activity.
To provide additional insight into how the mutations affected CalB enzymatic activity, we calculated the free energy (⌬G ‡ ), enthalpy (⌬H ‡ ), and entropy (⌬S ‡ ) of activation (Table 3). ⌬G ‡ is the difference in free energy between the transition state intermediate and the reactant. The ⌬G ‡ value of each mutant was lower than that of wild type at 35°C, suggesting that the rate of formation of mutant intermediates was faster than that of wild type.
Crystal Structure Analysis and MD Simulation-To gain insight into how the conformation produced by the various mutations increased CalB stability, we performed x-ray crystallography on each mutant. The data for wild-type and mutant CalB (except D223G segment 141-146, which was missing) are summarized in Table 4. The resolution was 1.5, 1.6, 1.6, and 1.5 Å for wild-type CalB and D223G, L278M, and D223G/L278M mutants, respectively. Superimposing the mutant structure onto the wild-type structure revealed that the root mean square deviation (RMSD) of the C␣ position of D223G, L278M, and D223G/L278M was similar (0.266, 0.063, and 0.121, respectively) (Fig. 4A). The D233G mutation was located in front of the catalytic His 224 residue, whereas the L278M mutation resided in the middle of the long C-terminal ␣10 helix, which lines the channel leading into the active site (42). Compared with wild type, no obvious increase in interaction was observed in the mutant structures (Table 5). Even the most stable mutant, D223G/L278M, exhibited fewer hydrogen bonds and ionic interactions despite the additional aromatic-sulfur interaction. Therefore, our data indicate that enhanced stability in the mutants was not the result of a greater number of interactions. Instead, we found that the critical difference between the mutants and wild-type CalB involved the types of interactions introduced into the active site. The L278M mutation resulted in a slight shift of the main chain of surrounding residues that induced a new aromatic-sulfur interaction between Met 278 and Trp 104 while simultaneously eliminating the aromatic-sulfur interaction between Met 298 and Tyr 300 and introducing two new hydrogen bonds (Fig. 4, B and C, and Table 6). In D223G/ L278M, the mutations caused tighter compaction of the ␣10 helix. Compared with wild-type CalB (Fig. 4B)  Enzymes were incubated at various temperatures for 1 min prior to activity measurement at the same temperature after the addition of substrate. The temperature-k cat profile according to the Arrhenius plot is also shown (inset). f, wild type; •, D223G; OE, L278M; , D223G/L278M. Error bars represent S.D.

TABLE 3 Kinetic constants of wild-type and mutant CalB
Values represent the mean of three independent sets of experiments with S.D. Ͻ5%. chain nitrogen atom of Ala 276 , Leu 277 , and Ala 279 shifted ϳ0.2-0.5 Å. Together with the change in the dihedral angle, these subtle structural adjustments induced a new hydrogen bond network involving six pairs of hydrogen bonds on segment 268 -281 that strengthens its rigidity ( Fig. 4D and Table 6). Meanwhile, an extra aromatic-sulfur interaction between Met 278 and Trp 104 allowed simultaneous maintenance of the other six aromatic-sulfur interactions ( Fig. 4E and Table 5). Calculation of the B factor from x-ray crystallographic data is frequently used to determine the distribution of flexible regions within a protein. Because crystallization conditions and resolution affect the B factor, we used a relative B factor value to indicate the relative rigidity of peptide segments within the entire protein. The residue with the highest B factor was set as 100%, and the relative B factors for the other residues were calculated accordingly. In D223G/L278M, we found significantly lower relative B factors in regions 211-226 and 257-281 (Fig. 5), indicating enhanced rigidity in these two segments. The enhanced rigidity at 257-281 correlated well with the formation of a new hydrogen bond network with the 268 -281 segment. On the contrary, only a subtle change in relative B factor was observed in L278M (data not shown).

Mutant
Next, we conducted MD simulation at 300 and 330 K to further assess the improved stability of the CalB mutants. The RMSD values of the backbone atoms of the CalB variants are shown in Fig. 6 for which the reference structure was obtained from the equilibration step performed immediately before the MD simulation run. The RMSD values of wild-type CalB demonstrated much greater fluctuation between 300 and 330 K than the mutants, which is consistent with the fact that the mutants possessed greater thermostability.
RMSF values often reflect the fluctuation of individual residues during the MD simulation process. As shown in Fig. 7, wild-type CalB possesses two major unstable regions, namely a segment of the ␣5 helix (142-148) and the initial portion of the ␣10 helix (267-286). The ␣5 helix resides on the lid domain, whose intrinsically high flexibility is beneficial for binding substrate and promoting catalysis. This helix exhibited high flexibility in wild-type CalB and the D223G/L278M mutant. Nevertheless, the behavior of the ␣10 helix was different between the two enzymes. In wild-type CalB, fluctuation of this region was higher at 330 K than at 300 K, suggesting that it is a highly flexible region during heat treatment. However, the D223G/ L278M mutant showed considerably greater structural rigidity compared with wild type at both 300 and 330 K (Fig. 7), which is consistent with their relative B factor profiles (Fig. 5). These data reflect a novel approach for enhancing the stability of CalB that involves increasing the local rigidity of its active site.

DISCUSSION
Low stability often limits the usefulness of an enzyme industrially. In this study, we conducted iterative saturation mutagenesis on the structurally flexible residues of the active is the mean intensity of the observations I(hkl) i of reflection hkl. c R work ϭ ⌺ʈF obs ͉ Ϫ ͉F calc ʈ/⌺͉F obs ͉ where F obs and F calc are the observed and calculated structure factor amplitudes, respectively. R free was calculated as R work using a randomly selected subset consisting of ϳ10% of unique reflections not used for structure refinement. . Additionally, the dynamic and chemical sta-bility of the mutant improved without sacrificing activity. To our knowledge, this is the first report demonstrating that increasing the rigidity of its active site can enhance the kinetic stability of an enzyme. In our previous work, we found that local interaction at the N-terminal region or on the interdomain interface could significantly affect the stability of a hyperthermophilic acylpeptide hydrolase from the archaeon Aeropyrum pernix K1 (4,12). Koudelakova et al. (43) recently reported that the stability and resistance to organic cosolvent of haloalkane dehalogenase DhaA from Rhodococcus rhodochrous NCIMB 13064 could be adjusted by modifying the rigidity of the access tunnel. Because the kinetic stability of an enzyme is dependent on a subtle balance between the flexibility and rigidity of its active site, we reasoned that optimizing the rigidity of the active site may be an efficient method for enhancing kinetic stability of an enzyme. This modification may protect the enzyme against irreversible inactivation under harsh conditions. Moreover, it allows for    the generation of mutants that perform better at elevated temperatures.

Local Active Site Rigidity Impacts Enzyme Kinetic Stability
Indeed, data from this study showed that CalB mutants with significantly enhanced stability could be created using this new strategy. The best mutant, D223G/L278M, showed a greater improvement in stability than most previously published work (21,22). It should be noted that Zhang et al. (32) screened a random mutagenesis library to identify a CalB mutant (A281E) that exhibited as much as 20-fold greater thermostability at 70°C. However, we did not find any stabilization effect at position Ala 281 by site-directed mutagenesis or screening of the site saturation mutagenesis library (Table 1). Our results suggest that focusing the mutagenesis effort on the active site creates "smarter" libraries that are more likely to yield positive results. Moreover, this approach is more feasible than other methods that require an in-depth knowledge of the structural features of the enzyme.
Data from high resolution crystal structures of wild-type and mutant CalB also provided important insights into the source of stability in the mutants. No obvious changes in interaction were observed in the L278M and D223G single mutants. However, when both mutations were present, their synergistic actions introduced an extra hydrogen bond network on segment 269 -281 that increased the rigidity of the segments (Fig. 4D and Table 6).
Long MD simulations of CalB in aqueous solvent confirmed high mobility in the regions lining the channel that lead into the active site, especially the ␣5 and ␣10 helices (42). We postulated that the segment between 268 and 287, which exhibited great flexibility, is susceptible to disruption at elevated denaturing conditions. This movement exposes the active site to the external environment, thereby inactivating the enzyme. Enhanced rigidity in this segment could help the enzyme maintain its correct conformation at high temperature, prevent exposure of the active site, and improve its kinetic stability.  This hypothesis was corroborated by relative B factor profiles and RMSF analysis (Figs. 5 and 7). The segment exhibited greater flexibility than other parts of the protein but displayed a notable decrease in flexibility (due to increased rigidity) in the presence of both mutations. This result demonstrates that even minimal structural modification can create sufficient conformational change to enhance thermostability. Our data also demonstrate the formation of an extra hydrogen bond network between atoms on the main chain of the protein rather than the side chains. For many years, enhanced enzymatic stability was achieved by introducing new side chain interactions because the main chain residues exhibited little change following mutagenesis (14,19,44). However, our results indicate that the accumulation of subtle hydrogen bond interactions between the main chain residues also contributes to enzyme stabilization. It should also be noted that the flexible secondary structure could be made even more rigid by further tightening the main chain interactions.
Enhancing stability while retaining enzymatic activity is a demanding task for protein engineering, particularly when improvements to stability are achieved through mutagenesis of residues within the active site. Because of high flexibility and the conservation of structure and sequences, any mutation within the active site can inactivate the enzyme. In this study, we successfully utilized a two-step screening method to increase the stability of CalB without sacrificing its activity. It is notable that the catalytic efficiency (k cat /K m ) of all the mutants was not compromised in lieu of enhanced stability at all temperatures. Indeed, the relative B factor and RMSF profiles indicated that, with the exception of the flexible ␣10 helix (257-281), the backbone dynamic of the mutant did not change relative to the wild type (Figs. 5 and 7). This simultaneous improvement in stability and activity, which is reportedly uncommon, provides insight into how such mutations should be designed. While screening the iterative saturation mutagenesis libraries, we discovered that A281F and I285F exhibited even more specific activity than the wild type. The specific activity was 100.3 and 100 units/mg for A281F and I285F, respectively, compared with that of wild-type CalB, which was 20 units/mg. Both sites are located within the substrate-binding tunnel and are involved in enzyme-substrate binding. The introduction of more hydrophobic, larger side chains at these positions may facilitate substrate binding and/or catalysis, thereby leading to improved enzymatic activity. This result highlights the fact that mutagenesis of the structurally flexible residues within the active site can enhance kinetic stability and increase enzymatic activity.

CONCLUSIONS
In this study, we demonstrate the enhancement of C. antarctica lipase B kinetic stability by increasing the rigidity of its active site. Although some reports revealed that mutagenesis of high B factor sites/regions may enhance enzyme stability, our specific focus on the flexible residues within the active site allowed us to understand how local rigidity affects enzyme kinetic stability. The increased intermolecular interactions, particularly the main chain hydrogen bond network, contribute to higher kinetic stability and thus the maintenance of enzy-matic activity at elevated temperatures. MD simulation in combination with analysis of the relative B factors suggests that a balance between flexibility (␣5 helix) and rigidity (␣10 helix) in the active site is essential for catalysis. Additional structural and biochemical assays will be performed on other enzymes to reveal the generality of this local stabilization strategy. Nevertheless, our results will help elucidate the properties that confer thermostability in enzymes and thus provide insight into how to design more efficient and thermostable biocatalysts.