Effect of Heavy Atoms on the Thermal Stability of α-Amylase from Aspergillus oryzae

Currently, there are no versatile and established methods for improving stability of proteins. In an entirely different approach from conventional techniques such as mutagenesis, we attempted to enhance enzyme stability of α-amylase from Aspergillus oryzae using a heavy-atom derivatization technique. We evaluated changes in stability using differential scanning calorimetry (DSC). Candidate heavy atoms were identified using the Heavy-Atom Database System HATODAS, a Web-based tool designed to assist in heavy-atom derivatization of proteins for X-ray crystallography. The denaturation temperature of α-amylase derivatized with gadolinium (Gd) or samarium (Sm) ions increased by 6.2 or 5.7°C, respectively, compared to that of the native protein (60.6°C). The binding of six Gd ions was confirmed by X-ray crystallography of the enzyme at 1.5 Å resolution. DSC and dynamic light-scattering data revealed a correlation between stability and the aggregation state upon addition of Gd ions. These results show that HATODAS search is an effective tool for selecting heavy atoms for stabilization of this protein.


Introduction
Enzymes are currently used in many different industrial products and processes [1]. To enhance thermal stability of highly useful enzymes for industrial applications, protein-engineering techniques such as site-directed mutagenesis, random mutagenesis, recombination, and directed-evolution techniques have been successfully employed for various proteins [2,3]. To date, most stability improvements have been based on the introduction of point mutations into proteins. In general, however, there are no established methods that can be used to predict the effect of mutations on thermal stability, because even single amino-acid substitutions affect various stabilization factors depending on the surrounding environment, substituted residues, and structural changes due to the mutations [4][5][6]. Therefore, a versatile and simple method for thermal stabilization of enzymes is required.
As well as the site directed mutagenesis, techniques for protein stability by using additives such as covalent cross-linkers and ligands have been well studied since early times. The covalent cross-linking technique by bifunctional chemical modification reagents [7,8] as an additive has been successfully used to enhance the stability of proteins [9]. The cross-linking in proteins reduces the configurational backbone chain entropy of the unfolded polypeptide, resulting in increase in conformational stability [10][11][12]. On the other hand, the binding of ligand increases the stability of proteins in response to Schellman's binding theory [13,14]. The stabilization mechanism can be explained stoichiometrically by the shift of the folding-unfolding equilibrium toward the folded state caused by the higher affinity of ligand to the folded state [15]. Thus these techniques using the additives have the advantages of being simple and quick, and are useful for enhancing protein stability efficiently.
Metal ions that bind with high affinity to specific sites often stabilize the conformation of proteins [16][17][18][19][20][21]. In protein X-ray crystallography, experimental phasing from heavy atom-derivatized crystals is a major technique in structure determination [22]. The identification of heavy atoms that bind proteins can be useful both for improving crystallization success rates and in some cases for enhancing the thermal stability of these proteins [23]. However, determination of candidate heavy atoms is laborious and time-consuming. Recently, to facilitate the heavy-atom derivatization process, we developed the Heavy-Atom Database System (HATODAS, http://hatodas.harima.riken.jp), which suggests candidate heavy-atom reagents for a target protein based on a search against a database of known heavy atom-liganded proteins, using the amino-acid sequence and crystallization conditions as the query terms [24,25]. We believe that HATODAS search could also contribute to the efficient selection of candidate heavy atoms for use in protein stabilization using the derivatization technique.
Here we present such a method, which utilizes derivatization with heavy atoms identified through HATODAS search. This study was designed to investigate whether there is a relationship between heavy-atom-derivatized protein and its conformational stability; to date, there has been no detailed study of the effect of heavy atoms on protein stability. To elucidate thermal behavior of a derivatized protein, we used differential scanning calorimetry (DSC) to investigate the stability of Aspergillus oryzae a-amylase (Ao aamylase) containing heavy atoms. This enzyme catalyzes the hydrolysis of the a-1,4 glycosidic linkages in raw and soluble starches, and is used in numerous industrial applications for starch conversion in value-added products. The stability of Ao a-amylase containing heavy atoms was evaluated over a wide range of concentrations by heat denaturation, and the structures of the derivatized proteins were determined by X-ray crystallography. We also discuss the thermostabilization mechanism of heavy atom-derivatized Ao a-amylase.
Because heavy atoms sometimes act as inhibitors, we first investigated whether derivatized proteins retained their enzymatic activity. We examined the activity of Ao a-amylase with each heavy atom at room temperature; digestion of starch was assayed by monitoring the disappearance of the starch-iodine color. In the presence of the six heavy atoms, activity was preserved, indicating that none of these heavy atoms is a potent inhibitor of the enzyme. In general, heavy atoms bind to characteristic residues: Hg, Au, and Ag tend to bind to Cys or His; Pt binds to Met or His; and Pb and lanthanides bind to Asp or Glu [25]. To avoid inactivation of the enzyme, heavy atoms should be selected based on the types of amino acids constituting the protein active site.
To further characterize these heavy atoms, we examined the stability of Ao a-amylase derivatized with each candidate. Solutions of each derivatized Ao a-amylase were incubated in a water bath at ,70uC for 5 min, and the remaining activity was evaluated at room temperature. Only Sm, the fourth of six candidate atoms in terms of stabilization potential, had a significant effect on thermal stability of the enzyme; the other samples did not exhibit enzymatic activity after heat treatment. Although we confirmed that the Ao a-amylase crystals had been successfully derivatized with Hg ion, which HATODAS predicted would yield the greatest stabilization, this ion did not in fact stabilize this protein. In any case, use of Hg ions should be avoided because the accumulation of Hg from industrial uses poses serious environmental problems.
Based on these findings, we excluded heavy atoms other than Sm from further characterization, since they did not enhance the stability of this enzyme. However, because structure determination of the Sm-binding protein was not successful (see Materials and Methods), we performed additional HATODAS searches for lanthanide ions as a substitute for Sm, resulting in the identification of gadolinium (Gd) as a new candidate. Activity evaluation after heat treatment revealed that derivatization with Gd ion yielded heat-stabilization similar to that observed with Sm ion. Furthermore, we were able to successfully determine the crystal structure of Gd-derivatized protein. Therefore, we elected to further characterize the Gd-derivatized Ao a-amylase.

Differential scanning calorimetry of Ao a-amylase
Based on the initial findings regarding enzymatic activity after heat treatment, we selected two heavy atoms, Gd and Sm, for further study of thermal stabilization of Ao a-amylase. Enzymes derivatized with GdCl 3 or SmCl 3 were examined by DSC to detect differences in their thermal behavior (Fig. 1A). In the absence of heavy atoms, the denaturation temperature (T d ) for Ao a-amylase was 60.6uC. On the other hand, T d increased to 66.8uC or 66.3uC, respectively, when a 0.2 mg/mL solution of Ao a-amylase was incubated with 0.2 mM of either GdCl 3 or SmCl 3 . Thus, the T d of Ao a-amylase increased by ,6uC in the presence of these heavy atoms, indicating that the stability of the protein was remarkably improved by both ions.
We next conducted DSC experiments to evaluate the changes in stability under different concentrations (0-2.0 mM) of GdCl 3 , using a 0.4 mg/mL solution of Ao a-amylase (Fig. 1B, Table 1). The T d increased with increasing Gd concentration between 0 and 0.1 mM GdCl 3 . At 0.1 mM GdCl 3 , Ao a-amylase exhibited the highest T d value, 67.3uC; this value is 6.3uC higher than that of the native protein. The DSC curves show that T d declined with increasing concentration from 0.11 to 2.0 mM GdCl 3 , due to protein aggregation, as described below (Table 1). In the DSC measurements of the cooling and reheating processes, no excess heat capacity curve was observed in the presence or absence of Gd ions, suggesting that the heat denaturation of Ao a-amylase was irreversible under the conditions of the experiment.

Crystal structure of Gd derivatized Ao a-amylase
To elucidate the stabilization mechanism of Ao a-amylase with a T d of 67.3uC, we determined the structures of native and Gdderivatized proteins by X-ray crystallography. The crystal structure of the Gd-derivatized protein is shown in Fig. 2A. The structure consists of three domains: a central main domain, A (residues 1-121 and 177-380), formed by an N-terminal catalytic (b/a) 8 -barrel; a small b-pleated domain, B (122-176), protruding between b3 and a3; and a C-terminal globular domain, C (residues 381-478), consisting of a b-structure with a Greek-key motif [26][27][28][29][30][31][32]. This structure is similar to previously reported Xray structures of Ao a-amylase.
N-acetyl glucosamine (NAG)-and calcium-binding sites were present in the crystal structures of both the native and the derivatized protein ( Fig. 2B and 2C). The positions of these two ligands are identical to those in the reported structure of a-amylase (PDB code: 2gvy) [32]. Since no NAG was added during the crystallization of Ao a-amylase, this endogenous NAG may have been incorporated during the preparation steps. A calciumbinding site is located at the interface between the catalytic A domain and the B domain (Fig. 2C). This calcium ion is recognized by interactions with the side-chains of Asn121 and Asp175, the main-chain atoms of residues Glu162 and His210, and three water molecules. A characteristic feature of a-amylases is their requirement for calcium ions for activity and conformational stability [33].
In the derivatized protein structure, we found six bound Gd ions ( Fig. 2A). The Gd1, Gd2, Gd3, and Gd4 ions are located in between the residues of the asymmetric chain and those of a neighboring symmetry-related chain (Fig. 3A-C). The Gd1-4 ions may act as an interchain linker, resulting in oligomerization of Ao a-amylase molecules. Gd1, Gd3, and Gd4 ions exclusively mediate interactions between the proteins, as opposed to within a single protein molecule. However, the interactions involving Gd1, Gd3, and Gd4 make no contribution to stability, because dynamic lightscattering analysis of the stabilized protein in solution suggested that it exists in a monomeric state (see below and Table 1). The position of the Gd2 ion, which interacts with both Asp233 of domain A and Glu156 of domain B, suggests that Gd2 contributes to the conformational stability of Ao a-amylase (Fig. 3A). The Gd5 ion is bound to Asp47 on the molecular surface (Fig. 3D). One Gdbinding site, Gd6, was present on the catalytic pocket of domain A (Fig. 3E). The binding of Gd6 ion to Asp206 and Glu230 was weak, with a site occupancy of 0.2. Thus, the derivatized enzyme may retain activity because of the low occupancy of the Gd6 site.
To confirm the structural changes due to derivatization, we performed a C a superposition analysis using the program LSQKAB [34]. The C a r.m.s.d. (root mean square deviation) between the derivative and native Ao a-amylase structures was 0.26 Å , indicating that the overall crystal structure of the derivatized protein was highly similar to the native structure. In addition, to obtain additional information regarding structural changes in solution with Gd ions, we generated near-UV circular dichroism (CD) spectra in the ranges from 250 to 320 nm (data not shown). The CD spectrum of Ao a-amylase with Gd ions was indistinguishable from that of the protein without Gd ions, further supporting that the tertiary structure of Gd-derivatized Ao aamylase is similar to that of the native protein. From these results, we conclude that the Gd ions do not affect the overall conformation of the Ao a-amylase molecule.
In this study, it still remains unclear whether the Gd2 ion plays a significant role in stabilizing the Ao a-amylase structure. It might be difficult to specify effects of Gd ions by site-directed mutagenesis because an introduction of the point mutation in proteins affects not only the stabilizing factors of the mutation site but also those of other parts far from the substitution site [4][5][6]. The stabilization of Ao a-amylase was mediated by Gd ions. The ligand binding is important to the increase in thermal stability in solution [13,14]. In a study for thermal stability of tryptophan synthase, Ogasahara et al. have shown that the simulated interaction between the subunits with the order of 10 8 M 21 of binding constant remarkably enhances the thermal stability of the protein without conformational change [15]. The increase in T d of Gd-binding Ao a-amylase without conformational change could be caused by the increase in the binding constant. On the other hand, covalent cross-linking in proteins by disulfide or chemical cross-linkers restrains the folding and unfolding of the protein, resulting in a decrease in the conformational entropy of the unfolded polypeptide [10][11][12]. Therefore, the stabilization mechanism of the Gd-binding protein is completely different from that of the cross-linked protein. Our technique using heavy atoms is   Effect of Heavy Atoms on Stability of a-Amylase PLOS ONE | www.plosone.org one of those methods for thermal stabilization of proteins, suggesting that the application of a variety of techniques is essential to eliminate the time consuming procedure during the trial and error process.

Dynamic light-scattering (DLS) experiment
A DLS experiment revealed that Ao a-amylase exists in a monomeric state in the absence of Gd ions (Table 1). However, the monomeric state shifted toward the oligomeric state upon addition of Gd ions. For GdCl 3 concentrations between 0.02 and 0.11 mM, bimodal peaks were observed (98.6-99.8% mass in the monomeric state and 0.2-1.4% mass in the oligomeric state). For Gd concentrations between 0.12 and 2.0 mM, a bimodal analysis revealed an estimated molecular mass of 10 4 -10 6 kDa, suggesting aggregation of Ao a-amylase in solution. Thus, the inclusion of excess Gd ions actually appears to promote the aggregation of this enzyme.
At 0.1 mM GdCl 3 , i.e., the condition that confers maximum T d , 99% of the protein is in the monomeric state ( Table 1), indicating that the Gd ions in the monomeric crystal structure (Fig. 3) contribute predominantly to the stabilization of this protein. The Gd ion concentrations associated with protein aggregation correspond to the concentrations at which the T d decreases. The formation of aggregates might be due to Gd ions acting as interchain linkers (Gd1, 3, 4 in Fig. 3). Although the addition of excess Gd ions negatively affects the stability of this protein, the T d in the aggregated state under in 2.0 mM GdCl 3 is still approximately 4uC higher than that of the native protein. These results indicate that the stability of Ao a-amylase does not deteriorate significantly upon aggregation. In the crystal structure, one molecule of aamylase binds to six Gd ions. However, the maximum T d is observed at a 1:13 molar ratio of protein to Gd (Table 1), implying that Gd ions interact with Ao a-amylase in the equilibrium state with weak binding constants.
In summary, we examined the applicability of heavy-atom derivatization as a stabilization technique, using Ao a-amylase protein in the presence of lanthanide ions. The T d of Ao a-amylase increased in the presence of Gd ion, up to a maximum at 0.1 mM GdCl 3 , and decreased relative to this maximum at higher concentrations. At the maximum, the T d value of the Gdderivatized protein was 6uC higher than that of the native protein.
The stabilization mechanism of the enzyme by Gd was elucidated by analysis of crystal structure solved at 1.5 Å resolution, and by observation of its physicochemical properties (such as changes in T d ) and its oligomeric state in solution. The results suggest that the bound heavy atoms contribute substantially to the thermal stability of this enzyme. A similar stabilization by the heavy-atom derivatization technique, using ions selected by the HATODAS software, has been observed in a putative a-ribazole-59-phosphate phosphatase from Thermus thermophilus HB8. The T d of that protein in the presence of Pt ions was 95.6uC, 16.4 uC higher than that of the native protein (Sugahara et al., in preparation).
The results presented here confirm that HATODAS is a powerful tool for identifying heavy atoms for use in stabilization of two examined proteins, and suggest that the derivatization technique is useful for rapid and effective stabilization of proteins. One clear advantage of the heavy-atom derivatization technique is its simplicity: all that is required is the addition of heavy atoms to protein solutions. This method may, therefore, be applicable to a wide range of proteins.

Heavy-atom selection for thermostabilization
As a test protein for this work, we used commercially available a-amylase (endo-1,4-a-D-glucan glucohydrolase) from Aspergillus oryzae (EC 3.2.1.1, MW = 52.4610 3 , Sigma, Cat. No. 10065) without further purification. The optimal conditions for a-amylase in industrial application are pH and no Ca 2+ [35,36]. The program HATODAS (http://hatodas.harima.riken.jp/) identified six candidate heavy atoms based on the amino-acid sequence of Ao a-amylase and pH 5.8 as the query terms. Ao a-amylase activity with the selected heavy atoms was examined using the starch-iodine assay method [37]. A 0.1 mL aliquot of Ao aamylase at 0.4 mg/mL in 0.1 M MES-NaOH (pH 5.8) with 1.0 mM heavy atoms was incubated for 5 min at 70uC in a water bath; this temperature was chosen because T d for Ao a-amylase is 61uC in the absence of heavy atoms.

Crystallization and diffraction data collection
Diffraction-quality crystals of Ao a-amylase were obtained using the oil microbatch method [38] on the Autolabo automatic crystallization system [39]. A crystallization drop of 1.0 mL, in the presence of synthetic zeolite molecular sieves as heteroepitaxic nucleants [40,41], was created by mixing 1:1 mixture of 28.0 mg/ mL protein solution in 0.02 M MES-NaOH (pH 5.8) and precipitant solution composed of 40% (w/v) polyethylene glycol (PEG) 8,000, 0.2 M CaCl 2 , and 0.1 M MES-NaOH (pH 5.8) in a well of a Nunc HLA crystallization plate (Nalge Nunc International) which was then covered with 20 mL of paraffin oil. The resulting Ao a-amylase crystals were submitted to a heavy-atom derivatization experiment. The Gd derivative was prepared by soaking native crystals with 10 mM GdCl 3 , 40% (w/v) PEG 8,000, 0.2 M CaCl 2 , 0.1 M MES-NaOH (pH 5.8). Although the SmCl 3 derivatization was attempted using concentrations in the range 0.1-10 mM, this treatment caused the crystals to crack, and structure determination was therefore impossible.
All crystals were directly mounted in a cryoloop from the crystallization drop and flash-cooled at 100 K in a nitrogen gas stream. Complete diffraction data sets were collected using an inhouse Rigaku R-AXIS VII image-plate detector with Cu Ka radiation and a Rigaku R-AXIS V image-plate detector with synchrotron radiation at BL26B1 of SPring-8, Japan [42]. All data were processed using the program HKL-2000 [43].

Structure determination
Positioning of one Ao a-amylase molecule in the asymmetric unit was accomplished using the molecular-replacement method as implemented in the program MOLREP [44], based on the crystal structure deposited in the Protein Data Bank (PDB code 2taa). The structure of the Gd-bound protein was isomorphous to that of the native form and was determined by difference Fourier analysis using the corresponding model of the native structure. Manual model revision was performed using QUANTA2000 software (Accelrys Inc.).
Bound ions were observed in the structure of the Gd-soaked crystal. Based on comparison of temperature factors to those of neighboring atoms, and from their coordination with ion-binding residues, the bound ions are most likely to be Gd ions from the heavy-atom soaking reagent. This interpretation is in agreement with the fact that strong signals are observed at the ion sites in the anomalous Fourier map at a wavelength of 1.000 Å , prepared using the program FFT in CCP4 suite [45].
The program CNS [46] was used for structure refinement and electron-density map calculation. Each cycle of refinement with bulk solvent and overall anisotropic B-factor corrections consisted of rigid-body refinement, simulated annealing incorporating the slow-cool protocol, positional refinement, and B-factor refinement (individual or group). Several cycles of model revision and refinement yielded the final models. The stereochemical quality of the final structures was verified using the program PROCHECK [47]. Statistics of the data collection and refinement are shown in Table 2. The structural data are available in the Protein Data Bank under the accession numbers 3VX0, 3VX1.

Dynamic light-scattering study
Ao a-amylase was examined by dynamic light-scattering experiment using a DynaPro MS/X (Protein Solutions) instrument at a protein concentration of 0.4 mg/mL in 50 mM MES-NaOH (pH 5.8) containing 0-2.0 mM GdCl 3 . Several measurements were taken at 293 K and analyzed using the program DYNAMICS v.5.26.60 (Protein Solutions). The average values of two measurements of mass at each GdCl 3 concentration are shown in Table 1.

CD spectra
Circular dichroism (CD) spectra of Ao a-amylase were recorded on a Jasco J-725 spectropolarimeter (Jasco Co., Japan). Near-UV CD spectra in the wavelength range from 250 to 320 nm were scanned 16 times at a scan rate of 20 nm/min, using a time constant of 0.25 sec. The light-path length of the cell was 10 mm in the near-UV region. For the measurements at 25uC, the protein concentrations were 0.75 mg/mL in 50 mM MES-NaOH (pH 5.8) with 0.04 mM or 0.2 mM GdCl 3 . Values in parentheses are for the outermost shell.
{ R merge = g hkl g i |I i (hkl)2,I(hkl).|/g hkl g i I i (hkl), where I i (hkl) is the ith observation of reflection hkl and ,I(hkl). is the weighted average intensity for all observations i of reflection hkl.
1 R cryst = g hkl ||F obs |2|F calc ||/g hkl |F obs |, where |F obs | and |F calc | are the observed and calculated structure-factor amplitudes, respectively. R free was calculated with 5% of the reflections chosen at random and omitted from refinement. doi:10.1371/journal.pone.0057432.t002