Crystal Structure of Calcium-free α-Amylase from Bacillus sp. Strain KSM-K38 (AmyK38) and Its Sodium Ion Binding Sites
2003; Elsevier BV; Volume: 278; Issue: 27 Linguagem: Inglês
10.1074/jbc.m212763200
ISSN1083-351X
AutoresTsuyoshi Nonaka, M. Fujihashi, Akiko Kita, Hiroshi Hagihara, Katsuya Ozaki, Susumu Ito, Kunio Miki,
Tópico(s)Enzyme Structure and Function
ResumoThe crystal structure of a calcium-free α-amylase (AmyK38) from Bacillus sp. strain KSM-K38, which resists chelating reagents and chemical oxidants, has been determined by the molecular replacement method and refined to a crystallographic R-factor of 19.9% (R-free of 23.2%) at 2.13-Å resolution. The main chain folding of AmyK38 is almost homologous to that of Bacillus licheniformis α-amylase. However, neither a highly conserved calcium ion, which is located at the interface between domains A and B, nor any other calcium ions appear to exist in the AmyK38 molecule, although three sodium ions were found, one of which is located at the position corresponding to that of a highly conserved calcium ion of other α-amylases. The existence of these sodium ions was crystallographically confirmed by the structures of three metal-exchanged and mutated enzymes. This is the first case in which the structure of the calcium-free α-amylase has been determined by crystallography, and it was suggested that these sodium ions, instead of calcium ions, are used to retain the structure and function of AmyK38. The crystal structure of a calcium-free α-amylase (AmyK38) from Bacillus sp. strain KSM-K38, which resists chelating reagents and chemical oxidants, has been determined by the molecular replacement method and refined to a crystallographic R-factor of 19.9% (R-free of 23.2%) at 2.13-Å resolution. The main chain folding of AmyK38 is almost homologous to that of Bacillus licheniformis α-amylase. However, neither a highly conserved calcium ion, which is located at the interface between domains A and B, nor any other calcium ions appear to exist in the AmyK38 molecule, although three sodium ions were found, one of which is located at the position corresponding to that of a highly conserved calcium ion of other α-amylases. The existence of these sodium ions was crystallographically confirmed by the structures of three metal-exchanged and mutated enzymes. This is the first case in which the structure of the calcium-free α-amylase has been determined by crystallography, and it was suggested that these sodium ions, instead of calcium ions, are used to retain the structure and function of AmyK38. Amylases, which are glucosidic-bond-hydrolyzing enzymes, are classified into endo and exo types. The former contains α-amylase (EC 3.2.1.1), pullulanase (EC 3.2.1.41), and isoamylase (EC 3.2.1.68), whereas the latter contains exo-1,4-α-d-glucosidase (EC 3.2.1.3), β-amylase (EC 3.2.1.2), exo-isomaltotriohydrolase (EC 3.2.1.95), exo-maltotetraohydrolase (EC 3.2.1.60), and exo-maltohexaohydrolase (EC 3.2.1.98). α-Amylases hydrolyze α-1,4-d-glucosidic-bonds of glycogen, starch, related polysaccharides, and some oligosaccharides. α-Amylases are found in microorganisms, animals, and plants. The three-dimensional structures of several α-amylases such as Bacillus licheniformis α-amylase (BLA) 1The abbreviations used are: BLA, B. licheniformis α-amylase; MES, 2-(N-morpholino)ethanesulfonic acid; TVA, Thermoactinomyces vulgaris R-47 α-amylase. (1Machius M. Declerck N. Huber R. Wiegand G. Structure. 1998; 6: 281-292Google Scholar, 2Machius M. Wiegand G. Huber R. J. Mol. Biol. 1995; 246: 545-559Google Scholar), TAKA α-amylase (Aspergillus oryzae α-amylase) (3Matsuura Y. Kusunoki M. Harada W. Tanaka N. Iga Y. Yasuoka N. Toda H. Narita K. Kakudo M. J. Biochem. (Tokyo). 1980; 87: 1555-1558Google Scholar, 4Boel E. Brady L. Brzozowski A.M. Derewenda Z. Dodson G.G. Jensen V.J. Petersen S.B. Swift H. Thim L. Woldike H.F. Biochemistry. 1990; 29: 6244-6249Google Scholar, 5Swift H.J. Brady L. Derewenda Z.S. Dodson E.J. Dodson G.G. Turkenburg J.P. Wilkinson A.J. Acta Crystallogr. Sect. B Struct. Crystallogr. 1991; 47: 535-544Google Scholar), barley malt α-amylase (6Kadziola A. Abe J. Svensson B. Haser R. J. Mol. Biol. 1994; 239: 104-121Google Scholar), and porcine pancreatic α-amylase (7Buisson G. Duee E. Haser R. Payan F. EMBO J. 1987; 6: 3909-3916Google Scholar, 8Qian M. Haser R. Payan F. J. Mol. Biol. 1993; 231: 785-799Google Scholar, 9Larson S.B. Greenwood A. Cascio D. Day J. McPherson A. J. Mol. Biol. 1994; 235: 1560-1584Google Scholar), have been determined. The structures of these α-amylases commonly consist of three domains, a structurally conserved (β/α)8-barrel domain first observed in triose phosphate isomerase (Domain A), an additional domain inserted within Domain A (Domain B), and the C-terminal domain (Domain C) (10Banner D.W. Bloomer A.C. Petsko G.A. Phillips D.C. Pogson C.I. Wilson I.A. Corran P.H. Furth A.J. Milman J.D. Offord R.E. Priddle J.D. Waley S.G. Nature. 1975; 255: 609-614Google Scholar, 11Janecek S. Svensson B. Henrissat B. J. Mol. Evol. 1997; 45: 322-331Google Scholar). Three completely conserved catalytic residues (Asp-231, Glu-261, and Asp-328 according to the sequence of BLA) are located in Domain A (12Nielsen J.E. Borchert T.V. Biochim. Biophys. Acta. 2000; 1543: 253-274Google Scholar). All known α-amylases contain calcium ions that contribute to stabilization of the structures (1Machius M. Declerck N. Huber R. Wiegand G. Structure. 1998; 6: 281-292Google Scholar, 2Machius M. Wiegand G. Huber R. J. Mol. Biol. 1995; 246: 545-559Google Scholar, 4Boel E. Brady L. Brzozowski A.M. Derewenda Z. Dodson G.G. Jensen V.J. Petersen S.B. Swift H. Thim L. Woldike H.F. Biochemistry. 1990; 29: 6244-6249Google Scholar, 13Vallee B.L. Stein E.A. Summerwill W.N. Fisher E.H. J. Biol. Chem. 1959; 234: 2901-2905Google Scholar, 14Savchenko A. Vieille C. Kang S. Zeikus J.G. Biochemistry. 2002; 41: 6193-6201Google Scholar). Furthermore, all known α-amylases, except for cyclodextrins- and pullulan-hydrolyzing α-amylases containing an additional domain (15Kamitori S. Abe A. Ohtaki A. Kaji A. Tonozuka T. Sakano Y. J. Mol. Biol. 2002; 318: 443-453Google Scholar, 16Kamitori S. Kondo S. Okuyama K. Yokota T. Shimura Y. Tonozuka T. Sakano Y. J. Mol. Biol. 1999; 287: 907-921Google Scholar), have a common site for the highly conserved calcium ion at the interface between two domains (A and B) to keep the functioning structure (12Nielsen J.E. Borchert T.V. Biochim. Biophys. Acta. 2000; 1543: 253-274Google Scholar). The role of the conserved calcium ion is mainly to retain the structural rigidity of the α-amylase molecules (1Machius M. Declerck N. Huber R. Wiegand G. Structure. 1998; 6: 281-292Google Scholar, 7Buisson G. Duee E. Haser R. Payan F. EMBO J. 1987; 6: 3909-3916Google Scholar, 9Larson S.B. Greenwood A. Cascio D. Day J. McPherson A. J. Mol. Biol. 1994; 235: 1560-1584Google Scholar). α-Amylases are widely used for desizing textiles as well as for producing glutinous starch syrup and sugar, detergent for automatic dishwashing machines, and so on. Because detergents usually display their washing function at a pH range between 8 and 11, it is desirable for the enzymes used in detergents to be alkaliphilic (12Nielsen J.E. Borchert T.V. Biochim. Biophys. Acta. 2000; 1543: 253-274Google Scholar, 17Ito S. Kobayashi T. Ara K. Ozaki K. Kawai S. Hatada Y. Extremophiles. 1998; 2: 185-190Google Scholar). When alkaline α-amylase is used as a component of detergents, it is unfavorable that its structural rigidity depends on the calcium ions included in the α-amylase molecule because the chelating reagents usually contained in detergents easily remove calcium and zinc ions (12Nielsen J.E. Borchert T.V. Biochim. Biophys. Acta. 2000; 1543: 253-274Google Scholar). However, known alkaline α-amylases usually contain structurally essential calcium ions and are often inhibited by chelating agents such as zeolite, EDTA, and EGTA (18Moranelli F. Yaguchi M. Calleja G.B. Nasim A. Biochem. Cell Biol. 1987; 65: 899-908Google Scholar, 19Violet M. Meunier J.C. Biochem. J. 1989; 263: 665-670Google Scholar, 20Kim C.H. Kim Y.S. Eur. J. Biochem. 1995; 227: 687-693Google Scholar, 21Igarashi K. Hatada Y. Ikawa K. Araki H. Ozawa T. Kobayashi T. Ozaki K. Ito S. Biochem. Biophys. Res. Commun. 1998; 248: 372-377Google Scholar, 22Lecker D.N. Khan A. Biotechnol. Prog. 1998; 14: 621-625Google Scholar). AmyK38 was found in an alkaliphilic Bacillus sp. strain KSM-K38 (23Hagihara H. Igarashi K. Hayashi Y. Endo K. Ikawa-Kitayama K. Ozaki K. Kawai S. Ito S. Appl. Environ. Microbiol. 2001; 67: 1744-1750Google Scholar, 24Hagihara H. Hayashi Y. Endo K. Igarashi K. Ozawa T. Kawai S. Ozaki K. Ito S. Eur. J. Biochem. 2001; 268: 3974-3982Google Scholar) (accession number of DNA data bank of Japan, AB051102; registration number of National Institute of Bioscience and Human Technology Agency, FERM BP-6946). This α-amylase prefers alkaline conditions and resists oxidative reagents. This enzyme hydrolyzes soluble starch, amylopectin, glycogen, amylose, and dextrin to oligosaccharides or glucose and does not hydrolyze dextran, pullulan, or α-, β-, and γ-cyclodextrins (23Hagihara H. Igarashi K. Hayashi Y. Endo K. Ikawa-Kitayama K. Ozaki K. Kawai S. Ito S. Appl. Environ. Microbiol. 2001; 67: 1744-1750Google Scholar). The amino acid sequence homology with BLA, a typical known bacterial α-amylase from B. licheniformis, is ∼63% (Fig. 1). However, it has been shown by activity measurements and elemental analysis that AmyK38 has no Ca2+ ions and is not inhibited by chelating reagents. It also appears that its enzymatic activity depends on the existence of Na+ ions (23Hagihara H. Igarashi K. Hayashi Y. Endo K. Ikawa-Kitayama K. Ozaki K. Kawai S. Ito S. Appl. Environ. Microbiol. 2001; 67: 1744-1750Google Scholar, 24Hagihara H. Hayashi Y. Endo K. Igarashi K. Ozawa T. Kawai S. Ozaki K. Ito S. Eur. J. Biochem. 2001; 268: 3974-3982Google Scholar). These characteristics of AmyK38 are advantageous for the use of alkaline α-amylases as a component of detergents. We have determined the crystal structure of AmyK38 from an alkaliphilic Bacillus sp. strain KSM-K38 to prove crystallographically the calcium independency of this enzyme and to clarify how the structural rigidity is maintained in the absence of calcium on the basis of the three-dimensional structures of the wild-type, metal-exchanged, and mutated forms of AmyK38. Expression and Purification—The recombinant wild-type AmyK38 from alkaliphilic Bacillus sp. strain KSM-K38 was produced and purified as described previously (23Hagihara H. Igarashi K. Hayashi Y. Endo K. Ikawa-Kitayama K. Ozaki K. Kawai S. Ito S. Appl. Environ. Microbiol. 2001; 67: 1744-1750Google Scholar, 24Hagihara H. Hayashi Y. Endo K. Igarashi K. Ozawa T. Kawai S. Ozaki K. Ito S. Eur. J. Biochem. 2001; 268: 3974-3982Google Scholar). The N289H mutant of AmyK38, in which Asn-289 was replaced with His, was constructed using the method of splicing by overlap extension (25Horton R.M. White B.A. Methods in Molecular Biology. 15. Humana Press, Totowa, NJ1993: 251-261Google Scholar). This mutation was based on the correspondence between that Asn-289 at the Na III site (see below) of wild-type AmyK38 and His-289 of BLA. Expression and purification were carried out in the same manner as with wild-type AmyK38 (23Hagihara H. Igarashi K. Hayashi Y. Endo K. Ikawa-Kitayama K. Ozaki K. Kawai S. Ito S. Appl. Environ. Microbiol. 2001; 67: 1744-1750Google Scholar, 24Hagihara H. Hayashi Y. Endo K. Igarashi K. Ozawa T. Kawai S. Ozaki K. Ito S. Eur. J. Biochem. 2001; 268: 3974-3982Google Scholar). Crystallization and Preparation of the Metal-exchanged Enzymes— Protein solutions of the wild-type AmyK38 and its N289H mutant were dialyzed against 10 mm Tris-HCl buffer solution (pH 7.5) and concentrated to 10 mg/ml by centrifugal ultrafiltration. Both the wild-type and N289H enzymes were crystallized by the sitting-drop vapor diffusion method at 293 K using the reservoir solutions (pH 6.8) containing 0.085 m cacodylate-NaOH, 25.5% (w/v) polyethylene glycol 8000, 0.17 m sodium acetate, and 15% (v/v) glycerol (Crystal Screen Cryo™ number 28, Hampton Research). The wild-type enzyme was also crystallized in the presence of CaCl2 (2 mm) added to the protein solution. The mixtures of the same volume of the protein solutions and the reservoir solutions were equilibrated against the reservoir solutions. Sodium ions of the wild-type AmyK38 were replaced with rubidium, potassium, and lithium ions by the soaking method for crystals because the co-crystallization from the metal-containing solutions was not successful. The components of the soaking buffer solutions were: 0.085 m MES-NaOH, pH 6.8, 26% (w/v) polyethylene glycol 8000, 0.2 m rubidium acetate, and 15% (v/v) glycerol for rubidium; 0.085 m MES-KOH, pH 6.8, 26% (w/v) polyethylene glycol 8000, 0.2 m potassium acetate, and 15% (v/v) glycerol for potassium; 0.085 m MES-LiOH, pH 6.7, 25.5% (w/v) polyethylene glycol 8000, 0.2 m lithium acetate, and 15% (v/v) glycerol for lithium. Excess volumes of these buffer solutions were added into the drops containing wild-type AmyK38 crystals for soaking. The crystals were frozen in a nitrogen gas stream at 100 K after the soaking times of 5 min, overnight, and 2 min for rubidium, potassium, and lithium, respectively. Data Collection, Phasing, and Refinement—X-ray diffraction data from the crystals of the wild-type, metal-exchanged, and N289H mutant enzyme were collected at 100 K at the BL40B2, BL44B2, and BL45PX beamlines of SPring-8. The x-ray wavelength of 0.81 Å was selected in the data collection for Rb+-containing crystals to observe the large anomalous dispersion effect as the K absorption edge of rubidium is 0.8155 Å. These data were processed by the programs of the HKL2000 suite (26Otwinowski Z. Minor W. Methods Enzymol. 1997; 276: 307-326PubMed Google Scholar). The molecular replacement method for the wild-type AmyK38 was carried out using the atomic coordinates of BLA (Protein Data Bank accession code 1bli) as a search model by the program AMoRe (27Navaza J. Acta Crystallogr. Sect. A. 1994; 50: 157-163Google Scholar) from the CCP4 program suite (28Collaborative Computational Project Number 4 Acta Crystallogr. Sect. D Biol. Crystallogr. 1994; 50: 760-763Google Scholar). The refined structure of the wild-type AmyK38 was used as the search model in the molecular replacement method of other data sets of the N289H mutant and metal-exchanged enzymes. All of the structures were refined by the programs O and CNS (29Jones T.A. Zou J.Y. Cowan S.W. Kjeldgaard Acta Crystallogr. Sect. A. 1991; 47: 110-119Google Scholar, 30Brünger A.T. Adams P.D. Clore G.M. DeLano W.L. Gros P. Grosse-Kunstleve R.W. Jiang J.S. Kuszewski J. Nilges M. Pannu N.S. Read R.J. Rice L.M. Simonson T. Warren G.L. Acta Crystallogr. Sect. D Biol. Crystallogr. 1998; 54: 905-921Google Scholar). Replacements of Na+ ions by Rb+, K+, or Li+ ions were assessed by electron density maps, anomalous difference Fourier maps and coordinates, and B-factors of refined structures. In the present study, we determined the crystal structures of five crystal forms of AmyK38, the wild-type enzyme, three metal-exchanged enzymes with rubidium, potassium, and lithium, and the N289H mutated enzyme. Crystallization, Data Collection, Molecular Replacement, and Structure Refinement—The quadrilateral bipyramidal-shaped crystals grew in ∼2 weeks to an approximate size of 0.2 × 0.2 × 0.2 mm. The crystals were found to belong to the cubic space group P23 (a = 132.1 Å for the wild-type crystal) and to diffracted X-rays beyond 2.2-Å resolution. Statistics for data collection and structure refinement are shown in Table I. The molecular replacement method was successfully performed for all crystals. The structure of the wild-type AmyK38 was refined to a crystallographic R-factor of 19.9% (R-free of 23.2%) at 2.13-Å resolution. Root mean square deviations from ideal bond distances and angles were 0.006 Å and 1.3°, respectively (31Engh R.A. Huber R. Acta Crystallogr. Sect. A. 1991; 47: 392-400Google Scholar). In a Ramachandran plot, 87.6% of the non-glycine residues were in the most favored regions of a ϕ-ψ plot (32Ramachandran G.N. Sasisekharan V. Adv. Protein Chem. 1968; 23: 283-438Google Scholar). The structure refinement of the wild-type crystals grown from the CaCl2-containing solution afforded R-factor of 19.4% and R-free factor of 22.2%. The structures of the metal-exchanged enzymes were also refined at 2.50–2.88-Å resolution to R-factors between 21.0 and 22.4% (R-free factors between 25.1 and 27.2%). The structure refinement of the N289H mutant converged with the R-factor to 23.4% (R-free of 26.4%) at 2.15-Å resolution.Table IData collection and refinement statisticsCrystalWild-type (Ca-free)aGrown from calcium-free solutions.Wild-type (CaCl2)bGrown from 2 mM CaCl2-containing solutions.Rb+ soakingcSoaked into Rb+-, K+-, and Li+-containing solutions, respectively.K+ soakingcSoaked into Rb+-, K+-, and Li+-containing solutions, respectively.Li+ soakingcSoaked into Rb+-, K+-, and Li+-containing solutions, respectively.N289H mutantData collectionResolution (Å)dValues for the outermost resolution shell are in parentheses.100-2.13 (2.17-2.13)100-2.15 (2.23-2.15)100-2.70 (2.80-2.70)100-2.50 (2.59-2.50)100-2.88 (2.98-2.88)100-2.15 (2.19-2.15)Cell constant a (Å)132.1133.7132.6132.5132.6132.5Wavelength (Å)1.020.980.811.001.001.02Rmerge (%)dValues for the outermost resolution shell are in parentheses.,eRmerge = Σ|Ii - |/Σ , where Ii is the observed intensity and is the average intensity over symmetry equivalent measurements.10.5 (28.8)15.7 (30.0)9.3 (26.2)9.9 (27.2)13.7 (29.7)7.7 (29.6)Completeness (%)dValues for the outermost resolution shell are in parentheses.98.1 (95.6)96.7 (91.9)95.0 (88.1)95.4 (89.9)96.4 (90.3)96.7 (91.0)No. of observed diffractions (> 1σ)783,382465,200260,419404,246363,444345,465No. of unique reflections (> 1σ)42,44442,15520,57425,84617,22740,862Redundancy18.511.012.715.621.18.5I/σ(I)dValues for the outermost resolution shell are in parentheses.23.5 (6.79)11.67 (4.43)27.4 (10.9)30.8 (12.9)23.5 (12.8)19.5 (3.42)RefinementR factor (%)fR = Σ ∥Fobs - Fcalc∥/Σ|Fobs|. Rfree is the same as R, but for a 5% subset of all reflections that were never used in crystallographic refinement.19.919.422.421.521.023.4Rfree factor (%)fR = Σ ∥Fobs - Fcalc∥/Σ|Fobs|. Rfree is the same as R, but for a 5% subset of all reflections that were never used in crystallographic refinement.23.222.227.225.126.526.4a Grown from calcium-free solutions.b Grown from 2 mM CaCl2-containing solutions.c Soaked into Rb+-, K+-, and Li+-containing solutions, respectively.d Values for the outermost resolution shell are in parentheses.e Rmerge = Σ|Ii - |/Σ , where Ii is the observed intensity and is the average intensity over symmetry equivalent measurements.f R = Σ ∥Fobs - Fcalc∥/Σ|Fobs|. Rfree is the same as R, but for a 5% subset of all reflections that were never used in crystallographic refinement. Open table in a new tab Overall Structure of the Wild-type AmyK38 —The overall structure of the wild-type AmyK38 is shown in Fig. 2. The folding of the wild-type AmyK38 is essentially similar to that of BLA, with root mean square deviations of 0.90 Å for the main chain atoms of the common 473 amino acid residues. The AmyK38 molecule consists of three domains (A, B, and C) containing a (β/α)8-barrel motif (Domain A) commonly observed in various glycosidases (Fig. 2). In addition to the protein molecule, three significant peaks were observed in the electron density map of the wild-type AmyK38 (assigned to sites I, II, and III). The peak heights in the solvent-omit Fo – Fc map (11.0σ, 9.5σ, and 12.3σ for the sites I, II, and III, respectively), which are only slightly larger than those of the most well identified water oxygen (9.0σ), are too small to assign to the calcium ions. Two of these (sites I and II), surrounded by one carboxyl group of an aspartic acid residue, amido groups of asparagine residues, carbonyl oxygen atoms of the main chain, and a few water molecules, are located at the sites corresponding to the Ca2+ ions (Ca I and Ca III) in the BLA structure (Fig. 3, a and b). These ions are hexacoordinated, and the geometry is described as a distorted quadrilateral bipyramidal octahedron (Fig. 3, a and b), which is often observed in the metal ion binding sites of metalloproteins. The third site (site III) is observed only in the wild-type AmyK38 structure and is occupied by the side chain of a histidine residue in the BLA structure (Fig. 3c). The structure of the wild-type AmyK38 was determined using two types of crystals grown from the Ca2+-free and Ca2+-containing (2 mm) crystallization solutions (Table I). However, no significant differences of electron density between these two crystal structures were observed in these metal binding sites (Table II). In addition, it has already been shown that the enzymatic activity of AmyK38 depends on Na+ ions (not on Ca2+ ions) (23Hagihara H. Igarashi K. Hayashi Y. Endo K. Ikawa-Kitayama K. Ozaki K. Kawai S. Ito S. Appl. Environ. Microbiol. 2001; 67: 1744-1750Google Scholar, 24Hagihara H. Hayashi Y. Endo K. Igarashi K. Ozawa T. Kawai S. Ozaki K. Ito S. Eur. J. Biochem. 2001; 268: 3974-3982Google Scholar). The peak heights of the electron density at three positions are also the most reasonable to be assigned to the sodium ions. These observations imply that the wild-type AmyK38 contains not Ca2+ ions but Na+ ions, at the three positions mentioned above, a hypothesis that was confirmed by the subsequent metal-exchange experiments.Table IIPeak heights and B-factors of the metal ions and bond distances between the metal ions and the ligand oxygen atomsCrystalWild-type (Ca-free)aGrown from calcium-free solutions.Wild-type (CaCl2)bGrown from 2 mM CaCl2-containing solutions.Rb+ soakingcSoaked into Rb+ containing solutions. Three metal positions for sites I, II, and III were refined as Na+, Na+, and Rb+ ions, respectively.K+ soakingdSoaked into K+ -containing solutions. Three metal positions (sites I, II, and III) were refined as either Na+ (left) or K+ (right), respectively.Li+ soakingeSoaked into Li+ -containing solutions. Three metal positions were refined as Na+ ions.Peak heightfPeak heights in the solvent omit Fo — Fc map.Site I11.0σ12.0σ7.7σ8.4σ/8.5σ7.5σSite II9.5σ9.2σ5.9σ10.7σ/11.0σ5.9σSite III12.3σ13.1σ17.6σ13.7σ/14.0σ6.2σB-factors (Å2)Site I12.810.414.717.0/40.419.6Site II20.014.933.19.0/28.932.4Site III14.311.434.01.0/18.928.0Bond distance (Å)gAveraged bond distances between the metal ion and the ligand oxygen atoms.Site I2.382.392.382.44/2.542.45hThis metal ion site is penta-coordinated, whereas all the other sites are hexa-coordinated.Site II2.462.432.492.64/2.702.51Site III2.322.312.672.56/2.602.86a Grown from calcium-free solutions.b Grown from 2 mM CaCl2-containing solutions.c Soaked into Rb+ containing solutions. Three metal positions for sites I, II, and III were refined as Na+, Na+, and Rb+ ions, respectively.d Soaked into K+ -containing solutions. Three metal positions (sites I, II, and III) were refined as either Na+ (left) or K+ (right), respectively.e Soaked into Li+ -containing solutions. Three metal positions were refined as Na+ ions.f Peak heights in the solvent omit Fo — Fc map.g Averaged bond distances between the metal ion and the ligand oxygen atoms.h This metal ion site is penta-coordinated, whereas all the other sites are hexa-coordinated. Open table in a new tab Asp-161, Asp-183, Asp-200, and Asp-204 around the calcium-sodium-calcium triad containing the highly conserved Ca I in BLA (1Machius M. Declerck N. Huber R. Wiegand G. Structure. 1998; 6: 281-292Google Scholar) are replaced with Asn-161, Asn-183, Asn-200, and Ser-204 at the corresponding site around site I of AmyK38, respectively (Fig. 3a). Asp-430 at the Ca III site of BLA is replaced with Asn-427 at site II of AmyK38 (Fig. 3b). These replacements of amino acid residues reduce the negative charges of the cation binding sites and induce binding a monovalent metal ion such as Na+ rather than a divalent metal ion such as Ca2+ to these sites. A similar exchange of metal ions has been reported in the cyclodextrin- and pullulan-hydrolyzing α-amylases, TVA I and TVA II (15Kamitori S. Abe A. Ohtaki A. Kaji A. Tonozuka T. Sakano Y. J. Mol. Biol. 2002; 318: 443-453Google Scholar, 16Kamitori S. Kondo S. Okuyama K. Yokota T. Shimura Y. Tonozuka T. Sakano Y. J. Mol. Biol. 1999; 287: 907-921Google Scholar). Although TVA I and TVA II also contain Ca2+ ions, these ions are not highly conserved among α-amylases. The amino group Nζ of Lys-295 in TVA II, corresponding to Asp-200 in BLA, occupies the site of the highly conserved Ca2+ ion (16Kamitori S. Kondo S. Okuyama K. Yokota T. Shimura Y. Tonozuka T. Sakano Y. J. Mol. Biol. 1999; 287: 907-921Google Scholar). The replacement of negatively charged amino acid residues with neutral or positively charged residues might be a common way to change or remove metal ions from α-amylases and other metalloproteins by site-directed mutagenesis. Crystallographic Evidence for the Existence of the Monovalent Metal Ions in AmyK38 —As already mentioned, it is reasonable to assume that the three metal ions found in AmyK38 are not assigned to calcium, but to sodium, from several reasons, including the peak heights in the electron density map (Fig. 3, a–c). To obtain crystallographic evidence for the replacement of divalent Ca2+ with monovalent Na+ ions, the crystal structures of the metal-exchanged AmyK38 (to Rb+, K+, and Li+) were determined. In the crystal of the Rb+-exchanged AmyK38 prepared by the soaking method in the Rb+-containing buffer, the structure showed that sites I and II were not replaced with Rb+, but that site III was replaced with Rb+. A significant peak (over 18 σ) was observed at site III in the anomalous Fourier map (amplitude, F(+) – F(–); phase, Fcalc+90 degrees) with the results indicating that site III was replaced with Rb+ ions. In addition, three significant anomalous Fourier peaks (over 10 σ) were also found on the surface of the AmyK38 molecule, and these were refined as Rb+ ions. In the native AmyK38 structure, no metal ions are observed at the corresponding Rb+ positions on the surface of AmyK38. The replacement with Rb+ at site III provides evidence that site III is exchangeable by a monovalent metal ion such as Na+. Therefore, it is most likely that site III contains a Na+ ion in native AmyK38. Sites I and II may not be replaced with Rb+ ions due to the large ion radius of Rb+. A similar observation has also been reported for the sodium binding site of BLA (1Machius M. Declerck N. Huber R. Wiegand G. Structure. 1998; 6: 281-292Google Scholar). Moreover, sites I and II correspond to the sites for the Ca I and Ca III of BLA (Fig. 3, a and b). This result also suggests that the binding affinity of site III for monovalent metal ions is weak. In the K+-exchanged AmyK38 prepared by the soaking method in the K+-containing buffer, no large differences were observed in the electron density of the crystal structure (Table II). However, B-factors of atoms at sites II and III refined as Na+ ions were obviously decreased from 20.0 and 14.3 Å2 (the native AmyK38) to 9.0 and 1.0 Å2 (the K+-containing AmyK38 refined as Na+ ions), respectively (Table II). B-factors of atoms at sites II and III refined as K+ ions were converged to 29.0 and 18.9 Å2, respectively (Table II). In comparison with B-factors of surrounding residues, these values at sites II and III refined as K+ ions are reasonable (Table II). On the other hand, the peak height of site I is similar to that of the native AmyK38, and the B-factor of the atom at site I refined as a Na+ ion was increased from 12.9 Å2 (the native AmyK38) to 18.5 Å2 (the K+-containing AmyK38) (Table II). Moreover, coordination distances between these three metal peaks and surrounding residues are slightly lengthened (Table II). We therefore conclude that the metal ions at sites II and III were partially or fully replaced with K+ ions and that the metal ion at site I may have been partially replaced with a K+ ion. Therefore, these results also support the hypothesis that the metal ions at sites I, II, and III tend to be replaced with monovalent metal ions and that the metal ions in native AmyK38 are assigned to Na+ ions. In the Li+-exchanged AmyK38 prepared by the soaking method in the Li+-containing buffer, significant differences were observed in the electron density of the crystal structure. At site I, the electron density of the metal ion was obviously decreased, indicating that the electron density could be assigned to Li+ (Table II). The electron density of the water molecule coordinated to the metal ion had disappeared in the Li+-exchanged AmyK38, and the coordination geometry around the metal ion was consequently changed. At the same time, the electron densities at sites II and III were also obviously decreased. These findings indicate that sites I, II, and III were almost fully replaced with Li+ ions. It is therefore concluded that sites I, II, and III are replaceable with monovalent metal ions and can be assigned as Na+ ions in the native AmyK38. The coordination geometry and the electron density in the structure of the K+- and Li+-exchanged AmyK38 show that the binding of the Na+ ion at site I is tight, as compared with that at sites II and III. Moreover, this result is in accord with the correspondence between site I and the highly conserved Ca2+ ion binding site reported in almost all α-amylases (Fig. 3a). These ion-exchange experiments show that sites I, II, and III are certainly Na+ ion binding sites (Na I, Na II, and Na III, respectively) and that Na I is bound more tightly than Na II and is the most important to the AmyK38 structure. In addition, Na III, which is surrounded by only four atoms, Oδ1 or Nδ2 of the amido group of Asn-289, Oδ1 of the carboxyl group of Asp-325, the carbonyl oxygen atom of Val-324, and the carbonyl oxygen atom of Ser-337, is loosely bound (Fig. 3c). The structure of N289H mutant AmyK38 was refined to a crystallographic R-factor of 23.4% in the resolution range between 100 and 2.15 Å (Table I) to investigate the role of site III. In this mutant, where Asn-289 is replaced by histidine, a corresponding residue in BLA, the metal ion at site III has disappeared, with the position being occupied by the side chain of His-289 (Fig. 3d). Consequently, the arrangement of amino acid residues at site III is almost identical to that of BLA, as shown in Fig. 3d. The activity measurements show that this mutation hardly affects the specific activity and affords only 10% decrease of the sodium ion dependence (Fig. 4). The site III is far from the catalytic residues and not conserved in other α-amylases. These results show that the Na III is dispensable to retain the structure and function of AmyK38. Conclusion—Neither the highly conserved Ca2+ ion nor any other Ca2+ ion exists in AmyK38. However, there are three Na+ ions at Na I, Na II, and Na III sites in the AmyK38 molecule. These ions are located far from the catalytic residues of AmyK38 and probably do not affect the catalytic mechanism. Two of the three ions located at Na I and Na II sites correspond to Ca2+ ions observed in other α-amylase structures. In particular, the Na+ ion at the Na I site corresponds to the highly conserved Ca2+ ion in almost all α-amylases. It has been reported that residues 178–199, which are disordered in apo-BLA, are dynamically moved and ordered by the binding of metal ions, including the highly conserved Ca2+ ion in holo BLA (1Machius M. Declerck N. Huber R. Wiegand G. Structure. 1998; 6: 281-292Google Scholar). Some of these ordered residues form parts of sugar subsites. Therefore, Na+ ions at Na I and Na II sites must be important to retain the functioning structure of the AmyK38 molecule, which is free from any calcium ions. Recently, α-amylase from Pyrococcus furiosus, which had been initially reported as the case of calcium-free α-amylase (33Dong G. Vieille C. Savchenko A. Zeikus J.G. Appl. Environ. Microbiol. 1997; 63: 3569-3576Google Scholar), was indicated to be a calcium- and zinc-containing enzyme (14Savchenko A. Vieille C. Kang S. Zeikus J.G. Biochemistry. 2002; 41: 6193-6201Google Scholar). AmyK38 is the first case in which the structure of the α-amylase has no Ca2+ ions and in which the Na+ ions instead of Ca2+ play an important role in retaining the structure of the α-amylase. Such metal substitutions occur by the charge-changing replacements of metal ions surrounding the residues. This amylase could be employed as a component of detergents used industrially, and this finding could prove useful with regard to developing further applications for α-amylases and to the protein engineering of metalloproteins. We thank Dr. Keiko Miura (SPring-8 BL40B2, proposal number 2001B0357), Dr. Shin-ichi Adachi (SPring-8 BL44B2), and Dr. Yoshiaki Kawano (SPring-8 BL45PX) for help during data collection.
Referência(s)