Crystal structure of amylomaltase from Thermus aquaticus, a glycosyltransferase catalysing the production of large cyclic glucans1

Crystal structure of amylomaltase from Thermus aquaticus, a glycosyltransferase catalysing the production of large cyclic glucans1

doi:10.1006/jmbi.1999.3503 available online at http://www.idealibrary.com on J. Mol. Biol. (2000) 296, 873±886 Crystal Structure of Amylomaltase fro...

856KB Sizes 0 Downloads 34 Views

doi:10.1006/jmbi.1999.3503 available online at http://www.idealibrary.com on

J. Mol. Biol. (2000) 296, 873±886

Crystal Structure of Amylomaltase from Thermus aquaticus, a Glycosyltransferase Catalysing the Production of Large Cyclic Glucans Ingo Przylas1, Koji Tomoo1, Yoshinobu Terada2, Takeshi Takaha2 Kazutoshi Fujii2, Wolfram Saenger1 and Norbert StraÈter1* 1

Institut fuÈr Chemie, Abteilung Kristallographie, Freie UniversitaÈt Berlin, Takustraûe 6, 14195 Berlin, Germany 2

Biochemical Research Laboratory, Ezaki Glico Co. Ltd., 4-6-5 Utajima Nishiyodogawa-ku, Osaka 5558502, Japan

Amylomaltase is involved in the metabolism of starch, one of the most important polysaccharides in nature. A unique feature of amylomaltase is its ability to catalyze the formation of cyclic amylose. In contrast to the well studied cyclodextrin glucanotransferases (CGTases), which synthesize cycloamylose with a ring size (degree of polymerization or DP) of 6-8, the amylomaltase from Thermus aquaticus produces cycloamyloses with a DP of 22 and higher. The crystal structure of amylomaltase from Ê resolution. It is a member of Thermus aquaticus was determined to 2.0 A the a-amylase superfamily of enzymes, whose core structure consists of a (b, a)8 barrel. In amylomaltase, the 8-fold symmetry of this barrel is disrupted by several insertions between the barrel strands. The largest insertions are between the third and ®fth barrel strands, where two insertions form subdomain B1, as well as between the second and third barrel strands, forming the a-helical subdomain B2. Whereas part of subdomain B1 is also present in other enzyme structures of the a-amylase superfamily, subdomain B2 is unique to amylomaltase. Remarkably, the C-terminal domain C, which is present in all related enzymes of the a-amylase family, is missing in amylomaltase. Amylomaltase shows a similar arrangement of the catalytic side-chains (two Asp residues and one Glu residue) as in previously characterized members of the a-amylase superfamily, indicating similar mechanisms of the glycosyl transfer reaction. In amylomaltase, a conserved loop of around eight amino acid residues is partially shielding the active center. This loop, which is well conserved among other amylomaltases, may sterically hinder the formation of small cyclic products. # 2000 Academic Press

*Corresponding author

Keywords: (b, a)8 barrel; glucanotransferase; disproportionating enzyme; a-amylase family; protein crystallography

Introduction Carbohydrates are essential components of all living organisms and form the most abundant class of biological molecules. Polysaccharides such Present address: K. Tomoo, Department of Physical Chemistry, Osaka University of Pharmaceutical Sciences, 4-20-1 Nasahara, Takatsuki, Osaka 569-1064, Japan. Abbreviations used: CA, cycloamylose; CGTase, cyclodextrin glucanotransferase; D-enzyme, disproportionating enzyme; DP, degree of polymerization; MIR, multiple isomorphous replacement; r.m.s.d., root-mean-square-deviation. E-mail address of the corresponding author: [email protected] 0022-2836/00/030873±14 $35.00/0

as starch are an important food reserve in plants and a major nutrient for animals. Whereas higher plants synthesize starch, bacteria, lower eukaryotes and animals accumulate glycogen. Due to the important biological role of these polysaccharides for energy storage and uptake, selective hydrolysis and formation of glycosidic bonds are critical steps for all organisms. Thus, various enzymes have been identi®ed to act on starch. They are classi®ed on the basis of the catalyzed reaction and substrate speci®city (IUBMB, 1992) into glycosyl hydrolases (EC 3.2.1.x) and glycosyl transferases (EC 2.4.x.y). According to further classi®cation based on evolutionary relationships, sequence homology and common catalytic mechanisms into families and # 2000 Academic Press

874 clans, amylomaltase belongs to family-13 (a-amylases), which includes about 20 different enzymes (Henrissat, 1991; Henrissat & Bairoch, 1996; Janecek et al., 1999; Kuriki & Imanaka, 1999). Crystal structures of the family-13 enzymes are known for a-amylase (Matsuura et al., 1984; Boel et al., 1990; Brady et al., 1991; Swift et al., 1991; Qian et al., 1993; Kadziola et al., 1994; Larson et al., 1994; Brayer et al., 1995; Machius et al., 1995; Ramasubbu et al., 1996; Fujimoto et al., 1998; Strobl et al., 1998 Kamitori et al., 1999), cyclodextrin glucanotransferase (Klein & Schulz, 1991; Kubota et al., 1991; Lawson et al., 1994; Harata et al., 1996; Knegtel et al., 1996), maltotetrahydrolase (Morishita et al., 1997), oligo-1,6-glucosidase (Kizaki et al., 1993; Watanabe et al., 1997) and isoamylase (Katsuya et al., 1998). From these structural studies the presence of three domains (A-C) emerged as a common structural feature, although additional domains are present in some enzymes. Domain A consists of a (b, a)8 barrel (TIM-barrel) and forms the core of the protein structure. The C-terminal domain C contains two sandwiched b-sheets. The function of this domain is unknown, although different studies demonstrate that it is necessary for catalytic activity (Desseaux et al., 1991; Holm et al., 1990; Vihinen et al., 1994). Domain B is formed by an insertion between the third and fourth strand of the barrel. In contrast to domains A and C, this domain varies considerably in both length and sequence, depending on the enzyme and source (Janecek et al., 1997; Jesperson et al., 1993). Domain B is mainly built by b-sheets. It is located close to the active site, forms part of the substrate binding cleft and is essential for catalytic activity (Desseaux et al., 1991). It contains an almost invariant aspartate residue (Asp175 in Taka-amylase), which is involved in the binding of a calcium ion at the interface of domains A and B, close to the active site (Janecek, 1995). Amylomaltase (EC 2.4.1.25) catalyzes the transglycosylation reaction of a-1,4-glucans and is proposed to be a member of the family-13 enzymes (Janecek, 1997). It was ®rst found in Escherichia coli, but seems to be distributed in various bacterial species with different physiological functions. In E. coli, amylomaltase is expressed with glucan phosphorylase from the same operon. It is proposed to be a member of the maltooligosaccharide transport and utilization system (Schwartz, 1987) and plays apparently a role in converting short maltooligosaccharides into longer chains upon which glucan phosphorylase can act (Takaha & Smith, 1999). However, the genomes of Haemophilus in¯uenzae and Aquifex aeolicus seem to lack genes coding for maltose transport proteins. In these organisms, the genes for amylomaltase are part of the glycogen operon, indicating that the enzyme might be involved in glycogen metabolism (Takaha & Smith, 1999). A similar enzyme, termed disproportionating enzyme (D-enzyme), is present in plants. The physiological role of D-enzyme, which is located in plastids, has not been clari®ed,

Crystal Structure of Amylomaltase

but the enzyme is assumed to be involved in starch turnover (Takaha & Smith, 1999). Amylomaltase and potato D-enzyme show strong similarities in their primary sequence and their catalytic properties (Takaha & Smith, 1999). They catalyze a transglycosylation reaction in which one a-1,4-glucan is transferred to another or to glucose (see Figure 1(a)): …a-1; 4-glucan†m ‡ …a-1; 4-glucan†n ˆ …a-1; 4-glucan†mÿx ‡ …a-1; 4-glucan†n‡x This intermolecular transglycosylation is readily reversible and often called ``disproportionating reaction''. Amylomaltase and potato D-enzyme also catalyse intramolecular transglycosylation which creates a cyclic glucan from a single linear glucan molecule: …a-1; 4-glucan†m ˆ cyclic…a-1; 4-glucan†x ‡ a-1; 4-glucan†mÿx Besides transglycosylation activity, these enzymes also have a weak hydrolytic activity: …a-1; 4-glucan†m ‡ H2 O ˆ …a-1; 4-glucan†x ‡ …a-1; 4-glucan†mÿx How these enzymes control hydrolysis versus transglycosylation activity and intramolecular versus intermolecular transglycosylation by their active site structures remains to be determined. Since cyclodextrin glucanotransferase (CGTase) (EC 2.4.1.19) catalyzes all the reactions above, it shows a close similarity with amylomaltase in the catalytic actions. The only detectable biochemical difference is the size of the smallest cyclic glucan molecule produced from amylose. When CGTase, amylomaltase or D-enzyme are incubated with amylose, they preferentially form larger cyclic glucans in the initial stage of reaction and smaller cyclic glucans, subsequently (Takaha et al., 1996; Terada et al., 1997, 1999). The smallest cyclic products by CGTase action are cycloamyloses (CAs) with a degree of polymerization (DP) of 6-8, termed a, b and g-cyclodextrin. In contrast, the amylomaltases from E. coli and Thermus aquaticus and potato D-enzyme produce CA17, CA22 and CA17, respectively, as smallest CAs (Takaha & Smith, 1999). The mechanism which determines the size of the smallest cyclic glucan is not known, but of great interest as CAs have found industrial applications. The doughnut-shaped cyclodextrins can complex guest molecules in their central hydrophobic cavity and have found many applications in the food, pharmaceutical and cosmetics industry and also as column material in chromatographic separations (Hedges, 1998). A crystal structure analysis of a cycloamylose with 26 glucose residues revealed that this macrocycle folds into two short left-handed V-amylose helices in antiparallel arrangement. A hydrophobic channel of

Crystal Structure of Amylomaltase

875

Figure 1. Amylomaltase activity. (a) Scheme for the reaction pathway of the hydrolysis and transglycosylation activity of amylomaltase. Whether or not the reaction proceeds via a covalent intermediate, as shown here, is still a matter of debate. (b) Structure of CA-26, a cycloamylose with 26 glucose rings (Gessler et al., 1999). The molecule contains two antiparallel helices forming two hydrophobic cavities, which are ®lled with partially occupied water molecules.

Ê radius runs along the axis of the V-helix 5.0-5.5 A (Figure 1(b)) (Gessler et al., 1999). Since this channel might complex a large variety of hydrophobic guest molecules, it appears likely that larger cycloamyloses may ®nd industrial applications similar to those of the cyclodextrins. Despite the similarities in the catalyzed reaction, CGTases and amylomaltases have no signi®cant sequence similarity, except for four conserved short motifs comprising the catalytic residues, including two aspartate and one glutamate (Takaha & Smith, 1999). Within the a-amylase superfamily these four motifs and the three acidic residues are highly conserved. Indeed, based on sequence comparisons amylomaltase appears to be the most distantly related member of the a-amylase superfamily (Janecek, 1997). In contrast, a-amylase and CGTase may be the most closely related enzymes within the a-amylase superfamily. Here, we report the crystal structure of amylomaltase from the thermophilic bacterium T. aquatiÊ resolution. The monomeric cus determined at 2.0 A enzyme contains 500 amino acid residues (molecular mass 57 kDa). Comparison of this novel enzyme structure to the previously characterized structures of other enzymes of the a-amylase superfamily reveals signi®cant differences with respect to the protein topography and active site that might help us to understand the factors governing the reaction speci®city and lead to improved variants for the production of large cyclic glucans, by rational enzyme design.

Results and Discussion Protein fold The structure of amylomaltase from T. aquaticus was determined by the method of multiple isomorphous replacement (MIR) and re®ned to a crystal-

Ê resolution. lographic R-factor of 19.6 % at 2.0 A The ®nal model contains all 500 amino acid residues. Amylomaltase essentially has a one-domain structure, its core consists of a (b, a)8 barrel. Several insertions between strands of the central barrel form subdomains located at the C-terminal edge around the barrel (Figure 2). Only between the ®fth and seventh barrel strand no additional secondary structure elements are inserted. Although the assignment of subdomains might appear somewhat arbitrary, we have chosen to divide the structure into four subdomains (A, B1, B2 and B3), to facilitate comparison to related structures. Insertions between the third and fourth strand (b5 and b8) of the central barrel (subdomain A) and between the fourth and ®fth strand (b8 and b11) form subdomain B1. Subdomain B2 consists of a large insertion between the second and third barrel strand (b3 and b5). The remaining insertions, between the ®rst and second (b1 and b3), between the seventh and eighth (b13 and b14), and after the eighth barrel strand (b14) build subdomain B3. Thus, subdomains B1-B3 form an almost continuous ring around the C-terminal edge of the barrel and might participate in binding the large amylose substrates. Subdomain B1 consists of a highly twisted four-stranded antiparallel b-sheet (strands b6, b7, b9 and b10) which is formed by two strands from each of the two insertions building this subdomain. In addition to the b-sheet, two ahelices (a8 and a9) are present. Helix a9 packs against one side of the sheet, while helix a8 and the loop between this helix and strand b6 interact with subdomain B2. Subdomain B2 has predominantly an a-helical structure. Most of these helices are packed around the N-terminal half of the long helix a7, which forms part of the barrel subdomain A and this subdomain. Helix a5 and the loop between helices a5 and a6 interact with

876

Crystal Structure of Amylomaltase

Figure 2. Fold of amylomaltase from Thermus aquaticus. (a) Topography diagram. b-Strands are indicated by triangles, pointing up if facing towards the viewer. Helices are marked by circles. The ®rst and last residue numbers of each secondary structure element are shown. (Upper numbers mark the residue which is closer to the viewer). The large numbers (1-8) in the center of the barrel refer to the position of the ®rst to eighth barrel strand. The (b, a)8 barrel core structure (subdomain A) is coloured in green, insertions between the ®rst and ®fth strand of the barrel (subdomain B2 and B1) are painted in a gradient going from yellow to red. Additional small insertions are shown in blue (subdomains B3). (b) Ribbon representation (stereo view) of the fold of amylomaltase coloured as in (a) (programs MOLSCRIPT (Kraulis, 1991) and RASTER3D (Meritt & Murphy, 1994)).

the neighbouring subdomain B1. The interface between subdomain B2 and the three helices of subdomain B3 is formed by a short antiparallel three-stranded b-sheet (strands b2, b4 and b15). The strands of this sheet are derived from three different insertions between the ®rst and second (b1 and b3), the second and the third (b3 and b5), and after the eighth barrel strand (b14).

intramolecular glycosyl transfer, and for the product ring size in the case of the cyclization reaction, we compared the crystal structure of amylomaltase with related enzymes of the a-amylase superfamily. In a comparison of the amylomaltase structure against the structures deposited in the Protein Data Bank using the DALI server (Holm & Sander, 1993), different members of the a-amylase superfamily appear as the ten most similar structures (Table 1). b-Glucuronidase and tryptophan synthase, which do not belong to the a-amylase superfamily, show up at the 11th and 12th position with similarity scores below ten compared to 13.5 and higher for the a-amylase superfamily members. Below position ten only proteins

Comparison to related enzymes To characterize the topographical differences in the a-amylase superfamily which might be responsible for the reaction speci®cities, e.g. hydrolysis versus transglycosylation or intermolecular versus

Table 1. Structural similarity of amylomaltase to other enzymes of the a-amylase superfamily Enzymea a-Amylase II (1bvz) Taka-amylase (7taa) Isoamylase (1bf2) Oligo-1,6-glucosidase (1uok) Maltotetrahydrolase (1jdc) CGTase (1cxl) a-Amylase (1bli) a-Amylase (1bag) a-Amylase (1smd) a b c

Organism

Residues

Aligned residues

r.m.s.d.b

Sequence identity (%)

Z-scorec

Thremoactinomyces vulgaris R-47 Aspergillus oryzae Pseudomonas amyloderamosa

585 476 750

298 288 319

2.9 3.4 3.4

13 10 11

20.6 19.5 17.1

Bacillus cereus Pseudomonas stutzeri Bacillus circulans Bacillus licheniformis Bacillus subtilis Homo sapiens

558 418 686 481 425 496

315 268 283 273 275 276

3.2 3.4 3.4 3.4 3.2 3.5

12 11 10 12 17 12

16.3 16.0 15.8 15.7 14.8 13.5

The PDB identi®er is given in parentheses. Root-mean-square deviation of the Ca atoms of the aligned residues. Z-score, i.e. strength of structural similarity in standard deviations above expected (Holm & Sander, 1993).

Crystal Structure of Amylomaltase

which do not belong to the family of a-amylases are found. These proteins have relatively high similarity scores, because they also contain a TIM-barrel. Within the a-amylase superfamily, a-amylase II from Thermoactinomyces vulgaris and Taka-amylase are most similar to amylomaltase, followed by isoamylase, oligo-1,6-glucosidase, 1,4-a maltotetrahydrolase and CGTase. However, all of these structures have similar Z-scores (15.8 to 20.6) and r.m.s. Ê ) for around 270-320 resideviations (2.9 to 3.4 A dues (of a total of 420 to 750 residues), which can be superimposed. The sequence identity is, in agreement with the primary sequence alignments, between 10 and 13 %.

877 The most striking difference between amylomaltase and the other members of the a-amylase superfamily is the absence of a C-terminal domain (domain C, Figures 2 and 3). In all other structures of these enzymes determined so far, a C-terminal domain is present which contains two sandwiched antiparallel b-sheets. Additional domains have been found at the C terminus and at the N terminus in some enzymes. In CGTase two additional C-terminal domains (D and E) are present, which also consist of sandwiched antiparallel b-sheets. The functions of these domains are not clear (Jespersen et al., 1991), however, it has been speculated that domain D is involved in carbohydrate

Figure 3. Comparison of amylomaltase to a-amylase and CGTase. Topography diagrams of (a) pancreatic pig a-amylase and of (b) CGTase from Bacillus circulans. The domain colours are the same as described in the legend to Figure 2. The additional C-terminal domains (C, D and E) are coloured in purple. Note that domains A-C of these two enzymes are related and found in all previously characterized enzymes of the a-amylase superfamily. Domain B varies greatly in sequence and structure. (c) Superposition of the Ca-traces of amylomaltase and CGTase from Bacillus circulans. Amylomaltase is coloured as in Figure 2 and CGTase is shown in black. The domains of CGTase are labeled. The two structures have been superimposed using the DALI server (Holm & Sander, 1993).

878 binding, whereas domain E is believed to be important for attaching the enzyme to raw starch granules (Svensson et al., 1989; Villette et al., 1992). Additional N-terminal domains have been found in isoamylase (Katsuya et al., 1998) and in a-amylase II from T. vulgaris (Kamitori et al., 1999). Both domains consist of anti-parallel b-sheets, but appear not to be related. Domain B has been found to vary strongly in both length and sequence and is often hardly alignable. The insertion between the third and fourth barrel strands (b5 and b8) forming the main part of subdomain B1 in amylomaltase is related to this domain B. Similar to the situation in the previously determined structures of a-amylases, this subdomain in amylomaltase also shows no close alignment to any of the known structures. Furthermore, the two additional antiparallel b-strands (b9 and b10) of this subdomain have not been observed in related structures. Another remarkable feature of the amylomaltase structure in comparison to the related family-13 enzymes is the presence of subdomain B2. Inser-

Crystal Structure of Amylomaltase

tions of longer loops or secondary structure elements between the second and third barrel strands have been previously observed in other family members, e.g. in pig pancreatic a-amylase (Qian et al., 1993) and Bacillus circulans CGTase (Klein & Schulz, 1991) (Figure 3(a) and (b)). These insertions often interact with domain B. However, the large a-helical subdomain B2 is unique to amylomaltase. In a superposition, this subdomain has no overlap with domains of the other enzymes, except for domain N of isoamylase (Katsuya et al., 1998), which partially overlaps (data not shown). Thus, subdomain B2 may have a unique function in amylomaltase. To characterize topographical differences which might be responsible for the reaction speci®city and for the product ring size, we compared the amylomaltase structure to those of an a-amylase and a CGTase for which enzyme-substrate complexes have been structurally characterized: pig pancreatic a-amylase isozyme II bound to a linear maltooligosaccharide derivative (Machius et al., 1996) and a mutant CGTase bound to a b-cyclo-

Figure 4. Molecular surfaces (program GRASP (Nicholls et al., 1991)) of (a) a-amylase isozyme II of porcine pancreas in complex with a maltohexaose (part of a larger inhibitor) (Machius et al., 1996), (b) CGTase from Bacillus circulans strain 8 complexed with b-cyclodextrin (Schmidt et al., 1998) and (c), (d) amylomaltase from Thermus aquaticus with a modeled binding mode of a maltohexaose. The surfaces are coloured according to the distance to the center of mass. The domains and subdomains are labeled. The bound or modeled inhibitors are shown in green. Possible binding paths for a cycloamylose product to amylomaltase are indicated as broken green lines in (c) and (d). The model for the maltohexaose binding mode was obtained by superimposing the inhibitor bound to porcine a-amylase isozyme II onto the active site of amylomaltase. Also labeled are the locations of two loops, the 250s loop and the 460s loop, which might prevent the formation of small cyclic products by steric hindrance. The active center of amylomaltase is located at the center of the modeled oligosaccharide in (c) and (d). The three enzymes shown in (a)-(c) have approximately the same orientation as in the superposition shown in Figure 3(c).

879

Crystal Structure of Amylomaltase

dextrin derivative (Schmidt et al., 1998) (Figures 3 and 4). The a-amylase has a relatively open active site cleft, which is formed by residues of domains A and B (Figure 4(a)). In CGTase, domain B also forms part of the active site pocket (Figure 4(b)). In a co-crystal structure with a b-cyclodextrin derivative, a tyrosine or phenylalanine side-chain (e.g. Tyr195 in CGTase from Bacillus circulans strain 8) from domain B is oriented nearly perpendicular to the plane de®ned by the cyclodextrin ring (Figure 4(b)). This hydrophobic side-chain has been shown to affect the cyclization reaction and it has been proposed to favour the synthesis of cyclodextrins by forming a non-polar core around which the a-glucan could wrap (Nakamura et al., 1994; Penninga et al., 1995). The tyrosine or phenylalanine residue is part of domain B and is replaced by a glycine, serine, or valine residue in a-amylases. A tryptophan residue (Trp258) is near this residue in amylomaltase, which might have a similar function. In amylomaltase the active-site cleft is partially covered by a long extended loop (250s loop) formed by residues 247-255 between strands b6 and b7 of subdomain B1 (Figures 2(a), 4(c) and (d)). Two hydrophobic side-chains, Tyr250 and Phe251, are located at the tip of the loop and point towards subdomain B3. This loop has only weak electron density (Figure 5) and re®ned to B-factors Ê 2, which is signi®cantly higher than around 60-70 A the average B-value of the amylomaltase molecule Ê 2). This indicates a ¯exibility of the loop (31.9 A conformation, which might be important for binding of substrates and dissociation of products. It is also clear that the formation of small cylic products like cyclodextrins is sterically hindered by the presence of this loop near the active site. If the cyloamylose product wraps around the 250s loop during the cyclization reaction, the minimum ring size might be restricted to about 18 residues by the size of this loop (Figure 4(d)).

Figure 5. Omit electron density maps. (a) Density of the 250s loop, contoured at 4.5 srms. A break in the main-chain density and weak side-chain density for Tyr250 and Phe251 indicate ¯exibility of this loop which might be important for the catalytic function. (b) Density of selected active site residues, contoured at 5.5 srms. In addition to the three catalytic acidic side-chains, Trp258 is shown.

On the other side of the active site groove the loop between a17 and b15 (460s loop) and Tyr54 derived from the loop between b3 and a2 might restrict the formation of smaller cyclic products (Figure 4(c)). In addition to residues Tyr250, Phe251 and Tyr54, the hydrophobic side-chains of Tyr101 and Tyr465 are solvent exposed and located near the catalytic cleft along an alternative glucan binding groove indicated in Figure 4(c). The path indicated in Figure 4(c) has a length of about Ê . Larger CA consisting of n glucose units 110 A which form a planar ring have a length of 4.6n Ê ) and a radius of 0.73n (A Ê ), assuming that the (A distance of two neighbouring glucose units is Ê . Therefore, an extended CA22 ring has about 4.6 A Ê and a length of 100 A Ê. a radius of about 16 A These side-chains may be involved in stacking interactions with the hydrophobic face of the glucan rings of the substrate. Active site structure Superimposition of the active sites of a-amylase, CGTase and amylomaltase reveals interesting similarities as well as differences in the active site structure (Figure 6). The catalytic residues Asp293, Glu340 and Asp395 are well de®ned in the electron density maps (Figure 5(b)) and show a similar location and orientation as in the related enzymes, although the side-chains do not have identical conformers. However, the side-chain conformation might change upon formation of the Michaelis complex. The presence of these core catalytic residues supports a similar reaction mechanism for amylomaltase and the other enzymes of this family, as previously indicated by homologous signature motifs of the amino acid sequences. In a mechanism involving a covalent intermediate, Glu340 protonates the glycosidic oxygen atom of the scissile bond and a planar oxocarbenium-like transition state is formed. The Asp293 is the nucleophile which attacks the C1 atom of the substrate under formation of the covalent intermediate. The Asp395 presumably exerts strain on the substrate in the Michaelis complex and speci®cally stabilizes the planar oxocarbenium-like transition state (Uitdehaag et al., 1999). Obviously, the environment of the three acidic residues plays an important role in governing reaction speci®city. Four additional amino acid residues are conserved and have a similar orientation in the three aligned structures: Tyr59, Asp213, Arg291 and His394. The tyrosine residue helps to orientate the sugar in the catalytic subsite ÿ1 by forming a stacking interaction with the hexose ring, while His394 and Arg291 interact with the O2-atom of the substrate hexose in the ÿ1 subsite. Asp213 is part of a hydrogen bonding network that shows some ¯exibility in the substrate-bound and intermediate structures (Uitdehaag et al., 1999). In some a-amylases and CGTases this aspartate residue is ligand to a calcium ion. The absence of a calcium ion in the amylomaltase structure is in

880

Crystal Structure of Amylomaltase

Figure 6. Active site structure (programs MOLSCRIPT (Kraulis, 1991) and RASTER3D (Meritt & Murphy, 1994)). (a) Catalytic center of amylomaltase with a modeled maltose (green). The model for maltose was obtained as described for Figure 4. (b) Superimposition of the active centers of amylomaltase, a-amylase from pig pancreas (Machius et al., 1996) and CGTase from Bacillus circulans strain 8 (Schmidt et al., 1998). The three native structures were superimposed by their (b, a)8 barrel core structure. The a-amylase is in blue, CGTase in red and amylomaltase in green. The core catalytic residues are the acidic side-chains Asp293, Glu340 and Asp395 and residues Tyr59, Asp213, Arg291 and His394, which are strictly conserved among the family-13 enzymes.

agreement with biochemical data, which suggest that the activity of this enzyme is not in¯uenced by metal ions (T.T. et al., unpublished results). Divalent metal ions were not present during the isolation or crystallization procedure. The structural superposition shows that Asp213 is the conserved aspartate residue of region 1 (Figure 7), in contrast to a previous alignment in which region 1 was misaligned due to weak sequence homology (Janecek, 1997; Takaha & Smith, 1999). These seven residues build up the core of the catalytic cleft. Except for Tyr59, they are part of the four conserved motifs. The sequence homology is too weak to detect if Tyr59 is conserved in all a-amylase superfamily members, as are the other six residues. These seven residues probably form the minimum necessary active site environment for the family-13 enzymes.

In addition to the seven residues discussed above, active site residues Trp258, His294, Leu342 and Asn464 are also highly conserved within the currently known sequences of amylomaltases and the related plant D-enzymes (Figure 7), but not among the other family-13 enzymes (Figure 6). If we compare these residues to the other members of the a-amylase superfamily, an interesting difference is seen in Trp258 of subdomain B1, as it has no direct counterpart in CGTase and a-amylase. This hydrophobic side-chain might be important for the formation of cyclic products, similar to the role of Tyr195 in CGTase, which is also derived from domain B. Indeed, the Ca atom of the tyrosine of CGTase (not shown in Figure 6) is near the Ca of Trp258 of amylomaltase in the superposition, although the side-chains point in differ-

Crystal Structure of Amylomaltase

881

Figure 7. Sequence alignment generated with CLUSTAL W 1.74 (Thompson et al., 1994) for selected amylomaltases and a plant D-enzyme (potato). The residue numbers refer to the amino acid sequence from Thermus aquaticus. Secondary structure elements are derived from the X-ray structure of amylomaltase and coloured according to the domain structure shown in Figure 2. Homologous residues are shown in yellow and identical residues in orange. Catalytic side-chains are marked in red. The four conserved regions are marked by black triangles. This Figure was generated using ALSCRIPT (Barton, 1993). In addition to the sequences shown here, homologous genes have been found in other microorganisms and plants (Takaha & Smith, 1999).

ent directions. However, this region does not superimpose well enough to suggest that Trp258 and Tyr195 are homologoues residues. Substitution of Tyr195 in CGTase in¯uences the product ring size of the cyclodextrins (Nakamura et al., 1994; Penninga et al., 1995; Sin et al., 1994). His294 of amylomaltase is not conserved in the related a-amylase enzymes; however, the nearby His233 of CGTase (201 of a-amylase) might have a similar function (Figure 6). Leu342 of amylomaltase is related to Phe259 in CGTase and Ile235 in a-amylase, respectively. These residues form part of the subsite ‡1 and may in¯uence the reaction speci®city (hydrolysis versus transglycosylation), as it was shown that a change in hydrophobicity of the entrance path for a water molecule in¯uences the hydrolytic

activity and the equilibrium between hydrolysis and transglycosylation (Kuriki et al., 1996; Matsui et al., 1991; Nakamura et al., 1994). Asn464 might form contacts to glucose residues in subsite ÿ1 or ÿ2. This position is occupied by Arg375 in CGTase (Strokopytov et al., 1996) and by Trp58 in a-amylase (Qian et al., 1994; Machius et al., 1996). These three residues are derived from non-homologous sequence positions. Arg375 in CGTase is hydrogen-bonded to O2 and O3 of the glucose in subsite ÿ2, which could also be the function of Asn464 in amylomaltase. However, Trp58 in a-amylase does not interact with the substrate. Phe217 of amylomaltase is related to a strongly conserved histidine side-chain in CGTase (His140) and a-amylase (His101) in region 1. In a substrate

882 complex of CGTase, His140 binds to O6 of the hexose in subsite ÿ1, whereas this residue has lost its contact to the sugar in a complex with a trapped covalently bound reaction intermediate (Uitdehaag et al., 1999). However, Phe217 has a different conformation in amylomaltase and it is not conserved among other amylomaltases and plant D-enzymes. Thus, this residue probably has no critical catalytic function in amylomaltase. In addition, the residue corresponding to the conserved asparagine residue (Asn139 in CGTase) before His140 is deleted in amylomaltase. Not shown in Figure 6 are residues Ser57, Pro58, Asp341, Gly343, Thr393 and Pro466, which are conserved among other amylomaltases (Figure 7) and form part of the cleft around the active center. In addition to the discussed residues and the four regions that are conserved among all a-amylase superfamily enzymes, the residues that build up the 250s loop form an even higher conserved stretch (Figure 7). Eight out of 17 residues are conserved of the 250s loop and the ¯anking b-strands b6 and b7. The conserved glycine (Gly245, Gly255) and proline (Pro247, Pro248, Pro261) residues are probably important for the shape and conformation of the loop. Interestingly, Tyr250 and Phe251, which are highly solvent exposed at the tip of the loop, are replaced by other hydrophobic side-chains. This ®nding con®rms the view that these residues could have a functional role. Gln256 points toward the active-site cleft and might be involved in substrate binding. Asp249 is solvent exposed and points away from the presumed substrate binding groove. Although this residue does not appear to have a structural role, it is conserved in all known amylomaltase sequences. Another conserved residue is Trp302 of subdomain B1 which supports the 250s loop conformation by stacking with Pro247. Trp258 is located close to the active center, as described above. In summary, this structure analysis con®rms that amylomaltase differs in several points from other members of the a-amylase superfamily, most notably in the presence of the a-helical subdomain B2 and in the absence of a C-terminal (sub)domain. Differences are also seen in the active-site structure, although the three catalytic acidic residues Asp293, Glu340 and Asp395 are conserved. Nevertheless, the enzyme satis®es all requirements for being a member of the a-amylase superfamily (Takata et al., 1992). It acts on a-glucosidic linkages by hydrolysis or transglycosylation using a retaining mechanism. The conserved active site amino acids are located within the four conserved regions of its primary sequence. The core structure of the enzyme is a (b, a)8 barrel. The structure of amylomaltase reported here forms the basis for future biochemical studies, especially on mutant variants, as well as for further structural studies on enzyme-inhibitor complexes to characterise the enzyme function. Knowledge of

Crystal Structure of Amylomaltase

the enzyme structure also enables a rational protein design in order to enhance the cyclisation yield and in¯uence the minimum product ring size.

Materials and Methods Protein expression and purification Amylomaltase was expressed in E. coli strain MC1062 carrying the recombinant plasmid pFQG8 (Terada et al., 1999). For crystallization the protein was puri®ed as described by Terada et al. (1999), with the following modi®cation. After hydrophobic interaction chromatography the enzyme was dialyzed against 20 mM Tris-HCl (pH 7.2). Only one single band remained in SDS-PAGE. However, after application to high-resolution, anionexchange chromatography (HR 10/10 MonoQ, Amersham Pharmacia Biotech, Uppsala, Schweden) with a smooth linear gradient from 2 mM to 300 mM sodium chloride, the enzyme eluted as three separate peaks. All three peaks have amylomaltase activity, but contain proteins with different pI values. Only the ®rst (major) peak was pooled, dialyzed against 5 mM Tris-HCl (pH 7.6), 1 mM DTT, 100 mM NaCl and concentrated to 10 mg/ ml for crystallization experiments. Crystals obtained from a preparation of the pooled fractions of the second peak revealed no signi®cant differences in the protein structure.

Crystallization and heavy-atom derivatives The enzyme was crystallized using the hanging drop vapour-diffusion method. The best crystals were obtained within three weeks at 291 K by mixing equal volumes of protein and reservoir solution containing 17 % (w/v) PEG 8000, 100 mM Hepes (pH 7.5) and 10 % (v/v) ethylene glycol. For cryo data collection crystals were transfered to mother liquor containing 30 % ethylene glycol, by stepwise increase of the cryoprotectant. Crystals belong to space group P64 with cell parameters Ê and c ˆ 64 A Ê. a ˆ b ˆ 154 A Six heavy-atom derivatives were obtained by soaking the crystals in mother liquor containing 20 mM PCMBS (four days), 1 mM HgCl2 (two days), 5 mM K2PtCl4 (two days), 5 mM KAu(CN)2, (four days) 1 mM K2Pt(SCN)6 (two days) and 5 mM (CH3)3PbAc (seven days).

Data collection High-resolution native data were collected at 100 K at the BW7a beamline of DESY/EMBL (Hamburg, Germany) using cryocrystallographic techniques. The native data set used for phasing and derivative data sets were collected at room temperature using Cu-Ka radiation from a rotating anode source. In all cases, the data were measured using a MarResearch imaging plate detector and processed with Denzo/Scalepack (Otwinowski, 1993) (Table 2). The high-resolution data set of the native enzyme, which was used for re®nement of the structure, has a redundancy of 3.4. The I/sI ratio is 15.1 for the whole data set and 87.5 % of the re¯ections have an I/sI > 1 (72.9 % for the outer shell).

883

Crystal Structure of Amylomaltase Table 2. Data collection and phasing statistics for native and derivative data sets Data collection Temperature (K) Max. resolution Ê) (A Collected reflections Unique reflections Completenessa (%) Rsym a,b (%) Riso c (%)

Native-1

Native-2

PCMBS

HgCl2

K2PtCl4

KAu(CN)2

K2Pt(SCN)6

(CH3)3PbAc

100

298

298

298

298

298

298

298

2.0

2.8

3.5

3.7

3.5

3.0

3.5

3.5

347,606

100,007

57,818

34,439

53,559

98,961

44,573

51,148

56277

21895

9293

9435

10607

17354

9568

7572

94.0 (93.8) 7.4 (33.7)

94.4 (85.6) 9.3 (46.0)

80.4 (83.8) 10.6 (22.6) 17.1

92.1 (95.9) 7.7 (13.4) 16.2

88.6 (90.7) 7.7 (16.0) 13.3

92.8 (96.9) 9.9 (46.2) 14.0

80.2 (81.3) 10.0 (23.0) 11.5

62.6 (65.0) 13.9 (33.1) 20.0

1 0.87/0.88

1 0.84/0.87

1 0.84/0.87

1 0.86/0.89

1 0.93/0.93

1 0.86/0.90

Phasing to 3.0 AÊ Number of sites Phasing powerd Combined FOMe

0.49 Ê ). In parentheses are the values for the outer resolution shell (2.07-2.00 A b Rsym ˆ jIobs ÿ hIij/hIi. c Riso ˆ jF2ph ÿ F2pj/(F2ph ‡ F2p). d Phasing power ˆ r.m.s. f/Eiso, root-mean-square heavy atom F/lack of closure error. The ®rst number refers to the centric re¯ections and the second number to the acentric re¯ections. e FOM, Figure of merit. FOM ˆ hcos(ah)i, where ah is the error in the phase angle for re¯ection h. a

MIR phasing MIR phases were derived from the heavy-atom derivatives in combination with the room temperature native data set (native-2). Examination of the derivative isomorphous Patterson map using the CCP4 package (CCP4, 1994) revealed a single heavy-atom site for each derivative. These sites were re®ned and phases were calculated with the program SHARP (De La Fortelle & Bricogne, 1997). Phases were further improved by solvent ¯attening using SOLOMON (CCP4, 1994) (Table 2). The resulting map was of good quality, allowing us to build an initial model with O (Jones et al., 1991). Crystallographic refinement After initial re®nement against the room temperature data set (native-2), re®nement was continued using the higher resolution cryo data set (native-1). The re®nement procedure included simulated anneal-

Table 3. Crystal parameters and re®nement statistics Space group Ê) Cell dimensions (A Protein molecules per AUa Solvent content (%) s cut-off Ê) Resolution range (A R/Rfree (%) Reflections in working/test set Number of protein atoms Number of water molecules Ê 2)i hB protein (A Ê 2)i hB solvent (A Ê) r.m.s.d. bond length (A r.m.s.d. angles (deg.) a

AU, asymmetric unit.

P64 a ˆ b ˆ 155.7, c ˆ 64.2 1 65 Not applied 40-2.0 19.6/22.6 53,416/2848 4065 739 31.9 48.9 0.006 1.22

ing and conjugate gradient energy minimization against maximum likelihood targets as implemented in the program CNS (BruÈnger et al., 1998). A bulk solvent correction was applied. The ®nal model comprised all 500 amino acid residues and 739 water molecules. Water molecules were picked automatically from Fo ÿ Fc difference electron density maps and checked manually at a graphics station. Individual Bfactors were re®ned for all non-hydrogen atoms. The ®nal R (Rfree) value was 19.6 % (22.6 %) (Table 3) and Ê was estimated by a Luza coordinate error of 0.23 A zati plot: 90.6 % of all amino acid residues are in the most favoured regions and 1.0 % in the generously allowed (Asp370, Glu253) or disallowed regions (Phe251, Ser252) (Program PROCHECK (Laskowski et al., 1993)). Protein Data Bank accession code The atomic coordinates (1CWY) and structure factors have been deposited in the RCSB Protein Data Bank.

Acknowledgments We thank W. Rypniewski, V. Lamzin and S. Popov for help with data collection at the EMBL beamlines at DESY, Hamburg. This work is supported by a grant of the Deutsche Forschungsgemeinschaft to N.S. and by a grant for the development of the next generation of bioreactor systems from the Society for Techno-Innovation of Agriculture, Forestry and Fisheries (STAFF), Japan, to Ezaki Glico Co.

References Barton, G. J. (1993). ALSCRIPT: a tool to format multiple sequence alignments. Protein Eng. 6, 37-40.

884 Boel, E., Brady, L., Brzozowski, A. M., Derewenda, Z., Dodson, G. G., Jensen, V. J., Petersen, S. B., Swift, H., Thim, L. & WoÈldike, H. F. (1990). Calcium binding in a-amylase: an X-ray diffraction study at Ê resolution of two enzymes from Aspergillus. 2.1 A Biochemistry, 29, 6244-6249. Brady, R. L., Brzozowski, A. M., Derewenda, Z. S., Dodson, E. J. & Dodson, G. G. (1991). Solution of the structure of Aspergillus niger acid a-amylase by combined molecular replacement and multiple isomorphous replacement methods. Acta Crystallog. sect. B, 47, 527-535. Brayer, G. D., Luo, Y. & Withers, S. G. (1995). The strucÊ resolture of human pancreatic a-amylase at 1.8 A ution and comparisons with related enzymes. Protein Sci. 4, 1730-1742. BruÈnger, A. T., Adams, P. D., Clore, G. M., Delano, W. L., Gros, P. O., Grose-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T. & Waren, G. L. (1998). Crystallography & NMR System: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905-921. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760-767. De La Fortelle, E. & Bricogne, G. (1997). Maximumlikelihood heavy-atom parameter re®nement for multiple isomorphous replacement and multiwavelength anomalous diffraction methods. Methods Enzymol. 276, 472-494. Desseaux, V., Payan, F., Ajandouz, E. H., Svensson, B., Haser, R. & Marchis-Mouren, G. (1991). Effect of limited proteolysis in the eighth loop of the barrel and of antibodies on porcine pancreas amylase activity. Biochim. Biophys. Acta, 1080, 237-244. Fujimoto, Z., Takase, K., Doui, N., Momma, M., Matsumoto, T. & Mizuno, H. (1998). Crystal structure of a catalytic-site mutant a-amylase from Bacillus subtilis complexed with maltopentaose. J. Mol. Biol. 277, 393-407. Gessler, K., UsoÂn, I., Takaha, T., Krauss, N., Smith, S. M., Okada, S., Sheldrick, G. M. & Saenger, W. (1999). V-Amylose at atomic resolution: X-ray structure of a cycloamylose with 26 glucose residues (cyclomaltohexaicosaose). Proc. Natl Acad. Sci. USA, 96, 42464251. Harata, K., Haga, K., Nakamura, A., Aoyagi, M. & Yamane, K. (1996). X-ray structure of cyclodextrin glucanotransferase from alkalophilic Bacillus Sp. 1011. Comparison of two independent molecules at Ê resolution. Acta Crystallog. sect. D, 52, 11361.8 A 1145. Hedges, A. R. (1998). Industrial applications of cyclodextrins. Chem. Rev. 98, 2035-2044. Henrissat, B. (1991). A classi®cation of glycosyl hydrolases based on amino acid sequence similarities. Biochem. J. 280, 309-316. Henrissat, B. & Bairoch, A. (1996). Updating the sequence-based classi®cation of glycosyl hydrolases. Biochem. J. 316, 695-696. Holm, L. & Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123-138. Holm, L., Koivula, A. K., Lehtovaara, P. M., Hemminki, A. & Knowles, J. K. (1990). Random mutagenesis used to probe the structure and function of Bacillus stearothermophilus a-amylase. Protein Eng. 3, 181-191.

Crystal Structure of Amylomaltase International Union of Biochemistry Molecular Biology (IUBMB) (1992). Enzyme Nomenclature: Recommendations of the Nomenclature Committee of the International Union of Biochemistry and Molecular Biology, Academic Press, San Diego. Janecek, SÏ. (1995). Close evolutionary relatedness among functionally distantly related members of the (a, b)8-barrel glycosyl hydrolases suggested by the similarity of their ®fth conserved sequence region. FEBS Letters, 377, 6-8. Janecek, SÏ. (1997). a-Amylase family: molecular biology and evolution. Prog. Biophys. Mol. Biol. 67, 67-97. Janecek, SÏ. , Svensson, B. & Henrissat, B. (1997). Domain evolution in the a-amylase family. J. Mol. Evol. 45, 322-331. Janecek, SÏ., MacGregor, E. A. & Svensson, B. (1999). ALAMY: an ALpha-AMYlase database for sequences, structures and evolution of proteins from the a-amylase family. In Third Carbohydrate Bioengineering Meeting 11-14 April 1999, Newcastle upon Tyne, UK. Jespersen, H. M., MacGregor, E. A., Sierks, M. R. & Svensson, B. (1991). Comparison of the domainlevel organization of starch hydrolases and related enzymes. Biochem. J. 280, 51-55. Jesperson, H. M., MacGregor, E. A., Henrissat, B., Sierks, M. R. & Svensson, B. (1993). Starch- and glycogen-debranching and branching enzymes: prediction of structural features of the catalytic (b, a)8-barrel domain and evolutionary relationship to other amylolytic enzymes. J. Protein Chem. 12, 791-805. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for the building of protein models. In Crystallographic Computing (Moras, D., Podany, A. D. & Thierry, J. C., eds), pp. 413432, Oxford University Press, Oxford. Kadziola, A., Abe, J., Svensson, B. & Haser, R. (1994). Crystal and molecular structure of barley a-amylase. J. Mol. Biol. 239, 104-121. Kamitori, S., Kondo, S., Okuyama, K., Yokota, T., Shimura, Y., Tonozuka, T. & Sakano, Y. (1999). Crystal structure of Thermoactinomyces vulgaris R-47 a-amylase II (TVAII) hydrolysing cyclodextrins and Ê resolution. J. Mol. Biol. 287, 907pullulan at 2.6 A 921. Katsuya, Y., Mezaki, Y., Kubota, M. & Matsuura, Y. (1998). Three-dimensional structure of Pseudomonas Ê resolution. J. Mol. Biol. 281, 885isoamylase at 2.2 A 897. Kizaki, H., Hata, Y., Watanabe, K., Katsube, Y. & Suzuki, Y. (1993). Polypeptide folding of Bacillus cereus ATCC7064 oligo-1,6-glucosidase revealed by Ê resolution X-ray analysis. J. Biochem. 113, 6463.0 A 649. Klein, C. & Schulz, G. E. (1991). Structure of cyclodexÊ resolution. trin glycosyltransferase re®ned at 2.0 A J. Mol. Biol. 217, 737-750. Knegtel, R. M., Wind, R. D., Rozeboom, H. J., Kalk, K. H., Buitelaar, R. M., Dijkhuizen, L. & Dijkstra, B. W. Ê resolution and (1996). Crystal structure at 2.3 A revised nucleotide sequence of the thermostable cyclodextrin glycosyltransferase from Thermoanaerobacterium thermosulfurigenes EM1. J. Mol. Biol. 256, 611-622. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallog. 24, 946-950.

Crystal Structure of Amylomaltase Kubota, M., Matsuura, Y., Sakai, S. & Katsube, Y. (1991). Molecular structure of B. stearothermophilus cyclodextrin glucanotransferase and analysis of substrate binding site. Denpun Kagaku, 38, 141-146. Kuriki, T. & Imanaka, T. (1999). The concept of the aamylase family: structural similarity and common catalytic mechanism. J. Biosci. Bioeng. 87, 557-565. Kuriki, T., Kaneko, H., Yanase, M., Takata, H., Shimada, J., Handa, S., Takada, T., Umeyama, H. & Okada, S. (1996). Controlling substrate preference and transglycosylation activity of neopullulanase by manipulating steric constraint and hydrophobicity in active center. J. Biol. Chem. 271, 17321-17329. Larson, S. B., Greenwood, A., Cascio, D., Day, J. & McPherson, A. (1994). Re®ned molecular structure Ê resolution. of pig pancreatic a-amylase at 2.1 A J. Mol. Biol. 235, 1560-1584. Laskowski, R. A., MacArthur, M. W., Moss, D. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283-291. Lawson, C. L., van Montfort, R., Strokopytov, B., Rozeboom, H. J., Kalk, K. H., de Vries, G. E., Penninga, D., Dijkhuizen, L. & Dijkstra, B. W. (1994). Nucleotide sequence and X-ray structure of cyclodextrin glycosyltransferase from Bacillus circulans strain 251 in a maltose-dependent crystal form. J. Mol. Biol. 236, 590-600. Machius, M., Wiegand, G. & Huber, R. (1995). Crystal structure of calcium-depleted Bacillus licheniformis aÊ resolution. J. Mol. Biol. 246, 545amylase at 2.2 A 559. Machius, M., VeÂrtesey, L., Huber, R. & Wiegand, G. (1996). Carbohydrate and protein-based inhibitors of porcine pancreatic a-amylase: structure analysis and comparison of their binding characteristics. J. Mol. Biol. 260, 409-421. Matsui, I., Ishikawa, K., Miyairi, S., Fukui, S. & Honda, K. (1991). An increase in the transglycosylation activity of Saccharomycopsis a-amylase altered by site-directed mutagenesis. Biochim. Biophys. Acta, 1077, 416-419. Matsuura, Y., Kusunoki, M., Harada, W. & Kakudo, M. (1984). Structure and possible catalytic residues of Taka-amylase A. J. Biochem. 95, 697-702. Meritt, E. A. & Murphy, M. E. P. (1994). Raster 3D Version 2.0. A program for photorealistic molecular graphics. Acta Crystallog. sect. D, 50, 869-873. Morishita, Y., Hasegawa, K., Matsuura, Y., Katsube, Y., Kubota, M. & Sakai, S. (1997). Crystal structure of a maltotetraose-forming exo-amylase from Pseudomonas stutzeri. J. Mol. Biol. 267, 661-672. Nakamura, A., Haga, K. & Yamane, K. (1994). Four aromatic residues in the active center of cyclodextrin glucanotransferase from alkalophilic Bacillus sp. 1011: effects of replacements on substrate binding and cyclization characteristics. Biochemistry, 33, 9929-9936. Nicholls, A., Sharp, K. A. & Honig, B. (1991). Protein folding and association: insights from the interfacial and thermodynamic properties of hydrocarbons. Proteins: Struct. Funct. Genet. 11, 281-296. Otwinowski, Z. (1993). Oscillation data reduction program. In Proceedings of the CCP4 Study Weekend: Data Collection and Processing (Sawyer, L., Isaacs, N. & Bailey, S., eds), pp. 56-62, SERC, Daresbury Laboratory, Warrington, UK. Penninga, D., Strokopytov, B., Rozeboom, H. J., Lawson, C. L., Dijkstra, B. W., Bergsma, J. & Dijkhuizen, L.

885 (1995). Site-directed mutations in tyrosine 195 of cyclodextrin glycosyltransferase from Bacillus circulans strain 251 affect activity and product speci®city. Biochemistry, 34, 3368-3376. Qian, M., Haser, R. & Payan, F. (1993). Structure and molecular model re®nement of pig pancreatic aÊ resolution. J. Mol. Biol. 231, 785amylase at 2.1 A 799. Qian, M., Haser, R., Buisson, G., Duee, E. & Payan, F. (1994). The active center of a mammalian a-amylase. Structure of the complex of a pancreatic a-amylase with a carbohydrate inhibitor re®ned to Ê resolution. Biochemistry, 33, 6284-6294. 2.2 A Ramasubbu, N., Paloth, V., Luo, Y., Brayer, G. D. & Levine, M. J. (1996). Structure of human salivary Ê resolution: implications for its a-amylase at 1.6 A role in the oral cavity. Acta Crystallog. sect. D, 52, 435-446. Schmidt, A. K., Cottaz, S., Driguez, H. & Schulz, G. E. (1998). Structure of cyclodextrin glycosyltransferase complexed with a derivative of its main product b-cyclodextrin. Biochemistry, 37, 5909-5915. Schwartz, M. (1987). In Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F. D., ed.), pp. 1482-1502, American Society for Microbiology, Washington, DC. Sin, K.-A., Nakamura, A., Masaki, H., Matsuura, Y. & Uozumi, T. (1994). Replacement of an amino acid residue of cyclodextrin glucanotransferase of Bacillus ohbensis doubles the production of g-cyclodextrin. J. Biotechnol. 32, 283-288. Strobl, S. & Glockshuber, R., et al. (1998). Crystal strucÊ ture of yellow meal worm a-amylase at 1.64 A resolution. J. Mol. Biol. 278, 617-628. Strokopytov, B., Knegtel, R. M. A., Penninga, D., Rozeboom, H. J., Kalk, K. H., Dijkhuizen, L. & Dijkstra, B. W. (1996). Structure of cyclodextrin glycosyltransferase complexed with a maltononaose Ê resolution. Implications for proinhibitor at 2.6 A duct speci®city. Biochemistry, 35, 4241-4249. Svensson, B., Jespersen, H. M., Sierks, M. R. & MacGregor, E. A. (1989). Sequence homology between putative raw-starch binding domains from different starch-degrading enzymes. Biochem. J. 264, 309-311. Swift, H. J., Brady, L., Derewenda, Z. S., Dodson, E. J., Dodson, G. G., Turkeberg, J. P. & Wilkinson, A. J. (1991). Structure and molecular model re®nement of Aspergillus oryzae (TAKA) a-amylase: an application of the simulated-annealing method. Acta Crystallog. sect. B, 47, 535-544. Takaha, T. & Smith, S. M. (1999). The function of 4-aglucanotransferases and their use for the production of cyclic glucans. Biotechnol. Genet. Eng. Rev. 16, 257-280. Takaha, T., Yanase, M., Takata, H., Okada, S. & Smith, S. M. (1996). Potato D-enzyme catalyzes the cyclization of amylose to produce cycloamylose, a novel cyclic glucan. J. Biol. Chem. 271, 2902-2908. Takata, H., Kuriki, T., Okada, S., Takesada, Y., Iizuka, M., Minamiura, N. & Imanaka, T. (1992). Action of neopullanase. Neopullulanase catalyzes both hydrolysis and transglycosylation at a-1,4 and a1,6-glucosidic linkages. J. Biol. Chem. 267, 1844718452. Terada, Y., Yanase, M., Takata, H., Takaha, T. & Okada, S. (1997). Cyclodextrins are not the major cyclic a-1,4-glucans produced by the initial action of

886

Crystal Structure of Amylomaltase

cyclodextrin glucanotransferase on amylose. J. Biol. Chem. 272, 15729-15733. Terada, Y., Fuji, K., Takaha, T. & Okada, S. (1999). Thermus aquaticus ATCC33923 amylomaltase gene cloning and expression and enzyme characterization: production of cycloamylose. Appl. Environ. Microbiol. 65, 910-915. Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994). CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-speci®c gap penalties and weight matrix choice. Nucl. Acids Res. 22, 4673-4680. Uitdehaag, J. C. M., Mosi, R., Kalk, K. H., van der Veen, B. A., Dijkhuizen, L., Withers, S. G. & Dijkstra, B. W. (1999). X-ray structures along the reaction pathway of cyclodextrin glycosyltransferase eluci-

date catalysis in the a-amylase family. Nature Struct. Biol. 6, 432-436. Vihinen, M., Peltonen, T., IitiaÈ, A., Suominen, I. & MaÈntsaÈlaÈ, P. (1994). C-terminal truncations of a thermostable Bacillus stearothermophilus a-amylase. Protein Eng. 7, 1255-1259. Villette, J. R., Krzewinski, F. S., Looten, P. J., Sicard, P. J. & Bouquelet, S. J.-L. (1992). Cyclomaltodextrin glucanotransferase from Bacillus circulans E192. IV. Evidence for a raw starch-binding site and its interaction with b-cyclodextrin copolymer. Biotechnol. Appl. Biochem. 16, 57-63. Watanabe, K., Hata, Y., Kizaki, H., Katsube, Y. & Suzuki, Y. (1997). The re®ned crystal structure of Ê resolBacillus cereus oligo-1,6-glucosidase at 2.0 A ution: structural characterization of proline-substitution sites for protein thermostabilization. J. Mol. Biol. 269, 142-153.

Edited by R. Huber (Received 13 October 1999; received in revised form 28 December 1999; accepted 28 December 1999)