Isomaltulose Synthase (PalI) of Klebsiella sp. LX3

Isomaltulose synthase from Klebsiella sp. LX3 (PalI, EC 5.4.99.11) catalyzes the isomerization of sucrose to produce isomaltulose ( -D-glucosylpyranosyl-1,6-Dfructofuranose) and trehalulose ( -D-glucosylpyranosyl1,1-D-fructofuranose). The PalI structure, solved at 2.2-Å resolution with an R-factor of 19.4% and Rfree of 24.2%, consists of three domains: an N-terminal catalytic ( / )8 domain, a subdomain between N 3 and N 3, and a Cterminal domain having seven -strands. The active site architecture of PalI is identical to that of other glycoside hydrolase family 13 members, suggesting a similar mechanism in substrate binding and hydrolysis. However, a unique RLDRD motif in the proximity of the active site has been identified and shown biochemically to be responsible for sucrose isomerization. A two-step reaction mechanism for hydrolysis and isomerization, which occurs in the same pocket is proposed based on both the structural and biochemical data. Selected Cterminal truncations have been shown to reduce and even abolish the enzyme activity, consistent with the predicted role of the C-terminal residues in the maintenance of enzyme conformation and active site topology.

The molecular mechanism of isomaltulose synthase that controls sucrose isomerization has not been fully characterized, except the recent prediction by Veronese and Perlot (2,8), in which sucrose binding, hydrolysis, and isomerization depend on the charges provided by residues in a closed shell.
To understand the mechanism of sucrose isomerization at the molecular level and identify the key amino acids involved in the enzyme reaction, we recently cloned the palI gene encoding isomaltulose synthase (PalI) from the bacterial isolate Klebsiella sp. LX3 (6). Sequence alignment and secondary structure prediction revealed that PalI is a novel member of glycoside hydrolase family 13. Family 13 contains enzymes that act on starch such as ␣-amylase and cyclodextrin glycosyltransferase (CGTase) as well as enzymes specific for the cleavage of other glycosidic linkage such as ␣-1,6and ␣-1,1-bonds (9). The basic structural characteristics of this family of enzymes is that the catalytic core domain contains a (␤/␣) 8 -fold. The potential catalytic triad (Asp 241 , Glu 295 , and Asp 369 ) and two histidine residues (His 145 and His 368 ) in PalI are highly conserved in ␣-amylase and glycosyltransferase (6). These residues, as found in oligo-1,6-glucosidase from Bacillus cereus (Asp 199 , Glu 255 , Asp 329 , His 103 , and His 328 ) (10) and in amylosucrase from Neisseria polysacchareais (Asp 286 , Glu 328 , Asp 393 , His 187 , and His 392 ) (11), form a catalytic pocket that binds the substrate and hydrolyzes the glycosidic bond. The similarity of the active site architecture strongly suggests that PalI adopts the same molecular mechanism for hydrolysis of the glycosidic bond and formation of the glucosyl-enzyme complex. The reaction mechanism occurs via a general acid catalysis, as do those of all glucoside hydrolases (12). In addition, PalI adopts the same mechanistic scheme in the formation of enzyme-substrate intermediate. As shown in Scheme 2, the glycosidic bond is protonated by a proton donor and the anomeric carbon of the glucose moiety is attacked by the nucleophilic acid, simultaneously, leading to the formation of the covalently linked enzyme-substrate intermediate. The glucosyl moiety can be transferred to a water molecule for hydrolysis and to the fructose moiety for sucrose isomers synthesis. In the isomerization step, the structure of fructose determines the balance in the formation of two sucrose isomers, isomaltulose and trehalulose (8).
To understand the specific structural features required for isomerization in PalI, the PalI protein was overexpressed, purified, and crystallized (13). Here we describe, at 2.2-Å resolution, the crystal structure of PalI and the potential function of an RLDRD motif, at the structural level, in determining the mechanism of sucrose hydrolysis and isomerization. The function of the C-terminal domain in the enzymatic activity is also reported. Based on these data, we present a mechanism to explain the sucrose isomerization process. We believe this paper is the first structure-function report on PalI, a sucrose isomerase that converts sucrose to isomaltulose and trehalulose simultaneously.

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Site-directed Mutagenesis-Escherichia coli DH5␣ were used as host cells for plasmid propagation and protein overexpression. Site-directed mutation of palI was performed by using the QuikChange TM site-directed mutagenesis kit (Stratagene). The double-stranded DNA vector pGEK (6) was used as the template, and two synthetic oligonucleotides containing the desired mutation were used as primers. Five mutants, PalI:D241A, PalI:E295A, PalI:D369A, PalI: H145A, and PalI:H368A, were created in which the indicated residues were replaced by Ala. For the C-terminal deletion mutants PalI:⌬587, PalI:⌬572 and PalI:⌬545, a stop codon was separately introduced after the indicated amino acid positions. All point mutations (Table I) were confirmed by DNA sequencing, using the dideoxy chain termination procedure. The digestion of DNA with restriction endonucleases, agarose gel electrophoresis, and transformation of E. coli DH5␣ were carried out according to standard procedures.
Enzyme Purification and Assays-Overexpression and purification of enzymes (PalI and its mutant versions) were carried out as described previously (6). Further purification was performed by gel filtration chromatography with a HiPrep Sephacryl S-200, 16/60 column (Amersham Biosciences) at a flow rate of 0.5 ml min Ϫ1 with a buffer of 150 mM NaCl, 10 mM Hepes, pH 7.5, and 1 mM dithiothreitol. The purified enzymes were further examined by sodium dodecyl sulfate-polyacrylamide (10%) gel electrophoresis (SDS-PAGE). The enzyme activity was assayed as described previously (6). One unit of enzyme is defined as the amount of enzyme required to catalyze the formation of 1 mol of isomaltulose in 1 min under assay conditions. The data presented are the means of three individual experiments.
Crystallization and Data Collection-Crystallization of wild type PalI (residues 29 -598) was performed as described by Li et al. (13), using the hanging drop vapor diffusion method at 22°C. Crystals that were 1.2 mm in length were cut to an optimum size, treated with a cryo-protectant (reservoir solution with 4% increased precipitant and supplemented with 10% glycerol) for 3 min, and flash-cooled in liquid nitrogen. The diffraction data were collected at Spring8, Japan (beamline BL40B2, ADSC Quantum4 CCD detector, Ϫ173°C). The data were indexed, processed, and scaled using the programs DENZO and SCALEPACK (14).
Structure Determination and Refinement-The primary amino acid sequence of PalI shows 47% identity and 65% similarity to that of oligo-1,6-glucosidase (6). Moreover, these two proteins have the highest structural similarity using the DALI server (15). Consequently, the structure of PalI was determined by the method of molecular replacement by the MOLREP program (16) using oligo-1,6-glucosidase as the search model. The asymmetric unit contains one protein molecule. The resulting electron density map was solvent flattened using PHASES (17) and model building was carried out with the help of O (18). Electron density for the first 15 N-terminal amino acids was not observed in the map. The structure was refined with the CNS program suite (19) and the geometry of the molecule was checked with PROCHECK (20). No residue was found to lie in the disallowed region. All drawings were prepared with MOLSCRIPT (21) and RASTER 3D (22) programs. Table  II shows the data and refinement statistics.
Coordinates-The atomic coordinates and structure factors for PalI have been deposited at the Protein Data Bank (accession code 1M53).

RESULTS AND DISCUSSION
Overall Structure-As shown in Fig. 1A, the PalI molecule consists of three domains: the N-terminal catalytic (␤/␣) 8 domain (residues 43-146 and 216 -521, colored blue), a subdomain (residues 147-215, magenta), and the C-terminal domain (residues 522-598, red), as reported for oligo-1,6-glucosidase ( Fig. 1B) and amylosucrase (Fig. 1C). The (␤/␣) 8 domain, as the main body of the structure, is sandwiched between the subdomain and C-terminal domain. It consists of the well characterized (␤/␣) 8 barrel with eight alternating ␤-strands (N␤1-N␤8) and ␣-helices (N␣1-N␣8). This domain contains the residues involved in catalysis and substrate-binding. The cavity contains the acidic residues (Asp 241 , Glu 295 , and Asp 369 : PalI numbering) that are highly conserved in members of ␣-amylase family 13. The active site is shown in Fig. 2A as an example of the quality of the 2F o Ϫ F c electron density. This active site cleft is surrounded by a loop (residues 321-340), forming a pocket with the dimensions of 20 ϫ 20 ϫ 25 Å, large enough to accommodate a sucrose molecule. The overall surface around the active site pocket of PalI is highly negatively charged (data not shown). This highly negative character of the (␤/␣) 8 barrel domain is important for sugar-protein interactions. The subdomain, which is inserted between N␤3 and N␣3, consists of two ␣-helices and three anti-parallel ␤-strands (Fig. 1A) and has no known function either in PalI or in other family 13 members. Only one strong salt bridge (Lys 248 N . . . Asp 211 O␦2 with a distance of 2.6 Å) connects the subdomain and the N-terminal domain. The C-terminal domain is made of two antiparallel ␤-sheets. Five ␤-strands (C␤1, C␤2, C␤3, C␤5, and C␤7) form the larger ␤-sheet, and two strands (C␤4 and C␤6) form the smaller one. The six loop segments that are present between the pairs of adjacent ␤-strands in the C-terminal domain are named as Clp1-Clp6, respectively. A network of salt bridges and hydrogen bonds between the two C-terminal ␤-sheets as well as the N-and C-terminal domains ensure the conformational stability of the structure in general and the active pocket in particular.
Active Site Architecture-Structural alignment results from the DALI server (15) show that the tertiary structures of PalI and oligo-1,6-glucosidase can be aligned in four parts (residues 43-257, 260 -381, 383-566, and 576 -598) with 46.4% sequence identity. A total of 544 C␣ atoms could be superimposed with root mean square deviations in the range 0.46 -1.69 Å. The SCHEME 1. Hydrolysis and isomerization of sucrose as catalyzed by PalI. SCHEME 2. Possible mechanism of sucrose isomerization. Sucrose (top left panel) forms intermediates with PalI. Pathway A shows the hydrolysis of sucrose and pathways B and C show the isomerization of sucrose to form isomaltulose and trehalulose, respectively. alignment between PalI and amylosucrase is very poor, resulting in several shorter fragments. However, superimposition of the five conserved amino acids (His 141 , Asp 241 , Glu 295 , Asp 369 , and His 368 in PalI) in the substrate-binding pocket of PalI, oligo-1,6-glucosidase, and amylosucrase ( Fig. 2B; oligo-1,6-glucosidase data not shown) shows a high degree of structural similarity of the active site architecture with a root mean square deviation of 0.41 Å between PalI and oligo-1,6-glucosidase and 0.89 Å between PalI and amylosucrase. To verify the importance of these conserved residues, the five residues (His 141 , Asp 241 , Glu 295 , Asp 369 , and His 368 ) were replaced individually by Ala, and the created mutant PalI proteins were purified (Table I). With the specific activity of the wild type enzyme (335.3 units mg Ϫ1 ) defined as 100%, the remaining activity of the mutants PalI:D241A, PalI:E295A, PalI:D369A, PalI:H145A, and PalI:H368A is only 0.2, 0, 1.23, 0.65, and 2.68% of the native PalI, respectively, strongly suggesting that these conserved residues are essential for PalI activity.
The substrate recognition scheme and binding sites have been identified in CGTase (23), TAKA-amylase with substrate analogs (24), amylosucrase with D-glucose, and mutated amylosucrase with sucrose (25,26). Structural comparison of PalI with amylosucrase is of particular interest as these two enzymes use sucrose as their sole substrate. The superimposition of the structure of PalI with the complex of amylosucrase and sucrose (Fig. 2B) clearly indicates that the active site pocket in PalI is closely similar with that of amylosucrase and also suitable for containing one sucrose molecule. In PalI, Glu 295 (equivalent to Glu 328 in amylosucrase) acts as the general acid catalyst to protonate the oxygen of the glycosidic linkage for substrate hydrolysis; Asp 241 (Asp 286 in amylosucrase), the attacking nucleophile, forms a bond with C1 to form ␤-glucosylenzyme intermediate, whereas Asp 369 (Asp 393 in amylosucrase) forms hydrogen bonds to O2 and O3. Arg 239 in PalI forms a salt Enzyme activity (units/mg) was assayed as described previously (6). Briefly, 2 g of purified protein was mixed with 400 l of 0.1 M citratephosphate buffer, pH 6.5, containing sucrose (40 g l Ϫ1 ) and incubated at 35°C for 15 min with gentle agitation. One unit of enzyme is defined as the amount of protein that form 1 mol of reducing sugar (with isomaltulose as standard) per min under the conditions specified.  1. Structures of PalI, oligo-1,6-glucosidase, and amylosucrase. All molecules are shown in the same orientation. A, structure of PalI. The N-terminal catalytic (␤/␣) 8 barrel is drawn in blue, the subdomain in magenta, and the C-terminal domain in red. The isomerization region (residues 321-340) in PalI (and the equivalent regions in oligo-1,6-glucosidase and amylosucrase) are drawn in yellow. B, structure of oligo-1,6-glucosidase (10). C, structure of amylosucrase (11). Its extra N-terminal portion is shown in green.
bridge to O␦1 of Asp 241 , which is essential for the correct positioning of the nucleophile. Similarly, His 145 forms a hydrogen bond to O6 and His 368 to O 2 , as is the case for the equivalent residues His 187 and His 392 in amylosucrase (11,25). In addition, a salt bridge between Asp 102 and Arg 456 is formed in PalI. The equivalent salt bridges between Asp 144 and Arg 509 in amylosucrase (25) and Asp 60 and Arg 415 in oligo-1,6-glucosidase (10) have been reported. One notable feature in amylosucrase, when compared with TAKA-amylase, is that the ϩ1 subsite is modified from Lys to Ala 289 (amylosucrase numbering), providing the specificity of amylosucrase for the furanosyl ring of sucrose (11). The residues in PalI at the equivalent positions follow those of amylosucrase, mainly with Ala 244 , implying similar modifications of PalI at the ϩ1 subsite to accommodate sucrose as the major substrate. Evidently, PalI adopts a mechanism similar to that in amylosucrase for sucrose binding, hydrolysis, and formation of covalent intermediate.
A Motif Influencing Sucrose Isomerization-The hydrolysis of sucrose by PalI constitutes only a minor part of the reactions mediating the synthesis of sucrose isoforms (6). Amylosucrase catalyzes the transfer of a D-glucopyranosyl moiety in the active site cleft to an acceptor molecule in a ravine formed by its domain BЈ that plays the pivotal role in transferase reaction (11,26). In PalI, however, breakage of ␣-1,2-linkage in sucrose and formation of ␣-1,6and ␣-1,1-linkages occur in the same pocket. To elucidate the mechanism of isomerization of PalI, the crucial structural features that interact with fructofuranose at the active site cleft and determine the change of fructofuranose to fructopyranose must be determined. This is because the conversion of fructofuranose to fructopyranose is the key step for trehalulose formation (8). Two residues in amylosucrase, Asp 394 and Arg 446 , directly interact with the fructosyl ring of sucrose through hydrogen bonds (25). The equivalent residues in PalI are Asn 370 and Arg 333 , respectively. Amylosucrase and oligo-1,6-glucosidase significantly differ from PalI in that PalI contains a flexible loop region from FIG. 2. Catalytic pocket, isomerization region, and N-C termini interactions in PalI. A, the 2F o Ϫ F c electron density map at the catalytic pocket is drawn at the 2.5 level. The five conserved residues that participate in substrate binding and hydrolysis are labeled in green, and the five residues that are involved in the isomerization of sucrose are labeled in red. B, the superimposition of PalI on the amylosucrase-sucrose complex based on the atoms of the five conserved hydrolysis residues. PalI is blue, amylosucrase is gray, and sucrose is magenta. The residues in amylosucrase that interact with sucrose and their corresponding residues in PalI are shown. Note the deviation of the helix from the substrate in amylosucrase and the ap- Phe 321 to Ser 340 against a more rigid ␣-helical structure in oligo-1,6-glucosidase and amylosucrase (Fig. 2B). The unique 325 RLDRD 329 sequence of PalI is located in this loop adjacent to the active site cleft (Fig. 2, A and B). Not surprisingly, the RLDRD motif is not present in oligo-1,6-glucosidase and amylosucrase, as these two enzymes are functionally different from PalI. Notably, all known isomaltulose synthases contain the RLDRD sequence at equivalent regions (5,27). This specific region has also been identified by sequence alignment (28). The crystal structural analysis further indicates its unique location and possible interactions with the substrate. The role of the charged residues in this motif was investigated by creating the mutant versions of PalI and analyzing the relative amount of glucose, fructose, isomaltulose, and trehalulose synthesized (28). The isomaltulose content is decreased and the trehalulose content is increased in all of the PalI mutants, compared with the native PalI. However, mutation of Asp 327 , Arg 328 , and Asp 329 did not significantly affect the ratio of sucrose hydrolysis to sucrose isomerization activity (Х0.06), but resulted in a 13-25-fold increase in trehalulose production. Only the mutation of residue Arg 325 enhanced both sucrose hydrolysis and trehalulose formation. Evidently, the charge distributions in the isomaltulose synthase motif influence the stability of glucose and fructose binding to the enzyme and ␣-1,1and ␣-1,6glucosidic bond formation. As such, the unique location of the 325 RLDRD 329 motif highlights its importance in the isomerization process and therefore in the control of PalI product specificity.
Mechanistic Implication of Sucrose Isomerization-Comparison of PalI with the amylosucrase-sucrose complex structure should reveal the mechanism of interactions between the side chains of active site residues and the glucosyl or the fructosyl moiety. In PalI, as shown in Scheme 2, A, Glu 295 interacts with a bound sucrose molecule by protonating the glycosidic bond. It, therefore, serves to activate Asp 241 to nucleophilically attack C1 to form the ␤-glucosyl-enzyme intermediate, as described for other members of glycosyl hydrolase family 13 (9,10,11). The residues that most probably interact with glucosyl moiety at Ϫ1 subsite include the conserved active residues, Asp 369 , Arg 239 , His 145 , and His 368 and the salt bridge residues, Asp 102 and Arg 456 . These remote glucose-binding subsites may prevent the release of glucose and thus confer less hydrolase activity on PalI.
The mechanism of sucrose isomerization is speculative, although the biochemical functions of the motif 325 RLDRD 329 provide molecular evidence that the isomerization of sucrose is controlled by the charged residues in the proximity of the active site cleft. Based on our data, we propose that enzyme-bound sucrose interacts with the conserved residues of the active site and the isomaltulose synthase motif. The fructofuranose conformation of the fructose moiety is tightly preserved by the charged residues of the RLDRD motif so that isomaltulose is the main isomer synthesized (Scheme 2, B). Disruption of the charge distribution balance by mutations of the 325 RLDRD 329 motif or the pH changes enhances the tautomerization of fructofuranose to fructopyranose (6), thereby forming the sucrose isomer, trehalulose (Scheme 2, C). Molecular dynamics analysis suggests that the fructose moiety bound to the mutant enzyme displays variable conformation (29). This electrostatic shift in the active site pocket also appears to cause the movement of 6Ј-OH group toward C2Ј to form fructopyranose as well as the rotation of the C1Ј-OH toward C1 of glucosyl ring. Although direct evidence for the proposed mechanism is not yet available, further structural studies of PalI-substrate complex would allow us to present a complete isomerization scheme.
C-terminal Domain-The C-terminal domain of PalI inter-acts with the N-terminal domain by forming salt bridges and hydrogen bonds (Fig. 2C). All hydrogen bonds between the two domains are clustered in two regions. In the first region, residues of N␣6 and the loops formed by residues 398 -401 and 515-521 in the N-terminal domain make hydrogen bonds with residues in C␤1-Clp1-C␤2 segment. The second region involves residues 379 -384 and 497-501 in the N-terminal domain and Clp3 and Clp5 in the C-terminal domain. An interdomain salt bridge (Arg 381 N2 . . . Glu 553 O⑀1), which is not present in oligo-1,6-glucosidase (10), connects Clp3 in the C-terminal domain and the loop after the active site in the N-terminal domain (Fig. 2C).
To understand the influence of the C-terminal domain on PalI functions, we have constructed three C-terminal deletion mutants, PalI:⌬587, PalI:⌬572, and PalI:⌬545, which are truncated after the indicated residue. The truncated PalI mutants were overexpressed, purified, and the remaining activity has been assayed. As shown in Table I, the mutant PalI:⌬587, created by the deletion of Clp6 and C␤7, reduces the enzyme activity to 92.3% of the native PalI. Further deletion of C␤5 and C1p5 in PalI:⌬572 leads to about 90% loss of activity, and deletion of C␤3-C␤7 in PalI:⌬545 completely abolishes the enzyme activity. C-terminal deletions interrupt the interdomain interactions and may cause significant changes in the structure of the N-terminal domain.
The preceding discussion is based on the comparison of the PalI structure with oligo-1,6-glucosidase and amylosucrase, together with biochemical analysis of a series of mutant PalI proteins. The unique RLDRD motif in the loop region participates in sucrose isomerization and thus influences product specificity. The protein-substrate complex structure should allow us to elucidate the real interaction of identified key residues with substrate and the true mechanism of sucrose isomerization.