Structural Advantage of Sugar Beet α-Glucosidase to Stabilize the Michaelis Complex with Long-chain Substrate

Background: Most plant α-glucosidases prefer long-chain substrates. Results: Inhibitory and structural analyses using the unique 4–10-mer inhibitors identified the substrate binding mode of the enzyme far from the active-site pocket. Conclusion: The structure of the substrate-binding subsites was suitable for single helical conformation of amylose. Significance: The enzyme seems to ingeniously use the self-stabilizing property of the substrate to form a stable ES complex. The α-glucosidase from sugar beet (SBG) is an exo-type glycosidase. The enzyme has a pocket-shaped active site, but efficiently hydrolyzes longer maltooligosaccharides and soluble starch due to lower Km and higher kcat/Km for such substrates. To obtain structural insights into the mechanism governing its unique substrate specificity, a series of acarviosyl-maltooligosaccharides was employed for steady-state kinetic and structural analyses. The acarviosyl-maltooligosaccharides have a longer maltooligosaccharide moiety compared with the maltose moiety of acarbose, which is known to be the transition state analog of α-glycosidases. The clear correlation obtained between log Ki of the acarviosyl-maltooligosaccharides and log(Km/kcat) for hydrolysis of maltooligosaccharides suggests that the acarviosyl-maltooligosaccharides are transition state mimics. The crystal structure of the enzyme bound with acarviosyl-maltohexaose reveals that substrate binding at a distance from the active site is maintained largely by van der Waals interactions, with the four glucose residues at the reducing terminus of acarviosyl-maltohexaose retaining a left-handed single-helical conformation, as also observed in cycloamyloses and single helical V-amyloses. The kinetic behavior and structural features suggest that the subsite structure suitable for the stable conformation of amylose lowers the Km for long-chain substrates, which in turn is responsible for higher specificity of the longer substrates.

Glucans of various types are widely distributed in nature. Each type of glucan adopts a unique conformation, which is dependent on the type of glucosidic linkage in the molecule. For the effective degradation of a variety of glucans, various subsites are present in glycoside hydrolases (1). Most endo-type glycoside hydrolases such as ␣-amylase, dextranase, and cellulase contain cleft-shaped subsites (2)(3)(4)(5), which, in ␤-amylase, extend from a pocket-shaped active site (6). Cellobiohydrolase, a cellulose-hydrolyzing enzyme, binds the ␤-1,4-glucan chain within a tunnel-shaped subsite (7). In the case of Coprinopsis cinerea cellobiohydrolase, the conformation of the tunnelshaped subsite has been observed to change from an open to closed conformation in response to substrate binding (8). Such structures of enzyme subsites facilitate the loosening of the packed conformation of glucans via multiple interactions and contribute to the effective degradation of these carbohydrate polymers.
␣-Glucosidase is an exo-type enzyme, which catalyzes the hydrolysis of ␣-glucosidic linkage at the non-reducing termini of substrate molecules. In addition to other exo-type glycosidases, ␣-glucosidase has a pocket-shaped active site. A majority of ␣-glucosidases exhibits preference for disaccharides and trisaccharides as substrates (9,10). In contrast, several ␣-glucosidases belonging to glycoside hydrolase family 31 (GH31) 4 (11) are known to display specificity for substrates with a high degree of polymerization (DP) (12)(13)(14). Among them, the ␣-glucosidase from sugar beet exhibits the highest specificity for long-chain maltooligosaccharides and soluble starch due to low K m and high k cat /K m (15). The crystal structure of SBG in a complex with the pseudo-tetrasaccharide inhibitor, acarbose (AC4), was determined for structural analysis of its substrate specificity (16). The overall structure of SBG was found to be substantially similar to that of the other GH31 ␣-glucosidases of known structure, and comprises a catalytic domain with (␤/␣) 8 -barrel fold, and the N-and C-terminal domains with ␤-sandwich structures. SBG has a pocket-shaped active site, also found in other related ␣-glucosidases (9,10,(17)(18)(19). This pocket is formed by loops that exist between the ␤-strands and the ␣-helices of the catalytic domain as well as a long loop (des- ignated as N-loop) that protrudes from the N-terminal domain. The loops that follow the third and fourth ␤-strands of the catalytic domain contain short insertions named subdomains b1 and b2, respectively. The structural study (16) demonstrated that Phe 236 and Asn 237 on the N-loop play a role in the specificity of SBG for long-chain substrates. These residues are involved in substrate binding at subsites ϩ2 and ϩ3, and direct the reducing end of the substrate toward subdomain b2, where Ser 497 is poised to bind the substrate at subsite ϩ4 (16,20). The aforementioned studies provided important clues toward understanding the reason behind the specificities of SBG and the other GH31 enzymes for longer substrates; however, a complete understanding is yet to be achieved, particularly because of the binding of substrates at subsites remote from the active site. In the present study, to comprehend the binding of long-chain substrates to SBG, a series of unique long-chain inhibitors, acarviosyl-maltooligosaccharides (AC5-AC10, where the numeral represents DP), was employed as the ligands for structural resolution of the resulting complexes. The acarviosyl-maltooligosaccharides have longer maltooligosaccharide parts than the maltose unit of AC4 (Fig. 1), and were synthesized from commercially available AC4 and maltooligosaccharides with a disproportionating enzyme (21); this methodology resulted in the conversion of more than half of the AC4 into acarviosyl-maltooligosaccharides.
In the current study, the potential of the various acarviosylmaltooligosaccharides for the inhibition of SBG was evaluated, and the crystal structures of SBG bound with AC5ϪAC8 were determined. The structures of the complexes of SBG with acarviosyl-maltooligosaccharides elucidated the mechanism of substrate binding at subsites remote from the active site pocket. The specificity of SBG for long-chain substrates is likely due to the structure of remote subsites, which accommodate the stable single-helical conformation of long-chain amylose.

EXPERIMENTAL PROCEDURES
Materials-The substrates for kinetic analysis included a series of maltooligosaccharides ranging from maltose (Glc 2 ) to maltoheptaose (Glc 7 ) (Nihon Shokuhin Kako, Tokyo, Japan), amylose EX-I (Glc 18 , average DP of 18; Hayashibara, Okayama, Japan), and soluble starch (Nacalai Tesque, Kyoto, Japan). The concentration of non-reducing termini of the soluble starch was estimated as 0.136 mol/mg using the Smith degradation method (22). A series of acarviosyl-maltooligosaccharides ranging from acarviosyl-maltotriose (AC5) to acarviosylmaltooctaose (AC10) were enzymatically synthesized and purified by HPLC as previously reported (21). The purified acarviosyl-maltooligosaccharides were evaporated to dryness and dissolved in water, and their concentrations were determined on the basis of absorbance at 214 nm. For the nomenclature of the sugar rings of acarviosyl-maltooligosaccharides, the valienamine residue at the non-reducing end was termed as ring A, the 4-amino-4,6-dideoxy ␣-D-glucose residue as ring B, and the glucose residues in the maltooligosaccharyl moiety from the non-reducing toward the reducing side, as rings C, D, E, and so forth (Fig. 1).
Enzyme Purification and Preparation-Purification of native SBG from sugar beet seeds and preparation of endoglycosidase-F3-treated native SBG for crystallization were reported previously (16).
Crystal Structure Analyses-Co-crystallizations of endoglycosidase F3-treated native SBG individually with AC5-AC8 were performed by the hanging-drop vapor diffusion method at 25°C, with the following composition of the drops: 3 l of the enzyme (4.2 mg/ml), 3 l of reservoir solution (50 mM sodium acetate buffer (pH 4.5), 50 mM ammonium sulfate, and 16 -18% PEG monomethyl ether 2000), and 1 l of the ligand (33.3 mM AC5, 15.8 mM AC6, 7.79 mM AC7, or 4.44 mM AC8). X-ray diffraction data were collected on beamline BL41XU at SPring-8 (Hyogo, Japan) in the same manner as described previously (16). All diffraction data sets were indexed, integrated, scaled, and merged using XDS (23). Crystals of AC5 and AC8 complex belonged to the P2 1 2 1 2 1 space group and are isomorphous with crystal of AC4 complex, whereas crystals of AC6 and AC7 complex belonged to the I222 space group. Complex structures of AC5 and AC8 were determined by a molecular replacement method with phenix.automr (24,25) using the SBG-AC4 complex (Protein Data Bank code 3W37) as a search model and those of AC6 and AC7 were determined by rigid body refinement with phenix.refine (26) using the SBG-AC4 complex model. After several cycles of manual model corrections with Coot (27) and refinement with REFMAC5 (28) and phenix.refine, the refinement converged. The coordinates and structure factors have been deposited in the Protein Data Bank with codes 3WEL, 3WEM, 3WEN, and 3WEO for the AC5, AC6, AC7, and AC8 complexes, respectively. Data collection and refinement statistics are summarized in Table 1. Biochemical Assays-␣-Glucosidase activity, protein concentration, and the effects of pH were measured as described previously (20). The type of inhibition and inhibition constants (K i ) of AC4-AC10 for native SBG were determined using 1/ where K m app is the apparent K m in the presence of the inhibitor, K m is the actual K m in the absence of inhibitor, and [I], the concentration of the inhibitor. The values of mean Ϯ S.D. for K i at three inhibitor concentrations were calculated.
For the determination of kinetic parameters (k cat , K m , and k cat /K m ), the initial rates for eight substrate concentrations were measured, and the kinetic parameters k cat (s Ϫ1 ) and The correlations between log K i for AC4ϪAC7 and log(K m / k cat ) for the hydrolysis of Glc 4 ϪGlc 7 or between log K i for AC4ϪAC7 and log K m for Glc 4 ϪGlc 7 were analyzed for evaluating the transition state mimicry of the acarviosyl-maltooligosaccharides. The correlations were derived on the basis of the equation K i ϭ dK TS ϭ dk non (K m /k cat ), where d and k non represent proportionality and non-enzymatic reaction rate constants, respectively (29).

SBG Inhibition by Acarviosyl-maltooligosaccharides-The
type of inhibition and the inhibition constants of AC4-AC10 for the hydrolysis of native SBG, which was prepared from sugar beet seeds, were evaluated from 1/[S] versus 1/v plots using Glc 7 as a substrate. The 1/[S] versus 1/v plots of all the acarviosylmaltooligosaccharides showed linear correlation and intersected with each other on the y axis, indicating that the acarviosyl-maltooligosaccharides are competitive inhibitors of SBG. AC4 inhibited SBG with K i of 15.4 Ϯ 3.5 M, and the K i values decreased with increasing DP of the inhibitor ( Table 2). In particular, a greater difference was found between the K i of AC4 and AC5 compared with the other acarviosyl-maltooligosaccharides. The plots of log K i for AC4ϪAC7 against log(K m /k cat ) for the hydrolysis of Glc 4 ϪGlc 7 as well as log K i for AC4ϪAC7 against log K m for Glc 4 ϪGlc 7 showed linear correlation with correlation coefficients of r ϭ 0.964 (slope ϭ 1.44) and r ϭ 0.985 (slope ϭ 1.59), respectively (Fig. 2). These results suggest that the acarviosyl-maltooligosaccharides mimic both the transition and ground states.
Crystal Structures of Complexes of SBG with Acarviosylmaltooligosaccharides-SBG was co-crystallized individually with the acarviosyl-maltooligosaccharides AC5, AC6, AC7, and AC8, and the structures of the corresponding complexes were determined at 1.8, 2.6, 2.6, and 1.5 Å, respectively ( Table 1). Analysis of the structures of all these complexes revealed that SBG existed as a monomer in each asymmetric unit and had three N-glycans at Asn 404 , Asn 728 , and Asn 823 (Fig. 3A), as well as the previous crystal structures of SBG (16). The overall structures obtained in the present study were almost identical to the ligand-free and AC4-complex structures, and the root mean square deviations between every pair, as estimated by the Dali pairwise server (30), was within 0.5 Å. Co-crystals of sufficient size for structure determination could not be obtained for SBG with AC9 or AC10. The electron density corresponding to acarviosyl-maltooligosaccharides was observed only at the active site of each structure (Fig. 3B). The electron density of the reducing glucose residue of AC7 (ring G) was not completely observed. This might be because of the kinetically weak affinity at subsite ϩ6 (20) or the low-resolution structure. Fig. 3C shows the structural superposition of acarviosyl-maltooligosaccharides from all the complexes, including the previously determined structure of the AC4 complex. The conformations of the rings of acarviosylmaltooligosaccharides were observed to be nearly identical in all the structures; however, the conformations of ring G showed a slight difference between the AC7 and AC8 complexes. Compared with AC7, ring G of AC8 was observed to lie closer to the enzyme in a manner dependent on the interactions between ring H and subdomain b2 (see below). The acarviosine unit (rings A and B) was buried in the active-site pocket, whereas the maltooligosaccharide part was bound to the N-loop and subdomain b2. All residues in the N-loop and subdomain b2 had identical conformations, although a slight difference was observed in the side chain of Lys 493 in the structure of the AC8 complex (data not shown).
The conformations of rings AϪC were extended, and these rings, particularly the acarviosine unit, were tightly bound to the enzyme through several hydrogen bonds. In contrast, the conformations of rings D-H were similar to that of the native helical conformation through intramolecular hydrogen bonds with adjacent glucose residues (Fig. 4A). All glucose residues had cis-orientation and were connected by O2Ј-O3 hydrogen bonds, for instance, O2 (ring D)-O3 (ring E); only the rings F and G were trans-oriented, and two hydrogen bonds, O6 (ring F)-O3 (ring G) and O5 (ring F)-O3 (ring G), were observed.
Rings C, D, and E were bound to the N-loop and subdomain b2 (Fig. 4B). Phe 236 , Asn 237 , and Leu 240 on the N-loop contacted rings D and E, and directed the subsequent glucose residues (rings F-H) toward subdomain b2. Phe 236 and Asn 237 established interactions with ring D, as shown in our previous reports (16,20), whereas the side chain of Leu 240 established van der Waals contact with ring E. Ring E also bound the enzyme through hydrogen bonds between the O6 and O5 atoms and the hydroxy group of Ser 497 in subdomain b2. Rings F, G, and H were located at a region of subdomain b2 spanning Lys 493 to Pro 502 . The conformation of this region was tightly packed by 10 hydrogen bonds between the side chains and backbone, and this region interacted with rings F, G, and H via hydrogen bonds through the backbone and van der Waals contact through the side chains. Ring F interacted with the backbone carbonyls of Ser 497 and Gly 498 via a water molecule. Three hydrogen bonds were observed between ring H (O2 and axially oriented O1) and the backbone carbonyls of Arg 500 and Gly 499 . Val 501 was significantly close to rings F and G, and Lys 493 and Pro 502 were found in the vicinity of ring H. The multiple sequence alignment indicated that the region from Lys 493 to Pro 502 was highly conserved but Val 501 of SBG was atypical among plant ␣-glucosidases (Fig. 4C).
Site-directed Mutagenesis-The structures of these complexes suggested that the side chains of Leu 240 , Lys 493 , Val 501 , and Pro 502 are likely to be involved in substrate binding. The contributions of these residues to substrate binding were assessed using the mutant enzymes, L240A (Leu 240 3 Ala), K493A (Lys 493 3 Ala), V501A (Val 501 3 Ala), and P502A (Pro 502 3 Ala), which were generated using site-directed mutagenesis. In addition, the characterization of V501R (Val 501 3 Arg) was also performed to determine the functional role of the conserved arginine residue in the other plant ␣-glucosidases (Fig. 4C). The optimum pH (pH 4.9) for Glc 2 hydrolysis and pH stability (pH 3.0 -10.3) of all the mutant enzymes was nearly identical to those of the wild-type recombinant SBG (rSBG) (20). The kinetic properties were determined for a series of maltooligosaccharides, Glc 18 , and soluble starch ( Table 3). All recombinant enzymes including rSBG exhibited lower k cat values than the native SBG. This may be caused by excessive glycosylation observed in the recombinant enzymes produced by P. pastoris.  Both K493A and P502A exhibited almost the same K m values for all the substrates as rSBG, even though their k cat values decreased. The k cat /K m values for soluble starch were 80-(K493A) and 88-fold (P502A) higher than each for Glc 2 , namely the substrate specificities of the mutant enzymes were almost identical to rSBG.
L240A showed that its k cat for all the substrates equaled 59 -72% that of rSBG. In contrast, the extent of decrease in k cat /K m values was dependent on the DP of the substrates. For instance, k cat /K m for Glc 2 -Glc 4 substrates equaled 51-56% that of rSBG, whereas for Glc 5 -Glc 7 , the values equaled only 23%. Reduction in k cat /K m was also observed for Glc 18 and soluble starch (21% of rSBG values for both substrates). This reduction in k cat /K m was associated with an increase in K m . Significantly higher K m of L240A was observed for substrates with DP of more than 4; for instance, the K m for Glc 2 -Glc 4 was 1.1-1.3fold that of rSBG, whereas for Glc 5 -Glc 7 , it was 2.6 -2.8-fold. These results indicated that the mutation (Leu 240 3 Ala) decreased the binding affinity in the transition state and the ES complex at subsite ϩ4. This is consistent with structure-based analysis, which revealed that Leu 240 established contacts with ring E of acarviosyl-maltooligosaccharides.
The changes in kinetic parameters of the V501A and V501R mutants paralleled those of L240A, with the k cat values for all substrates were affected equally, but with greater reduction in k cat /K m values for substrates with DP of more than 4 as a consequence of 1.2-1.8-fold increased K m for these substrates.
These results suggest that the substitutions of Val 501 caused a reduction in affinity at subsite ϩ4, which is in contradiction with the structure-based analysis, which revealed that Val 501 is located close to rings F and G, occupying subsites ϩ5 and ϩ6.

DISCUSSION
In the current study, acarviosyl-maltooligosaccharides were employed for clarifying the mode of substrate binding at subsites remote from the active site pocket. The K i values of the acarviosyl-maltooligosaccharides AC5-AC10 for SBG were significantly lower compared with AC4, indicating that these acarviosyl-maltooligosaccharides are more effective inhibitors of SBG than AC4. The clear correlation observed between log K i for AC4ϪAC7 and log(K m /k cat ) for the hydrolysis of Glc 4 ϪGlc 7 suggests that the inhibitors are transition state analogs (29). Relative to the substrate, the tighter binding of the acarviosyl-maltooligosaccharides to the active site results in a value of K i , which is 3 orders of magnitude lower than K m (upon equating K m with K s ); this could be considered a consequence of the valienamine unit, which is considered to mimic the glycosyl cation-like transition state. The reduction in K i with an increase in DP can be accounted for by a decrease in the dissociation constant of the maltooligosaccharide unit from the enzyme, because the correlation of log K i versus log K m is similar to that of log K i versus log(K m /k cat ). In the hydrolysis reaction, the increase in k cat /K m with an increase in the DP of substrates is due to the decrease in K m for these substrates. The values of k cat are almost unaltered, but K m decreases with an increase in the DP of maltooligosaccharides from 2 to 7 ( Table  2). In other words, maltose binding at subsites Ϫ1 and ϩ1 provides sufficient binding energy for lowering the activation energy of SBG, and the binding energy at the subsequent sub-sites ϩ2, ϩ3, and so forth are chiefly employed for decreasing the dissociation constant. This kinetic behavior is very similar to the inhibitory behavior of the acarviosyl-maltooligosaccharides; therefore, the binding of the acarviosyl inhibitors could be considered to represent the binding mode of the substrates, and hence, serves as an adequate probe for characterization of the remote substrate-binding site. The observed slope (1.44) of the correlation between log K i and log(K m /k cat ) as opposed to the expected slope (1.0) suggests a less than ideal mimicry of the transition state by the analog (31). The optimal mimicry of the transition state by an inhibitor should result in a 10 Ϫ5 -fold or lower K i , which is described as K TS in the case of transition state analog, compared with the K m , considering the equation K TS ϭ k non (K m /k cat ) and the reported values of k non /k cat (29,32). The evaluated K i of the acarviosyl-maltooligosaccharides, of the range ((2.5Ϫ8.8) ϫ 10 Ϫ3 )-fold that of the K m for the substrate, indicates a lesser degree of mimicry despite the tighter binding of acarviosyl-maltooligosaccharides to the active site compared with the substrate. The slightly higher values of K i of acarviosylmaltooligosaccharides compared with the theoretical values are attributable to imperfect mimicry by the valienamine unit.
Crystal structure analyses of SBG bound with the acarviosylmaltooligosaccharides revealed the molecular basis of substrate binding at a site distant from the active site pocket. The N-loop and the region spanning Lys 493 to Pro 502 in subdomain b2 are involved in the binding of the maltooligosaccharide part of the acarviosyl-maltooligosaccharides, even though Lys 493 and Pro 502 have little contribution to decreasing K m values for all the substrates (Table 3). In a previous study from our group (16), the role of Ser 497 (in subdomain b2) in governing the affinity at subsite ϩ4 was demonstrated through site-directed mutagenesis. In agreement with the previous study, the structures of complexes with acarviosyl-maltooligosaccharides revealed the interaction of Ser 497 with O5 and O6 of ring E through hydrogen bonds. The glucose residues of the maltooligosaccharide part were primarily observed to make contact with remote subsites through van der Waals interactions. The aliphatic side chains of Leu 240 and Val 501 significantly interacted with rings E and F/G, respectively. Therefore, the mutant enzyme L240A had a substantially reduced affinity at subsite ϩ4. In contrast, the kinetic characteristics of V501A are in conflict with its structural attribute. Structural analysis suggested that Val 501 likely contributes to the affinity at subsites ϩ5/ϩ6; however, the mutant V501A exhibited reduced affinity at subsite ϩ4. This contradiction could be explained by the effect of the Val 501 mutation on Ser 497 . As shown in our previous report (16), the mobility of the side chain of Ser 497 is high, resulting in its observation as a dual conformation of the AC4 complex structure. The side chain of Val 501 is located close to the hydroxy group of Ser 497 , and the mutation (Val 501 3 Ala) may increase the mobility of the hydroxy group of Ser 497 ; such increased mobility is likely to diminish the interaction of the hydroxy group with the substrate, leading to lowered affinity at subsite ϩ4. Furthermore, V501R showed the same substrate specificity as V501A, indicating that an arginine residue at the Val 501 position, which is observed in other plant ␣-glucosidases, is not enough to increase affinity at subsite ϩ4. It is possible that Val 501 of SBG is one of the key residues to achieve its Amino acid residues and water molecules interacting with AC8 are represented by a stick model and red sphere, respectively. Two catalytic residues are indicated by asterisks. The dashed lines indicate hydrogen bonds. Color coding is similar to Fig. 3A. C, multiple sequence alignment at a region spanning Lys 493 to Pro 502 was produced by MUSCLE (36). SOG, spinach ␣-glucosidase (O04893); BWG, buckwheat ␣-glucosidase (H. Mori, unpublished data); ONG1, rice ␣-glucosidase isozyme 1 (Q653V4); ONG2, rice ␣-glucosidase isozyme 2 (Q653V7); BAG, barley ␣-glucosidase (Q43763). enormous specificity for long-chain substrates as compared with other plant enzymes.
The higher specificity of SBG for longer substrates, which is responsible for the lower dissociation constant of substrates with high DP from the enzyme, appears to be rationalized by the conformation of rings D-H. Of the rings A-H bound to SBG, rings A-C at the non-reducing terminus are extended and the energy gained by binding with the extended substrate contributes to decreasing the activation energy of the hydrolysis reaction. In contrast, rings D-H remain in the stable conformation, as also observed in cycloamyloses and the helical structures of V-amylose (33). SBG is unlikely to obtain the binding energy required for lowering the activation energy from the binding of rings D-H. The binding of substrates in a stable conformation appears to be advantageous compared with a strained conformation insofar as the decrease in dissociation constant is concerned, providing additional energy is unnecessary. As mentioned above, the higher specificity for longer substrates is responsible for the lower dissociation constant of the substrates with high DP from the enzyme. The subsite structure suitable for native substrate conformation is likely to result in lower K m for the substrates with high DP without the extra energy. Moreover, SBG is likely to ingeniously use the self-stabilizing property of amylose and soluble starch to form stable ES complexes with these long-chain substrates.
The high specificity of SBG for amylose and soluble starch is also attributable to the band-flip, which is found between rings F and G in structures of AC7 and AC8 complexes. The band-flip observed in the helical structures of the larger cycloamyloses and V-amylose has been generally accepted to be responsible for the relieving strain induced in the macrocycle (34). A likely hypothesis is that AC8 forms the band-flip to alleviate the strain caused, and the structure of SBG is fit to flip. However, the subsite structure of SBG has also been considered to provoke the band-flip; thereby, relieving the strain in long-chain helical substrates leading to the formation of a stable ES complex. In the latter case, the subsite structure likely allows SBG to handle ␣-1,6-branched structures in the substrate at the trans-oriented point. In the structures of complexes with acarviosylmaltooligosaccharides, the hydroxy groups of C6 atoms of the rings A-F are close to the enzyme surface, and ␣-1,6-branched structures are unlikely to be accommodated there. However, the trans-oriented bond results in the O6 atoms of rings G and H facing outward, which would allow the accommodation of ␣-1,6-branched structures. The extremely higher specificity of SBG for soluble starch (branched substrate) than Glc 18 (linear substrate) may be attributable to this mode of substrate binding.
The current study has highlighted the utility of acarviosylmaltooligosaccharides and provided structural insights into the mechanism that governs the high specificity of SBG for substrates with high DP. The acarviosyl-maltooligosaccharides proved to be useful as inhibitors of SBG. Moreover, kinetic analysis has demonstrated that the binding of the series of acarviosyl inhibitors represents the binding mode of the substrates. Analyses of the structure of the acarviosyl-maltooligosaccharide complexes revealed that subsites remote from the active-site pocket of the enzyme complement the native structure of amylose, which adopts a helical conformation through intramolecular hydrogen bonds. This mode of binding likely leads to lower K m and higher specificity for longer substrates. The study of acarviosyl-maltooligosaccharides is likely to prove useful for the study of other ␣-glycosidases as well. For example, glucoamylase shows notable specificity for longer maltoo- ligosaccharides, but the crystal structure of the enzyme has been solved with acarbose (35). The use of acarviosyl-maltooligosaccharides is likely to clarify the mode of substrate binding at subsites far from the active site of glucoamylase.