The remote substrate binding subsite -6 in cyclodextrin-glycosyltransferase controls the transferase activity of the enzyme via an induced-fit mechanism.

Cyclodextrin-glycosyltransferase (CGTase) catalyzes the formation of alpha-, beta-, and gamma-cyclodextrins (cyclic alpha-(1,4)-linked oligosaccharides of 6, 7, or 8 glucose residues, respectively) from starch. Nine substrate binding subsites were observed in an x-ray structure of the CGTase from Bacillus circulans strain 251 complexed with a maltononaose substrate. Subsite -6 is conserved in CGTases, suggesting its importance for the reactions catalyzed by the enzyme. To investigate this in detail, we made six mutant CGTases (Y167F, G179L, G180L, N193G, N193L, and G179L/G180L). All subsite -6 mutants had decreased k(cat) values for beta-cyclodextrin formation, as well as for the disproportionation and coupling reactions, but not for hydrolysis. Especially G179L, G180L, and G179L/G180L affected the transglycosylation activities, most prominently for the coupling reactions. The results demonstrate that (i) subsite -6 is important for all three CGTase-catalyzed transglycosylation reactions, (ii) Gly-180 is conserved because of its importance for the circularization of the linear substrates, (iii) it is possible to independently change cyclization and coupling activities, and (iv) substrate interactions at subsite -6 activate the enzyme in catalysis via an induced-fit mechanism. This article provides for the first time definite biochemical evidence for such an induced-fit mechanism in the alpha-amylase family.

High resolution x-ray structures are known for the CGTases from Bacillus circulans strain 8 (8) and strain 251 (BC251) (9), Thermoanaerobacterium thermosulfurigenes strain EM1 (10), Bacillus stearothermophilus (11), and alkalophilic Bacillus sp. 1011 (12). The structures of CGTases are organized in five domains (A-E). The N-terminal part consists of the catalytic (␤/␣) 8 -barrel fold (domain A) with a loop of ϳ60 residues protruding at the third ␤-strand (domain B). Domains A and B together form the substrate binding groove and contain the catalytic site residues (8,13). Domains C and E are involved in starch binding (14), whereas the function of domain D remains to be elucidated. The substrate binding groove of the BC251 CGTase consists of at least nine sugar binding subsites (13), labeled Ϫ7 to ϩ2, with bond cleavage occurring between subsites Ϫ1 and ϩ1. Fig. 1 gives an overview of the interactions between the enzyme and a maltononaose substrate and shows the binding mode of this maltononaose in the active site of CGTase.
CGTase uses an ␣-retaining double displacement mechanism to catalyze four different reactions, cyclization, coupling, disproportionation, and hydrolysis. The cyclization (and disproportionation) reactions start with the binding of a linear maltooligosaccharide substrate, followed by cleavage of the ␣-(1,4)glycosidic bond between the residues bound at subsites ϩ1 and Ϫ1, resulting in an intermediate that is covalently linked to Asp-229 (15,16). Subsequently, the non-reducing end moves from subsite Ϫ7 (for ␤-cyclization) into subsite ϩ1. This step is called circularization, which is followed by intramolecular bond formation. Circularization is most likely the rate-determining step in the cyclization reaction (17,18). In the disproportionation reaction the non-reducing end of a second sugar molecule is used as acceptor. CGTase also catalyzes the reverse reaction of cyclization, which is called the coupling reaction. In this reaction a cyclodextrin ring is opened, and the resulting covalently bound, linear oligosaccharide is transferred to a second sugar molecule, the acceptor. Besides these three transglycosylation reactions, CGTase catalyzes the hydrolysis of ␣-(1,4)-glycosidic bonds in starch. Interestingly, the hydrolysis activity of CGTase is much lower than its transglycosylation activities, making the enzyme an efficient transferase (3). Of all reaction types, the disproportionation reaction is most efficiently catalyzed by CGTase (17,19).
The high transferase activity of CGTase was investigated recently by comparing x-ray structures of CGTase representing different stages of its reaction cycle (20). From these studies it appeared that the protein backbone of CGTase can undergo small but significant conformational changes after binding of substrate sugars at the acceptor subsites ϩ1 and ϩ2 and at the donor subsites Ϫ3 and Ϫ6. The conformation of Asn-139 and His-140 is changed only if sugars are simultaneously bound at these subsites (21). This enables His-140 to make a hydrogen bond to the O-6 atom of the Ϫ1 sugar (see Fig. 1A) and helps to distort the Ϫ1 sugar toward transition state planarity (16,20,21). It was suggested that in this way, these distant sugar binding subsites communicate the presence of long oligosaccharide substrates and acceptors and ensure that they are preferentially processed. Support for this mechanism has come from site-directed mutagenesis experiments of the residues in the acceptor subsites and of His-140 (22)(23)(24). Subsite Ϫ6 has not been studied so far, but its position far from the catalytic site (see Fig. 1B) makes it unlikely that the mutants interfere directly with the catalytic process. Instead, they may affect substrate binding or the proposed induced-fit mechanism, providing an excellent opportunity to test whether distant subsites play a role in regulating transglycosylation activity.
At present there are no mutagenesis data concerning subsite Ϫ6. Because this subsite is identical in all known CGTases, subsite Ϫ6 must be important for the function and the unique characteristics of CGTase. We constructed mutants that block subsite Ϫ6 (G179L, G180L, and G179L/G180L) or that abolish interactions at subsite Ϫ6 (Y167F, N193G, and N193L). Here we report a kinetic analysis of these mutants. The results obtained show that subsite Ϫ6 has an important function in all three transglycosylation reactions. They provide new insights in the catalytic mechanism employed by CGTase.

EXPERIMENTAL PROCEDURES
Structure Determination-Crystals of mutant BC251 CGTases were grown from 60% (v/v) 2-methyl-2,4-pentanediol, 100 mM HEPES (pH 7.5) and 5% (w/v) maltose (9). Soaking of N193G crystals with acarbose was carried out as described earlier (13). For G179L data were collected to 1.94 Å at 120 K on an in-house MacScience DIP2030H image plate (Nonius, Delft, The Netherlands) using CuKa radiation from a Nonius FR591 rotating-anode generator with Franks' mirrors. Processing was done with DENZO and SCALEPACK (25). The structure of CGTase liganded with maltotetraose (PDB code 1CXF), with all waters and sugars removed, was used as a starting model. Refinement was done with the Crystallography & Nuclear Magnetic Resonance System (26) in a standard way. The compression of the longest cell axis of G179L compared with that of N193G (Table I) is because of a locally changed crystal packing at the maltose binding site near Trp-616 and Trp-662. This has improved the crystal quality, as shown by the increased resolution of the data at the in-house source and the low overall B-factor of structure and the low R-factors. For N193G data were collected to 2.43 Å at room temperature on a MacScience DIP2020 imaging plate mounted on an Elliot GX21 rotating-anode generator producing CuKa radiation. Data were reduced and scaled using the program XDS (27) and programs from the Groningen BIOMOL software package. Sugar ligands were manually placed in sigmaA-weighted 2F o Ϫ F c and F o Ϫ F c electron density maps with the program O (28). The atomic coordinates and the structure factors of the structures have been deposited in the Protein Data Bank (code IKCL for G179L and IKCK for N193G; www.rcsb.org).
Enzyme Assays and Enzyme Purification-CGTase proteins were produced and purified as described before (3). All enzyme assays were performed in 10 mM sodium citrate buffer (pH 6.0) at 50°C. Cyclization activities were determined by incubating 0.1-0.5 g/ml enzyme with 2.5% (w/v) Paselli SA2 starch (partially hydrolyzed potato starch with an average degree of polymerization of 50; AVEBE, Foxhol, The Netherlands), as described by Penninga et al. (14).
Cyclodextrin product specificity under industrial process conditions was measured by incubating 10% (w/v) Paselli WA4 starch (pregelatinized drum-dried starch with a high degree of polymerization; AVEBE) with 2 units/ml of enzyme activity (1 unit is mol min Ϫ1 ␤-cyclodextrinforming activity per mg of protein). Samples were taken at regular intervals, boiled for 10 min, and analyzed by HPLC, as described below.
Coupling activities were measured as described by Nakamura et al. (24), with some modifications (33), with ␣-, ␤-, and ␥-cyclodextrin as donor substrates and methyl-␣-D-glucopyranoside (M␣DG) as acceptor substrate, using 0.1-0.5 g/ml enzyme. Values of k cat and K m were determined by measuring rates at 5 donor and 5 acceptor substrate concentrations (25 conditions) ranging from 0.2 to 5 times the K m values.
Disproportionation activity was determined as described by Nakamura et al. (19), with some modifications (33), using 0.1-0.5 g/ml enzyme, 4-nitrophenyl-␣-D-maltoheptaoside-4 -6-O-ethylidene (EPS; Roche Molecular Biochemicals), or 4-nitrophenyl-␣-D-maltopentaoside (G5-pNP; Megazyme, County Wicklow, Ireland) as donor substrate and maltose as acceptor substrate. With the EPS substrate, values of k cat and K m were determined by measuring rates at 6 donor and 5 acceptor substrate concentrations (30 conditions) ranging from 0.2 to 5 times the K m values. With the G5-pNP substrate, values of k cat and K m were determined by measuring rates at 12 different donor concentrations at fixed maltose concentration (10 mM).
Hydrolyzing activity was determined as described before (3) by measuring the increase in reducing power upon incubation of 5 g of enzyme with 1% (w/v) soluble starch (Lamers & Pleuger, Wijnegen, Belgium).
HPLC Analysis-Products formed were analyzed by HPLC, using an Econosphere NH 2 5 u column (250 ϫ 4.6 mm) (Alltech Nederland bv; Breda, The Netherlands) linked to a refractive index detector. A mobile phase of acetonitrile/water (60:40) (v/v) at a flow rate of 1 ml/min was used.
Analysis of the Experimental Data-The results obtained for the coupling and disproportionating reactions were analyzed using Sig-maPlot (Jandel Scientific). The coupling reaction followed the random order ternary complex mechanism (17). The disproportionating reaction proceeded via the substituted enzyme mechanism (or ping-pong mechanism) (19).

Structures-
The G179L structure had a maltotetraose ligand bound from subsites ϩ2 to Ϫ2 with the glucose at subsite Ϫ1 in its ␤-anomeric configuration. Because the G179L crystals were not soaked with sugars, the maltotetraose sugar must be the remainder of ␣-cyclodextrin used for the purification of the enzyme. This mutant has indeed a very low coupling activity (see Table V). The G179L structure also had a glucose molecule bound at the surface near Gln-287, Arg-290, Arg-294, Asp-295, and Glu330, about 8 Å from subsite ϩ2. A sugar at this position was not seen before. Its functional relevance for the enzyme is not known. The / angles of the mutated residue 179 were 66/Ϫ151 compared with 97/Ϫ162 in the wild-type enzyme (PDB code 1CDG). The protein backbone conformation was hardly affected by this mutation, however. The N193G structure had an acarbose molecule bound from subsites ϩ2 to Ϫ2. The / angles of the mutated residue were hardly changed, Ϫ72/150 compared with Ϫ60/145 in wild-type (PDB code 1CDG), and the protein backbone conformation was not significantly altered. In both structures the Asn-139/His-140 conformation is identical to that of the unliganded wild-type CGTase (34), as expected for structures that have no sugar bound at the Ϫ6 subsite.
Cyclization Activities of Wild-type and Mutant CGTases-The cyclization activities of the (mutant) CGTases are summarized in Table II. Substrate affinity values are not reported, because at the low substrate concentrations needed the amount of cyclodextrin formed is too low for reliable activity measurements. Low starch concentrations are needed, because BC251 CGTase has a high affinity for starch (Ͻ0.5 mg/ml) (17). All subsite Ϫ6 mutants have reduced ␤-cyclodextrin-forming activities, most pronounced for mutations that introduce a leucine at position 180 (G180L and G179L/G180L) or remove a side chain at position 193 (N193G) ( Table II). Gly-180 is most important for ␤-, ␥-, and ␦-cyclodextrin formation, whereas the G179L mutation especially affects ␣-cyclization. Mutant N193L is only slightly affected in its cyclization activities. The N193G mutation, in contrast, specifically decreases ␤-cyclodextrin formation. Thus, subsite Ϫ6 plays an important role in the cyclization reactions catalyzed by CGTase.
Subsite Ϫ6 Mutations Affect the Disproportionation Reaction-All mutants show reduced disproportionating activities with the maltoheptaose EPS (Table III), most prominently for G179L, G180L, and G179L/G180L. With the shorter G5-pNP substrate, which cannot reach subsite Ϫ6, the mutants G179L, G179L/G180L, and N193G had decreased disproportionation activity, whereas the Y167F, G180L, and N193L mutants had wild-type activity (Table IV). Furthermore, the wild-type CGTase had a lower disproportionation activity with the shorter G5-pNP substrate than with EPS, indicating that substrate interactions at subsite Ϫ6 are important in this reaction. Mutation of Gly-179 and Gly-180 resulted in 4-to 5-fold increased K m , EPS values (Table III), demonstrating that introducing leucines at positions 179 and 180 negatively affects binding of the maltoheptaose compound EPS. This indicates that the wild-type enzyme has interactions with EPS at subsite Ϫ6. Indeed, product analysis of the disproportionation reaction showed that EPS is able to reach subsite Ϫ6 (data not shown). Mutation of Tyr-167 and Asn-193, in contrast, has no significant effect on the K m , EPS value (Table III). The specificity constants (k cat /K m , EPS ) ( Table III) also show that Gly-179 and Gly-180 are especially important for the disproportionation activity of CGTase. The apparent affinities for the acceptor  (Table III). In All Mutants Coupling Activity Is Decreased-All subsite Ϫ6 mutants have reduced coupling activities (Table V). The G179L mutant has a 4-fold decreased k cat value for ␤-cyclodextrin coupling. Unexpectedly, the coupling activities with ␣and ␥-cyclodextrin are fully abolished (Table V). Furthermore, the coupling activities of the G179L/G180L mutant are also virtually absent for each of the three cyclodextrins (Table V). In contrast, the Y167F, G180L, N193G, and N193L mutants retained significant coupling activity with all cyclodextrins tested (data not shown for ␣and ␥-cyclodextrin). The G180L and N193G mutants strongly reduced the coupling activity with ␤-cyclodextrin, whereas the effect was small for Y167F and N193L. Thus, subsite Ϫ6 is very important in the coupling reactions. The mutations at subsite Ϫ6 also have an effect on acceptor binding as shown by the apparent affinity constants for the acceptor substrate (Table V). Especially the G180L mutation drastically increases the K m,M␣DG values, whereas the effect is smaller for mutant N193G. This is unlikely to be a direct effect of the mutation, because Gly-180 and Asn-193 are positioned a large distance from the acceptor site (Fig. 1). This shows that subsite Ϫ6 influences the acceptor binding subsites.
Hydrolyzing Activities-The hydrolyzing activity of CGTase results in the formation of linear products from starch. Although the substrate used (starch) is able to reach subsite Ϫ6, the mutations at this subsite have no significant effect on the hydrolyzing activity (Table II), showing that it is possible to selectively alter one of the CGTase activities without affecting another activity.
Cyclodextrin Product Ratios of (Mutant) CGTases-In this production assay the ␣:␤:␥ ratio of formed cyclodextrins changes in time as the combined result of all four reactions described above. Fig. 2 shows this time dependence of cyclodextrin production for the wild-type and mutant enzymes. After 6 h, 30-37% of the starch has been converted into cyclodextrins (Table VI), with only small amounts (Ͻ3%) converted into linear products (not shown). Initially wild-type CGTase produces mainly ␤and ␥-cyclodextrin, whereas smaller amounts of ␣-cyclodextrin are formed. Compared with wild-type, the G180L, G179L/G180L, and N193G CGTases produce larger amounts of ␣and ␥-cyclodextrin in the first minutes of the reaction (Fig. 2). After 6 h all mutants have produced more ␣-cyclodextrin, with the exception of G179L, which produced significantly less ␣-cyclodextrin (see Fig. 2 and Table VI). The wild-type and the G179L and G179L/G180L mutants formed more ␣than ␥-cyclodextrin, whereas the other mutants formed more ␥than ␣-cyclodextrin after 6 h of incubation (Fig. 2). The data thus show that subsite Ϫ6 is involved in cyclodextrin product specificity. Although the mutant CGTases have lower or minimized coupling activities, they do not produce significantly more cyclodextrins than the wild-type enzyme.
Circularization-One of the striking effects of the mutations is their influence on the cyclodextrin product ratio. An explanation for this is suggested from circularization pathway calculations, which revealed that subsite Ϫ6 stabilizes intermediary stages of the circularization process by successively binding the 6th, 7th, and 8th (␥-cyclodextrin formation) glucose residue during the movement of the non-reducing end of the substrate toward the ϩ1 acceptor subsite (18). Thus, mutations in subsite Ϫ6 are expected to interfere with the cyclization reaction. Indeed, mutations in subsite Ϫ6 especially affect ␤-, ␥-, and ␦-cyclization. No negative effect is seen for the ␣-cyclization activity, which only involves binding of six glucose residues. An exception, however, is mutant G179L, of which the ␣-cyclization activity has been reduced to 50% of the wild-type activity.
Function of Subsite Ϫ6 in the Cyclization Reactions-The rate-limiting step in the ␤-cyclization reaction is most likely the 23-Å movement of the non-reducing end of the substrate from subsite Ϫ7 to subsite ϩ1 (circularization) (16, 18). The lower ␤-cyclization activities of the subsite Ϫ6 mutant CGTases thus indicate that substrate binding, or the circularization process  or both, are affected. Because subsite Ϫ6 provides several strong interactions with linear substrates, it has been suggested that this subsite selects for substrates of sufficient length for cyclodextrin formation (13). In addition, it was suggested that Gly-179 and Gly-180 are conserved in CGTases, because the absence of side chains is a requirement for substrate binding at subsite Ϫ6 (21). The increased K m values (Table III) for the maltoheptaose compound used in the disproportionation reaction (EPS) show that linear substrate binding is indeed hindered by mutations in Gly-179 and Gly-180. For G179L this is most likely caused by the presence of the leucine side chain, because the protein backbone conformation was hardly affected by this mutation. The especially strongly de-creased ␣-cyclization activity of the G179L mutant (Table II) can be explained by the assistance of subsite Ϫ7 in binding of the longer sugar chains required for ␤-, ␥-, and ␦-cyclization. This assistance of subsite Ϫ7 cannot occur in the ␣-cyclization reaction.
It has been derived that the ratio of k cat, ␤-cyclization /k cat, disproportionation can be used as an indicator for cyclization efficiency (18). For the mutants discussed here, this ratio is decreased most drastically by mutation G180L (Table III), indicating that this mutation especially hampers the circularization process. Thus, a side chain at position 180 interferes strongly with the circularization process, explaining the conservation of Gly-180 in CGTases.  (Fig. 1B) it is unlikely that the mutations directly affect bond cleavage. Indeed, the mutants Y167F, G180L, and N193L have wild-type activity with the shorter maltopentaose substrate (Table IV) but not with EPS (Table III). The decreased disproportionation activities of G179L and N193G with the maltopentaose substrate are most likely caused by changes in structural flexibility, because the G179L and N193G structures showed no significant difference compared with the wild-type CGTase. Furthermore, the wildtype enzyme has a higher k cat value with the longer EPS substrate than with the shorter G5-pNP substrate (see Table  III and Table IV). Together this supports the presence of an induced-fit mechanism that is operated by substrate binding at subsite Ϫ6, as suggested by x-ray structure comparisons (21). This induced-fit mechanism can explain the high transglycosylation activity of CGTase with longer sugar chains (20) and the conservation of subsite Ϫ6 in all known CGTases.
Function of Subsite Ϫ6 in the Coupling Reaction-Subsite Ϫ6 has no interactions with a cyclodextrin molecule bound in the active site of CGTase (21,40). Unexpectedly, mutants in this subsite showed decreased coupling activities, demonstrating that subsite Ϫ6 is important for the coupling reaction. Especially mutation G179L drastically reduces the coupling activities, whereas its effect on the cyclization reaction is much smaller. Mutation G180L, in contrast, especially decreases the cyclization activities. In the coupling reaction binding of the cleaved cyclodextrin molecule to subsite Ϫ6 is necessary for efficient transfer of the covalent intermediate to the acceptor as shown by the decreased coupling activities. The strongly reduced acceptor affinities of G180L and N193G indicate that acceptor binding at subsite ϩ1 is hampered in these mutants, although these subsite Ϫ6 residues are far from the acceptor subsite ϩ1 (Fig. 1B). Therefore, we propose that attachment of the opened cyclodextrin molecule to subsite Ϫ6, together with acceptor binding at subsite ϩ1, activates the enzyme in the coupling reaction. This proposal thus extends and strengthens a previous hypothesis based on structural results only (20).
Conclusions-This study shows that subsite Ϫ6 of CGTase is of great importance in all three transglycosylation reactions catalyzed by the enzyme but not in the hydrolysis reaction (Table II). The data provide for the first time definite biochemical support for a hypothesis based on x-ray crystallographic evidence (21) that substrate binding at subsite Ϫ6 activates an induced-fit mechanism. Such an induced-fit mechanism favors the processing of longer oligosaccharides. In addition, our results explain the conservation of Gly-180, because a larger residue interferes with the cyclization reaction. In addition, we provide clear evidence that it is possible to independently change the cyclization and coupling reactions.