Substrate recognition mechanism of a glycosyltrehalose trehalohydrolase from Sulfolobus solfataricus KM1

Abstract

Glycosyltrehalose trehalohydrolase (GTHase) is an α-amylase that cleaves the α-1,4 bond adjacent to the α-1,1 bond of maltooligosyltrehalose to release trehalose. To investigate the catalytic and substrate recognition mechanisms of GTHase, two residues, Asp252 (nucleophile) and Glu283 (general acid/base), located at the catalytic site of GTHase were mutated (Asp252→Ser (D252S), Glu (D252E) and Glu283→Gln (E283Q)), and the activity and structure of the enzyme were investigated. The E283Q, D252E, and D252S mutants showed only 0.04, 0.03, and 0.6% of enzymatic activity against the wild-type, respectively. The crystal structure of the E283Q mutant GTHase in complex with the substrate, maltotriosyltrehalose (G3-Tre), was determined to 2.6-Å resolution. The structure with G3-Tre indicated that GTHase has at least five substrate binding subsites and that Glu283 is the catalytic acid, and Asp252 is the nucleophile that attacks the C1 carbon in the glycosidic linkage of G3-Tre. The complex structure also revealed a scheme for substrate recognition by GTHase. Substrate recognition involves two unique interactions: stacking of Tyr325 with the terminal glucose ring of the trehalose moiety and perpendicularly placement of Trp215 to the pyranose rings at the subsites −1 and +1 glucose.

Introduction

Trehalose (O-α-d-glucopyranosyl-(1→1)-α-d-glucopyranoside) is used for energy storage and for protecting proteins and cell membranes from extreme temperatures and osmotic shock in plants, insects, and microorganisms.^1–3 A coupled enzyme pathway for the production of trehalose from soluble starch has been discovered in the Arthrobacter sp. Q36^{4, 5} as well as in the Sulfolobaceae family of thermoacidophilic archaebacteria.^6–10 This pathway comprises two enzymes: glycosyltrehalose synthase (GTSase; also called maltooligosyltrehalose synthase; MTSase), an intramolecular glycosyltransferase (1,4-α-glucan 4-α-glycosyltransferase, EC 2.4.1.25), and glycosyltrehalose trehalohydrolase (GTHase; also called maltooligosyltrehalose trehalohydrolase; MTHase), 4-α-d-((1→4)-α-d-glucano)trehalose trehalohydrolase, EC 3.2.1.141) which is one of α-amylases (α-1,4-d-glucanohydrolase, EC 3.2.1.1). These enzymes are classified within the glycoside hydrolase (GH) family 13¹¹(CAZy database¹²; http://www.cazy.org/). Furthermore, family GH13 is divided into subfamilies¹³ and GTHase is classified into the subfamily GH13_10, which is described as 4-α-(1,4-α-glucano)trehalose -trehalohydrolase. GTSase converts the glycosidic bond between the last two glucose residues of a maltooligosaccharide from an α-1,4 bond to an α-1,1 bond, making a nonreducing maltooligosyltrehalose (Scheme 1). GTHase then cleaves the α-1,4 bond adjacent to the α-1,1 bond, releasing a trehalose molecule and regenerating a substrate for GTSase. This pathway differs considerably from the previously characterized phosphate-dependent pathway¹⁴ in which trehalose is synthesized from glucose-6-phosphate (a high-energy glycolytic intermediate) and UDP- or GDP-glucose by trehalose-6-phosphate synthase (EC 2.4.2.15) and trehalose-6-phosphate phosphatase (EC 3.1.3.12).

Scheme 1.

Sulfolobus solfataricus KM1, a member of the Sulfolobaceae family of hyperthermophilic, acidophilic archaea, was isolated from an acid hot spring in Gunma Prefecture, Japan in 1993.⁶ It is a coccoid, gram-negative bacterium with optimal growth at 75−85°C and pH 3.5−4.5. GTSase and GTHase have been purified from S. solfataricus KM1 (KM1-GTSase and KM1-GTHase); each of the two enzymes have also been cloned, sequenced, and expressed in Escherichia coli (GTSase)¹⁵ and Candida utilis (GTHase).¹⁶ The biochemical and kinetic characteristics of these two enzymes are as follows.

KM1-GTHase (Mr, 64,644 Da; number of residues, 558) is an exoamylase that hydrolyzes maltooligosyltrehalose to release trehalose. It also hydrolyzes maltooligosaccharide at the reducing end to release glucose or maltose with ∼ 16-fold lower activity⁸; this side reaction decreases the purity of the trehalose, and the extra step of purification decreases the yield of industrial trehalose production. Therefore, the exact substrate recognition mechanism of this enzyme is of interest in order to increase its substrate specificity. The activity of KM1-GTHase places it in the α-amylase family, which includes α-amylases, cyclodextrin glucosyltransferases,¹⁷Thermoactinomyces vulgaris R-47 α-amylase II (TVAII),¹⁸ and neopullulanases.¹⁹ We have previously determined the crystal structure of wild-type KM1-GTHase by X-ray diffraction.²⁰ The structure of the wild-type KM1-GTHase comprises three major domains (A, C, and E) and two subdomains (B and D), and with an N-terminal extension forming a stably folded immunoglobulin type domain connected by an extended linker peptide to the (β/α)₈ barrel catalytic domain. The enzyme exists as a symmetric dimer covalently linked by a disulfide bond at Cys298.

To investigate the catalytic and substrate recognition mechanisms, we mutated two residues, Asp252 and Glu283, located at the active site of KM1-GTHase. Two single and one double amino acid mutant KM1-GTHase proteins were produced and crystal structures of the three mutant proteins in complex with their substrate were obtained. Here, we present the entire interaction scheme between KM1-GTHase and its substrate; our scheme includes a specific mechanism for recognition of the trehalose moiety in the +1 and +2 substrate binding subsites. The recognition mechanism to distinguish the substrate and product is discussed.

Results

Enzymatic activities of the E283Q, D252S, and D252E mutants of GTHase

The α-amylase reaction retains the α-anomeric configuration of the scissile glycosidic bond and is considered to proceed via general acid/base catalysis.^{17, 21} On the basis of the amino acid sequence similarities between the α-amylases, the two acidic residues Glu283 and Asp252 are predicted to be involved in the catalytic function of GTHase as a general acid/base and nucleophile, respectively. Glu283 was mutated to the corresponding amide (Gln) to abolish the general acid/base function and to inactivate the GTHase completely. Moreover, Asp252 was mutated to Ser to change its catalytic function from anomer retaining mechanism to inverting²² and also mutated to Glu with the expectation of formation of an enzyme-substrate adduct, as has been successfully performed in a β-glycosidase (T4-phage lysozyme).^23–25

The enzymatic activities of the wild-type and mutant GTHases were determined using maltotriosyltrehalose (G3-Tre) as a substrate (Table I). Under these conditions, native GTHase showed an enzymatic activity of 611 U/mg. The D252S mutant showed very little enzymatic activity (3.4 U/mg), and the D252E and E283Q mutants had almost no enzymatic activity. Detected enzymatic activities of the D252E and E283Q mutants were less than 0.3 U/mg (0.05% of the activity of wild-type GTHase), which is close to the limit of saccharide detection.

Table I.

Enzymatic Activities for the Mutant GTHases

	Enzymatic activity (U)	Protein amount (mg)	Relative activity (U/mg)
Wild type	501	0.82	611
E283Q	0.252	0.92	0.274
D252S	3.47	1.02	3.40
D252E	0.139	0.81	0.172

Structures of wild-type, D252S, D252E, and E283Q mutant GTHases with substrates

The structure of wild-type GTHase complexed with G3-Tre, where G3-Tre was soaked into the crystal, was determined at 2.65-Å resolution. The general structural features are the same as previously described²⁰; these comprise of domains A, B, and C, which are common within the α-amylase family, domains D and E, which are unique in GTHase, the N-terminal extension of the immunoglobulin type domain connected by an extended linker peptide to the (β/α)₈ barrel catalytic domain, and the covalently linked dimeric structures at Cys298. However, no additional electron density for G3-Tre was observed (Fig. 2).

Figure 2.

Extra electron densities observed in the mutant GTHases. (a) Wild-type. (b) E283Q mutant GTHase in complex with maltotriosyltrehalose (G3-Tre). (c) E283Q mutant GTHase in complex with maltoheptaose (G7). (d) D252S mutant GTHase cocrystallized with G3-Tre. |F_o−F_c| maps are colored in green mesh and sugar models fitted to the |F_o−F_c| map are colored in black bonds.

To further investigate the location of the substrate, the catalytic site mutants of GTHase were crystallized with G3-Tre. The E283Q mutant GTHase in complex with G3-Tre (E283Q-G3-Tre) and the E283Q mutant in complex with maltoheptaose (G7) (E283Q-G7) crystallized isomorphously to the wild-type GTHase (Table II), and their X-ray structures were determined at 2.60- and 2.65-Å resolutions, respectively. These structures revealed extra electron density at the substrate binding region [Figs. 1, 2(b,c)], which were assigned as G3-Tre in E238Q-G3-Tre and maltopentaose (G5) in E283Q-G7. The bound substrate was clearly observed in the E283Q-G3-Tre complex.

Table II.

Data Processing and Refinement Statistics of the Wild Type and Mutant KM1-Amylases

KM1-amylase	Wild type	E283Q	E283Q	D252S	D252S	D252E
Substrate	G3-Tre	G3-Tre	G7	none	G3-Tre	G3-Tre
Data Collection
X-ray source	SPring-8 (BL40B2)	SPring-8 (BL41XU)	SPring-8 (BL41XU)	SPring-8 (BL41XU)	SPring-8 (BL40B2)	SPring-8 (BL41XU)
Detector	ADSC Quantum 4R	Mar CCD 165	Mar CCD 165	Mar CCD 165	ADSC Quantum 4R	Mar CCD 165
Resolution (Å)	2.65 (2.74–2.65)	2.60 (2.69–2.60)	2.65 (2.74–2.65)	2.70 (2.80–2.70)	2.30 (2.38–2.30)	2.40 (2.49–2.40)
Total reflections	178,567	230,847	195,222	166,502	442,848	316,350
Unique reflections	29,375	31,666	29,402	27,427	45,920	40,919
Redundancy	6.1 (3.2)	7.3 (4.7)	6.6 (4.5)	6.1 (3.3)	9.6 (7.1)	7.7 (6.8)
<I/σ(I)>	10.2 (1.8)	13.2 (1.7)	9.3 (1.4)	18.1 (2.9)	40.5 (4.5)	30.5 (5.0)
Rmergea	0.097 (0.296)	0.099 (0.296)	0.086 (0.349)	0.096 (0.193)	0.087 (0.366)	0.060 (0.179)
Completeness (%)	97.3 (84.1)	97.9 (81.7)	97.3 (95.9)	95.7 (89.2)	99.9 (99.3)	99.9 (99.9)
Refinement
Resolution (Å)	2.65 (2.72–2.65)	2.60 (2.67–2.60)	2.66 (2.72–2.66)	2.70 (2.77–2.70)	2.30 (2.36–2.30)	2.40 (2.47–2.40)
used reflections	27856	30005	27909	26037	43509	38820
Rworkb	0.201 (0.341)	0.197 (0.300)	0.175 (0.311)	0.159 (0.219)	0.177 (0.236)	0.158 (0.201)
Rfreeb (5% random)	0.243 (0.366)	0.244 (0.322)	0.227 (0.342)	0.212 (0.295)	0.218 (0.299)	0.198 (0.247)
No. of total atoms	4721	4800	4769	4773	4888	4973
Protein	4513	4550	4550	4553	4553	4538
Water/GOL*/FLC*	177/18/13	163/18/13	144/6/13	165/42/13	252/36/13	392/30/13
Substrate	−	56 (G3-Tre)	56 (G5)	−	34 (G3)	−
Mean B value (Å2)	44.5	45.8	40.7	40.9	47.2	37.0
R.m.s.d. bonds (Å2)	0.007	0.007	0.018	0.021	0.017	0.014
R.m.s.d. angles (°)	1.085	1.134	1.868	1.961	1.694	1.537

^aR_merge = Σ|I −<I>|/ΣI, where I is the intensity of a reflection and is the average intensity.

^bR_work(free) = Σ||F_obs| − |F_calc||/Σ|F_obs|, where F_obs and F_calc are observed and calculated structure factor amplitude, respectively.

^*GOL and FLC are three-letter-codes of glycerol and citrate anion in PDB, respectively.

Figure 1.

Stereo view of the overall dimer structure of E283Q mutant GTHase. The dimer is crosslinked by an intermolecular disulfide bridge at Cys298. Domains (A−E) are colored by red, green, blue, yellow, and cyan, respectively. Green mesh represents |F_o−F_c| omit map contoured 3.0 σ around G3-Tre. Tyr325 is stacking with the subsite −1 glucose. The figures were prepared by PyMOL (http://www.pymol.org/).

The D252S and D252E mutants in complex with G3-Tre (D252S-G3-Tre, D252E-G3-Tre) also crystallized isomorphously to the wild-type GTHase (Table II), and their X-ray structures were determined at 2.30- and 2.40-Å resolution, respectively. The crystal structure of D252S-G3-Tre revealed extra electron density at the substrate binding region [Fig. 2(d)], which was assigned as maltotriose (G3), but no electron density corresponding to the substrate was observed in the structure of D252E-G3-Tre.

To investigate the location of the water molecule that may be activated by the newly introduced Ser252, the crystal structure of D252S mutant GTHase without substrate was determined at 2.70-Å resolution. The differences in the root mean square (RMS) distances for the Cα atoms in wild-type GTHase and in E283Q-G3-Tre, E283Q-G7, D252S-G3-Tre, and D252E-G3-Tre mutant GTHases were 0.26, 0.30, 0.32, and 0.40 Å, respectively. In the crystal structure of the D252S mutant, we could not observe density corresponding to the water molecule interacting with Ser252 because of occupation by glycerol, which was used as a cryoprotectant.

Interaction of the substrate in the E283Q mutant GTHase

From the structures of inactive mutant GTHases in complex with G3-Tre, the entire substrate-binding site was determined. The electron density belonging to the substrate was detected at five subsites, from −3 to +2, which is shown most clearly in the structure of E283Q-G3-Tre [Fig. 2(b)]. The G3-Tre molecule bound GTHase through 21 hydrogen-bond interactions as shown in Table III and Figure 3. The trehalose end of maltooligosyltrehalose and the reducing end of maltooligosaccharide bind towards the C-terminal side of the (β/α)₈ barrel. The nonreducing end of the substrate is expected to exit from the N-terminal side of the barrel at the interface with domain A.

Table III.

(a) Hydrophilic Interaction Between E283Q and G3-Tre; (b) Interaction Between E283Q and G5; (c) Interaction Between D252S and G3

Protein	Subsite number		Distance
(a) Hydrophilic interaction between E283Q and G3-Tre
Asp308	Oδ2	+2	O3	2.7
His312	Nε2	+2	O3	3.0
His312	Nε2	+2	O4	3.2
Gln378	Nε2	+2	O3	2.8
His255	Nε2	+1	O2	3.0
His255	Nε2	+1	O3	3.1
Gln283	Oε1	+1	O3	2.9
Gln283	Nε2	+1	O4	3.0
Asp377	Oδ1	+1	O4	3.1
His192	Nε2	−1	O6	3.3
Arg250	Nη2	−1	O2	3.1
Asp252	Oδ2	−1	O6	2.7
His376	Nε2	−1	O2	2.8
His376	Nε2	−1	O3	2.8
Asp377	Oδ1	−1	O2	2.7
Asp377	Oδ2	−1	O3	2.9
Asp153	Oδ1	−2	O6	2.8
Asp153	Oδ2	−2	O6	3.2
Asn381	Oδ1	−2	O2	2.7
Arg444	Nε	−2	O3	3.2
Pro214	O	−3	O3	3.3
(b) Interaction between E283Q and G5
His255	Nε2	+1	O2	3.2
Gln283	Oε1	+1	O2	2.9
Gln283	Nε2	+1	O3	3.1
Asp377	Oδ1	+1	O3	3.0
Arg250	Nη2	−1	O2	2.9
Asp252	Oδ1	−1	O2	3.3
Asp252	Oδ1	−1	O5	3.3
Asp252	Oδ2	−1	O6	2.6
Gln283	Nε2	−1	O2	3.3
His376	Nε2	−1	O2	2.8
His376	Nε2	−1	O3	2.7
Asp377	Oδ1	−1	O2	2.8
Asp377	Oδ2	−1	O3	2.6
Asp153	Oδ1	−2	O6	2.8
Asn381	Oδ1	−2	O2	2.6
Arg444	Nε	−2	O3	2.9
Glu447	Oε1	−3	O2	3.0
(c) Interaction between D252S and G3
His192	Nε2	−1	O6	3.3
Ser252	Oγ	−1	O1	3.0
Ser252	Oγ	−1	O5	3.2
Arg250	Nη2	−1	O2	3.0
Glu283	Oε1	−1	O1	2.3
His376	Nε2	−1	O2	2.9
His376	Nε2	−1	O3	3.1
Asp377	Oδ1	−1	O2	2.5
Asp377	Oδ2	−1	O3	2.6
Asp153	Oδ1	−2	O6	2.8
Asp153	Oδ2	−2	O6	3.3
Asn381	Oδ1	−2	O2	2.8
Arg444	Nη2	−2	O3	3.1
Pro214	O	−3	O3	3.0

Figure 3.

Schematic view of the hydrogen bonding interactions (dotted) between the E283Q mutant GTHase and G3-Tre. Key hydrophobic residues for substrate recognition, Trp215, Phe325, and Phe352 are also included.

The glucose molecule at subsite +2 of G3-Tre interacts with the distorted α-hairpin structure overhanging from the α6a−α6b and β6b−β6c loops [Fig. 2(b)]; in this interaction, Tyr325, Phe352, and Arg353 residues contribute to form a steric barrier at the +2 subsite [Fig. 2(b)]. The subsite +2 glucose and Tyr325 have potential stacking interactions in which the pyranose and hydroxyphenyl rings are nearly parallel and recognized by four hydrogen bonds; three between the O3 in the subsite +2 glucose moiety and Asp308, His312, and Gln378 and one between O4 in the subsite +2 glucose and His312 [Fig. 4(a)].

Figure 4.

Interaction of saccharide in each substrate-binding subsite of GTHase. (a) G3-Tre bound structure of E283Q mutant GTHase (E283Q-G3-Tre). (b) Structure of pig pancreatic α-amylase (PPA) with acarbose. E283Q-G3-Tre structure (thin bonds in green) is superposed as reference model. (c) Structure of E283Q mutant GTHase with G7 (E283Q-G7). (d) Structure of D252S mutant GTHase with G3-Tre (D252S-G3-Tre). (e) The trehalose bound structure of MTHase from Deinococcus radiodurans. E283Q-G3-Tre structure (thin bonds in green) is superposed as reference model.

At the +1 subsite, the glucose residue exhibits five hydrogen bonds; involving O2 and Nε2 of His255, O3 and Nε2 of His255, O3 and Oε1 of Gln283, O4 and Nε2 of Gln283, and O4 and Oδ1 of Asp377. The subsite +1 glucose makes a van der Waals contact with Trp215 in which the planes of the indole and pyranose rings are almost perpendicular [Figs. 4(a) and 5]. The indole ring of Trp215 is fixed by adjacent Pro214 [Fig. 4(a)].

Figure 5.

(a) Pyranose ring distortion observed at the −1 subsite. Thick bonds represent the structure of E283Q mutant MTHase, while thin bonds (colored in yellow) represent the bound G3-Tre. The glucose molecule in the standard conformation is represented by thin cyan bonds. (b) Superposition of D252S with glycerol and E283Q-G3-Tre. D252S, glycerol, and E283Q-G3-Tre are represented in pink, red, and white, respectively. (c) Superposition of D252E and E283Q-G3-Tre. D252E and E283Q-G3-Tre are colored in blue and white, respectively. A rotamer of Glu252 side chain as a model of covalent enzyme-substrate adduct is shown in the thin stick presentation.

The subsite −1 glucose appears distorted toward a half-chair form and is surrounded by the invariant residues of Tyr152, His192, Trp215, and His376 (Fig. 5). When a glucose with standard conformation was superposed using the C1, C2, and C3 atoms of the sugar ring (RMS deviation = 0.027 Å), the glucose ring distortion is hard to detect at this resolution, but was more obvious for the accompanying positional shift of O6 and C6 of the subsite −1 glucose (positional shift of C6 was 1.6 Å). This distortion is stabilized by forming seven hydrogen bonds, which is the largest number among the five subsites. Of these seven bonds, three are between O2 and Nη2 of Arg250, Nε2 of His376 and Oδ1 of Asp377, two are between O3 and Nε2 of His376 and Oδ2 of Asp377, and two are between O6 and Nε2 of His192 and Oδ2 of Asp252 (Table III and Fig. 5).

The subsite −2 glucose participates in three hydrogen bonds; one between O2 and Oδ1 of Asn381, one between O3 and Arg444, and one between O6 and Oδ1 or Oδ2 of Asp153. The subsite −3 glucose exhibited only one hydrogen bond between O3 of the subsite −3 glucose and O of Pro214. On the basis of all the structural information obtained for the inactive mutants of GTHase, at least five substrate recognition subsites (−3 to +2 subsites) were identified in GTHase.

To further investigate how GTHase distinguishes the correct substrate, G3-Tre and G7 (nonpreferred substrates) were included in the crystallization of the E283Q mutant GTHase. Electron density belonging to a moiety of G5 lying within the −3 to +1 subsites was sufficient to build saccharide structures, although the electron density at the −3 and −2 subsites were relatively weak [Fig. 2(c)]. No electron density corresponding to a saccharide moiety bound at the +2 subsite was observed, and the location of the subsite +1 glucose was shifted outwards from the saccharide-binding cleft as shown in Figure 4(c). Observed hydrogen bonds are listed in Table III.

Saccharide interactions of the D252S and D252E mutant GTHases

Asp252 in GTHase, similar to Asp197 in pig pancreatic α-amylase (PPA),²⁶ participates as a nucleophile that attacks the C1 carbon of the subsite −1 glucose. Therefore, it was reasoned that the substitution of Asp252 with serine (shorter side chain) or glutamate (longer side chain) may result in a change in the anomer form of the product or formation of a covalent adduct.

Electron densities corresponding to an enzymatic product, G3, were observed over the −3 to −1 subsites in the crystal structure of the D252S mutant GTHase, whereas no clear density corresponding to a G3-Tre molecule was observed in the structure of the D252E mutant GTHase. The product bound structure of the D252S is shown in Figure 2(d). Hydrogen bonding interactions were observed between Oγ of Ser252 and O1 of the saccharide molecule at the reducing end. The interactions between G3 and D252S mutant GTHase are listed in Table III. It was also found that the saccharide molecule at the reducing end showed the α-anomer as shown in Figure 4(d).

The D252E mutant was expected to form a covalent enzyme-substrate adduct in which the newly introduced glutamate residue would react directly with the G3-Tre; however, no electron density corresponding to the substrate was observed. Consistent with this interpretation, the location of the side chain of Trp215 was shifted inward, and the side chain occupied the substrate-binding cleft of GTHase accompanying the structural shift of the loop region from 211 to 218.

Discussion

It has been predicted that several substrate-binding subsites and some active site structures act as steric barrier to capture and recognize the substrate at the +2 subsite in GTHase.⁸ To determine the substrate recognition mechanism of GTHase, which is important for improving the substrate specificity and thus increasing the enzymatic yield for industrial trehalose production, we undertook two mutagenic approaches. One was complete inactivation by substitution of catalytic acid/base Glu283 with the corresponding amide, and the other was to trap the enzyme-substrate adduct by substituting Asp252 with two other nucleophiles, namely, serine and glutamate. Although our attempt to obtain an enzyme-substrate adduct was unsuccessful, the structure analysis of E283Q-G3-Tre revealed that at least 5 subsites from −3 to +2, including the +1 and +2 subsites, directly recognize the trehalose moiety of the substrate. This observation was consistent with the result that a substrate longer than G3-Tre shows a similar reaction rate and only one or two glucose residues can be cleaved from maltooligosaccharides.

The structure of bound trehalose reported for maltooligosyltrehalose trehalohydrolase (MTHase; another name of GTHase) derived from Deinococcus radiodurans²⁷ [Fig. 4(e)] (PDB ID: 2BHY) is quite different from our structure for E283Q-G3-Tre [Fig. 4(a)]. The glucose moiety of trehalose interacts with His310 and Glu376 in Deinococcus radiodurans MTHase [Fig. 4(e)] whereas the glucose moiety of G3-Tre at the +2 subsite was surrounded by several residues, Asp308, Tyr325, Phe352, and Arg353 in GTHase. The stacking interaction between the pyranose and hydroxyphenyl ring of Tyr325 observed at the +2 subsite in GTHase was nearly parallel, indicating a stacking interaction for trehalose recognition. In contrast, the stacking interaction with Tyr345, which corresponds to Tyr325 of GTHase, was not observed in the trehalose complex of Deinococcus radiodurans MTHase. Thus, the steric barrier in GTHase is attributed to the residues Tyr325, Phe352, and Arg353 located at the +2 subsite [Fig. 4(a)].

The structural analysis of E283Q-G7 also provided important information about the substrate specificities of KM1-GTHase. It was known that GTHase weakly binds G7 that has α-1,4 linked glucose residues.⁸ The structure analysis of E283Q-G7 revealed that the +2 subsite is vacant and G7 is bound through the +1 to -4 subsites [Figs. 2(c) and 4(c)], indicating the lack of affinity of the +2 subsite to glucose with α-1,4 linkage. Since the α-1,1 linkage of the trehalose moiety gives it a very different shape than the α-1,4 linkage gives to maltose, GTHase has a different stacking interaction of the aromatic side chains (Tyr325 in KM1-GTHase is similar to Tyr151 in PPA²⁶ [PDB ID: 1HX0]) to recognize the α-1,1 linked glucose moiety at the +2 subsite [Fig. 4(a,b)]. To compare recognition mechanisms between trehalose and maltose moieties, we chose PPA because, there is a crystal structure of PPA complexed with acarbose in high-resolution and PPA has high structure similarity against GTHase (Z = 47.5 with 3D library search using MATRAS²⁸).

For this observation to be more useful, we further investigated the conformation of the trehalose moiety in the GTHase-G3-Tre complex and compared it to the calculated minimum energy conformation for trehalose and maltose as shown in Figure 6. The angle of α-1,1 bond (ϕ = −26.8 and ψ = −46.9) was similar to that of the lowest energy conformation of trehalose (ϕ = −48.8, ψ = −48.8).²⁹ Furthermore, the low-energy conformations of maltose (ϕ = −2.31, ψ = −23.6)²⁹ are very different from those of trehalose. Tyr325 provides specific binding with a low energy conformation of the subsite +2 glucose in the trehalose moiety. Only very high-energy conformations of maltooligosaccharides could have the same stacking interaction with Tyr325; this is consistent with the finding of the cocrystallization experiment of E283Q-G7 in which the +2 subsite was empty. This result indicates that the +2 subsite is disfavored for α-1,4 linked glucose moiety relative to α-1,1 linked glucose moiety. It seems that an important aspect of the enzyme's specificity for maltooligosyltrehalose arises from the ability of the α-1,1 linked glucose residue at the +2 subsite to make a stacking interaction with Tyr325, which is then not available to α-1,4 linked glucose residues.

Figure 6.

(a) ϕ and ψ angle definitions of trehalose and maltose, respectively. (b) Comparison of the +1 and +2 conformation and minimum energy conformations of trehalose and maltose. The lowest (#1) and second lowest (#2) energy conformations of trehalose and maltose are shown. The definition of the ϕ and ψ angles are provided at the bottom left.

The mechanism of binding of the natural substrate (G3-Tre) to the inactive E283Q mutant KM1-GTHase, especially the +1 and −1 subsites, was successfully determined. The glucose moiety of G3-Tre bound to the +1 subsite was essentially in the same position and orientation as seen in PPA [Fig. 4(a,b)]. The glucose molecule appears to form hydrogen bonds with His192, Gln283, and Asp377 via its 2-, 3-, 4-, and 6-hydroxyls and has a van der Waals contact with Trp215. Trp215 interacts with both the glucose molecules at the −1 and +1 subsites; in this interaction, the indole ring of Trp215 is located almost perpendicular to the pyranose ring of the +1 subsite and also has contact with the C6 of the subsite −1 glucose (Fig. 5). This hydrophobic contribution of Trp215 to the substrate binding at the −1 and +1 subsites appears unique to GTHase because the indole ring of Trp215 is fixed by adjacent Pro214 and is replaced with two hydrophobic residues (Leu162 and Val163) in PPA, which may conserve the role of Trp215 [Fig. 4(b)]. The subsite −1 glucose of G3-Tre is also stabilized by the invariant residues Tyr152, His192, and His376 [Figs. 4(a) and 5]. Tyr152 makes a stacking interaction with the saccharide ring of the glucose at the −1 subsite (Fig. 5). His192 and His376 form hydrogen bonds with the 6-hydroxyl and the 2- or 3-hydroxyl of the glucose at the −1 subsite via their Nε atoms. These interactions are commonly conserved in α-amylases³⁰ and also shown by His101 and His299 in PPA complexed with acarbose. Together with the collision between the O6 atom and the indole ring of Trp215, these histidine residues seems to stabilize the distorted half-chair conformation proposed to be adopted by this glucose residue in the transition state. Indeed, the mutants of MTHase obtained from S. solfataricus ATCC 35092 that have substitution at Trp218 (W218A and W218F, Trp218 corresponds to Trp215 of KM1-GTHase) show a decrease in specific activities and catalytic efficiencies.³¹

The structure of E283Q with G3-Tre will provide clues as to how to convert the catalytic mechanism of this enzyme. Previously, we found that the location of nucleophile (Asp52) in hen egg white lysozyme is structurally equivalent to Thr26 in T4 phage lysozyme, and the replacement of these positions with a glutamate resulted in forming an enzyme-substrate adduct in both α-glycosidase (T4-phage lysozyme)^22–25 and β-glycosidase (hen egg white lysozyme).³² We also found from the structural analysis of a series of mutant T4 phage lysozymes²³ that the size of Ser/Thr with a bound water molecule is similar to that of a glutamate.³² Therefore, the conversion of a nucleophile at Asp252 to shorter side chains (Thr, Ser) may position the water molecule between the enzyme and substrate and result in changing a double displacement mechanism to a single displacement mechanism. The conversion to a longer side chain (Glu252) forms a stable covalent enzyme-substrate adduct.

Although we could not identify the location of the water molecule interacting with Ser252, the location of O1 of the bound glycerol (suggesting the oxygen position of water) was hydrogen bonded to OG of Ser252 and was located at a distance of 1.4 Å from the C1 of the subsite −1 glucose [Fig. 5(b)]. This suggests that the side chain of Ser252 may provide a catalytic water molecule to cleave the glycosidic bond between subsites −1 and +1 glucose through a single displacement mechanism. In this case, the enzymatic product of the D252S mutant GTHase should be a β-anomer, which is inconsistent with the α-anomer seen in the crystal structure of D252S-G3-Tre. Because of the relatively long period for crystallization, the β-anomer formed after hydrolysis may have spontaneously mutarotated to the more stable α-anomer.

The crystal structure of E283Q-G3-Tre also provided an explanation for the inactivation of the D252E mutant. The substrate-binding site of D252E was empty, despite the presence of G3-Tre in the crystallization solution. When the structure of D252E is superposed on the structure of E283Q-G3-Tre, the side chain of Glu252 forms an ion-pair with adjacent His192. We originally expected that the side chain of Glu252 would be located near the C1 of the subsite −1 glucose [relative rotamer position of Glu252 is represented by thin sticks in Fig. 5(c)]. The shift of the Glu252 side chain towards His192 observed in the crystal structure of D252E mutant appears to block the substrate-binding site at the −1 subsite, which may also lead to loss of the enzymatic activity in the D252E mutant (Table II).

Together with the crystal structures of the above-mentioned catalytic site mutants, the structural analysis of E283Q-G3-Tre suggests the specific substrate recognition mechanism, including the role of the structural features of the substrate-binding site in the recognition of the trehalose moiety and stabilization of the distorted substrate. These findings are important for improving the catalytic activity and substrate specificity of GTHase, especially for the industrial applications of this enzyme.

Materials and Methods

Materials

G3-Tre and G7 were prepared as reported previously⁶. Restriction endonucleases and T4 DNA ligase were obtained from Takara Shuzo (Japan). KOD DNA polymerase was obtained from TOYOBO (Japan). The concentration of purified protein was spectrophotometrically estimated at 280 nm using the previously reported extinction coefficient.²⁰

Mutagenesis

D252S, D252E, and E283Q variants were prepared by substituting the KM1-GTHase open reading frame of pGUSS2¹⁶ with the appropriate synthetic oligonucleotide. Replacement was performed using a mutagenesis kit (LA PCR in vitro Mutagenesis Kit, Takara, Japan) according to the protocol provided by the manufacturer. The XbaI-BglII fragment of pSS was ligated to the XbaI and BamHI sites of pGAPURA1 to construct pGUSS1. The pGAPURA1 construct was prepared by inserting a NotI-PstI fragment containing the promoter and terminator sequences of the GAP gene isolated from the plasmid pGAPPT10³³ into the NotI and PstI sites of pURAL10, which contains the CYHr gene as well as the 5′ and 3′ fragments of the URA3 gene from C. utilis.³³ The plasmid pGUSS2 for high-level production of α-amylase was constructed in the same manner using pURAL11.³³ The plasmid was used for transformation after cutting at the BglII sites to stimulate integration.

Cultivation

For production of α-amylase, seed cultures, grown in 100 mL of yeast extract peptone dextrose (YPD) medium in a 500-mL conical flask for 72 h at 30°C, were inoculated in 1.5 L of modified SD medium (3% glucose, 0.5% (NH₄)₂SO₄, vitamin, and rare metals) in a 2.5 L jar for fed-batch fermentation (Tokyo Rika, Japan). The fermentor was operated for 60 h at 30°C and the pH was maintained at 5.0 by adding NH₄OH. A concentrated medium containing glucose, (NH₄)₂SO₄, rare metals, and vitamins was successively fed to the growing cells to obtain a high cell density culture. The glucose concentration was kept between 1 and 5%.

Protein purification

Cells cultured in YPD medium were resuspended in 50 mM sodium acetate (pH 5.5) containing 0.3 mg/mL Zymolyase 100T (Seikagakukogyo, Japan) and incubated at 37°C for 30 min and then at 70°C for 1 h. Cells in suspension were disrupted by vortexing for 5 min with glass beads (diameter, 400−600 μm; Sigma Chemical). Cell debris was removed by centrifugation at 12,000g for 10 min. The soluble fractions were recovered for further purification.

The mutant GTHases were purified as previously described.¹⁵ In brief, the culture supernatant was loaded on the Shodex AsahiPak ES502C anion exchange column (7.5 × 100 mm², Showa Denko K.K, Japan). The protein was eluted by forming a linear gradient of 50 mM sodium acetate buffer (pH 5.0) in the same buffer containing 1.0M sodium chloride at a flow rate of 0.5 mL/min for 40 min.

Crystallization and data collection

Cocrystallization experiments were performed using a mixture of 15 mg/mL (0.25 mM) of GTHase and 5 mM of G3-Tre or G7. The mutant GTHases and G3-Tre/G7 were crystallized isomorphously with the wild-type enzyme at 4°C by the hanging drop vapor diffusion method by using 0.6−1.1 M sodium citrate, 0.1M HEPES at pH 7.5.

Diffraction data of wild-type and mutant KM1-GTHases were measured at beamlines BL40B2 and BL41XU by using SPring-8 (Hyogo, Japan). Reflections of wild-type and D252S and of E283Q and D252E were collected on Quantum 4R (Area Detector Systems Corporation, USA) and Mar165 (Rayonix/Mar, USA) CCD detectors at BL40B2 and BL41XU, respectively. Crystals were soaked in a solution containing 1.35M sodium citrate, 0.1M HEPES (pH 7.5), and 10% (v/v) glycerol and then flash-frozen in a nitrogen gas stream for data collection at 100 K. Datasets of the wild-type, E283Q with G3-Tre, E283Q with G7, D252S with G3-Tre, and D252E with G3-Tre were collected. The diffraction data were processed using the HKL2000 suite of programs.³⁴ All of the crystals belonged to space group P3₂21 with unit cell dimensions of a = b = 78 Å, c = 282 Å, and γ = 120°. Details of data collection statistics are shown in Table II.

Structure determination

Initial phase information was obtained by molecular replacement methods using the program MOLREP³⁵ with our previous structure, KM1-GTHase²⁰(PDB ID: 1EH9), as the search model. Each KM1-GTHase molecule was found in the asymmetric unit with a solvent content of 69%. The refinement process for each model was carried out by using the programs CNS 1.21³⁶ and REFMAC 5.5.³⁷ Modification of the model, the initial selection, and manual verification of water molecules were performed with the program Coot.³⁸ The refinement statistics are also given in Table II.

Enzymatic activity

Enzymatic activities of the wild-type and mutant GTHases were assayed as reported previously.^{8, 39} In brief, the KM-1 GTHase was incubated in 10 mM maltotriosyltrehalose in 50 mM sodium acetate buffer (pH 5.5) at 60°C. After the reaction was stopped by heating at 100°C for 5 min, the trehalose produced was analyzed by HPLC using a TSKgel Amide-80 column (4.6 × 250 mm², Tosoh, Japan). Trehalose was eluted by 72.5% acetonitrile-deionized water solution and detected with a differential refractometer (RID-6A, Shimadzu, Kyoto, Japan). One unit (U) of GTHase was defined as the amount of enzyme that would liberate 1 μmol of trehalose from maltotriosyltrehalose per min at 60°C.

Glossary

Abbreviations

G3

maltotriose

G3-Tre

maltotriosyltrehalose

G5

maltopentaose

G7

maltoheptaose

PPA

pig pancreatic α-amylase