Structure–function analysis of silkworm sucrose hydrolase uncovers the mechanism of substrate specificity in GH13 subfamily 17 exo-α-glucosidases

Abstract

The domestic silkworm Bombyx mori expresses two sucrose-hydrolyzing enzymes, BmSUH and BmSUC1, belonging to glycoside hydrolase family 13 subfamily 17 (GH13_17) and GH32, respectively. BmSUH has little activity on maltooligosaccharides, whereas other insect GH13_17 α-glucosidases are active on sucrose and maltooligosaccharides. Little is currently known about the structural mechanisms and substrate specificity of GH13_17 enzymes. In this study, we examined the crystal structures of BmSUH without ligands; in complexes with substrates, products, and inhibitors; and complexed with its covalent intermediate at 1.60–1.85 Å resolutions. These structures revealed that the conformations of amino acid residues around subsite −1 are notably different at each step of the hydrolytic reaction. Such changes have not been previously reported among GH13 enzymes, including exo- and endo-acting hydrolases, such as α-glucosidases and α-amylases. Amino acid residues at subsite +1 are not conserved in BmSUH and other GH13_17 α-glucosidases, but subsite −1 residues are absolutely conserved. Substitutions in three subsite +1 residues, Gln¹⁹¹, Tyr²⁵¹, and Glu⁴⁴⁰, decreased sucrose hydrolysis and increased maltase activity of BmSUH, indicating that these residues are key for determining its substrate specificity. These results provide detailed insights into structure–function relationships in GH13 enzymes and into the molecular evolution of insect GH13_17 α-glucosidases.

Sucrose, α-d-glucopyranosyl-(1→2)-β-d-fructofuranoside, is ubiquitously distributed in plants and is utilized as a carbon source by many organisms. In general, sucrose is hydrolyzed by glycoside hydrolases (GHs) to produce glucose and fructose, which are primary substrates for glycolysis (1, 2). Sucrose-hydrolyzing enzymes are largely divided into two types. β-Fructofuranosidase (invertase) recognizes a β-fructofuranosyl residue and hydrolyzes substrates via a covalent fructosyl-enzyme intermediate (3). Sucrose α-glucosidase (sucrase) recognizes an α-glucopyranosyl residue and hydrolyzes the α-glucosidic linkages of sucrose and maltose (4). According to the CAZy database (RRID:SCR_012909) (5), β-fructofuranosidases belong to GH family 32 and GH68 that form the clan GH-J and share five-bladed β-propeller folded catalytic domains (3). Sucrose α-glucosidases that show relaxed substrate specificity (e.g. sucrase-isomaltase in mammals) are categorized in GH31 (4), and sucrose-specific α-glucosidases are identified as GH13 from Xanthomonas bacteria and lepidopterans (6–9) and as GH100 from bacteria and plants (10, 11). GH13 and GH31 enzymes employ a retaining mechanism (12, 13), whereas GH100 enzymes are proposed to hydrolyze sucrose by an inverting mechanism analogous to other inverting α-glucosidases (11, 14–17). Sucrose is also a substrate for GH13 amylosucrase (18, 19) and GH70 glucansucrase (20, 21) that produce α-glucose polymers and for GH13 sucrose phosphorylase, which catalyzes the phosphorolysis of sucrose instead of hydrolysis (22, 23).

GH13 is a large GH family, with more than 90,000 protein sequences in the CAZy database, and comprises various glycosidases and transglycosylases active on α-glucosidic bonds, such as α-amylase, pullulanase, α-glucosidase, and cyclodextrin glucanotransferase. More than 100 GH13 enzyme structures have been determined, and they share a domain architecture comprised of three domains: A, B, and C. Domain A is the catalytic domain that displays a (β/α)₈-barrel fold (24). To date, this family is further divided into 42 subfamilies (GH13_1 to GH13_42) (25). GH13_17 is mainly composed of α-glucosidases active on maltooligosaccharides and their homologous proteins in insects only (26–30). Several hymenopteran and dipteran GH13_17 α-glucosidases have been cloned and enzymatically characterized, and some enzymes have activity for maltooligosaccharides and sucrose (31–35). Wang et al. (8) identified sucrose-specific hydrolases (SUHs) from lepidopterans Bombyx mori, Trilocha varians, and Samia cynthia ricini, which are homologous to GH13_17 α-glucosidases and hydrolyze sucrose but not other α-glucosides, such as maltose, isomaltose, and trehalose. SUHs are membrane-associated enzymes and are expressed in the midguts of these lepidopterans, where GH32 β-fructofuranosidase is also expressed, to digest sucrose (8, 36). There were very few studies on structure–function relationships of insect GH13 enzymes; the structures of GH13_15 α-amylase from yellow meal worm and most recently of the GH13_17 α-glucosidase (Cqm1) from a mosquito have been determined (37, 38). However, in the latter case, only an apo form of the enzyme is available; thus, the relationships between the structure and substrate specificity of GH13_17 enzymes are still unclear.

In this study, we examined the crystal structures of B. mori SUH (BmSUH) in an apo form and in complexes with ligands, including substrates and inhibitors. The structure of the covalent intermediate was also determined using a synthetic substrate, revealing conformational changes of the enzyme and the complete conformational itinerary of substrates during hydrolysis. Combined with mutational analysis, amino acid residues important for substrate specificity were identified. This study provides novel molecular insights into the catalytic mechanisms and substrate specificity of GH13 enzymes.

Results and discussion

Expression and characterization of recombinant BmSUH

The N-terminally His-tagged BmSUH without its transmembrane region (residues 1–29) was expressed in E. coli BL21 (DE3). Initially, the recombinant enzyme was induced in the host cultured in a Luria–Bertani (LB) medium without additive, and it showed low activity with a low yield (∼0.6 mg protein/liter). When host cells were cultured and induced in LB supplemented with 10 mm CaCl₂, the final yield of recombinant BmSUH reached ∼3 mg from 1 liter of the culture, suggesting that BmSUH may require calcium ion for proper folding. The optimum pH and temperature of the purified enzyme were 8.0 and 30 °C, respectively, when using sucrose as a substrate (Fig. S1). The enzyme was stable (>80% residual activity) up to 30 °C after a 30-min incubation and in a pH range of 6.0–11. BmSUH had rather strict substrate specificity toward sucrose and was slightly active on isomaltulose, 1-kestose, nystose, and maltooligosaccharides (from maltose to maltohexaose). Among synthetic substrates, α-glucosyl fluoride (GlcF) was hydrolyzed by BmSUH, whereas p-nitrophenyl α-glucopyranoside was not (Table 1). K_m, k_cat, and k_cat/K_m values for sucrose were 0.92 mm, 41.2 s⁻¹, and 44.7 s⁻¹ mm⁻¹, respectively (Table 2).

Table 1.

Substrate specificity of recombinant BmSUH

Substrate	Relative activitya
	%
Sucrose [Glc-α(1↔2)β-Fru]	100 ± 2a
Turanose [Glc-α(1→3)-Fru]	NDb
Isomaltulose [Glc-α(1→6)-Fru]	1.0 ± 0.1
1-Kestose [Glc-α(1↔2)β-Fru-(1←2)β-Fru]	1.6 ± 0.1
Nystose [Glc-α(1↔2)β-Fru-(1←2)β-Fru-(1←2)β-Fru]	0.5 ± 0.1
Trehalose [Glc-α(1↔1)α-Glc]	ND
Kojibiose [Glc-α(1→2)-Glc]	ND
Nigerose [Glc-α(1→3)-Glc]	ND
Maltose [Glc-α(1→4)-Glc]	0.05 ± 0.03
Maltotriose	0.6 ± 0.1
Maltotetraose	0.06 ± 0.02
Maltopentaose	0.06 ± 0.02
Maltohexaose	0.04 ± 0.03
Isomaltose [Glc-α(1→6)-Glc]	ND
Raffinose [Gal-α(1→6)-Glc-α(1↔2)β-Fru]	ND
α-Glucosyl fluoride	69 ± 4
p-Nitrophenyl α-glucopyranoside	ND

^a Hydrolytic activity toward sucrose was taken to be 100% (24.1 ± 0.5 μmol min⁻¹ mg⁻¹).

^b ND, not detected.

Table 2.

Kinetic parameters of recombinant BmSUH and its mutants for sucrose and maltotriose compared with GH13_17 α-glucosidases

Enzyme	Sucrose				Maltotriose				Reference
	kcat	Km	kcat/Km	Relative kcat/Kma	kcat	Km	kcat/Km	Relative kcat/Km
	s−1	mm	s−1 mm−1	-fold	s−1	mm	s−1 mm−1	-fold
BmSUH									This study
WT	41.2 ± 0.6	0.92 ± 0.06	44.7	1	0.29 ± 0.01	6.73 ± 0.66	0.043	1
Q191V	26.0 ± 0.7	1.36 ± 0.14	19.1	0.43	0.51 ± 0.01	3.09 ± 0.21	0.165	3.8
D247N (nucleophile)	NDb	ND	ND	ND
Y251H	40.4 ± 0.9	1.95 ± 0.17	20.7	0.46	2.09 ± 0.06	1.63 ± 0.39	1.28	30
E322Q (acid/base)	ND	ND	ND	ND
E440A	14.9 ± 0.5	0.89 ± 0.14	16.7	0.37	1.30 ± 0.02	1.56 ± 0.16	0.833	19
HBG-II	87.6	30.6	2.91		87.2	3.82	22.8		33
HBG-III	222	42.3	5.27		133	8.56	15.6		33
Cqm1	329	7.74	44.6		320	2.18	147		34

^a Normalized to the k_cat/K_m value of WT BmSUH toward each substrate.

^b ND, not detected.

Overall structure of BmSUH

The crystal structure of BmSUH was determined at a resolution of 1.85 Å using the molecular replacement method with Bacillus licheniformis GH13_29 trehalose-6-phosphate hydrolase (32% sequence identity, PDB entry 5BRQ) (39) as a search model because no GH13_17 structure was initially available during this study.

In addition, we determined eight structures complexed with ligands at 1.60–1.90 Å resolution, including WT enzyme complexed with glucose (BmSUH-Glc) and three inhibitors (BmSUH-DNJ, -DAB, and -ACR); catalytically inactive mutants D247N and E322Q complexed with sucrose (D247N-Suc and E322Q-Suc); E322Q covalent intermediate generated using GlcF (E322Q-GlcF); and its complex form with fructose (E322Q-GlcF-Fru) (Table 3). The details of mutants and ligand complexes are described below. All crystals belong to the space group P2₁2₁2₁ and contain two molecules in an asymmetric unit. The monomer of BmSUH contained four domains: a catalytic domain A (residues 30–146, 220–419, and 500–520); domain B (residues 147–219); domain B′ (residues 420–499); and domain C (residues 521–606) (Fig. 1A). The domains A, B, and C are generally conserved in GH13 enzymes: domain A adopts a (β/α)₈-barrel fold; domain B is inserted into each catalytic domain A and consists of five β-strands and one α-helix; and domain C adopts a β-sandwich fold. A structural homology search using the Dali server (40) reveals high structural similarity to Culex quinquefasciatus GH13_17 maltase (Cqm1; PDB entry 6K5P) (38), GH13_31 sucrose isomerases (41–44), GH13_31 α-1,6-glucosidases (45–49), GH13_31 α-1,4-glucosidases (50, 51), GH13_23 α-glucosidases (52), GH13_16 trehalose synthases (53–56), GH13_29 trehalose 6-phosphate hydrolase (39), and GH13_40 oligo-α-1,6-glucosidase (57) (Table S1). All of these enzymes are exo-glycosidases active on α-glucosides and have domain B′, which is inserted in each catalytic domain A (Fig. S2).

Table 3.

Data collection and refinement statistics

	BmSUH Apo	BmSUH-Glc	BmSUH-DNJ	BmSUH-DAB	BmSUH-ACR	D247N-Suc	E322Q-Suc	E322Q-GlcF	E322Q-GlcF-Fru
Data collection
Beamline	PF BL5A	PF BL5A	PF AR-NW12A	PF BL5A	PF AR-NW12A	PF BL5A	PF BL5A	PF AR-NW12A	PF AR-NW12A
Wavelength (Å)	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000	1.0000
Space group	P212121	P212121	P212121	P212121	P212121	P212121	P212121	P212121	P212121
Cell dimensions
a, b, c (Å)	65.9, 145.7, 153.8	64.7, 128.5, 154.0	65.6, 146.8, 154.0	65.1, 146.4, 153.3	65.3, 145.8, 152.9	64.9, 128.3, 154.8	64.7, 127.9, 154.3	64.4, 128.3, 154.4	65.4, 147.1, 153.5
Resolution range (Å)	50–1.85 (1.92–1.85)	50–1.70 (1.73–1.70)	50–1.90 (2.00–1.90)	50–1.75 (1.84–1.75)	50–1.75 (1.84–1.75)	50–1.85 (1.95–1.85)	50–1.84 (1.94–1.84)	50–1.70 (1.79–1.70)	50–1.60 (1.69–1.60)
Measured reflections	751,624	855,106	1,546,590	973,471	1,815,578	479,722	727,454	662,288	1,241,043
Unique reflections	127,062	141,639	117,900	148,023	147,455	105,020	111,648	140,237	194,405
Completeness (%)	100 (100)a	100 (100)	100 (100)	100 (100)	100 (100)	95.1 (92.9)	99.9 (100)	99.5 (99.1)	99.7 (100)
Redundancy	5.9 (5.4)	6.0 (6.0)	13.1 (12.8)	6.6 (6.7)	12.3 (11.9)	4.6 (3.9)	6.5 (6.7)	4.7 (4.3)	6.4 (6.6)
Mean I/σ(I)	19.5 (2.5)	13.5 (2.1)	18.4 (3.0)	15.4 (2.2)	18.7 (2.8)	6.5 (2.2)	12.4 (2.4)	7.9 (2.0)	16.9 (2.6)
Rmerge	0.083 (0.620)	0.094 (0.769)	0.099 (0.933)	0.074 (0.889)	0.080 (0.805)	0.120 (0.491)	0.097 (0.775)	0.127 (0.683)	0.061 (0.650)
CC1/2	0.992 (0.786)	0.998 (0.674)	0.999 (0.850)	0.999 (0.755)	0.999 (0.846)	0.985 (0.778)	0.998 (0.769)	0.989 (0.672)	0.999 (0.833)
Refinement statistics
Rwork/Rfree	0.157/0.186	0.155/0.182	0.173/0.197	0.162/0.184	0.199/0.228	0.192/0.216	0.180/0.211	0.183/0.211	0.178/0.207
RMSDb
Bond lengths (Å)	0.007	0.008	0.010	0.005	0.011	0.010	0.010	0.006	0.013
Bond angles (degrees)	1.392	1.458	1.516	1.301	1.695	1.600	1.600	1.345	1.780
No. of atoms
Protein	9,327	9,297	9,308	9,312	9,304	9,258	9,277	9,266	9,406
Ligand/Ion	45	43	51	64	94	52	52	26	85
Water	1,223	1,066	872	1,021	610	566	740	822	983
Average B (Å2)
Protein	23.6	17.7	31.6	30.4	32.1	24.7	27.9	26.6	31.2
Ligands	40.5	20.3	38.9	40.1	30.9	21.2	23.5	24.4	31.1
Water	31.6	27.2	35.7	37.2	33.6	25.1	31.1	31.5	38.0
Ramachandran plot
Favored (%)	98.2	97.8	97.6	97.1	97.2	97.9	97.5	98.1	97.6
Outliers (%)	0.1	0.1	0.1	0.1	0.0	0.2	0.2	0.2	0.1
PDB codes	6LGA	6LGB	6LGC	6LGD	6LGE	6LGF	6LGG	6LGH	6LGI

^a The values for the highest resolution shells are given in parentheses.

^b Root mean square deviation.

Figure 1.

Overall structure of BmSUH. A, ribbon model of the BmSUH monomer. The catalytic (β/α)₈ barrel A-domain is shown in red, domain B is green, domain B′ is cyan, and domain C is yellow. Calcium and magnesium ions are indicated as slate blue and light green spheres, respectively, and glucose at subsite −1 is shown as a black stick model. The N and C termini are indicated as Nt and Ct, respectively. B, molecular surface and ribbon models of BmSUH dimer. One protomer is shown in the same colors in A, and the other is shown in gray.

Electron density maps for two metal ions were found in domain A. One (site I) is in a solvent-accessible loop of domain A and is hexacoordinated with Asp⁶³, Asp⁶⁵, Asp⁶⁷, Asp⁷¹, Leu⁶⁹, and one water molecule (Fig. S3). The other site (site II) is located at the interface between domain A and domain B and is heptacoordinated with Asn¹⁴⁴, Asp²¹⁷, Tyr²⁵¹, Leu²⁵², Glu²⁵⁴, and two water molecules. Considering the experimental conditions and the electron density maps after refinement, the former metal was assigned as magnesium and the latter as calcium. Among the subfamilies structurally homologous to GH13_17, site I is conserved in subfamilies GH13_16, 23, 29, 31, and 40, whereas site II is conserved only in GH13_16 (Fig. S2).

The molecular masses of BmSUH calculated by its amino acid sequence and calibrated by gel filtration chromatography were 68.7 and 151.3 kDa, respectively (data not shown), suggesting that the enzyme forms a dimer in the solution. The analysis using the Protein Interfaces, Surfaces, and Assemblies (PISA) server (58) revealed that BmSUH is dimeric via 16 hydrogen bonds and six salt bridges (Fig. 1B and Fig. S4). The buried interface area is 1,131 Ų (5.0% of the monomer surface). Seventeen residues per monomer are involved in these interactions, with 11 residues located in domain A and the rest in domain C. The dimeric state is similar to that of Cqm1 (38), but BmSUH has another interaction interface between each loop (Glu²⁵⁴–Tyr²⁸⁷) inserted in the domain A (β/α)₈-barrel (Fig. S4).

Complex structures with substrates and products

To identify residues involved in substrate recognition and the hydrolytic mechanism of BmSUH, the crystal structure of the enzyme complexed with Glc (BmSUH-Glc) was determined. An electron density map for an α-glucose molecule was found at subsite −1. Glucose interacted with Asp¹⁰², His¹⁴⁵, Glu³²², His³⁸⁸, Asp³⁸⁹, and Arg⁴⁵⁵ residues via hydrogen bonds and with Tyr¹⁰⁵ by hydrophobic stacking (Fig. 2A). These residues are completely conserved in GH13_17 and related subfamilies GH13_16, 23, 29, 31, and 40 (Figs. S5 and S6). GH13 enzymes hydrolyze α-glucosidic linkages using a retaining mechanism. Similar to other similar enzymes, Asp²⁴⁷ and Glu³²² were identified as nucleophilic and acid/base catalytic residues, respectively. In support, mutants D247N and E322Q lost hydrolytic activity toward sucrose (Table 2).

Figure 2.

Active sites of BmSUH complexes with substrates, intermediates, and products. Active-site structures of BmSUH-Glc (A), D247N-Suc (B), E322Q-Suc (C), E322Q-GlcF (D), and E322Q-GlcF-Fru (E). The side chains of the amino acid residues and ligands are indicated as stick models, and water molecules interacting with ligands are shown as red spheres. |F_o| − |F_c| omit maps (contoured at 2 σ) for ligands and hydrogen bonds are shown as blue mesh and a dashed line, respectively. Labels of catalytic residues are highlighted in red. Colors used are as follows: amino acid residues (green), glucose and its covalent intermediate (yellow), sucrose (pink), and fructose (cyan).

Subsequently, the structures of D247N complexed with sucrose (D247N-Suc) and E322Q complexed with sucrose (E322Q-Suc) were determined. Clear electron density maps for sucrose were found at subsite −1 to +1 in both structures, and conformations of sucrose and interacting residues are almost identical (Fig. 2, B and C). Thus, the results using E322Q-Suc are used in the following discussions.

The orientation and ⁴C₁ conformation of the sugar ring of the glucose residue in E322Q-Suc are identical to those of the α-glucose in BmSUH-Glc. The fructose residue of sucrose forms hydrogen bonds with Gln³²² (acid/base), Asp³⁸⁹, and Glu⁴⁴⁰ (i.e. fewer bonds than the glucose residue). Tyr³²⁴ is located between the O1 atom of the fructose residue and the entrance of the active site, suggesting that longer substrates with a β-2,1-fructoside linkage (e.g. 1-kestose and nystose) have difficulty binding to the active site. This possibility is consistent with enzyme assays that show less activity toward such oligosaccharides (Table 1).

Trapping the covalent intermediate and conformational changes in the catalytic cycle

To completely understand the structural mechanism of BmSUH hydrolysis, the crystal structures of covalently bound intermediates, where glucose residue binds the nucleophilic catalytic residue of the enzyme, were determined by X-ray crystallography. Crystallizing E322Q in the presence of the synthetic substrate GlcF succeeded in trapping a covalent intermediate (E322Q-GlcF) at the active site (Fig. 2D). Furthermore, the E322Q crystal prepared in the presence of GlcF and fructose provided the structure of a covalent intermediate with fructose at subsite +1 (E322Q-GlcF-Fru) with the same orientation as the fructose residue in E322Q-Suc (Fig. 2E). In both structures, covalently bound glucose forms a ⁴C₁ conformation and interacts with the same residues as glucose molecules in BmSUH-Glc and E322Q-Suc, except that Gln²¹⁵ forms an additional hydrogen bond with the O6 atom of covalently bound glucose. Gln²¹⁵ is highly conserved among GH13_17 and other GH13 enzymes (Figs. S5 and S6), suggesting that it may have an important role in stabilizing the covalent intermediate.

Compared with ligand complex structures, the conformations of amino acid residues around subsite −1, including the catalytic residues, were remarkably different (Fig. 3A). The conformations of residues in the Michaelis complex (E322Q-Suc, E·S) are almost identical to the conformation of the ligand-free structure. By contrast, the conformation of Phe¹⁴¹, Val¹⁴², and Leu²⁴⁶ changes in the E322Q-GlcF-Fru complex (covalent intermediate, E-I·P₁ in Fig. 3A). In particular, the main chains of Phe¹⁴¹ and Val¹⁴² get closer to subsite −1—their Cα atoms move by 1.8 and 1.1 Å, respectively—and the orientation of the side chains changes accordingly. No remarkable difference was observed between covalent intermediates with (E-I·P₁) and without fructose (E322Q-GlcF, E-I). In BmSUH-Glc (E·P₂), the orientation of the catalytic acid/base Glu³²² changes, and the catalytic nucleophile Asp²⁴⁷ points away from the C1 atom of glucose. Accordingly, the side chains of Asp¹⁴⁰ and Arg²⁴⁵ move to avoid steric hindrance with Asp²⁴⁷. Thus, conformations in the catalytic cycle of BmSUH can be divided into three states: open, semi-closed, and fully closed (Fig. 3A). No such conformational changes in the catalytic site appear to have been reported for other GH13 enzymes, including exo- and endo-acting forms. The function of these conformational changes is not clear, but they may contribute to the stabilization of the covalent intermediate during hydrolysis.

Figure 3.

Complete structural mechanism of sucrose hydrolysis by BmSUH. A, conformational changes in the active site during sucrose hydrolysis. E, enzyme; S, substrate; I, covalent intermediate; P₁, product fructose; P₂, product glucose; A/B, acid/base catalyst; Nuc, nucleophilic catalyst. The amino acid residues of E (white), E·S (pink), E-I·P₁ (cyan), E-I (green), and E·P₂ (yellow) states are indicated as sticks and their ligands as thin sticks. The distance between an oxygen atom of Asp²⁴⁷ nucleophilic catalyst and C1 atom of the glucose residue of substrate in the Michaelis (E·S) complex is shown as a green dashed line. The stick models of amino acid residues in a preceding state are superposed for transparency, and arrows indicate conformational changes of the residues. B, conformational itinerary of glucose during BmSUH hydrolytic reaction.

Through all steps in the BmSUH hydrolysis, including the covalent intermediate, the pyranose ring of glucose adopts a ⁴C₁ conformation. The oxocarbenium ion in transition states before and after the covalent intermediate state may adopt a ⁴H₃ half-chair, supported by the QM/MM analysis of GH13 amylosucrase that hydrolyzes sucrose (59). The conformational itinerary of BmSUH hydrolysis is suggested to be as follows: ⁴C₁ → [⁴H₃] → ⁴C₁ → [⁴H₃] → ⁴C₁ (Fig. 3B). To date, the covalent intermediates of 10 GH13 enzymes have been identified using their substrates, covalent inhibitors (2-deoxy-2-fluoro-α-glycosyl fluorides and 5-fluoro-α-glycosyl fluorides), and a combination of α-glycosyl fluoride and catalytic acid/base mutagenesis (12, 23, 52, 60–66). Their sugar ring conformations take a ⁴C₁ conformation, except for the covalent intermediates of Bifidobacterium adolescentis GH13_18 sucrose phosphorylase and Chlamydomonas reinhardtii GH13_11 isoamylase 1, where the sugar ring was distorted toward ¹S₃ skew-boat and half-chair conformations, respectively (23, 65). These differences are perplexing, but the QM/MM analysis using GH13_4 amylosucrase that showed ⁴C₁ and E₃ conformations can be seen in the covalent intermediate (59). GH31 α-glucosidases are also retaining enzymes, and the sugar ring is distorted into a ¹S₃ skew-boat conformation in their glycosyl–enzyme intermediates (67, 68). Consequently, the difference in the conformational itinerary among GH13 subfamilies may depend on their active-site architectures and substrate structures.

Complexes with inhibitors

BmSUH and its lepidopteran orthologs are reportedly inhibited by 1-deoxynojirimycin (DNJ) and 1,4-dideoxy-1,4-imino-d-arabinitol (DAB) (Fig. 4, A and B), which are observed in the latex of mulberry (8, 69). BmSUH was reported to be less sensitive to these compounds than the other lepidopteran SUHs (8). However, inhibitory mechanisms had not been investigated. BmSUH activity was competitively inhibited by DNJ and DAB (Fig. 4, D and E), with a K_i value for DAB of 4.2 µm, considerably lower than the K_i for DNJ (290 µm). Interestingly, BmSUH was also competitively inhibited by acarbose (ACR) (Fig. 4, C and F), which is a maltotetraose mimic inhibitor, even though maltooligosaccharides were poor substrates for BmSUH (Table 1). ACR is a typical inhibitor toward α-amylases and α-glucosidases (34, 70–72) but not GH13_31 sucrose isomerase (41). The K_i of acarbose for BmSUH hydrolysis was 424 µm, which is higher than the K_i of GH13 α-amylases and α-glucosidases. These enzymes show a wide range of K_i values from nanomolar to micromolar levels (71, 73, 74).

Figure 4.

Inhibitors for sucrose hydrolysis by BmSUH. A–C, chemical structures of DNJ (A), DAB (B), and ACR (C). D–F, Lineweaver–Burk plots of BmSUH activity toward sucrose in the presence of DNJ (D), DAB (E), or ACR (F). Concentrations of inhibitors were as follows: 0 (open square), 0.5 (gray square), and 1.0 mm (black square) for DNJ and ACR; 0 (open square), 0.1 (gray square), and 0.5 mm (black square) for DAB. G–I, structures of the active sites in BmSUH-DNJ (G), BmSUH-DAB (H), and BmSUH-ACR (I) complexes. Colors are used in the same manner as in Fig. 2, and DNJ, DAB, and ACR are shown in magenta, brown, and orange, respectively.

To obtain structural insights into the inhibitory mechanism, the crystal structures of BmSUH complexed with three inhibitors at 1.75–1.90 Å resolutions (Table 3) were examined. Clear electron density maps for DNJ and DAB were observed at subsite −1, and ACR occupied subsites −1 to +3 (Fig. 4, G–I). Compared with the acarviosine moiety of ACR that is recognized by several amino acids via hydrogen bonds, the reducing-end maltose residue forms fewer hydrogen bonds with the carbonyl oxygen of Thr³⁵³ and the side chain of Asn³⁹⁰ and is exposed to the solvent outside the active site (Fig. 4I). Although Cqm1 structure in the complex with substrates has not been determined, the superimposition of BmSUH-ACR and E322Q-Suc reveals that the orientation of sugar rings at subsite −1 is identical, and an imino linkage of ACR (corresponding to an α-1,4-glycosidic linkage of maltooligosaccharide) is located at the proper position to interact with catalytic residues (Fig. 5, A and B). However, the second sugar residue of ACR is in 6-deoxy form, and not enough space is available for an additional 6-hydroxy group of maltooligosaccharides. Thus, subsite +1 architecture may not be suitable for maltooligosaccharide substrate, because of the steric interference, resulting in a low hydrolytic activity (Table 1).

Figure 5.

Amino acid residues important for substrate specificity. A, structural comparison of the active sites of BmSUH (green) and Cqm1 (slate blue) in stereo. The side chains of the amino acid residues around subsite +1 are indicated as sticks, and residues 428–457 of BmSUH and the corresponding region (residues 376–388) of Cqm1 are displayed as ribbon models. Sucrose (cyan) and ACR (orange) derived from E322Q-Suc and BmSUH-ACR are superimposed and indicated as thin stick models. B, molecular surface of the catalytic site of BmSUH. The side chains of the subsite +1 residues (green for BmSUH and slate blue for Cqm1), sucrose (cyan), and an acarviosine moiety of ACR (orange) are indicated. Solvent-accessible areas of Gln¹⁹¹, Tyr²⁵¹, and Glu⁴⁴⁰ are highlighted in red. C, sequence alignments of regions around Gln¹⁹¹, Tyr²⁵¹, and Glu⁴⁴⁰ and their corresponding regions of GH13_17 sucrose hydrolases and maltases. The conserved residues are highlighted in red; Gln¹⁹¹, Tyr²⁵¹, and Glu⁴⁴⁰ of BmSUH and conserved residues of the other GH13_17 enzymes are in cyan. D, hydrolytic activity of BmSUH and its mutants toward sucrose and maltooligosaccharides (from maltose to maltohexaose). Bar charts and error bars, means and S.D., respectively, from triplicate experiments. QV, Q191V; YH, Y251H; EA, E440A; QV-YH, Q191V/Y251H; QV-EA, Q191V/E440A; YH-EA, Y251H/E440A; QV-YH-EA, Q191V/Y251H/E440A.

Active-site residues important for substrate specificity

To identify the structural determinants for substrate specificity, site-directed mutations were generated in the catalytic site of BmSUH. The sequence alignment of GH13_17 sucrose hydrolases and maltases demonstrated that amino acid residues around subsite +1 are not completely conserved (Fig. 5, B and C). In dipteran and hymenopteran α-glucosidases, the corresponding residues of Gln¹⁹¹ and Tyr²⁵¹ of BmSUH are valine and histidine, respectively, except for Tyr²²⁷ of honeybee α-glucosidase III (HBG-III). Glu⁴⁴⁰ of BmSUH is completely conserved among lepidopteran SUHs but not in honeybee α-glucosidase II (HBG-II) and HBG-III. Furthermore, the region of residues 430–446 that includes Glu⁴⁴⁰ of BmSUH is lacking in Cqm1 (Fig. 5C). Q191V, Y251H, and E440A mutations were constructed, and their activities toward sucrose and maltotriose were analyzed. All mutations caused a decrease in sucrose hydrolysis activity and enhanced maltotriose-hydrolyzing activity (Table 2). The k_cat values of Q191V and E440A toward sucrose decreased, whereas the k_cat for Y251H was comparable with WT (Table 2). The K_m value of E440A for sucrose was similar to WT, but the other mutations showed higher K_m values, indicating that Glu⁴⁴⁰ and Tyr²⁵¹ influence a catalytic turnover and affinity for sucrose, respectively, and that Gln¹⁹¹ is important for both. These mutations raised k_cat and reduced K_m for maltotriose. Y251H showed the highest catalytic efficiency (k_cat/K_m), 30-fold greater than WT. Ngiwsara et al. (33) reported that a corresponding mutation (Tyr²²⁷ → His) in HBG-III also resulted in a decrease of sucrose hydrolysis and in an increase maltooligosaccharide hydrolysis, indicating that Tyr²⁵¹ is the most important residue for specificity toward sucrose.

Furthermore, double and triple mutations (combination of Q191V, Y251H, and E440A) were assessed for the activity toward sucrose, several maltooligosaccharides from maltose to maltohexaose. Double and triple mutants showed lower activity toward sucrose than the single mutants and higher activity toward maltooligosaccharides (Fig. 5D). Y251H/E440A mutant had 21- and 290-fold higher activity toward maltotriose and maltotetraose than WT, respectively. Interestingly, the triple mutant Q191V/Y251H/E440A had the highest activity toward longer substrates, namely, maltopentaose and maltohexaose, although the underlying reason is unclear. Unfortunately, no crystal structure of GH13_17 maltase complexed with substrates is available, but subsite +1 of BmSUH, which is composed of Gln¹⁹¹, Tyr²⁵¹, and Glu⁴⁴⁰ residues (Q/V/E motif), may be narrower than that for GH13_17 maltases (Fig. 5B) and suitable for the fructose residue of sucrose.

Distribution of sucrose hydrolases within GH13_17

The genomic analyses reveal that many insects, including lepidopteran species, possess several copies of GH13_17 (75). A phylogenetic analysis was completed using 142 sequences of GH13_17 proteins listed in the CAZy database or that were found in a PSI-BLAST search using BmSUH as a query sequence. The phylogenetic tree reveals that GH13_17 can be further divided into several clades (Fig. 6). The sucrose-specific motif (Q/Y/E) is highly conserved among the closest orthologs belonging to the same clades (Lepidoptera SUH1) as BmSUH, except for Danaus plexippus SUH1, where Tyr is substituted to Phe. Recent bioinformatics studies showed that some butterflies possess another paralog termed SUH2, and Dai et al. (9) reported that Papilio xuthus SUH2 has lower activity toward sucrose than P. xuthus SUH1. In the Lepidoptera SUH2 clade, the corresponding motif to the Q/Y/F of SUH1 is Q/H/N (Q, A), suggesting that enzymes belonging to SUH2 are not sucrose-specific and may have different substrate specificity. No protein that has the Q/Y/F motif was found in other clades that include dipteran and hymenopteran proteins. Moreover, lepidopterans have paralogs (Mal1, Mal2, Mal3, and Mal4) that have a different subsite +1 motif compared with SUH1 and SUH2. Taken altogether, lepidopterans may have evolved a unique digestion system for sugars, especially sucrose.

Figure 6.

Phylogenetic tree of GH13_17 proteins and amino acid residues related to substrate specificity. The 142 amino acid sequences were aligned using the MUSCLE program, and the phylogenetic tree constructed by the neighbor-joining method was visualized using the iTOL v5 server. Proteins used were enzymes listed in the CAZy database and their homologs (>40% identity) found using the PSI-BLAST search with BmSUH as a template. Bootstrap values based on 1,000 replicates are shown. Origins, abbreviations, and GenBank^TM ID are labeled and summarized in Table S3. GH13_17 proteins are divided into several clades with different colors based on amino acid residues corresponding to subsite +1 (Gln¹⁹¹, Tyr²⁵¹, and Glu⁴⁴⁰) in BmSUH. Proteins shown in green and blue are genetically identified and enzymatically characterized, respectively. BmSUH and Cqm1 are marked with red circles.

Conclusions

In this study, the crystal structure of GH13_17 sucrose hydrolase, BmSUH, is reported as the first such structure of an insect sucrose hydrolase. BmSUH adopts a domain architecture (domains A, B, B′, and C), such as enzymes belonging to GH13 exo-α-glucosidase subfamilies. BmSUH hydrolyzes sucrose with conformational changes in the active site never reported previously for GH13 enzymes. Subsite +1 residues Q/Y/E determine the strict specificity toward sucrose, and this motif is not found in insects other than lepidopterans. Further investigation, such as the enzymatic characterization, structural analysis, and physiological analysis of B. mori GH13_17 paralogs and other insect orthologs, will enable a complete understanding of carbohydrate digestion and the molecular evolution of related enzymes.

Experimental procedures

Materials and strains

Trehalose was obtained from Nacalai Tesque (Kyoto, Japan). Nigerose, maltose, 1-kestose, nystose, and raffinose were purchased from FUJIFILM Wako Pure Chemical Co. (Osaka, Japan). Maltotriose, maltotetraose, maltopentaose, and maltohexaose were obtained from Hayashibara Co. (Okayama, Japan). Kojibiose, ACR, DNJ, and DAB were purchased from Carbosynth (Compton, Berkshire, UK). Turanose, isomaltulose, and isomaltose were from Tokyo Chemical Industry Co. (Tokyo, Japan). p-Nitrophenyl-α-d-glucopyranoside was obtained from Merck (Darmstadt, Germany). α-d-Glucopyranosyl fluoride was prepared by deacetylation of its tetraacetate derivative (Merck). All other chemicals were reagent grade and obtained from standard commercial sources. E. coli strains DH5α and BL21 (DE3) were used for DNA manipulation and protein expression, respectively.

Cloning, expression, purification, and mutagenesis

First-strand cDNA was synthesized by reverse transcription with total RNA from fifth-instar larvae (Ehime Sanshu, Ehime, Japan) as described previously (76). A transmembrane region of BmSUH (GenBank^TM BAP18683.1) was predicted by the TMHMM server (RRID:SCR_014935). A DNA fragment coding BmSUH without the transmembrane region (Met¹–Leu²⁹) was amplified by PCR using cDNA as a template, KOD-Plus-Neo DNA polymerase (Toyobo, Osaka, Japan), and primers BmSUH_F and BmSUH_R (Table S2). The resultant DNA was digested with NdeI and XhoI (New England Biolabs, Ipswich, MA, USA) and ligated into a pET-28a vector (Merck), followed by DNA sequencing. The recombinant protein had an N-terminal His tag and a thrombin cleavage site (MGSSHHHHHHSSGLVPRGSHM-) prior to Ser³⁰. Site-directed mutagenesis was performed by inverse PCR with the desired primers (Table S2) using the recombinant BmSUH expression plasmid as a template.

E. coli BL21 (DE3) harboring the desired plasmids was grown at 37 °C in 1 liter of LB medium containing 10 mm CaCl₂ and 50 μg/ml kanamycin. When the culture reached an optical density of 0.6–0.8 measured at 600 nm, it was induced with isopropyl-β-d-thiogalactopyranoside at a final concentration of 0.1 mm and further incubated overnight at 20 °C. Cells were harvested by centrifugation at 10,000 × g for 5 min and resuspended in 30 ml of 50 mm sodium phosphate buffer (pH 8.0) containing 20 mm imidazole and 300 mm NaCl before disruption by sonication. The cell lysate was centrifuged at 20,000 × g for 20 min to remove insoluble debris. The supernatant was applied to a nickel (Ni²⁺) nitrilotriacetic acid–agarose (Qiagen, Hilden, Germany) column equilibrated with the same buffer. The column was washed with buffer, and recombinant proteins were eluted with 50 mm sodium phosphate buffer (pH 8.0) containing 250 mm imidazole and 300 mm NaCl. Enzymes were dialyzed against 20 mm Tris-HCl buffer (pH 7.5) and applied to a Mono Q 5/50 GL column (GE Healthcare) and eluted with a linear gradient of 300–600 mm NaCl. Fractions containing active enzymes were concentrated using an Amicon Ultra 30,000 molecular weight cut off (Merck) and further purified by gel filtration chromatography with a Superdex 200 Increase 10/300 column (GE Healthcare) and 20 mm sodium phosphate buffer (pH 7.0) containing 300 mm NaCl. The latter two purification steps were performed using an ÄKTAexplorer system (GE Healthcare). Protein purity was confirmed by SDS-PAGE. Protein concentration was determined by absorbance at 280 nm based on theoretical molar absorption coefficients (127,660 m⁻¹ cm⁻¹) calculated using the ExPASy ProtParam server (RRID:SCR_018087).

Enzymatic assays

The hydrolytic activity toward sucrose and other oligosaccharides was measured in 50-µl reaction mixtures containing 2.0 µg/ml of purified enzyme, 10 mm substrate, and 50 mm HEPES–NaOH buffer (pH 8.0) at 30 °C. After incubation for 15 min, reactions were quenched by boiling for 3 min, and the amount of glucose liberated was measured using the glucose oxidase–peroxidase method with a Glucose C-II Test Kit (Wako Pure Chemicals, Osaka, Japan).

The effect of pH was measured at 30 °C using a 50 mm Britton–Robinson buffer (sodium borate–phosphate–citrate, pH 3.0–8.0) and 10 mm sucrose as the substrate. The effect of temperature was assayed at 25–60 °C using 100 mm HEPES–NaOH buffer (pH 8.0). To test the pH stability, enzymes (1 mg/ml) were incubated at 4 °C for 24 h in 20 mm Britton–Robinson buffer (pH 3.0–8.0). To test thermal stability, enzymes (1 mg/ml) were incubated at 20–65 °C in 20 mm HEPES–NaOH buffer (pH 8.0) for 30 min The remaining activity toward sucrose was examined under standard conditions described above.

Kinetic studies

The initial velocities of hydrolytic reactions for sucrose and maltotriose were determined using the 50 mm HEPES–NaOH buffer (pH 8.0) and at least five concentrations of substrate (0.5–40 mm). Enzyme concentrations were 2.0 µg/ml for sucrose and 20 µg/ml for maltotriose. All kinetic assays were performed at 30 °C. Kinetic parameters were calculated by the nonlinear regression analysis using KaleidaGraph (Synergy Software, Reading, PA, USA). For the inhibition kinetic assay for inhibitors, the same reaction mixtures supplemented with at least four concentrations of each inhibitor were used. Inhibition constants were calculated according to a competitive inhibition model.

Crystallization, data collection, structure determination, and refinement

Before the crystallization, purified proteins were concentrated to 7–10 mg/ml using Amicon Ultra 30K ultrafiltration devices (Millipore). Proteins were crystallized at 20 °C using the hanging-drop vapor diffusion method, in which 1.0 µl of protein solution in 10 mm HEPES–NaOH buffer (pH 7.0) was mixed with an equal volume of a crystallization reservoir solution. Initial crystallization screening was performed using Crystal Screen, Crystal Screen 2, and PEG/Ion Screen kits (Hampton Research, Aliso Viejo, CA, USA). Well-diffracted BmSUH crystals were obtained with a crystallization solution containing 12–18% (v/v) PEG 3,350 (Hampton Research) and 200 mm magnesium acetate. The crystals of WT or mutant enzymes in complex with ligands were obtained by co-crystallization under the same condition in the presence of 10 mm sucrose, α-glucopyranosyl fluoride, fructose or DAB, or 2 mm DNJ or ACR. All crystals were cryoprotected with the reservoir solution supplemented with glycerol at a final concentration of 22% (v/v) and then flash-frozen in liquid nitrogen.

Diffraction data were collected at PF BL5A and PF AR NW12A beamlines (Photon Factory, Tsukuba, Japan). All data were processed and scaled using either HKL2000 (77), Mosflm (78), or XDS (79). Initial structure solutions were obtained using the automated molecular replacement program MrBUMP (80). The best solution was obtained when B. licheniformis GH13_29 trehalose-6-phosphate hydrolase (PDB entry 5BRQ) was used as a search model. Structures complexed with ligands were solved with the molecular replacement method using MOLREP (81), with the unliganded structure as a search model. Refinement was performed using REFMAC5 (82), and manual adjustment and rebuilding of the model were performed using Coot (83). Solvent molecules were introduced using ARP/wARP (84). Structure validation was performed using MolProbity (85). The data collection and refinement statistics are summarized in Table 3. Protein assembly was evaluated by the PISA server (RRID:SCR_015749) (58). Structural figures were prepared using PyMOL (Schrödinger LLC, New York). Coordinates and structure factors were deposited in the Worldwide Protein Data Bank under the accession codes listed in Table 3.

Molecular weight determination

The molecular weights of recombinant BmSUH were determined by gel filtration chromatography in the condition as described above. Calibration was performed using Gel Filtration Calibration Kit HMW (GE Healthcare) containing blue dextran 2,000 (2,000 kDa), thyroglobulin (669 kDa), ferritin (440 kDa), aldolase (158 kDa), conalbumin (75 kDa), and ovalbumin (44 kDa).

Sequence alignment and phylogenetics

Protein sequences were obtained using the CAZy database and PSI-BLAST search with BmSUH as a template. The primary sequence alignment was performed using the MUSCLE program (86). Alignment figures were generated by ESPript 3.0 (87). The phylogenetic analysis of GH13_17 proteins was performed with the neighbor-joining method (88) using multiple alignments prepared as above. The phylogenetic tree was constructed using the iTOL v5 server (RRID:SCR_018174) (89).

Data availability

The atomic coordinates and structure factors have been deposited in the Worldwide Protein Data Bank under accession codes 6LGA, 6LGB, 6LGC, 6LGD, 6LGE, 6LGF, 6LGG, 6LGH, and 6LGI. All other data are contained within the article.