Crystal structure of basic 7S globulin, a xyloglucan-specific endo-β-1,4-glucanase inhibitor protein-like protein from soybean lacking inhibitory activity against endo-β-glucanase


H. Hashimoto, Graduate School of Nanobioscience, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan
Fax: +81 45 508 7365
Tel: +81 45 508 7227


β-Linked glucans such as cellulose and xyloglucan are important components of the cell walls of most dicotyledonous plants. These β-linked glucans are constantly exposed to degradation by various endo-β-glucanases from pathogenic bacteria and fungi. To protect the cell wall from degradation by such enzymes, plants secrete proteinaceous endo-β-glucanases inhibitors, such as xyloglucan-specific endo-β-1,4-glucanase inhibitor protein (XEGIP) in tomato. XEGIPs typically inhibit xyloglucanase, a member of the glycoside hydrolase (GH)12 family. XEGIPs are also found in legumes, including soybean and lupin. To date, tomato XEGIP has been well studied, whereas XEGIPs from legumes are less well understood. Here, we determined the crystal structure of basic 7S globulin (Bg7S), a XEGIP from soybean, which represents the first three-dimensional structure of XEGIP. Bg7S formed a tetramer with pseudo-222 symmetry. Analytical centrifugation and size exclusion chromatography experiments revealed that the assembly of Bg7S in solution depended on pH. The structure of Bg7S was similar to that of a xylanase inhibitor protein from wheat (Tritinum aestivum xylanase inhibitor) that inhibits GH11 xylanase. Surprisingly, Bg7S lacked inhibitory activity against not only GH11 but also GH12 enzymes. In addition, we found that XEGIPs from azukibean, yardlongbean and mungbean also had no impact on the activity of either GH12 or GH11 enzymes, indicating that legume XEGIPs generally do not inhibit these enzymes. We reveal the structural basis of why legume XEGIPs lack this inhibitory activity. This study will provide significant clues for understanding the physiological role of Bg7S.

Coordinates and structure factors have been deposited in the Protein Data Bank Japan (PDBj) ( under the accession number 3AUP.

Structured digital abstract 


Aspergillus niger xylanase


accessible surface area


analytical ultracentrifugation




basic 7S globulin


extracellular dermal glycoprotein




glycoside hydrolase




inhibition loop 1


inhibition loop 2


Protein Data Bank


size exclusion chromatography


Tritinum aestivum xylanase inhibitor


xyloglucan-specific endo-β-1,4-glucanase


xyloglucan-specific endo-β-1,4-glucanase inhibitor protein


The cell wall of plants is composed of various polysaccharides, such as cellulose and hemicellulose. Cellulose is a major component of the plant cell wall, and cellulose microfibrils are linked via hemicellulose. The network of cellulose–hemicellulose provides tensile strength. In most dicotyledonous plants, hemicellulose comprises xyloglucan, which consists of a cellulosic backbone substituted with side chains. These β-linked glucans, namely cellulose and xyloglucan, are constantly exposed to degradation by various endo-β-glucanases, such as cellulase and xyloglucanase from pathogenic bacteria and fungi. To protect the cell wall from degradation by such enzymes, plants secrete proteinaceous inhibitors against endo-β-glucanases. The first endo-β-glucanase inhibitor protein to be discovered was the so-called xyloglucan-specific endo-β-1,4-glucanase inhibitor protein (XEGIP) [1], a tomato protein that inhibits fungal xyloglucan-specific endo-β-1,4-glucanase (XEG), an enzyme classified as a member of the glycoside hydrolase (GH)12 family in the CAZy database [2] ( Tomato XEGIP is a basic 51-kDa protein, and, as its name suggests, inhibits XEG by forming a tightly associated 1 : 1 complex with an inhibition constant (Ki) of ∼ 0.5 nm. XEGIPs have been discovered in various higher plants [3], and some of these proteins have been characterized. For example, carrot XEGIP is termed extracellular dermal glycoprotein (EDGP). It has been shown that EDGP also inhibits XEG from Aspergillus aculeatus [4]. Tobacco XEGIP, termed nectarin IV, has been shown to inhibit XEG and does not inhibit GH11 xylanases [5], although the structures of GH12 and GH11 are very similar.

XEGIPs are structurally related to Tritinum aestivum xylanase inhibitor (TAXI), a xylanase inhibitor protein isolated from wheat [6], because both XEGIP and TAXI have 12 cysteines in similar positions. These cysteines form six disulfide bonds in the tertiary structure of TAXI [7]. To date, four TAXI isomers have been identified in wheat (TAXI-IA, TAXI-IB, TAXI-IIA, and TAXI-IIB). TAXI inhibits GH11 xylanase, whereas it inhibits neither GH12 nor GH10 xylanase. A structural study has revealed that TAXI-IA adopts a pepsin fold lacking proteolytic activity [7]. The structure of TAXI-IA in complex with Aspergillus niger xylanase (ANXY), a GH11 xylanase from Aspergillus niger, coupled with functional studies, has revealed that His374 of TAXI-IA plays a significant role in the inhibition of ANXY, where His374 interacts with the catalytic Glu79 and Glu170 of ANXY [7,8]. Furthermore, it has been reported that the hydrophobic interaction of Leu292 of TAXI-IA with Pro294 of TAXI-IIB regulates the strength of inhibition and specificity for GH11 xylanases [9].

XEGIPs are also found in legumes, including lupin and soybean. γ-Conglutin is a XEGIP found in lupin [3]. In soybean, a XEGIP is the basic 7S globulin (Bg7S) [10]. Soybean Bg7S shares 38% and 37% amino acid identity with tomato XEGIP and EDGP, respectively. Bg7S is initially synthesized as a precursor protein with an N-terminal signal peptide. Bg7S is matured by post-translational modifications: cleavage of the N-terminal 24 residues, formation of disulfide bonds, and cleavage between Ser251 and Ser252, where the numbering starts from the first residues of the matured protein. Mature Bg7S consists of 403 amino acids, and has a molecular mass of 43 kDa; it is composed of 27-kDa (α) and 16-kDa (β) chains [10]. Although tomato XEGIP and EDGP are monomeric proteins, Bg7S exists as an oligomeric form [10,11]. Furthermore, it has been reported that Bg7S binds a 4-kDa hormone-like peptide, termed leginsulin, from soybean [11–13]. However, both the structure and function of Bg7S remain unknown. Here, we report the crystal structure of Bg7S from soybean, and functional analysis of Bg7S.

XEGIPs have been discovered in various plants, including potato (Uniprot ID Q7XJE7; sequence identity with Bg7S, 39%), Arabidopsis (Q8LF70, 38%), rice (A2Y4I2, 36%), and maize (B6UHL4, 26%). Thus, our structural and functional studies on Bg7S will shed light on XEGIPs which are widely conserved in various plants.

Results and Discussion

Structure of Bg7S from soybean

The crystal structure of soybean Bg7S was determined at 1.9-Å resolution. The asymmetric unit contained four Bg7S protomers (A, B, C and D molecules), and they formed a tetramer with pseudo-222 symmetry (Fig. 1A). The N-terminal moieties of the β-chains of the C and D molecules protrude into the AB dimer (Fig. 1A), whereas the corresponding regions of the A and B molecules are disordered. We have obtained a Bg7S crystal with different cell dimensions [14]: Bg7S also forms a tetramer in the same manner in the other crystal form (data not shown). This finding suggests that tetramer formation is not an artefact of crystal packing. The four protomers superimpose well, with an averaged rmsd value of 0.7 Å for comparable Cα atoms (Fig. 1B). This observation indicates that the structures of the four protomers are essentially identical, except for the N-terminal region of the β-chain. Thus, the structure of the A molecule is hereafter considered to be representative of the Bg7S protomer, unless otherwise noted.

Figure 1.

 Structure of Bg7S from soybean. (A) Top and side views of the Bg7S tetramer. A, B, C and D molecules in the asymmetric unit are shown as green, red, yellow and blue ribbon representations, respectively. (B) Superimposed structures of the Bg7S protomers are shown by wire representations. Colors correspond to those in (A). (C) The overall structure of the Bg7S protomer is shown by a ribbon representation. The structure of the A molecule is shown as an example. The N-terminus and C-terminus are labeled. The α-chain and β-chain are shown as green and light blue ribbon representations, respectively. The cysteines involved in the disulfide bonds are shown as stick representations and labeled in black. The disordered regions are shown as dashed lines. The black triangle indicates the post-translational cleavage position. The pseudo-active site of aspartic protease is indicated by the red triangle. (D) Superimposed structures of Bg7S and TAXI-IA (PDB ID 1T6G, chain A) are shown as green and light brown wire representations, respectively. The loops of TAXI-IA involved in interactions with ANXY are labeled IL-1 and IL-2.

Bg7S adopts a β-rich structure with several α-helices (Fig. 1C). Bg7S is post-translationally cleaved between Ser251 and Ser252, resulting in the α-chain and β-chain. Although these chains are intricately folded, the structure of Bg7S is roughly divided into the α-domain and β-domain. Bg7S has 12 cysteines in positions similar to those found in the primary structures of other XEGIPs and TAXIs, and these residues form six disulfide bonds. Because Bg7S is secreted from seeds in response to various stresses, such as heat [15], these disulfide bonds supposedly stabilize the three-dimensional structure of Bg7S. The Cys209–Cys418 bond seems to be significant for stabilization in particular, because it links the α-chain and β-chain (Fig. 1C).

A search for homologous structures of Bg7S by DALI [16] revealed that the structure of Bg7S is similar to those of the xylanase inhibitor TAXI-IA [Protein Data Bank (PDB) ID 1T6G, Z-score = 39.7] (Fig. 1D) and aspartic proteases such as pepsin (PDB ID 1MPP, Z-score = 29.7). Structure-based sequence alignment indicated that secondary structural elements are well conserved between Bg7S and TAXI-IA, whereas deletions and insertions in some loop regions are observed (Fig. 2A). In addition, although TAXI-IA also has 12 cysteines forming disulfide bonds, the positions of the disulfide bonds in Bg7S are different from those in TAXI-IA (Fig. 2B) [7]. Both Bg7S and TAXI-IA adopt a pepsin fold. The pseudo-active site of Bg7S corresponding to pepsin is located in the cleft between the α-domain and β-domain, as observed in TAXI-IA [7] (Fig. 1C). However, both Bg7S and TAXI-IA lack protease activity, because one aspartate corresponding to the catalytic residue of pepsin is replaced by Ser265 and Ser235 in Bg7S and TAXI-IA, respectively.

Figure 2.

 Primary structures of Bg7S (soybean) and TAXI-IA (wheat). (A) Sequence alignment of Bg7S and TAXI-IA. Identical and homologous residues are highlighted by black and gray backgrounds, respectively. All cysteines are highlighted by a yellow background. Bg7S shares 26% amino acid identity with TAXI-IA. The secondary structures of Bg7S and TAXI-IA are shown above and below the sequences, respectively. The β-strand, α-helix and 310-helix are shown in blue, red and magenta, respectively. (B) Disulfide bonds of Bg7S (upper) and TAXI-IA (lower).

Assembly of Bg7S in solution

In marked contrast to TAXI-IA, Bg7S forms a tetramer with pseudo-222 symmetry, as mentioned above. A number of water molecules are found in the protomer–protomer interfaces (Table 1), implying that assembly of Bg7S might be dynamically altered by solution conditions. To investigate the assembly of Bg7S in solution, we first performed an analytical ultracentrifugation (AUC) experiment, based on the sedimentation velocity method, at pH 7.4 (Fig. 3A). Sedimentation velocity analysis showed major and minor peaks corresponding to Bg7S tetramers and dimers, respectively. This observation indicates that there is an equilibrium between tetramers and dimers. Next, we performed sedimentation equilibrium analysis under the same buffer conditions (Fig. 3B). A tetramer–dimer self-association model was used for data analysis, and the dissociation constant (Kd) for dissociation of the Bg7S tetramer from the dimer was estimated to be 0.83 μm. We also performed size exclusion chromatography (SEC) to investigate the pH dependency of self-assembly of Bg7S (Fig. 3C). SEC analysis revealed the pH-dependent dynamic assembly of Bg7S in solution. At neutral pH (7.0), Bg7S formed a tetramer, a finding consistent with the results of AUC. In contrast, Bg7S was found to exist as a monomer at acidic pH (4.0). Interestingly, Bg7S seemed to form a dimer at both weakly acidic pH (6.0) and weakly basic pH (8–9).

Table 1.   ΔASA in dimer formation. The ΔASA of the AB dimer is defined as [(ASA of A) + (ASA of B) − (ASA of AB dimer)]/2. The number of water molecules in the dimer interface was detected with asv calculator [34]. ASA was calculated with a program kindly provided by M Maeda (National Institute of Agrobiological Sciences, Japan).
 ΔASA (Å2)No. of water molecules
AB dimer146218
BC dimer149324
CD dimer172725
DA dimer151122
Figure 3.

 Analysis of Bg7S assembly. (A) Sedimentation velocity analysis of Bg7S and EDGP. The sedimentation coefficient distributions of Bg7S and EDGP are indicated by the green and orange lines, respectively. EDGP is a monomeric standard. (B) Sedimentation equilibrium data are shown with the residuals from the best fit to a dimer–tetramer self-association model. Plots show data obtained at 5000 r.p.m. (red), 7000 r.p.m. (green), and 9000 r.p.m. (blue). (C) SEC elution profiles of Bg7S in various pH buffers are shown by the blue (9.0), light blue (8.0), green (7.0), yellow (6.0), red (5.0) and pink (4.0) lines. Absorbance at 280 nm is normalized. (D) Electrostatic potentials of the Bg7S A molecule (left) and the homology model of the γ-conglutin protomer (right). The blue and red surfaces indicate positive and negative potential, respectively. The B and D molecules of Bg7S are shown as loop representations. The colors of the B and D molecules of Bg7S correspond to those of Fig. 1A.

Because structural analysis revealed that Bg7S forms a tetramer with pseudo-222 symmetry (Fig. 1A), there are potentially two types of dimer formation, namely AB (or CD) and DA (or BC). The former and latter are designated face-to-face (FTF) and back-to-back (BTB) dimers, respectively. To assess which dimer is more plausible, the difference in accessible surface area (ΔASA) in each dimer was calculated (Table 1). It is conceivable that a dimer with larger ΔASA is more plausible. We found that the ΔASAs of the AB and DA dimers were comparable. Although the ΔASA of the CD dimer was slightly larger than the others, this was attributable to the N-termini of the β-chains (Fig. 1A). Those findings imply that both FTF and BTB dimers might be plausible. However, the electrostatic potential provided further insights into dimer formation (Fig. 3D, left panel). The FTF and BTB dimers utilize, respectively, acidic and basic surfaces during their formation. As a result, FTF and BTB dimers are supposed to be formed in weakly basic and weakly acidic conditions, respectively. Very recently, it has been reported that the formation of lupin γ-conglutin oligomers is dependent on pH [17]. γ-Conglutin undergoes a tetramer–dimer–monomer transition from neutral to acidic pH, which is consistent with our findings for Bg7S. Furthermore, Bg7S shares 63% amino acid identity with γ-conglutin. A homology model of γ-conglutin was build by swiss-model [18], using the structure of the Bg7S protomer as a template. In this homology model, the electrostatic potential of γ-conglutin is very similar to that of Bg7S (Fig. 3D, right panel). Thus, pH dependence of dynamic assembly might be a general feature of legume XEGIP proteins.

Bg7S does not inhibit GH11 or GH12 enzymes

XEGIP was originally found to inhibit GH12 enzymes and not to inhibit GH11 enzymes. Thus, on the basis of this analogy with XEGIP, we first investigated whether or not Bg7S inhibits GH12 enzymes (Fig. 4A,B). Surprisingly, Bg7S did not inhibit either XEG or FI-CMC, a GH12 carboxymethyl cellulase from A. aculeatus [19]. We further investigated the activity of the GH11 xylanase ANXY in the presence of Bg7S (Fig. 4C). As expected, Bg7S did not affect the activity of ANXY. Even in the presence of leginsulin, a Bg7S-binding peptide, Bg7S did not inhibit GH12 or GH11 enzymes. Recently, it has been reported that lupin γ-conglutin does not inhibit GH12 endo-β-glucanase [20]. Therefore, we extracted XEGIPs from several legume seeds (azukibean, yardlongbean, and mungbean), and tested whether these proteins inhibited GH12 and GH11 enzymes (Fig. 4). Like Bg7S, these legume XEGIPs did not affect the activities of GH12 and GH11 enzymes.

Figure 4.

 Inhibitory activities of legume XEGIPs against GH12 and GH11 enzymes. The enzymatic activities of XEG (A), FI-CMC (B) and ANXY (C) in the presence of various legume XEGIPs were measured with or without the 4-kDa peptide from soybean (leginsulin).

To date, structures of TAXI in complex with GH11 xylanase have been reported [7,9]. Structural superimposition of Bg7S on TAXI-IA in complex with ANXY (PDB ID 1T6G) provides significant insights into the structural basis of the lack of inhibition of GH11 enzymes by Bg7S (Fig. 5A). His374 and Leu292 of TAXI-IA, which are located in the loops termed, respectively, inhibition loop 2 (IL-2: residues 372–377) and inhibition loop 1 (IL-1: residues 290–294) in the present work, intrude into the active site of ANXY. His374 in IL-2 of TAXI-IA undergoes electrostatic interactions with the catalytic Glu79 and Glu170 of ANXY. In contrast, Leu292 in IL-1 of TAXI-IA undergoes a hydrophobic interaction with Tyr10 of ANXY. The interactions mimic those in the enzyme–substrate complexes (PDB ID 1BCX and 2QZ2) [7,9]. In addition, His374 of TAXI-IA interacts with Asp37 of ANXY. Bg7S lacks IL-1, and Leu292 of TAXI-IA is not conserved in Bg7S (Fig. 5A,B). Bg7S has His388 and His390 in IL-2 (residues 388–393). His390 is equivalent to His374 in IL-2 of TAXI-IA (Fig. 5B). However, the side chains of His388 and His390 do not face the protein exterior in the A molecule of the Bg7S tetramer. In the other protomers of the tetramer, the electron densities of IL-2 are ambiguous. This indicates that the IL-2 structure of Bg7S is potentially flexible, implying that these residues might interact with the catalytic residues of ANXY. However, sequence conservation in IL-2 between Bg7S and TAXI-IA is markedly lower than in any other region, and, furthermore, IL-2 of Bg7S is longer than that of TAXI-IA (Figs 2 and 5B). Thus, it is unlikely that IL-2 of Bg7S interacts with the active site.

Figure 5.

 Structural basis for the lack of inhibitory activity of Bg7S against GH12 and GH11 enzymes. (A) Structure of Bg7S superimposed on that of the TAXI-IA–ANXY complex (PDB ID 1T6G). The right panel shows a close-up view of the site of interaction between TAXI-IA and ANXY, roughly corresponding to the box in the left panel. Bg7S, TAXI-IA and ANXY are shown as green, light brown and gray ribbon representations, respectively. Residues that are significantly involved in the interaction between TAXI-IA and ANXY are shown as stick representations and labeled. His388 and His390 of Bg7S are also shown as stick representations. (B) Sequence alignment of IL-1 and IL-2 is shown in the upper and lower panels, respectively. IL-1 and IL-2 are indicated by light brown squares. Leu292 and His374 of TAXI-IA are highlighted in red. Homologous residues in IL-2 are highlighted by gray backgrounds.

The structure of XEGIP in complex with a GH12 enzyme has not been determined so far. However, the structures of both GH12 and GH11 enzymes adopt a similar β-jelly roll structure and have catalytic glutamates, indicating that Bg7S lacks inhibitory activity against GH12 enzymes for a similar reason. Recently, it has been reported that γ-conglutin, which also lacks IL-1, does not inhibit GH12 or GH11 enzymes [20]. Therefore, it is conceivable that legume XEGIPs in general do not inhibit either GH12 or GH11 enzymes.


In this work, we have determined the crystal structure of Bg7S, which is the first three-dimensional structure of XEGIP. Bg7S forms a tetramer in a pH-dependent manner. Our biochemical characterization revealed that Bg7S, in contrast to XEGIP or TAXI, lacks inhibitory activity against both GH12 and GH11 enzymes. Furthermore, our study clarifies the structural basis for the lack of legume XEGIP inhibitory activity against both GH12 and GH11 enzymes. However, our results do not exclude the possibility that Bg7S functions as an inhibitory protein against GH enzymes other than GH12 and GH11 enzymes. The biochemical and biophysical features of legume XEGIPs are significantly distinct from those of XEGIPs from other plants. Thus, legume XEGIPs might be categorized differently from others. The physiological functions of legume XEGIPs, including Bg7S and γ-conglutin, remain unclear, and further functional studies are therefore required. Our structural and functional studies will provide significant clues for understanding the physiological function of legume XEGIPs, and will pave the way for future analysis.

Experimental procedures

Preparation and crystallographic analysis of Bg7S

Preparation and crystallization of the Bg7S have been described previously [14]. In brief, Bg7S was extracted from mature soybeen seeds (Glycine max L. Merrill cv. Miyagishirome). The protein was purified by using HisTrap Crude (GE Healthcare, UK Ltd, Little Chalfont, UK), HiTrap SP (GE Healthcare) and EconoPac CM (Bio-Rad Laboratory, Hercules, CA, USA) columns. The orthorhombic crystal was obtained by the hanging-drop vapor-diffusion method under the form II crystallization condition [14]. X-ray diffraction data were collected at Photon Factory beamline BL-5A, with a Quantum 315 CCD detector (Area Detector Systems, Corporation, San Diego, CA, USA) All diffraction data were processed with the hkl2000 [21]. The structure was solved by a molecular replacement method with molrep [22], using the crystal structure of EDGP (Yoshizawa et al., unpublished work). Model building was performed with coot [23]. Structure refinement was performed at 1.9-Å resolution with cns [24] and refmac [25], and validated with procheck [26]. The data collection and refinement statistics are given in Table 2.

Table 2.   Data collection and refinement statistics. The values in parentheses are those for the highest-resolution shell (1.97–1.90 Å).
Data collection
 Wavelength (Å)1.0000
 Space groupP21212
 a (Å)135.2
 b (Å)161.2
 c (Å)84.8
 Resolution (Å)50.0–1.90
 Observed reflections720 554
 Unique reflections138 568
 R-merge (%)6.5 (36.5)
 Completeness (%)95.6 (81.0)
 <I>/σ<I>11.3 (1.9)
 Resolution (Å)20–1.91
 Refined reflections130 634
 Free reflections6532
 Protein atoms11 297
 Water molecules621
 R (%)21.1
 R-free (%)25.9
  Bond length (Å)0.018
  Bond angles (°)1.782
 Ramachandran plot
  Most favored (%)87.1
  Additional allowed (%)11.1
  Generously allowed (%)1.1
  Disallowed (%)0.7
 Averaged B-value (Å2)47.07
 PDB code3AUP

SEC and AUC experiments

SEC was performed with a Superdex 200 10/300 GL column (GE Healthcare). Bg7S was eluted with buffer solutions of various pH: 50 mm sodium acetate (pH 4.0, pH 5.0), 20 mm potassium phosphate (pH 6.0, pH 7.0), or 50 mm Tris/HCl (pH 8.0, pH 9.0), with 150 mm KCl. AUC was performed with an Optima XL-I analytical ultracentrifuge (Beckman Coulter, Brea, CA, USA). The concentrations of the loaded protein solutions in the sedimentation velocity experiment were 0.88 mg·mL−1 Bg7S or 0.91 mg·mL−1 EDGP in a reference buffer (20 mm potassium phosphate, pH 7.4, and 250 mm KCl). EDGP was purified from carrot callus tissue [4]. Absorbance (A280 nm) scans were collected during sedimentation at 182 000 g. Data analysis was performed with sedifit [27,28] and sednterp [29]. Sedimentation equilibrium experiments were performed in a six-channel centerpiece with quartz windows. The concentrations of the loaded protein solutions in the sedimentation equilibrium experiments were 0.18, 0.35 and 0.88 mg·mL−1 in the reference buffer (20 mm potassium phosphate, pH 7.4, and 250 mm KCl). Data were obtained at 1820, 3562 and 5896 g, respectively. Data analysis was performed by global analysis with ultraspin (MRC Center for Protein Engineering, Cambridge, UK;

Preparation of XEGIPs from various legume seeds

Legume XEGIPs were purified from various dry mature seeds. We used soybean (G.max L. Merrill cv. Miyagishirome), yardolongbean (Vigna unguiculata sesquipedalis L. Verdc), azukibean (Vigna angularis L. cv. Dainagon), and mungbean (Vigna radiata R. Wilczek). For each, mature seeds were ground with water in a food processor (Cuisinart, Stamford, CT, USA) and a Polytron homogenizer (Kinematica, Bohemia, NY, USA), and then filtered through Miracloth (Merck KGaA, Darmstadt, Germany). The residue was stirred in buffer (20 mm potassium phosphate, pH 7.4, and 0.5 m NaCl) overnight at 4 °C, and then centrifuged at 43 667 g for 30 min. The supernatant contained mostly legume XEGIP, and was therefore used for enzyme inhibition assays. The purity of the proteins was checked by SDS/PAGE (Fig. S1).

Preparation of XEG, FI-CMC, and ANXY

cDNA encoding XEG, FI-CMC or ANXY was obtained by PCR-based gene synthesis. The oligonucleotides were designed by using dnaworks 3.1 [30] ( The synthesized cDNAs were inserted into a pGEX6P-1 vector (GE Healthcare) at the BamHI–XhoI site. The resulting plasmid encoded XEG, FI-CMC or ANXY with a glutathione-S-transferase (GST)-tag at the N-terminus. The expression vector was introduced into Escherichia coli BL21(DE3). The cells were grown at 37 °C to a cell density of 0.6–0.8 at 660 nm, and then for a further 6 h at 25 °C after the addition of 1 mm isopropyl thio-β-d-galactoside before being harvested. XEG and FI-CMC were purified by procedures similar to those already published [31,32]. In brief, XEG was purified with a glutathione Sepharose 4B (GS4B) resin (GE Healthcare), HiTrap Q HP column (GE Healthcare), and HiLoad Superdex 75 26/60 column (GE Healthcare). FI-CMC was purified with a GS4B resin (GE Healthcare) and HiLoad Superdex 75 26/60 column (GE Healthcare). The N-terminal GST tags of XEG and FI-CMC were cleaved by HRV3C protease, after affinity purification with GS4B (GE Healthcare). GST-fused ANXY was purified with GS4B resin (GE Healthcare). Because removal of the GST-tag of GST–ANXY reduced the stability of the protein, GST–ANXY was used in the following inhibition assay.

Enzyme inhibition assay

The inhibitory activities of legume XEGIPs against GH enzymes were measured by the p-hydroxy-benzoic acid hydrazide method, where reducing sugar was detected by colorimetric reaction with p-hydroxy-benzoic acid hydrazide [33]. The assay for inhibition of XEG was performed in a 20-μL solution containing 50 mm sodium acetate (pH 4.6), 5 mg·mL−1 xyloglucan from tamarind seeds (DS Pharma, Osaka, Japan), 5 μg of legume XEGIP, and 100 ng of XEG. The assay for inhibition of FI-CMC was performed in a 50-μL solution containing 50 mm sodium acetate (pH 4.6), 5 mg·mL−1 carboxymethyl cellulose (Nacalai, Kyoto, Japan), 5 μg of legume XEGIP, and 100 ng of FI-CMC. The assay for inhibition of ANXY was performed in a 20-μL solution containing 50 mm sodium acetate (pH 4.6), 5 mg·mL−1 xylan (Sigma-Aldrich, St. Louis, MO, USA), 5 μg of legume XEGIP, and approximately 100 ng of GST–ANXY. In the assays in the presence of leginsulin, 0.5 μg of leginsulin was added to each reaction mixture including xyloglucan and XEGIP, and the solution was incubated for 10 min at room temperature. Then, each GH enzyme was added to the solution. The leginsulin used in the assay was chemically synthesized. The reaction mixtures were incubated at room temperature for 15 min, and the amount of reducing sugar was measured with a DU530 spectrometer (Beckman Coulter, Brea, CA, USA). The activity was measured at least three times for each sample. The average values are shown in Fig. 4.

Figure preparation

Figures 1, 3D and 5A were prepared with pymol ( All of the figures were modified with photoshop and illustrator (Adobe Systems, San Jose, CA, USA).


We acknowledge the kind support of the beamline staff of PF and SPring-8 for data collection. We also acknowledge the kind support of M Maeda (National Institute of Agrobiological Sciences, Japan) for calculation of ΔASA. This work was supported by KAKENHI (16770080, 17048023, and 19036025), the Protein 3000 Project and Target Protein Research Programs to M. Sato, T. Shimizu and H.H. from MEXT to M Sato, T Shimizu and H Hashimoto.