research papers\(\def\hfill{\hskip 5em}\def\hfil{\hskip 3em}\def\eqno#1{\hfil {#1}}\)

Journal logoSTRUCTURAL
BIOLOGY
ISSN: 2059-7983

Structural insight into the hydrolase and synthase activities of an alkaline α-galactosidase from Arabidopsis from complexes with substrate/product

crossmark logo

aLife Science Group, Scientific Research Division, National Synchrotron Radiation Research Cente, Hsinchu 30076, Taiwan, bInstitute of Tropical Plant Sciences and Microbiology, National Cheng Kung University, Tainan City 701, Taiwan, cInstitute for Protein Research, Osaka University, Suita, Osaka 565-0871, Japan, dDepartment of Biotechnology and Bioindustry Sciences, National Cheng Kung University, Tainan City 701, Taiwan, eDepartment of Physics, National Tsing Hua University, Hsinchu 30013, Taiwan, and fDepartment of Biological Science and Technology, National Chiao Tung University, Hsinchu 30010, Taiwan
*Correspondence e-mail: cjchen@nsrrc.org.tw

Edited by G. Kurisu, Osaka University, Japan (Received 7 January 2022; accepted 3 January 2023; online 20 January 2023)

The alkaline α-galactosidase AtAkαGal3 from Arabidopsis thaliana catalyzes the hydrolysis of α-D-galactose from galacto-oligosaccharides under alkaline conditions. A phylogenetic analysis based on sequence alignment classifies AtAkαGal3 as more closely related to the raffinose family of oligosaccharide (RFO) synthases than to the acidic α-galactosidases. Here, thin-layer chromatography is used to demonstrate that AtAkαGal3 exhibits a dual function and is capable of synthesizing stachyose using raffinose, instead of galactinol, as the galactose donor. Crystal structures of complexes of AtAkαGal3 and its D383A mutant with various substrates and products, including galactose, galactinol, raffinose, stachyose and sucrose, are reported as the first representative structures of an alkaline α-galactosidase. The structure of AtAkαGal3 comprises three domains: an N-terminal domain with 13 antiparallel β-strands, a catalytic domain with an (α/β)8-barrel fold and a C-terminal domain composed of β-sheets that form two Greek-key motifs. The WW box of the N-terminal domain, which comprises the conserved residues FRSK75XW77W78 in the RFO synthases, contributes Trp77 and Trp78 to the +1 subsite to contribute to the substrate-binding ability together with the (α/β)8 barrel of the catalytic domain. The C-terminal domain is presumably involved in structural stability. Structures of the D383A mutant in complex with various substrates and products, especially the natural substrate/product stachyose, reveal four complete subsites (–1 to +3) at the catalytic site. A functional loop (residues 329–352) that exists in the alkaline α-galactosidase AtAkαGal3 and possibly in RFO synthases, but not in acidic α-galactosidases, stabilizes the stachyose at the +2 and +3 subsites and extends the catalytic pocket for the transferase mechanism. Considering the similarities in amino-acid sequence, catalytic domain and activity between alkaline α-galactosidases and RFO synthases, the structure of AtAkαGal3 might also serve a model for the study of RFO synthases, structures of which are lacking.

1. Introduction

α-Galactosidases (α-D-galactoside galactohydrolases; EC 3.2.1.22), which are widely found in microorganisms, plants, animals and humans, catalyze the removal of the nonreducing α-D-galactose moiety from galacto-oligosacharides, polysaccharides, galactolipids and glycoproteins (Henrissat et al., 2001[Henrissat, B., Coutinho, P. M. & Davies, G. J. (2001). Plant Mol. Biol. 47, 55-72.]). Fabry disease is a rare genetic disease in humans that is due to a deficiency of α-galactosidase A, which is needed to catabolize the α-D-galactosyl moiety of sphingolipids in the lysosome (Calhoun et al., 1985[Calhoun, D. H., Bishop, D. F., Bernstein, H. S., Quinn, M., Hantzopoulos, P. & Desnick, R. J. (1985). Proc. Natl Acad. Sci. USA, 82, 7364-7368.]; Weidemann et al., 2003[Weidemann, F., Breunig, F., Beer, M., Sandstede, J., Turschner, O., Voelker, W., Ertl, G., Knoll, A., Wanner, C. & Strotmann, J. M. (2003). Circulation, 108, 1299-1301.]). The hydrolytic specificity of α-galactosidases enables their application in the food industry to remove raffinose and to increase the yield of sucrose in the sugar beet industry (Shibuya et al., 1995[Shibuya, H., Kobayashi, H., Park, G. G., Komatsu, Y., Sato, T., Kaneko, R., Nagasaki, H., Yoshida, S., Kasamo, K. & Kusakabe, I. (1995). Biosci. Biotechnol. Biochem. 59, 2333-2335.]), to improve the gelling properties of galactomannans for use as food thickeners and to degrade the raffinose family of oligosaccharides (RFOs) in food and feed materials and in enzymotherapy (Guimarães et al., 2001[Guimarães, V. M., de Rezende, S. T., Moreira, M. A., de Barros, E. G. & Felix, C. R. (2001). Phytochemistry, 58, 67-73.]).

Based on their glycoside hydrolase (GH) signatures, α-galactosidases and RFO synthases belong to GH families 27 and 36, which are characterized by two conserved motifs: KxD and RxxxD (Henrissat & Romeu, 1995[Henrissat, B. & Romeu, A. (1995). Biochem. J. 311, 350-351.]; Henrissat, 1991[Henrissat, B. (1991). Biochem. J. 280, 309-316.]; Carmi et al., 2003[Carmi, N., Zhang, G., Petreikov, M., Gao, Z., Eyal, Y., Granot, D. & Schaffer, A. A. (2003). Plant J. 33, 97-106.]; Lee et al., 2004[Lee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281-295.]; https://www.cazy.org/). Plant α-galactosidases can be further divided into acidic and alkaline forms, according to their optimal pH value for enzymatic activity. Alkaline α-galactosidases have optimal activities at neutral–alkaline pH values, whereas acidic α-galactosidases have optimal pH values of 3–6.5 (Lee et al., 2004[Lee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281-295.]). Another major difference between these two types of α-galactosidase is that the total number of amino acids in alkaline α-galactosid­ases is about twice that in acidic α-galactosidases. A phylogenic analysis showed that the amino-acid sequences of alkaline α-galactosidases share a higher similarity to those of RFO synthases than to those of acidic α-galactosidases (Lee et al., 2004[Lee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281-295.]). Both alkaline α-galactosidases and RFO synthases are unique to plants. Most studies have focused on acidic α-galactosidases (Keller & Pharr, 1996[Keller, F. & Pharr, D. M. (1996). Photoassimilate Distribution in Plants and Crops: Source-Sink Relationships, edited by E. Zamski & A. A. Scharffer, pp. 157-184. New York: Marcel Dekker.]; Dey & Pridham, 1972[Dey, P. M. & Pridham, J. B. (1972). Adv. Enzymol. Relat. Areas Mol. Biol. 36, 91-130.]; Herman & Shannon, 1985[Herman, E. M. & Shannon, L. M. (1985). Plant Physiol. 77, 886-890.]). Acidic α-galactosidases are involved in the mobilization of seed reserves of RFOs and the degradation of cell-wall galactomannan during seed germination (Corchete & Guerra, 1987[Corchete, M. P. & Guerra, H. (1987). Phytochemistry, 26, 927-932.]). It has been suggested that these enzymes also play a role in cell-wall modification during fruit ripening (Kang & Lee, 2001[Kang, H.-C. & Lee, S.-H. (2001). Phytochemistry, 58, 213-219.]; Nunan et al., 2001[Nunan, K. J., Davies, C., Robinson, S. P. & Fincher, G. B. (2001). Planta, 214, 257-264.]).

The first alkaline α-galactosidase was uncovered in the Cucurbitaceae family (Gaudreault & Webb, 1986[Gaudreault, P. R. & Webb, J. A. (1986). Plant Sci. 45, 71-75.]; Gao & Schaffer, 1999[Gao, Z. & Schaffer, A. A. (1999). Plant Physiol. 119, 979-988.]). High alkaline α-galactosidase activity was mostly found in roots and immature leaves in the early development stage; the activity declined with increasing leaf maturity in Cucurbita pepo (Gaudreault & Webb, 1986[Gaudreault, P. R. & Webb, J. A. (1986). Plant Sci. 45, 71-75.]). High activity in an immature organ is due to the translocation of RFOs, the major galactosyl-sucrose oligosaccharides, from their source organs to developing organs such as leaves (Gaudreault & Webb, 1986[Gaudreault, P. R. & Webb, J. A. (1986). Plant Sci. 45, 71-75.]) and fruits (Gao & Schaffer, 1999[Gao, Z. & Schaffer, A. A. (1999). Plant Physiol. 119, 979-988.]). Alkaline α-galactosidases can be found in more than one form in the Cucurbitaceae family, with varied substrate specificities. Alkaline α-galactosidases from Cucurbita pepo, Cucumis sativus and Cucurbita maxima prefer stachyose to raffinose, whereas in Cucumis melo fruit alkaline α-galactosidases of forms I and II exhibit a higher activity towards raffinose and stachyose, respectively (Gao & Schaffer, 1999[Gao, Z. & Schaffer, A. A. (1999). Plant Physiol. 119, 979-988.]). The maximum activity of maize alkaline α-galactosidase was detected in mature dehydrated, germinating and germinated seeds (Zhao et al., 2006[Zhao, T. Y., Corum, J. W. III, Mullen, J., Meeley, R. B., Helentjaris, T., Martin, D. & Downie, B. (2006). Seed Sci. Res. 16, 107-121.]); ZmAGA1, with maximum catalysis at pH 7.5, was solely involved in seed germination. Elevated levels of alkaline α-galactosidase gene expression were detected when seed germination in maize was interrupted by heat, cold or dehydration (Zhao et al., 2006[Zhao, T. Y., Corum, J. W. III, Mullen, J., Meeley, R. B., Helentjaris, T., Martin, D. & Downie, B. (2006). Seed Sci. Res. 16, 107-121.]) and when New Zealand spinach was grown under drought conditions (Hara et al., 2008[Hara, M., Tokunaga, K. & Kuboi, T. (2008). Plant Biotechnol. 25, 497-501.]). A rice chloroplast alkaline α-galactosidase (OsAkαGal) has been suggested to exhibit a function in degrading digalactosyl diacylglycerol, a major glycolipid in the chloro­plast thylakoid membrane, during leaf senescence (Lee et al., 2004[Lee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281-295.], 2009[Lee, R.-H., Hsu, J.-H., Huang, H.-J., Lo, S.-F. & Chen, S.-C. G. (2009). New Phytol. 184, 596-606.]).

The Arabidopsis thaliana (Arabidopsis) genome contains genes for three alkaline α-galactosidases (At1g55740, At3g57520 and At5g20250), two RFO synthases (At4g01970 and At5g40390) and four acidic α-galactosidases (At3g26380, At5g08370, At5g08380 and At3g56310). Here, the proteins encoded by At1g55740, At3g57520 and At5g20250 are referred to as AtAkαGal1, AtAkαGal2 and AtAkαGal3, respectively (Lee et al., 2004[Lee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281-295.]). A BLAST search showed that AtAkαGal3 is related to probable galactinol–sucrose galactosyltransferases from various species, such as Arabidopsis lyrata subsp. lyrate (XP_020876652), Camelina sativa (XP_010493135), Capsella rubella (XP_006287127), Raphanus sativus (XP_018444981), Brassica oleracea var. oleracea (XP_013612073.1) and Prunus persica (XP_020424207), with 76–94% sequence identity. Moreover, AtAkαGal3 shares sequence identities of less than 30% with all acidic α-galactosidases from Arabidopsis and other plants.

To date, only structures of acidic α-galactosidases have been reported, including rice acidic α-galactosidase (PDB entry 1uas; Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]), chicken α-N-acetylgalactos­aminidase (PDB entry 1ktb; Garman et al., 2002[Garman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425-434.]) and human α-galactosidase (PDB entry 1r46; Garman & Garboczi, 2004[Garman, S. C. & Garboczi, D. N. (2004). J. Mol. Biol. 337, 319-335.]). The structures of rice acidic α-galactosidase and the closely related chicken α-N-acetylgalactos­aminidase revealed a double-displacement catalytic mechanism and the mode of substrate binding (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]; Garman et al., 2002[Garman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425-434.]). However, no structure of any alkaline α-galactosidase or RFO synthase was available prior to this work. Here, we report structures of AtAkαGal3 and its D383A mutant in complex with various substrates and products, including the natural substrate/product stachyose, revealing four complete substrate-binding subsites (–1 to +3) at the catalytic site for catalysis. The dual-functional AtAkαGal3 was also discovered to exhibit stachyose synthase activity in this work; the new structure of AtAkαGal3 thus potentially serves as a model for the study of raffinose/stachyose synthases.

2. Methods

2.1. Cloning, expression and purification

Materials and chemicals were obtained from Sigma unless specified otherwise. A. thaliana L. Herynh plants (Col-0) were used in this study. AtAkαGal3 full-length open frame sequences (2273 bp) were obtained using database searches and sequence analyses using OsAkαGal as a query (Lee et al., 2004[Lee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281-295.]). The S3 stage of senescent leaves containing about 30–50% chlorophyll compared with full-grown green rosette leaves (100% chlorophyll) was used for total RNA extraction. Chlorophyll determination was carried out as described previously (Lee & Collins, 2001[Lee, D. W. & Collins, T. M. (2001). Int. J. Plant Sci. 162, 1141-1153.]). Total RNA was isolated using the method described previously (Chang et al., 1993[Chang, S., Puryear, J. & Cairney, J. (1993). Plant Mol. Biol. Rep. 11, 113-116.]), and an Agilent 2100 Bioanalyzer (ThermoFisher, USA) was used to assess its quality and quantity. First-strand cDNA was synthesized from 1 µg of total RNA using the SuperScript III First-Strand Synthesis System, as described by the manufacturer (ThermoFisher, USA). A DNase (Promega, USA) digestion step was conducted prior to reverse transcription in a total volume of 20 µl. AtAkαGal3 full-length open frame sequences (2273 bp) containing KpnI and SalI sites were amplified from 1 µl cDNA solution by 30 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s and extension at 72°C for 1 min in 20 µl PCR reaction mixture. The primer pair used was At5gKpnIF2 (GATCTGGGTACCATGACGATTAAACCGGCGGT) and At5gnostopSalIR (CTGCAGGTCGACTAACTCAACTTGGATCAGA). The AtAkαGal3 full-length open frame sequences were cloned into the expression vector pET-30a(+) (Novagen) at KpnI and XhoI sites with a 6×His tag at the N-terminus and were transformed into Escherichia coli BL21(DE3) cells for expression. The bacteria were cultured at 37°C until the OD600 reached ∼0.6 prior to induction with isopropyl β-D-thiogalactopyranoside (IPTG; final concentration 0.15 mM) in Luria–Bertani (LB) medium at 20°C for 20 h with shaking, and were then harvested by centrifugation at 8000g for 25 min at 4°C. The cell pellet from 1 l culture was suspended in lysis buffer (35 ml) consisting of 50 mM Tris–HCl pH 8.0 and was subjected to cell disruption by ultrasonication using a pulse cycle of 2 s on and 3 s off with a total duration of 20 min at 40% energy on ice. The soluble protein extract was collected by centrifugation at 12 000g for 30 min at 4°C. The target AtAkαGal3 was purified using a nickel-immobilized metal ion-affinity chromatography (Ni-IMAC) column with a low imidazole concentration (50 mM), thus avoiding protein contamination at higher concentrations that affects crystallization, in 20 mM Tris–HCl buffer pH 7.8 containing 500 mM NaCl and was further dialyzed against 20 mM Tris–HCl buffer pH 8.2 containing 1 mM dithiothreitol (DTT) and 1 mM galactose overnight. The eluted AtAkαGal3 was concentrated and subjected to size-exclusion chromatography on an S200 column previously equilibrated with the abovementioned dialysis buffer. The target protein was then purified using the same buffer condition as the dialysis buffer to improve the purity.

Two AtAkαGal3 mutants, D383A and D447A, were generated by a site-directed mutagenesis method (QuikChange kit, Stratagene) using the pET-30a-AtAkαGal3 recombinant vector as a template. The forward primers for the mutants are as follows: D383A, 5′-GGACGGTGTGAAAGTGGCTGTGCAGTGTGTATTGG-3′; D447A, 5′-TGATTAGAGCATCAGATGCTTTCTATCCACGGGATCC-3′. The D383A and D447A mutants were confirmed by sequencing before expression. After Ni-IMAC purification, both mutants were dialyzed against the same dialysis buffer as wild-type AtAkαGal3 protein with and without galactose for various substrate and product complex studies. Selenomethionine-labeled AtAkαGal3 (SeMet-AtAkαGal3) protein was prepared following the method of Van Duyne (Doublié, 2007[Doublié, S. (2007). Methods Mol. Biol. 363, 91-108.]). Briefly, a single colony of recombinant AtAkαGal3 in E. coli BL21(DE3) cells was inoculated into 5 ml LB culture at 37°C for 12 h and then inoculated into 100 ml M9 medium in a 250 ml flask and cultured overnight. The bacterial cells from this overnight culture were collected and inoculated into 1 l M9 medium; a mixture of amino acids (100 mg lysine, phenylalanine and threonine; 50 mg isoleucine, leucine and valine; 60 mg SeMet) was added when the OD reached 0.6. After bacterial growth for a further 15 min, 0.15 mM IPTG was added to induce protein production. The rice raffinose synthase from O. sativa L. var. Nipponbare (XP_015621501) was overexpressed, purified and characterized using the protocol described previously (Li et al., 2007[Li, S., Li, T., Kim, W. D., Kitaoka, M., Yoshida, S., Nakajima, M. & Kobayashi, H. (2007). Biotechnol. Lett. 29, 635-640.]).

2.2. Activity assay using thin-layer chromatography

The purified proteins, wild-type AtAkαGal3 and its D383A and D447A mutants, as well as rice raffinose synthase, at equal amounts of 4 × 10−12 mol were incubated with the substrate raffinose at a final concentration of 10 mM at 35°C for 1 h. Each protein reaction and standard oligosaccharides (galactose, sucrose, raffinose and stachyose) were spotted (2 µl) onto a silica thin-layer plate (Merck Millipore); the sample was developed with a solvent consisting of a mixture of chloroform, acetic acid and water in a 6:7:1 ratio by volume (Farag, 1978[Farag, S. (1978). J. Am. Soc. Sugar Beet Technol. 20, 251-254.]). The results of the reaction were examined by spraying ethanol containing sulfuric acid (10%) onto the plate and baking it on a hotplate until brown spots developed.

2.3. Crystallization and collection of X-ray data

The purified SeMet-AtAkαGal3, wild-type and mutant AtAkαGal3 protein fractions from size-exclusion chromatography were concentrated to 40–50 mg ml−1 in a buffer consisting of 20 mM Tris–HCl pH 8.2, 1 mM DTT with or without 1 mM galactose before crystallization, which was performed using a microbatch method. The protein crystals appeared in 2–3 months using the best crystallization condition (0.1 M ammonium sulfate, 0.3 M sodium formate, 0.1 M Tris pH 7.8, 10% PEG 2000) at 18°C. Among the crystallization trials with various AtAkαGal3 proteins, crystals of SeMet-AtAkαGal3, wild-type AtAkαGal3 and the D383A mutant (referred to in the following as D383A) in various substrate or product complexes could be obtained. No satisfactory crystals of the D447A mutant were obtained. In addition, without galactose neither wild-type nor mutant crystals could be obtained. Crystals of D383A–substrate complexes were obtained by co-crystallization and subsequent crystal soaking with various substrates, including galactinol, galactose, raffinose and stachyose (final concentration 5 mM), in the reservoir solution at 25°C for 10 min.

The crystals were harvested, transferred to reservoir solution containing glycerol (20%) as a cryoprotectant and immediately cooled with liquid nitrogen for data collection. The X-ray diffraction experiments were performed at various wavelengths ranging from 0.9 to 1.0 Å on synchrotron beamlines TPS 05A (with a Rayonix MX300HS detector) and TLS 15A (with a Rayonix MX300HE detector) at the National Synchrotron Radiation Research Center (NSRRC) in Taiwan and on BL44XU (with a Dectris EIGER X 16M detector) at SPring-8 in Japan. The data sets for SeMet-AtAkαGal3, AtAkαGal3–galactose, D383A–galactose, D383A–stachyose and D383A–galactose–sucrose collected on the TPS 05A and TLS 15A beamlines were processed using HKL-2000 (Otwinowski & Minor, 1997[Otwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307-326.]) and the D383A–galactinol and D383A–rafffinose data sets collected on the BL44XU beamline were processed with XDS (Kabsch, 2010[Kabsch, W. (2010). Acta Cryst. D66, 125-132.]).

2.4. Structure determination and refinement

The initial crystal structure of wild-type AtAkαGal3 was determined by the selenomethionine multi-wavelength anomalous diffraction (Se-MAD) phasing method. 29 selenomethionine sites were first located using SOLVE (Terwilliger & Berendzen, 1999[Terwilliger, T. C. & Berendzen, J. (1999). Acta Cryst. D55, 849-861.]) to generate reasonable initial phases with a figure of merit of 0.34 for data in the resolution range 30–3.0 Å. The phases were further calculated and improved with AutoSol (Zhao et al., 2006[Zhao, T. Y., Corum, J. W. III, Mullen, J., Meeley, R. B., Helentjaris, T., Martin, D. & Downie, B. (2006). Seed Sci. Res. 16, 107-121.]) in Phenix and the direct phase selection method (Chen et al., 2014[Chen, C.-D., Huang, Y.-C., Chiang, H.-L., Hsieh, Y.-C., Guan, H.-H., Chuankhayan, P. & Chen, C.-J. (2014). Acta Cryst. D70, 2331-2343.]). A model of most of the structure was autobuilt with ARP/wARP (Langer et al., 2008[Langer, G., Cohen, S. X., Lamzin, V. S. & Perrakis, A. (2008). Nat. Protoc. 3, 1171-1179.]), and further model building and adjustment were undertaken with Coot (Emsley et al., 2010[Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486-501.]). The structures of the D383A mutant in its various complexes were determined by molecular replacement with MOLREP (Vagin & Teplyakov, 2010[Vagin, A. & Teplyakov, A. (2010). Acta Cryst. D66, 22-25.]) from the CCP4 suite (Winn et al., 2011[Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A. & Wilson, K. S. (2011). Acta Cryst. D67, 235-242.]) using the previously solved structure of wild-type AtAkαGal3 as a model. All refinements were performed with REFMAC (Murshudov et al., 2011[Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355-367.]) from the CCP4 suite. The structural models of galactose, galactinol, raffinose and starchyose were obtained from eLBOW (Moriarty et al., 2009[Moriarty, N. W., Grosse-Kunstleve, R. W. & Adams, P. D. (2009). Acta Cryst. D65, 1074-1080.]) in Phenix before building and refinement.

2.4.1. Model validation

The final refined structures of wild-type AtAkαGal3 and the various D383A–substrate and D383A–product complexes were validated with MolProbity (Williams et al., 2018[Williams, C. J., Headd, J. J., Moriarty, N. W., Prisant, M. G., Videau, L. L., Deis, L. N., Verma, V., Keedy, D. A., Hintze, B. J., Chen, V. B., Jain, S., Lewis, S. M., Arendall, W. B. III, Snoeyink, J., Adams, P. D., Lovell, S. C., Richardson, J. S. & Richardson, D. C. (2018). Protein Sci. 27, 293-315.]). The ligand-binding interaction at the catalytic binding site of the AtAkαGal3 structures was analyzed by PLIP (Salentin et al., 2015[Salentin, S., Schreiber, S., Haupt, V. J., Adasme, M. F. & Schroeder, M. (2015). Nucleic Acids Res. 43, W443-W447.]).

2.4.2. Accession codes

All of the coordinates and structural factors in this work have been deposited in the Protein Data Bank as PDB entries 7exf, 7exg, 7exh, 7exj, 7exr and 7exq for WT AtAkαGal3–galactose, D383A–galactose, D383A–galactinol, D383A–rafffinose, D383A–stachyose and D383A–galactose–sucrose, respectively.

3. Results and discussion

3.1. Protein purification

Both the recombinant wild-type AtAkαGal3 and the D383A mutant were passed through an Ni-IMAC column and a size-exclusion column (S200) for purification. All purified AtAkαGal3 and mutant proteins showed a single band corresponding to a molecular mass of ∼82 kDa on SDS–PAGE. The target proteins eluted from size-exclusion chromatography at a molecular mass in the range 80–100 kDa, indicating that the biological unit of AtAkαGal3 is a monomer.

3.2. Hydrolytic activity assay

A hydrolytic reaction mixture containing AtAkαGal3 and its natural substrate raffinose was incubated and subsequently analyzed by thin-layer chromatography (TLC; Supplementary Fig. S1). For a comparison of activity, a similar assay was simultaneously performed with purified recombinant rice raffinose synthase (O. sativa L. var. Nipponbare; Li et al., 2007[Li, S., Li, T., Kim, W. D., Kitaoka, M., Yoshida, S., Nakajima, M. & Kobayashi, H. (2007). Biotechnol. Lett. 29, 635-640.]). After the hydrolysis reactions with rice raffinose synthase and with AtAkαGal3, the TLC plate showed that the raffinose substrate was hydrolyzed to galactose and sucrose (lane 5) and to galactose, sucrose and stachyose (lane 6), respectively, while excess raffinose substrate remained. The results indicate that the hydrolase activity of AtAkαGal3 is greater than that of rice raffinose synthase towards raffinose as a substrate. Besides hydrolase activity, AtAkαGal3 exhibits stachyose synthase activity by transferring a galactose moiety to the raffinose substrate, in which both the galactose donor and acceptor are raffinose. Moreover, the TLC results showed that AtAkαGal3 appears to be incapable of synthesizing raffinose as a raffinose synthase according to an activity comparison between AtAkαGal3 and rice raffinose synthase with galactinol as the galactose donor and sucrose as the acceptor (data not shown). The activity assay thus shows that alkaline α-galactosidase exhibits a dual function of hydrolase and stachyose synthase activities.

3.3. Overall structure of Arabidopsis alkaline α-galactosidase

The crystal structure of wild-type AtAkαGal3 was first solved ab initio by the selenomethionine multi-wavelength anomalous diffraction (Se-MAD) method at a resolution of 2.8 Å and was subsequently refined to a high resolution of 2.17 Å using the native crystal with the best diffraction quality, which contains two molecules in the asymmetric unit (Supplementary Fig. S2a). The crystal structures of complexes of the D383A mutant with galactose, galactinol, raffinose and stachyose were subsequently determined using the wild-type AtAkαGal3 structure as a search model by the molecular-replacement method (Table 1[link]). The crystal structure of AtAkαGal3 comprises 724 of a total of 749 amino acids that fold into three domains: N-terminal, catalytic and C-terminal (Supplementary Fig. S2b). The first four residues are absent due to a lack of electron density. The N-terminal domain (residues 5–187) consists of 13 antiparallel β-strands connected by variable loops with an α-helix at the end. The catalytic domain (residues 206–518) folds into an (α/β)8-barrel structure with eight parallel β-strands placed around a central axis and surrounded by eight α-helices, which can be seen in α-galactosidases of both the GH27 and GH36 families. The C-terminal domain comprises 16 β-strands that form two Greek-key motifs comprising residues 537–653 and 654–749, respectively, which are presumably involved in structural stability. Two regions of residues, 103–119 at the N-terminal domain and 254–263 at the catalytic domain, are disordered without electron density, possibly due to loop flexibility on the protein surface (Supplementary Fig. S2b).

Table 1
Data-collection and refinement statistics

  SeMet-AtAkαGal3            
  Peak Inflection AtAkαGal3–galactose D383A–galactose D383A–galactinol D383A–rafffinose D383A–stachyose D383A–galactose–sucrose
PDB code     7exf 7exg 7exh 7exj 7exr 7exq
Data collection
 Beamline TPS 05A TLS 15A TPS 05A BL44XU BL44XU TPS 05A TLS 15A
 Space group P212121 P212121 P212121 P212121 P212121 P212121 P212121 P212121
a, b, c (Å) 95.75, 96.00, 104.59 104.69, 183.76, 183.74 94.94, 103.64, 181.21 94.03, 103.74, 182.70 97.52, 103.65, 182.43 92.60, 103.30, 181.44 97.59, 103.75, 182.37 97.71, 104.10, 182.42
 Wavelength (Å) 0.97918 0.97935 1.000 0.999 0.900 0.900 0.999 1.000
 Resolution (Å) 30–3.10 30–3.50 30–2.17 30–2.05 30–2.63 30–2.47 30–2.0 30–2.2
 No. of observed reflections 331082 245197 568408 714794 326607 370537 785125 500553
 No. of unique reflections 35194 26054 95207 111655 105231 118865 124558 92281
 Completeness (%) 98.6 (100) 99.0 (100) 98.2 (83.7) 99.3 (95.4) 98.6 (99.1) 98.5 (96.6) 99.9 (99.5) 97.2 (100)
 Multiplicity 9.4 (9.7) 9.4 (9.8) 6.0 (6.0) 6.4 (5.8) 3.1 (3.2) 3.1 (3.3) 6.3 (5.4) 5.4 (5.6)
 〈I/σ(I)〉 21.0 (3.3) 15.7 (4.4) 16.8 (2.2) 26.06 (3.3) 9.28 (2.3) 10.56 (1.8) 26.87 (2.4) 22.39 (2.6)
Rmerge (%) 11.7 (78.0) 14.2 (63.0) 7.8 (77.6) 7.5 (55.4) 11.8 (66.2) 7.4 (60.1) 9.3 (59.7) 8.5 (78.9)
 CC1/2     0.829 0.906 0.763 0.812 0.855 0.880
Refinement
 Resolution (Å)     30.0–2.17 30.0–2.05 30.0–2.63 30.0–2.47 30.0–2.00 30.0–2.20
Rwork/Rfree     0.21/0.26 0.20/0.25 0.17/0.23 0.19/0.24 0.18/0.21 0.19/0.25
 No. of atoms
  Protein     11246 11220 11232 11240 11240 11226
  Ligand     24 24 46 68 90 70
  Water     177 579 52 16 544 416
B factors (Å2)
  Protein     61.3 31.1 53.5 71.6 42.6 40.4
  Ligand     53.0 20.5 60.0 73.4 55.5 49.3
  Water     44.2 31.8 41.0 56.2 44.8 38.0
 R.m.s.d.
  Bond lengths (Å)     0.014 0.018 0.017 0.013 0.020 0.017
  Bond angles (°)     1.728 2.015 1.948 1.733 2.125 1.936

3.4. Protein sequence alignment

A BLAST search with the amino-acid sequence of AtAkαGal3 shows that the highest sequence similarities of 75–94% correspond to various probable galactinol–sucrose galactosyltransferases from different species, such as A. lyrata subsp. lyrate (94%), C. rubella (92%), C. sativa (91%), R. sativus (87%), O. sativa L. var. Nipponbare, B. oleracea (87%) and P. persica (76%). However, Lee et al. (2009[Lee, R.-H., Hsu, J.-H., Huang, H.-J., Lo, S.-F. & Chen, S.-C. G. (2009). New Phytol. 184, 596-606.]) analyzed 31 plant α-galactosidases and RFO synthases based on sequence alignments and identities and found that they could be classified into acidic α-galactosidases, alkaline α-galactosidases and RFO synthases, among which AtAkαGal3 belongs to the alkaline α-galactosidase group. Some structures of acidic α-galactosidases have been reported, including α-galactosidase from Thermotoga maritima (Ren et al., 2018[Ren, W., Pengelly, R., Farren-Dai, M., Shamsi Kazem Abadi, S., Oehler, V., Akintola, O., Draper, J., Meanwell, M., Chakladar, S., Świderek, K., Moliner, V., Britton, R., Gloster, T. M. & Bennet, A. J. (2018). Nat. Commun. 9, 3243.]), chicken α-N-acetylgalactos­aminidase (Garman et al., 2002[Garman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425-434.]), rice α-galactosidase (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]), human α-galactosidase (Garman & Garboczi, 2004[Garman, S. C. & Garboczi, D. N. (2004). J. Mol. Biol. 337, 319-335.]) and Trichoderma reesei α-galactosidase (Golubev et al., 2004[Golubev, A. M., Nagem, R. A. P., Brandão Neto, J. R., Neustroev, K. N., Eneyskaya, E. V., Kulminskaya, A. A., Shabalin, K. A., Savel'ev, A. N. & Polikarpov, I. (2004). J. Mol. Biol. 339, 413-422.]), which share low sequence identities of 32%, 29%, 27%, 25% and 24%, respectively, with AtAkαGal3, whereas other α-galactosidases from Nicotiana benthamiana (Kytidou et al., 2018[Kytidou, K., Beekwilder, J., Artola, M., van Meel, E., Wilbers, R. H. P., Moolenaar, G. F., Goosen, N., Ferraz, M. J., Katzy, R., Voskamp, P., Florea, B. I., Hokke, C. H., Overkleeft, H. S., Schots, A., Bosch, D., Pannu, N. & Aerts, J. M. F. G. (2018). J. Biol. Chem. 293, 10042-10058.]), Bacteroides fragilis (Joint Center for Structure Genomics, unpublished work) and Saccharomyces cerevisiae (Fernández-Leiro et al., 2010[Fernández-Leiro, R., Pereira-Rodríguez, A., Cerdán, M. E., Becerra, M. & Sanz-Aparicio, J. (2010). J. Biol. Chem. 285, 28020-28033.]) show no reasonable identity to AtAkαGal3.

3.5. The catalytic residues and binding sites for various natural substrates

A sequence alignment of acidic α-galactosidases from rice (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]), chicken (Garman et al., 2002[Garman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425-434.]), human (Garman & Garboczi, 2004[Garman, S. C. & Garboczi, D. N. (2004). J. Mol. Biol. 337, 319-335.]), N. benthamiana (Kytidou et al., 2018[Kytidou, K., Beekwilder, J., Artola, M., van Meel, E., Wilbers, R. H. P., Moolenaar, G. F., Goosen, N., Ferraz, M. J., Katzy, R., Voskamp, P., Florea, B. I., Hokke, C. H., Overkleeft, H. S., Schots, A., Bosch, D., Pannu, N. & Aerts, J. M. F. G. (2018). J. Biol. Chem. 293, 10042-10058.]), B. fragilis, a Gram-negative bacterium (Joint Center for Structure Genomics, unpublished work), S. cerevisiae (Fernández-Leiro et al., 2010[Fernández-Leiro, R., Pereira-Rodríguez, A., Cerdán, M. E., Becerra, M. & Sanz-Aparicio, J. (2010). J. Biol. Chem. 285, 28020-28033.]) and T. reesei (Golubev et al., 2004[Golubev, A. M., Nagem, R. A. P., Brandão Neto, J. R., Neustroev, K. N., Eneyskaya, E. V., Kulminskaya, A. A., Shabalin, K. A., Savel'ev, A. N. & Polikarpov, I. (2004). J. Mol. Biol. 339, 413-422.]) with known structures indicates the presence of two conserved motifs KXD and RxxxD, where X can be Y, F, L or I and x is any variety of amino acid. The aspartic acid residues in the KXD and RxxxD motifs serve as the nucleophile and the catalytic acid (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]; Garman et al., 2002[Garman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425-434.]), respectively (Supplementary Fig. S3). A sequence comparison of AtAkαGal3 with acidic α-galactosidases further confirmed that Asp383 and Asp447 are presumably critical residues that serve as the nucleophile and the catalytic acid/base, respectively. In addition, a sequence alignment of AtAkαGal3 with alkaline α-galactosidases from tomato (AF512549) and melon (AY114164), seed imbibition protein from Arabidopsis (AtAkαGal1) and RFO synthases from japonica rice (XP_015621501), cucumber (AAD02832), soybean (XP_006576826), yoshino cherry (PQQ02596) and clementine (XP_006444535) shows these two residues, Asp383 and Asp447 in AtAkαGal3, to be conserved in these motifs: 381KVD383, except for raffinose synthase from cucumber where the valine is replaced by isoleucine, and 443RxxDD447 in the catalytic domain. Based on the alignment, Arg443 and Ala444 of the 443RASDD447 motif in AtAkαGal3 and most alkaline α-galactosidases are evidently conserved compared with those in RFO synthases (the RxxxD motifs contain the nucleophile and catalytic acid; Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]; Garman et al., 2002[Garman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425-434.]; Supplementary Fig. S4).

To confirm that these two residues, Asp383 and Asp447 in AtAkαGal3, play important roles in catalysis, we generated two single mutants D383A and D447A by replacing Asp383 and Asp447 with alanine by site-directed mutagenesis. The TLC results showed that the hydrolase activity of both mutants towards raffinose was abolished (data not shown).

3.6. Interactions of galactose at the binding site of wild-type AtAkαGal3 and the D383A mutant

Before the crystallization of wild-type AtAkαGal3, galactose was added to the protein solution via dialysis in order to obtain suitable crystals. Our first solved structure of wild-type AtAkαGal3 was thus in complex with galactose at a resolution of 2.17 Å. The galactose moiety, with a chair conformation, is located in the −1 subsite of the catalytic binding site (Fig. 1[link]a). The structure of the AtAkαGal3–galactose complex shows that the OH group at C1, the anomeric C atom, of the galactose moiety is in a β-anomeric form and points towards the bottom side of the catalytic binding site, forming a direct hydrogen bond (distance 3.06 Å) to the nucleophile Asp383 and to the OD1 group of the acid/base catalytic residue Asp447 through a water molecule, with distances of 2.66 and 2.87 Å, respectively. Moreover, the OD1 group of Asp447 makes a strong hydrogen bond (distance 2.54 Å) to the OH2 group of the galactose. The other hydroxyl groups of the galactose are also stabilized by numerous hydrogen-bond interactions with various surrounding side chains of the residues at the −1 binding site, such as Arg443, Lys381, Asp243, Asp244 and Trp307. Furthermore, three water molecules help to stabilize the galactose moiety through water-mediated hydrogen-bonding interactions with Trp78, Lys381, Arg443, Asp447 and Asp479.

[Figure 1]
Figure 1
Active site of the crystal structure of AtAkαGal3. (a) The crystal structure of wild-type AtAkαGal3. The detailed interactions of galactose (yellow sticks) in a β-anomeric form with residues (yellow sticks) at the −1 subsite are shown. OH1 of galactose makes a hydrogen-bond interaction with the side chain of Asp383 and a water molecule (cyan sphere), with distances of 3.06 and 2.66 Å, respectively. Three water molecules (cyan spheres) help to stabilize the galactose moiety through water-mediated hydrogen-bonding interactions with Trp78, Lys381, Arg443, Asp447 and Asp479. (b) A galactose bound at the catalytic site of the D383A mutant forms an interaction network with residues similar to that in wild-type AtAkαGal3, except that there is a direct interaction between OD2 of Asp447 and O1 of the galactose moiety; two additional water molecules (cyan spheres) help to stabilize O1 and O6 of galactose as Asp383 is replaced by alanine.

The structure of the D383A mutant in complex with galactose was also determined at a resolution of 2.05 Å (Fig. 1[link]b). Interestingly, superposition of the two galactose moieties at the catalytic binding sites of the wild-type AtAkαGal3–galactose and D383A–galactose complex structures shows that the galactose moiety in the D383A mutant exhibits a chair form with a slight distortion; the O1 hydroxyl group at the C1 carbon is in an α-anomeric form (Supplementary Fig. S5). This particular structural form thus places the O1 hydroxyl group of the galactose in a position nearer the OD2 group of Asp447 to form a direct hydrogen-bond interaction with a distance of 2.87 Å. This effect might be due to the loss of an interaction with the carboxyl group of Asp383, which is replaced by alanine with a short side chain. Except for O1 at the anomeric C atom, which exhibits a conformational alteration from a β-form to an α-form, the positions of the remaining C atoms and hydroxyl groups of galactose are conserved and make unchanged interactions around the catalytic site compared with the wild-type enzyme. In the D383A structure, two additional water molecules interacting with O1 and O6 of galactose help to stabilize the galactose moiety compared with the wild-type AtAkαGal3 structure. Using the Cremer–Pople calculator, the galactose molecules complexed in the D383A and D383D mutants show Cremer–Pople parameters φ, θ, Q of 88.829°, 24.021°, 0.713 and 123.014°, 6.867°, 0.646, respectively, which might indicate an induced fit of the substrate to push it towards the transition state.

The residues involved in product-binding interactions with the galactose moiety at the −1 subsite of AtAkαGal3 are similar to those in rice acidic α-galactosidase (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]), except for Trp78 in AtAkαGal3, which is structurally related to Trp164 of rice acidic α-galactosidase and to Trp65 of T. maritima α-galactosidase (Ren et al., 2018[Ren, W., Pengelly, R., Farren-Dai, M., Shamsi Kazem Abadi, S., Oehler, V., Akintola, O., Draper, J., Meanwell, M., Chakladar, S., Świderek, K., Moliner, V., Britton, R., Gloster, T. M. & Bennet, A. J. (2018). Nat. Commun. 9, 3243.]), which positions its side chain farther away from the galactose moiety. Trp164 of rice acidic α-galactosidase, which is located on the loop at the end of the β5 strand, directly interacts with the O1 hydroxyl group of the galactose at the −1 subsite and is a conserved residue among acidic α-galactosidases from plants, bacteria and yeast, suggesting that it plays a role in recognizing the galactose substrate moiety (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]). In contrast, Trp78 of AtAkαGal3, which is located in a short β-turn (residues 75–78) between two β-strands of the N-terminal domain, contributes part of the catalytic binding site together with the catalytic domain. Except for water-mediated interactions with the O1 hydroxyl group of the galactose and the side chain of Asp447, Trp78 seems to make no direct interactions with the galactose moiety at the −1 subsite in AtAkαGal3, but rather to make a major interaction at the +1 subsite, which is discussed in the following section.

3.7. Interactions of galactinol in the binding site of the D383A mutant

The structure of D383A in complex with galactinol was determined at a resolution of 2.63 Å from D383A crystals by a combination of co-crystallization and soaking with galactinol as described in Section 2[link]. The 5R-OH group (O6′) of the myo-inositol moiety of galactinol at the +1 subsite is stabilized by the NE1 group of Trp78 together with OD1 of Asp446 by hydrogen bonds. In addition, 4S-OH is stabilized by the NZ group of Lys75 by hydrogen bonds (Fig. 2[link]a). The conformation of the galactosyl moiety of galactinol differs from that of galactose bound to wild-type AtAkαGal3 and the D383A mutant (Fig. 2[link]b), but the galactosyl moiety remains stabilized at the −1 subsite by mostly the same residues, including Asp447, Arg443, Lys381 and Asp244, but not Asp243 and Trp307, making interactions with galactose. Two water molecules are located in the catalytic binding site, one of which interacts with the galactosyl moiety and the other with the myo-inositol moiety and the side chain of Asp346, to help stabilize the galactinol. This Asp346 residue seems to play a major role in galactinol binding at the +1 subsite (Fig. 2[link]a).

[Figure 2]
Figure 2
Interactions of galactinol at the active site of the D383A mutant compared with galactose binding to Asp383 in the wild type. The structure of D383A in complex with galactinol is shown. (a) The galactosyl moiety of galactinol (yellow sticks) is bound at the −1 subsite. OD2 of Asp447 makes a direct interaction with O2 of the galactosyl moiety at a distance of 3.36 Å. Lys75, Trp78, Asp346, Asp446 and Tyr449 (green sticks) are involved in stabilization of the myo-inositol moiety by hydrogen-bonding interactions at the +1 subsite. Water molecules are shown as cyan spheres. (b) The carbon ring of the galactosyl moiety (yellow) of galactinol is distorted from that of galactose (green) bound to Asp383; the O3 and O4 groups hence interact with a different residue, Lys381, at the catalytic site as shown in (a).

3.8. Interactions of raffinose in the binding site of the D383A mutant

The crystal structure of D383A in complex with raffinose was determined at a resolution of 2.47 Å. Lys75, Tyr78, Asp446 and Tyr449 help to stabilize the glucose moiety of the raffinose substrate at the +1 subsite (Fig. 3[link]). The NE1 group of Trp78 and OD1 of Asp446 form hydrogen bonds to O4 of the hydroxyl group with distances of 3.08 and 3.47 Å, respectively. NZ of the Lys75 side chain stabilizes O3 of the hydroxyl group through a hydrogen bond with a distance of 3.54 Å. Tyr449 stabilizes the glucose at the +1 subsite and the fructose moiety at the +2 subsite of raffinose via hydrogen bonds from its side-chain OH group to the O3 hydroxyl groups of both galactose and fructose with distances of 3.52 Å (Fig. 3[link]). One water molecule that forms a hydrogen bond (distance of 2.70 Å) to the side chain of Asp447 seems to be involved in stabilization of the fructose moiety through weak hydrogen bonding. The interactions of the residues at the −1 subsite with the galactosyl moiety of raffinose are similar to those with the galactosyl moiety of galactinol bound in the D383A complex.

[Figure 3]
Figure 3
The active-site structure of the complex of D383A with raffinose (yellow sticks). Various residues (green sticks) are involved in binding inter­actions with glucose at the +1 subsite and fructose at the +2 subsite, as well as the galactosyl moiety at the −1 subsite. The catalytic residue Asp447 and the D383A mutation are also shown. One water molecule (cyan sphere) is located at the catalytic binding site to help stabilize the fructose moiety through hydrogen bonding.

3.9. Interactions of the galactose and sucrose products of hydrolyzed raffinose in the binding site

During the crystallization of the D383A–raffinose complex, co-crystallization of the D383A mutant and the raffinose substrate was occasionally set up for over two months. Unexpectedly, we observed both the galactose and sucrose products in the catalytic binding site and an additional cavity in the crystal structure after a sufficient reaction time (Fig. 4[link]a). The D383A mutant seems to be capable of gradually hydrolyzing the raffinose substrate into galactose and sucrose products that remain in the catalytic site. The D383A mutant in complex with hydrolyzed raffinose reveals a secondary product-binding site for sucrose; this site is stabilized and formed in part by a unique short loop (residues 329–352) composed of three short helices at the surface of the catalytic pocket, which equip Thr342 and Asp346 to interact with the glucose and fructose moieties of sucrose through hydrogen bonds (Fig. 4[link]b). This distinct extra loop (residues 329–352) in AtAkαGal3 can also be found in raffinose/stachyose synthases based on sequence alignment (Fig. 4[link]c), but not in acid α-galactosidase-related proteins. Lys75 from a β-turn (residues 75–78) in the N-terminal domain also helps to stabilize the glucose and fructose moieties of sucrose by hydrogen bonds with a distance of 2.80 Å. In addition, Trp78 and Asp447 help to stabilize both the sucrose product and the galactose moiety at the −1 subsite through a water molecule (Fig. 4[link]b).

[Figure 4]
Figure 4
The electron density of galactose and sucrose products at the active site and sequence alignment of the loop region. (a) Both the galactose and sucrose products (yellow sticks) are clearly revealed with electron densities (gray mesh) at the −1 subsite and the secondary product-binding site, respectively, after co-crystallization of D383A and raffinose for two months. The secondary product-binding site is near the surface of the catalytic binding cavity, which is formed in part by a unique loop with three short helices (magenta) and a β-turn (residues 75–78). (b) The interaction network with distances between galactose, sucrose (yellow sticks) and residues (green and magenta sticks) is shown. Water molecules are presented as cyan spheres. (c) A sequence alignment shows that the loop region (residues 329–352 in the green box) exists exclusively in AtAkαGal3 and possibly in the RFO synthase family from different plants, including rice (ABF99470; Rice_seed_imbibition), soybean (XP_006576826; Raf_Soybean), pea (RFS_PEA; Raf_Pea), jute (OMO49898; Raf_Corchorus), cucumber (AAD02832; Raf_Cucumis), cacao (EOY02480; Raf_Cacao), japonica rice (XP_015621501; Raf_Oryza sativa), melon (NP_001284472; Raf_II_Melon), yoshino cherry (PQQ02596; STA_Prunus), clementine (XP_006444535; STA_Citrus) and Arabidopsis (NP_192106; STA_A. thaliana). Residue numbers are based on AtAkαGal3.

3.10. Interactions of stachyose in the binding site of the D383A mutant

The crystal structure of the complex of D383A with stachyose determined at a resolution of 2.0 Å further reveals the complete subsites (–1, +1, +2 and +3) of AtAkαGal3 (Fig. 5[link]). The galactose at the +1 subsite of stachyose is stabilized by hydrogen bonds to both NE of the Lys75 side chain and NH1 of the Trp78 side chain with distances of ∼3 Å. Even though Trp77 makes no direct interaction with galactose at the +1 subsite, it helps to stabilize galactose through a water molecule. The extensive water network through the side chain of Arg451 helps to stabilize the glucose moiety at the +2 subsite through a water molecule. In addition, water-mediated hydrogen-bond interactions greatly stabilize both the glucose and fructose moieties at the +2 and +3 subsites.

[Figure 5]
Figure 5
The active-site structure of D383A in complex with stachyose. The complexed structure reveals the complete subsites (–1, +1, +2 and +3) in the catalytic binding site of AtAkαGal3. The glucose in the +2 subsite, stabilized by Arg451, Tyr215 and Trp211 through water-mediated hydrogen-bond interactions, is located nearly at the edge of the pocket cavity. The binding interaction of fructose in the +3 subsite, which is located at the outer edge of the cavity, is utilized only by Asp346 (magenta sticks) and one water molecule at the outer edge of the cavity. Stachyose, protein residues and water molecules are shown as yellow sticks, green sticks and cyan spheres, respectively.

3.11. Structural and sequence comparison explain the dual function of AtAkαGal3

The GH36 family contains enzymes, including RFO synthases and alkaline α-galactosidases, that generally exhibit transglycosylation and hydrolase activities. The first reported α-galactosidase structure was that of T. maritima α-galactosidase (Ren et al., 2018[Ren, W., Pengelly, R., Farren-Dai, M., Shamsi Kazem Abadi, S., Oehler, V., Akintola, O., Draper, J., Meanwell, M., Chakladar, S., Świderek, K., Moliner, V., Britton, R., Gloster, T. M. & Bennet, A. J. (2018). Nat. Commun. 9, 3243.]). Structural alignment of T. maritima α-galactosidase with the related GH27 family enzymes shows that T. maritima α-galactosidase and rice acidic α-galactosidase (Fujimoto et al., 2003[Fujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313-20318.]) share the greatest structural similarity. A structural comparison of the (α/β)8-barrel domain of our alkaline α-galactosidase AtAkαGal3 (residues 201–531) with those of rice acidic α-galactosidase (residues 6–271) and T. maritima α-galactosidase (residues 165–481) shows poor structural similarity, with r.m.s.d.s of 3.74 and 11.37 Å, respectively. The structure of wild-type AtAkαGal3 shows that the catalytic residues Asp383 and Asp447 are separated by approximately 6.7 Å, which is also generally found for the double-replacement retaining mechanism in members of clan D of the GH27 enzymes, with an average distance of 6.5 Å, and in T. maritima α-galactosidase, with a distance of 6.3 Å (Comfort et al., 2007[Comfort, D. A., Bobrov, K. S., Ivanen, D. R., Shabalin, K. A., Harris, J. M., Kulminskaya, A. A., Brumer, H. & Kelly, R. M. (2007). Biochemistry, 46, 3319-3330.]). A structural alignment of the (α/β)8-barrel domains of AtAkαGal3, rice acidic α-galactosidase and T. maritima α-galactosidase based on the AtAkαGal3 structure shows that most catalytic binding residues at the −1 subsite are conserved. In addition, Trp78 of AtAkαGal3 is not part of the (α/β)8-barrel of the catalytic domain like the corresponding Trp of rice acidic α-galactosid­ase, but is located in the N-terminal domain like Trp65 of T. maritima α-galactosidase. Moreover, Lys75 and Trp78 in the N-terminal domain share hydrogen-bond interactions that stabilize the second galactose moiety in the +1 subsite (Table 2[link]).

Table 2
Structural alignment of the (α/β)8-barrel domain of AtAkαGal3 with rice α-galactosidase and T. maritima α-galactosidase and a sequence alignment between AtAkαGal3 and the raffinose/stachyose (Raf/Sta) synthases show the conserved, variable and potential hydrogen-bond interactions (distances within 3.5 Å) at the −1, +1, +2 and +3 subsites among these glycosidase families

Bold labels indicate the key catalytic residues at the active site. Substrates and products: Gal, galactose; Gol, galactinol; Raf, raffinose; Sta, stachyose.

Substrate/product        
Gal Gol Raf Sta AtAkαGal3 residues 75–78, 201–531 (PDB entries 7exf, 7exg, 7exh, 7exj, 7exr, 7exq) Rice α-galactosidase (GH27) residues 6–271 (PDB entry 1uas) T. maritima α-galactosidase (GH36) residues 165–481 (PDB entry 6gvd) Raf/Sta synthases (no structure, based on sequence)
      +3 Waters     Asp/Val
    +2 +2 Tyr449     Trp
        Asp447(w)     Asp
        Asp451(w)     Thr, (Arg)/Gln, (Arg, Glu)
  +1 +1 +1 Lys75     Lys
        Trp78   Trp65 Trp
        Asp446 Trp164 (−1)   Trp
              Asp/1E
−1 −1 −1 −1 Asp243 Asp51 Asp220 Asp
        Asp244 Asp52 Asp221 Asp
        Trp307 Lys128 Trp257 Trp
        Lys381   Lys325 Lys
        Asp383 Asp130 Asp327 Asp
        Arg443 Arg181 Arg383 Arg
        Asp447 Asp185 Asp387 Asp

A sequence alignment (ESPript3.0; Robert & Gouet, 2014[Robert, X. & Gouet, P. (2014). Nucleic Acids Res. 42, W320-W324.]) of plant alkaline α-galactosidases with plant raffinose and stachyose synthases shows that the FRSK75xW77W78 region (referring to AtAkαGal3) located at the β-turn between two β-strands of the N-terminal domain is conserved among these families (Fig. 6[link]). Furthermore, two residues, Arg451 and Tyr449 of AtAkαGal3, which are also found in raffinose and stachyose synthases, help to stabilize the binding of the galactose moiety at the +2 subsite through water molecules and directed interaction, respectively. These observations from both structural and sequence alignments lead to the possibility of a dual function of the alkaline α-galactosidase AtAkαGal3, as the identified residues involved in product and substrate binding at the −1, +1 and +2 subsites of the alkaline α-galactosidase AtAkαGal3 are identical to the functional residues in both the GH27 and GH36 families as well as the raffinose and stachyose synthase families. However, the structure of D383A complexed with stachyose shows that water molecules seem to be important to help to stabilize the fructose moiety at the +3 subsite (Table 2[link]).

[Figure 6]
Figure 6
Sequence alignment of the WW box region of AtAkαGal3 with related enzymes from plant sources. The sequence alignment of alkaline α-galactosidases from rice (AF251068; Alk_Oryza_sativa), melon (AAM75139; Alk_I_Melon) and tomato (NP_001234763; Alk_Tamato), seed imbibition proteins from Arabidopsis thaliana (AtAkαGal1; A.thaliana_seed_imbibition) and rice (ABF99470; Rice_seed_imbibition) and raffinose/stachyose synthases from melon (NP_001284472; Raf_II_Melon), soybean (XP_006576826; Raf_Soybean), rice (XP_015621501; Raf_Oryza_sativa), cucumber (AAD02832; Raf_cucumis), yoshino cherry (PQQ02596; STA_Prunus) and clementine (XP_006444535; STA_Citrus) shows a conserved WW box region (FRSK75xW77W78) among alkaline α-galactosidases and the RFO synthases. The sequence numbering refers to AtAkαGal3.

3.12. N-terminal domain and catalytic site of AtAkαGal3

Our work confirms that the N-terminal domain of the alkaline α-galactosidase AtAkαGal3 is important for protein function because it contributes part of the core (α/β)8-barrel catalytic site for substrate binding. The C-terminal domain helps to stabilize this structural architecture through inter­actions with both the N-terminal domain and the (α/β)8-barrel domain. Sequence alignment indicates that the RFO synthases might also use their N-terminal domain for a similar function. The stachyose bound in the catalytic pocket of AtAkαGal3 shows that the depth and dimensions of the binding pocket can accommodate four sugar moieties, whereas the catalytic binding sites of the acidic α-galactosidases rice α-galactosidase and T. maritima α-galactosidase are relatively shallow and are likely to accommodate only one galactose. Moreover, the extra loop (residues 329–352) that only exists in the alkaline α-galactosidase AtAkαGal3 provides two key residues Asp346 and Thr342 to help stabilize the fructose and glucose moieties of the sucrose product in the +2 and +3 subsites, respectively (Figs. 4[link] and 7[link]).

[Figure 7]
Figure 7
Superimposition of catalytic (α/β)8-barrel domains. (a) Superimposition of the catalytic (α/β)8-barrel domains of the alkaline α-galactosidase AtAkαGal3 (green), rice α-galactosidase (gray) and T. maritima α-galactosidase (yellow) with the product galactose at the respective −1 subsite. An extra loop (residues 329–352, magenta) exists exclusively in AtAkαGal3. (b) The complex of D383A with stachyose reveals that an extra loop helps to stabilize the product or stachyose substrate at the +2 and +3 subsites in the alkaline α-galactosidase AtAkαGal3. An electrostatic surface is shown (blue, positive charge; red, negative charge). (c, d) Docking of stachyose from the D383A complex into the structures of rice α-galactosidase (c) and T. maritima α-­galactosidase (d) shows that both catalytic binding sites are shallow such that they can accommodate only one galactose, compared with AtAkαGal3 that can hold four-sugar moieties such as stachyose.

Our D383A mutant can gradually hydrolyze raffinose into galactose and sucrose. A structural superimposition of D383A in complex with raffinose, with stachyose and with hydrolyzed raffinose in the catalytic binding pocket shows that the wider binding-site pocket of AtAkαGal3 provides a secondary product-binding site to stabilize the sucrose after raffinose has been hydrolyzed into galactose and glucose. This product-binding site contains Thr342 and Asp346 on a short loop (residues 329–352) that interact with the glucose and fructose moieties of sucrose through hydrogen bonds. Lys75 in the N-terminal domain also helps to stabilize the glucose moiety of sucrose by hydrogen bonding (Fig. 8[link]).

[Figure 8]
Figure 8
Active-site structures of D383A in complex with raffinose. The structures of D383A in complexes with raffinose (magenta), with stachyose (green) and with hydrolyzed raffinose, i.e. galactose and sucrose (cyan), in the catalytic binding pocket shown with the electrostatic surface indicate that the wider binding-site pocket of the alkaline α-galactosidase AtAkαGal3 provides a new secondary product-binding site to stabilize the sucrose moiety (cyan).

3.13. Dual function of AtAkαGal3

We compared the hydrolase activities of the alkaline α-galactosidase AtAkαGal3 and rice (O. sativa L. var. Nipponbare) raffinose synthase towards only the natural substrate raffinose, without galactinol, which serves as a galactose donor for raffinose or stachyose synthase activity. The TLC assay showed that AtAkαGal3 exhibits a greater hydrolase activity towards the raffinose substrate considering the TLC spot intensities of the galactose and sucrose products. AtAkαGal3 can also produce a stachyose product without galactinol as a galactose donor. This observation indicates that AtAkαGal3 seems to act with the transferase mechanism observed for fructosyl transferases and xyloglucan endotransferases. Our crystal structures of wild-type AtAkαGal3 and its D383A mutant in complex with galactose and with hydrolyzed raffinose, i.e. galactose and sucrose, reveal one water molecule at the conserved position near the anomeric C atom and Asp447. This water is replaced by O4 of galactose in the +1 subsite of raffinose and stachyose in the complexes of D383A with raffinose and stachyose. The only difference between the hydrolase and transferase reactions is the acceptor type after the removal of the terminal glycosyl moiety. The acceptor for a hydrolase is a water molecule, whereas a transferase uses a saccharide or an oligosaccharide, such as sucrose, raffinose or stachyose, as the acceptor (Henrissat et al., 2001[Henrissat, B., Coutinho, P. M. & Davies, G. J. (2001). Plant Mol. Biol. 47, 55-72.]; Carmi et al., 2003[Carmi, N., Zhang, G., Petreikov, M., Gao, Z., Eyal, Y., Granot, D. & Schaffer, A. A. (2003). Plant J. 33, 97-106.]; Chuankhayan et al., 2010[Chuankhayan, P., Hsieh, C.-Y., Huang, Y.-C., Hsieh, Y.-Y., Guan, H.-H., Hsieh, Y.-C., Tien, Y.-C., Chen, C.-D., Chiang, C.-M. & Chen, C.-J. (2010). J. Biol. Chem. 285, 23251-23264.]). In AtAkαGal3, the conserved water molecule identified above might thus serve as an acceptor in the hydrolase step, releasing galactose as a product, while sucrose and raffinose act as an acceptor for the transferase mechanism, releasing raffinose and stachyose as products. In addition, the unique extra loop (residues 329–352) in AtAkαGal3, which contributes to the depth enhancement of the catalytic pocket, presumably supports the transferase mechanism to help to stabilize the oligomer products.

3.14. Catalytic and synthetic mechanism

The structure of D383A in complex with hydrolyzed raffinose reveals the products galactose at the −1 subsite and sucrose at the secondary binding site. Unexpectedly, after raffinose has been hydrolyzed between galactose and glucose at subsites −1 and +1, respectively, the sucrose shifts and rotates by 180°. The glucose moiety of sucrose in the +1 subsite is thus replaced by a fructose moiety and is bound in a different orientation and position to the related moieties of raffinose and stachyose in other D383A–substrate complexes (Fig. 8[link]). We specify this newly discovered binding position as the secondary product-binding site. A 180° rotation of the sucrose product after the hydrolysis of raffinose is possible because the catalytic pocket is wide, which provides sufficient space for flipping and binding (Fig. 9[link]). After rotation, Thr342 and Asp346 help to stabilize the glucose and fructose of the sucrose product, respectively. Furthermore, Lys75, which helps to stabilize the sugar moiety in the +1 subsite, also makes a strong interaction with the glucose of the sucrose product in the +3 subsite, indicating that the conserved residues Lys75 and Trp78 in the WW box region are important for stabilization of both the substrate and the product in the +1 subsite, according to the proposed mechanism in Fig. 10[link]. Lys381 might be the key stabilizer during the catalytic process.

[Figure 9]
Figure 9
The active site for galactose and the secondary active site for sucrose. (a) A side view of the superimposed substrate raffinose (magenta) and products galactose and sucrose (cyan) from hydrolyzed raffinose in the catalytic binding pocket shown as an electrostatic surface. An empty space is found when the substrate raffinose is bound; this space is then occupied by the sucrose product after raffinose has been hydrolyzed. (b) A top view clearly shows that after raffinose has been hydrolyzed the sucrose product is flipped 180° and occupies the secondary product-binding site.
[Figure 10]
Figure 10
The proposed catalytic mechanism of the alkaline α-galactosidase AtAkαGal3. Schematic presentation of the proposed catalytic mechanism of the alkaline α-galactosidase AtAkαGal3 with raffinose as a donor and an acceptor. The main interactions of residues involved in substrate recognition and the catalytic process are shown: the nucleophile (Asp383), acid/base catalyst (Asp447), Lys381 and Arg443 for the −1 subsite, Trp78, Lys75, Asp446 and Tyr449 for the +1 subsite, Lys75, Trp77, Thr342 and Asp346 for sucrose at the secondary product-binding site, Asp346 for the +3 subsite and water molecules that help to stabilize the +2 subsite. (a) Raffinose as a donor substrate binds in the catalytic binding site. (b) After the glycosidic bond of raffinose has been cleaved, the sucrose product rotates 180° and binds to the secondary product-binding site. The sucrose is subsequently released and galactose remains in the catalytic pocket. (c) Another raffinose, which acts as an acceptor substrate, enters the pocket and moves towards the galactose that remains at the −1 subsite. (d) The acceptor raffinose is engaged with galactose, resulting in the production of stachyose.

In summary, the alkaline α-galactosidase AtAkαGal3 from A. thaliana exhibits a dual function: it not only catalyzes the hydrolysis of α-D-galactose from galacto-oligosaccharides under alkaline conditions but can also synthesize stachyose using raffinose, instead of galactinol, as the galactose donor. Structural analyses of AtAkαGal3 and its D383A mutant in complex with various substrates and products, including galactose, galactinol, raffinose, stachyose and sucrose, elucidate four complete subsites (–1 to +3) and a new secondary product-binding site. The AtAkαGal3 structure in this study is the first representative structure of an alkaline α-galactosidase and may also serve a structural model for the raffinose family of RFO synthases.

Acknowledgements

We are indebted to the staff of beamlines TPS 05A and TLS 15A at the National Synchrotron Radiation Research Center (NSRRC) in Taiwan, Eiki Yamashita at BL44XU and the staff of the Taiwan beamline BL12B2 at SPring-8 in Japan for technical assistance. We are grateful for travel support from the International Collaborative Research Program of the Institute for Protein Research, Osaka University. We are also grateful to Ming-Hong Lai for the preliminary crystallization conditions of AtAkαGal3 and to Chien-Ming Chiang for discussion. We thank Shinya Fushinobu for the online Cremer–Pople parameter calculator.

Funding information

This work was supported by National Science and Technology Council (NSTC) grants 105-2311-B-213-001-MY3, 107-2923-B-213-001-MY3, 108-2311-B-213-001-MY3 and 111-2311-B-213-001 and NSRRC grants to C.-J. Chen.

References

First citationCalhoun, D. H., Bishop, D. F., Bernstein, H. S., Quinn, M., Hantzopoulos, P. & Desnick, R. J. (1985). Proc. Natl Acad. Sci. USA, 82, 7364–7368.  CrossRef CAS PubMed Web of Science Google Scholar
First citationCarmi, N., Zhang, G., Petreikov, M., Gao, Z., Eyal, Y., Granot, D. & Schaffer, A. A. (2003). Plant J. 33, 97–106.  Web of Science CrossRef PubMed CAS Google Scholar
First citationChang, S., Puryear, J. & Cairney, J. (1993). Plant Mol. Biol. Rep. 11, 113–116.  CrossRef CAS Google Scholar
First citationChen, C.-D., Huang, Y.-C., Chiang, H.-L., Hsieh, Y.-C., Guan, H.-H., Chuankhayan, P. & Chen, C.-J. (2014). Acta Cryst. D70, 2331–2343.  Web of Science CrossRef IUCr Journals Google Scholar
First citationChuankhayan, P., Hsieh, C.-Y., Huang, Y.-C., Hsieh, Y.-Y., Guan, H.-H., Hsieh, Y.-C., Tien, Y.-C., Chen, C.-D., Chiang, C.-M. & Chen, C.-J. (2010). J. Biol. Chem. 285, 23251–23264.  Web of Science CrossRef CAS PubMed Google Scholar
First citationComfort, D. A., Bobrov, K. S., Ivanen, D. R., Shabalin, K. A., Harris, J. M., Kulminskaya, A. A., Brumer, H. & Kelly, R. M. (2007). Biochemistry, 46, 3319–3330.  Web of Science CrossRef PubMed CAS Google Scholar
First citationCorchete, M. P. & Guerra, H. (1987). Phytochemistry, 26, 927–932.  CrossRef CAS Web of Science Google Scholar
First citationDey, P. M. & Pridham, J. B. (1972). Adv. Enzymol. Relat. Areas Mol. Biol. 36, 91–130.  CAS PubMed Web of Science Google Scholar
First citationDoublié, S. (2007). Methods Mol. Biol. 363, 91–108.  PubMed Google Scholar
First citationEmsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. (2010). Acta Cryst. D66, 486–501.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationFarag, S. (1978). J. Am. Soc. Sugar Beet Technol. 20, 251–254.  CrossRef Google Scholar
First citationFernández-Leiro, R., Pereira-Rodríguez, A., Cerdán, M. E., Becerra, M. & Sanz-Aparicio, J. (2010). J. Biol. Chem. 285, 28020–28033.  Web of Science PubMed Google Scholar
First citationFujimoto, Z., Kaneko, S., Momma, M., Kobayashi, H. & Mizuno, H. (2003). J. Biol. Chem. 278, 20313–20318.  Web of Science CrossRef PubMed CAS Google Scholar
First citationGao, Z. & Schaffer, A. A. (1999). Plant Physiol. 119, 979–988.  Web of Science CrossRef PubMed CAS Google Scholar
First citationGarman, S. C. & Garboczi, D. N. (2004). J. Mol. Biol. 337, 319–335.  Web of Science CrossRef PubMed CAS Google Scholar
First citationGarman, S. C., Hannick, L., Zhu, A. & Garboczi, D. N. (2002). Structure, 10, 425–434.  Web of Science CrossRef PubMed CAS Google Scholar
First citationGaudreault, P. R. & Webb, J. A. (1986). Plant Sci. 45, 71–75.  CrossRef CAS Web of Science Google Scholar
First citationGolubev, A. M., Nagem, R. A. P., Brandão Neto, J. R., Neustroev, K. N., Eneyskaya, E. V., Kulminskaya, A. A., Shabalin, K. A., Savel'ev, A. N. & Polikarpov, I. (2004). J. Mol. Biol. 339, 413–422.  Web of Science CrossRef PubMed CAS Google Scholar
First citationGuimarães, V. M., de Rezende, S. T., Moreira, M. A., de Barros, E. G. & Felix, C. R. (2001). Phytochemistry, 58, 67–73.  Web of Science PubMed Google Scholar
First citationHara, M., Tokunaga, K. & Kuboi, T. (2008). Plant Biotechnol. 25, 497–501.  Web of Science CrossRef CAS Google Scholar
First citationHenrissat, B. (1991). Biochem. J. 280, 309–316.  CrossRef PubMed CAS Web of Science Google Scholar
First citationHenrissat, B., Coutinho, P. M. & Davies, G. J. (2001). Plant Mol. Biol. 47, 55–72.  Web of Science CrossRef PubMed CAS Google Scholar
First citationHenrissat, B. & Romeu, A. (1995). Biochem. J. 311, 350–351.  CrossRef CAS PubMed Web of Science Google Scholar
First citationHerman, E. M. & Shannon, L. M. (1985). Plant Physiol. 77, 886–890.  CrossRef PubMed CAS Web of Science Google Scholar
First citationKabsch, W. (2010). Acta Cryst. D66, 125–132.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationKang, H.-C. & Lee, S.-H. (2001). Phytochemistry, 58, 213–219.  Web of Science CrossRef PubMed CAS Google Scholar
First citationKeller, F. & Pharr, D. M. (1996). Photoassimilate Distribution in Plants and Crops: Source–Sink Relationships, edited by E. Zamski & A. A. Scharffer, pp. 157–184. New York: Marcel Dekker.  Google Scholar
First citationKytidou, K., Beekwilder, J., Artola, M., van Meel, E., Wilbers, R. H. P., Moolenaar, G. F., Goosen, N., Ferraz, M. J., Katzy, R., Voskamp, P., Florea, B. I., Hokke, C. H., Overkleeft, H. S., Schots, A., Bosch, D., Pannu, N. & Aerts, J. M. F. G. (2018). J. Biol. Chem. 293, 10042–10058.  Web of Science CrossRef CAS PubMed Google Scholar
First citationLanger, G., Cohen, S. X., Lamzin, V. S. & Perrakis, A. (2008). Nat. Protoc. 3, 1171–1179.  Web of Science CrossRef PubMed CAS Google Scholar
First citationLee, D. W. & Collins, T. M. (2001). Int. J. Plant Sci. 162, 1141–1153.  Web of Science CrossRef CAS Google Scholar
First citationLee, R.-H., Hsu, J.-H., Huang, H.-J., Lo, S.-F. & Chen, S.-C. G. (2009). New Phytol. 184, 596–606.  Web of Science CrossRef PubMed CAS Google Scholar
First citationLee, R.-H., Lin, M.-C. & Chen, S.-C. G. (2004). Plant Mol. Biol. 55, 281–295.  Web of Science CrossRef PubMed CAS Google Scholar
First citationLi, S., Li, T., Kim, W. D., Kitaoka, M., Yoshida, S., Nakajima, M. & Kobayashi, H. (2007). Biotechnol. Lett. 29, 635–640.  Web of Science CrossRef PubMed CAS Google Scholar
First citationMoriarty, N. W., Grosse-Kunstleve, R. W. & Adams, P. D. (2009). Acta Cryst. D65, 1074–1080.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationMurshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355–367.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationNunan, K. J., Davies, C., Robinson, S. P. & Fincher, G. B. (2001). Planta, 214, 257–264.  Web of Science CrossRef PubMed CAS Google Scholar
First citationOtwinowski, Z. & Minor, W. (1997). Methods Enzymol. 276, 307–326.  CrossRef CAS PubMed Web of Science Google Scholar
First citationRen, W., Pengelly, R., Farren-Dai, M., Shamsi Kazem Abadi, S., Oehler, V., Akintola, O., Draper, J., Meanwell, M., Chakladar, S., Świderek, K., Moliner, V., Britton, R., Gloster, T. M. & Bennet, A. J. (2018). Nat. Commun. 9, 3243.  Web of Science CrossRef PubMed Google Scholar
First citationRobert, X. & Gouet, P. (2014). Nucleic Acids Res. 42, W320–W324.  Web of Science CrossRef CAS PubMed Google Scholar
First citationSalentin, S., Schreiber, S., Haupt, V. J., Adasme, M. F. & Schroeder, M. (2015). Nucleic Acids Res. 43, W443–W447.  Web of Science CrossRef CAS PubMed Google Scholar
First citationShibuya, H., Kobayashi, H., Park, G. G., Komatsu, Y., Sato, T., Kaneko, R., Nagasaki, H., Yoshida, S., Kasamo, K. & Kusakabe, I. (1995). Biosci. Biotechnol. Biochem. 59, 2333–2335.  CrossRef CAS PubMed Web of Science Google Scholar
First citationTerwilliger, T. C. & Berendzen, J. (1999). Acta Cryst. D55, 849–861.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationVagin, A. & Teplyakov, A. (2010). Acta Cryst. D66, 22–25.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationWeidemann, F., Breunig, F., Beer, M., Sandstede, J., Turschner, O., Voelker, W., Ertl, G., Knoll, A., Wanner, C. & Strotmann, J. M. (2003). Circulation, 108, 1299–1301.  Web of Science CrossRef PubMed CAS Google Scholar
First citationWilliams, C. J., Headd, J. J., Moriarty, N. W., Prisant, M. G., Videau, L. L., Deis, L. N., Verma, V., Keedy, D. A., Hintze, B. J., Chen, V. B., Jain, S., Lewis, S. M., Arendall, W. B. III, Snoeyink, J., Adams, P. D., Lovell, S. C., Richardson, J. S. & Richardson, D. C. (2018). Protein Sci. 27, 293–315.  Web of Science CrossRef CAS PubMed Google Scholar
First citationWinn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A. & Wilson, K. S. (2011). Acta Cryst. D67, 235–242.  Web of Science CrossRef CAS IUCr Journals Google Scholar
First citationZhao, T. Y., Corum, J. W. III, Mullen, J., Meeley, R. B., Helentjaris, T., Martin, D. & Downie, B. (2006). Seed Sci. Res. 16, 107–121.  Web of Science CrossRef CAS Google Scholar

This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.

Journal logoSTRUCTURAL
BIOLOGY
ISSN: 2059-7983
Follow Acta Cryst. D
Sign up for e-alerts
Follow Acta Cryst. on Twitter
Follow us on facebook
Sign up for RSS feeds