Novel α-l-Arabinofuranosidase from Cellulomonas fimi ATCC 484 and

Mar 23, 2015 - In the process of gene mining for novel α-l-arabinofuranosidases (AFs), the gene Celf_3321 from Cellulomonas fimi ATCC 484 encodes an ...
0 downloads 6 Views 8MB Size
Article pubs.acs.org/JAFC

Novel α‑L‑Arabinofuranosidase from Cellulomonas fimi ATCC 484 and Its Substrate-Specificity Analysis with the Aid of Computer Ying Yang,† Lujia Zhang,† Mingrong Guo,† Jiaqi Sun,† Shingo Matsukawa,§ Jingli Xie,*,†,‡ and Dongzhi Wei†,‡ †

State Key Laboratory of Bioreactor Engineering, Department of Food Science and Technology, School of Biotechnology, East China University of Science and Technology, Shanghai 200237, People’s Republic of China ‡ Shanghai Collaborative Innovation Center for Biomanufacturing, Shanghai 200237, People’s Republic of China § Department of Food Science and Technology, Tokyo University of Marine Science and Technology, Tokyo 108-8477, Japan S Supporting Information *

ABSTRACT: In the process of gene mining for novel α-L-arabinofuranosidases (AFs), the gene Celf_3321 from Cellulomonas fimi ATCC 484 encodes an AF, termed as AbfCelf, with potent activity, 19.4 U/mg under the optimum condition, pH 6.0 and 40 °C. AbfCelf can hydrolyze α-1,5-linked oligosaccharides, sugar beet arabinan, linear 1,5-α-arabinan, and wheat flour arabinoxylan, which is partly different from some previously well-characterized GH 51 AFs. The traditional substrate-specificity analysis for AFs is labor-consuming and money costing, because the substrates include over 30 kinds of various 4-nitrophenol (PNP)-glycosides, oligosaccharides, and polysaccharides. Hence, a preliminary structure and mechanism based method was applied for substratespecificity analysis. The binding energy (ΔG, kcal/mol) obtained by docking suggested the reaction possibility and coincided with the experimental results. AbfA crystal 1QW9 was used to test the rationality of docking method in simulating the interaction between enzyme and substrate, as well the credibility of the substrate-specificity analysis method in silico. KEYWORDS: α-L-arabinofuranosidase, substrate specificity, binding energy, docking



substrate specificities.9−11 This classification was useful to study evolutionary relationship, mechanistic information, and structural features of these enzymes.12 However, AFs display so different substrate specificities that within any given GH family, the substrate specificity of individual members could differ, and the GH 51 and 54 AFs were believed to exhibit more flexible substrate specificity.13 The GH 51 family contains endo-β-1,4-xylanase (EC 3.2.1.8), endoglucanase (EC 3.2.1.4), β-xylosidase (EC 3.2.1.37), and AFs. Activity of the endo-β-1,4-xylanase was tested by using birchwood xylan, carboxymethyl cellulose (CMC), Avicel, laminarin, and lichenan as substrates.14 The characterization of endoglucanase always included determining the activity toward barley β-glucan, CMC, xylan, laminarin, and dextrin.15 Thus, for the characterization of substrate specificity of GH 51 AFs, not only the arabinose-containing oligosaccharides and polysaccharides were used as substrate, but also xylan and CMC.16,17 Although GH 51 AFs were reported to be capable of hydrolyzing arabinose-containing oligosaccharides and polysaccharides, the catalytic efficiency against the substrates even varied greatly among different AFs of the same family. For example, a GH 51 AF, AbfATK4 from Geobacillus caldoxylolyticus TK4 showed no activity toward arabinoxylan, which was one of the main substrate for AFs,16 while ABF2 from Penicillium purpurogenum and another GH 51

INTRODUCTION Many enzymes are involved in the degradation of the plant cell wall, including endo-β-1,4-D-xylanase, β-xylosidase α-D-glucuronidases, acetylxylan esterases, and phenol aid esterases, among which α-L-arabinofuranosidases (EC 3.2.1.55; AFs) are accessory enzymes. AFs catalyze the hydrolysis of α-1,2-, α1,3-, or α-1,5-L-arabinofuranoside linkages and act synergistically with other hemicellulases or pectic enzymes for the complete degradation of hemicellulose and arabinose-containing polysaccharides.1,2 The complete extraction of heteropolysaccharides from natural biomass requires the concerted action of many different enzymes, including polygalacturonase, rhamnogalacturonase, α-1,5-L-arabinanase (EC 3.2.1.99; ABN), and AFs.3,4 And AFs play a key role in this synergistic process, since they are responsible for the cleavage of the arabinose side chains, allowing ABNs to cleave the glycosidic bonds of the backbone and the links between protopectin and cellulose, thereby promoting the solubilization of protopectin.1,5 Therefore, AFs are widely used in the food industry, such as wine and juice industry.6 Adding AFs into the dough will produce free pentoses (mainly arabinose and xylose) and increase the availability of soluble carbohydrates, thereby prolonging the shelf life of bread.7 Furthermore, AFs are essential for efficient production of L-arabinose from different agricultural waste and biomass. The L-arabinose is often used as a bioactive sweetener to prevent postprandial hyperglycaemia in diabetic individuals.6,8 AFs have been grouped into seven glycoside hydrolase (GH) families, GH 1, 3, 10, 43, 51, 54, and 64, depending on their amino acid sequences at their active sites rather than their © 2015 American Chemical Society

Received: Revised: Accepted: Published: 3725

December March 16, March 23, March 23,

11, 2014 2015 2015 2015 DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry

Strain, Media, and Plasmids. The strain used in this study, C. fimi ATCC 484, was obtained from own preservation, which was subcultured from a stock culture (50% glycerol) made from the original isolate. E. coli DH5α, E. coli BL21 (DE3), and the plasmid pET28a (+) were used for cloning and expression. Transformants were grown in Luria−Bertani (LB) medium (1% peptone, 1% NaCl, and 0.5% yeast extract) with 100 μg/mL kanamycin. Sequence Analysis. Multiple nucleotide sequence alignment of gene abf Celf was performed with the DNAMAN program. The molecular weight and pI were predicted by using the Compute pI/Mw tool on the ExPASy Bioformatics Resources Portal (http://web. expasy.org/compute_pi/). Cloning, Expression, and AbfCelf Purification. The coding sequence of abf Celf was amplified by polymerase chain reaction (PCR) with the following forward and reverse primers: forward primer for abf Celf, 5′-TTTGGATCCCCTCAGCGCGCCACCGT-3′; reverse primer for abf Celf, 5′-TTTGAATTCGCGGGTCAGGGTCGCCG-3′. The products of PCR reaction were cloned in the plasmid pET28a (+) containing both C- and N-terminal His6-tags for purification of AbfCelf. The recombinant plasmid was transformed into E. coli BL21 (DE3) and grown on solid LB medium. The transformation of E. coli strains was performed as described previously.26 The target protein was purified by using Ni−Sepharose through eluting with imidazole in a linear gradient from 10 to 500 mM. The active fractions were identified by 12.5% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS−PAGE). Then, imidazole was ultrafiltered by using a 15 mL 10 kDa Millipore’s Amicon Ultra-15 centrifugal filter device (Millipore, MA, U.S.A.). The concentrated active fractions were stored in 100 mM phosphate buffer (pH 7.0) containing 10% glycerol in several aliquots at −20 °C. One aliquot was used for enzyme activity verification as described in the following section. AF Assay and Protein Measurement. The reaction system for AF activity determination consisted of 30 μL of PNP-Abf (7 mM) in 205 μL of sodium acetate buffer. An amount of 15 μL of diluted enzyme sample (10−50 dilution volume by 100 mM phosphate buffer according to the enzyme protein concentration) was incubated at 40 °C for 15 min, then the reaction was stopped by adding 850 μL of 1 M Na2CO3. Finally, absorbance at 405 nm was read by using a UV-1200 spectrophotometer (MAPADA, Shanghai, China). One unit of AF activity was defined as the amount of enzyme that liberates 1 μmol of 4-nitrophenol (PNP) per min in the reaction mixture under these assay conditions. The protein content was determined using Bradford reagent with bovine serum albumin (BSA) as standard. Standard curves of PNP and BSA are shown in Figure S1 in the Supporting Information. The molecular mass of the enzyme AbfCelf was determined by SDS−PAGE on a 12.5% polyacrylamide gel. Proteins were visualized by staining with the Coomassie brilliant blue R 250. All assays were performed in triplicate batches, and the results are presented as means ± standard deviation (SD). Enzymatic Properties. The optimum pH was determined by detecting the enzyme activities in sodium acetate buffer (pH 4.0−6.0), sodium phosphate buffer (pH 6.0−8.0), and Tris−HCl buffer (pH 8.0−9.0) at 40 °C for 15 min. The optimum temperature was determined by measuring the enzyme activities of reactions incubated at temperatures ranging from 5 to 70 °C in 100 mM sodium acetate buffers (pH 6.0) for 15 min. To estimate pH stability, the enzyme was incubated in buffers ranging from pH 4.0 to 8.0 at 40 °C for various periods. And for temperature stability, the enzyme was incubated in 100 mM sodium acetate buffers (pH 6.0) at temperatures ranging from 40 to 60 °C for various phases. Then, the residual enzyme activity was measured under optimum condition. The kinetic parameters, apparent Km, and Vmax values were determined through the Lineweaver−Burk plot method under optimum condition by using PNP-Abf at concentrations ranging from 0.02 to 2 mM for 5 min of reaction.27 Substrate Specificity. Substrate specificity was estimated with different synthetic PNP-glycosides, oligosaccharides, and polysacchar-

AF from Streptomyces chartreusis GS901 demonstrated slight reacting with this substrate.2,18 In contrast, AbfD3 from Thermobacillus xylanilyticus, also a member of GH 51, was found to be extremely active on arabinoxylan.19 Accordingly, it is difficult to find regularity in the substrate specificity of AFs. Consequently, the definition of substrate specificity of certain AFs needs amount of experimental attempts; however, this is cost- and labor-intensive. Structure understanding and theoretical computation are popular assisting means in the research of various enzymes, including analysis of substrate specificity, rational design of stable protein, and mutation prediction on proteins.20,21 For example, the programs like PREDIKIN and CaSPredictor were especially written for the prediction of substrate specificity in protein serine/threonine kinases, cytochrome P450, and caspases, achieving medium to good performances.22−24 In the absence of computational tools, substantial and reliable crystal structures of substrate−enzyme complex were utilized with no hesitation for substrate-specificity analysis with the assist of the automated protocols like SOLVE, ARP/wARP, and REFMAC.25 In the situation no experimental resulted structure is feasible; the closely related structure can be relied on for molecular modeling of the enzyme under study. And substrate specificity can also be analyzed according to the information obtained from ligand docking through software like Autodock. We found some AF genes from different strains including Celf_3321, “AF domain-containing protein” from Cellulomonas fimi ATCC 484, by gene mining on the NCBI Web site (http://www.ncbi.nlm.nih.gov/). In this paper, we described the cloning and expression of this gene in Escherichia coli (E. coli), termed as abf Celf. The expressed enzyme AbfCelf displayed a relatively higher AF activity among the mined AFs. Then, AbfCelf was characterized, and the catalytic activities toward different substrates were determined after purification due to that the substrate specificity of GH 51 AFs was different. In case of no ready-made professional computational tools and crystal structure are available for AbfCelf, structure and mechanism based methods were applied to analyze substrate specificity to the greatest extent. The structure of AbfCelf was built by automated protein structure homology modeling from the Swiss Model server (http://swissmodel. expasy.org/). Substrates bind with the structure of AbfCelf by using Autodock 4.2, and supplementary analysis was conducted by using the discovery studio 4.0. The experimental results were compared with docking values, which helped to analyze substrate specificity of AbfCelf.



MATERIALS AND METHODS

Chemicals and Reagents. Substrates 4-nitrophenyl-α-L-arabinofuranoside (PNP-Abf), PNP-α-D-xylopyranoside, PNP-β-D-xylopyranoside, L-(+)-arabinose, and xylan (from beechwood) were purchased from Sigma-Aldrich Co., Ltd. (St. Louis, MO, U.S.A.). Sugar beet linear 1,5-α-arabinan, sugar beet red debranched arabinan, wheat flour arabinoxylan (medium viscosity), larch arabinogalactan, arabinan (sugar beet), arabinobiose, arabinotriose, and arabinotetraose were obtained from Megazyme International Ireland Co., Ltd. (Wicklow, Ireland). Gum arabic power and soluble starch were from Shanghai Hushi Co., Ltd. (Shanghai, China). Genomic DNA extraction kits, plasmid extraction kits, gel extraction kits, and Ni−Sepharose resin were purchased from Tiangen Biotech Co., Ltd. (Beijing, China). All other molecular cloning reagents were from Takara Biotechnology Co., Ltd. (Dalian, China). Other chemicals used in this study were of analytical grade or higher. 3726

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry

Figure 1. Multiple sequence alignment of AbfCelf with some GH 51 AFs. AbfCelf was compared with AFs from R. albus 8 (AEE64774), B. subtilis 168T+ (CAA99595), G. caldoxylolyticus TK4 (ABI34800), G. stearothermophilus T-6 (AAD45520.2), T. petrophila RKU-1(ABQ46651), and T. xylanilyticus T3 (CAA76421). The catalytic residues are marked with asterisks. Residues that are identical in all sequences are shaded in blue. included α-1,5-linked oligosaccharides, xylan from beechwood, linear 1,5-α-arabinan, wheat flour arabinoxylan, larch arabinogalactan, arabinan (sugar beet), gum arabic power, and starch soluble solutions. The substrate level was always 2% w/v. Homology Modeling. The structure of AbfCelf was built by homology modeling from Swiss Model server with the crystal structure of AbfA from Geobacillus stearothermophilus T-6 (1QW9, http://www. rcsb.org/pdb/explore/explore.do?structureId=1QW9) as template. The structure, especially loops, was optimized for energy minimization by Modeler 9.12. Ramachandran plot (RC plot) analysis was used to validate the structure of AbfCelf by using PROCHECK.28

ides. PNP-α-D-xylopyranoside and PNP-β-D-xylopyranoside were used as substrates with the method similar to AF assay. For nonchromogenic substrates, the hydrolysis products were determined by high-performance anion-exchange chromatography (HPAEC) under the following conditions: system, Agilent 1100 series system (Agilent, CA, U.S.A.); column, Waters Sugar-pak1 column (300 mm × 6.5 mm; Waters, MA, U.S.A.); mobile phase, doubledistilled deionized and degassed water; flow rate, 0.5 mL/min; column temperature, 80 °C. The eluted compounds were detected by refractive index. One unit of enzyme activity was defined as 1 μmol of L-arabinose produced per min. The nonchromogenic substrates 3727

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry

Figure 2. Expression and purification of AbfCelf. (a) Analysis of the protein expression (lanes: M, molecular marker; 1, crude cell lysate after cold osmotic shock of wild BL21; 2, crude cell lysate after cold osmotic shock; 3, soluble fraction of the crude cell lysate after cold osmotic shock; 4, insoluble fraction of the crude cell lysate after cold osmotic shock); (b) recombinant AbfCelf purified by Ni−Sepharose column. Docking. Autodock 4.2 was selected as the docking tool to predict protein binding partners based on initial-stage flexible-body docking algorithm, according to the User’s Guide.29 The structure of AbfCelf was used as macromolecular receptor in the docking simulation. The structure of ligands was generated with Chemdraw software and the energy-minimized with ChemBio3D Ultra 11. The ligands were sugar molecules, including arabinobiose, arabinotriose, arabinotetraose, α-Larabinofuranose, α-D-xylopyranose, β-D-xylopyranose, PNP-α-L-arabinofuranose, PNP-α-D-xylopyranose, and PNP-β-D-xylopyranose. The amino acids of AbfCelf molecule which may form hydrogen bonds with ligands were selected as the flexible amino acid, and the grid box was selected as a certain radius, 40 Å, with the center coordinates of x, 68.676; y, 43.846; z, 75.444. The ligands were docked into AbfCelf with these parameters by using Autodock 4.2 with MGLTools. Binding energy (ΔG, kcal/mol) of Autodock, about ligand−receptor combination, was used as the final criterion to identify the best docking mode.

Cloning, Expression, and AbfCelf Purification. The gene abf Celf was cloned and expressed in pET28a (+). Parts of the recombinant AbfCelf overexpressed in E. coli BL21 (DE3) were soluble and active (Figure 2a). The enzyme purified by Ni−Sepharose column is shown in Figure 2b, which indicated that the molecular weight of native protein after purification was about 66 kDa. The final preparation of purified AbfCelf was 0.19 mg/mL, and the specific activity measured under optimum condition was 19.4 U/mg. Enzymatic Properties. The optimum reaction temperature of purified AbfCelf was at 40 °C (Figure 3a). AbfCelf could maintain most of the initial activity after incubation at 40 °C for 20 h while losing most activity mostly when incubated at 60 °C for 30 min (Figure 3b). AbfCelf reached the highest activity at pH 6.0, 40 °C in 100 mM sodium acetate buffer and achieved about 57.9% and 72.1% of the maximal activity at pH 5.0 and pH 7.0, respectively (Figure 3c). AbfCelf was stable at pH 6.0 compared with no activity remaining after 4 h of incubation at pH 4.0 (Figure 3d). Kinetic parameters were measured at 40 °C, pH 6.0 for 5 min by the Lineweaver−Burk plot, which is shown in Figure S2 in the Supporting Information. The Km, Vmax, and kcat toward PNP-Abf are 1.5 mM, 112.4 U/mg, and 123.6 s−1, respectively. The Km’s of AbfD3 from T. xylanilyticus, ABF2 from P. purpurogenum, and AbfATK4 from G. caldoxylolyticus are 0.7 mM, 98.6 μM, and 0.17 mM, respectively.16,18,19 The Vmax’s of AF from B. pumilus PS213 and AbfD3 are 52.9 and 456 U/mg, respectively.19,30 Compared with these reported GH 51 AFs, AbfCelf had an average kinetic value, indicating the affinity of AbfCelf toward PNP-Abf was in a medium level. Substrate Specificity. The substrate specificity of the enzyme was tested by using several oligosaccharides, polysaccharides, heteropolysaccharide, and PNP-glycosides. Sugar beet arabinan, linear 1,5-α-arabinan, and wheat flour arabinoxylan were degraded by AbfCelf when the reaction time was prolonged to 2 h. In contrast, beechwood xylan, soluble starch, gum arabic, and larch arabinogalactan were resistant to the action of AbfCelf even though the enzyme was added in excess. Table 1 displays the specific activity of AbfCelf on different



RESULTS Sequence Analysis of abf Celf . The gene abf Celf (NC_015514) had an open reading frame of 1521 bp encoding a protein of 507 amino acids. The protein has a calculated molecular mass of 55.6 kDa and a calculated isoelectric point of 4.87. AbfCelf was classified as a member of the GH 51 based on its protein sequence, which was also included in CAZy database (Carbohydrate Active enZYmes database, http://www.cazy. org). Moreover, the amino acid sequence of AbfCelf displayed 20.3%, 45.2%, 44.6%, 45.2%, 29.7%, and 22.7% identity with the sequences of AFs from Ruminococcus albus 8, Bacillus subtilis 168T+, G. caldoxylolyticus TK4, G. stearothermophilus T-6, Thermotoga petrophila RKU-1, and T. xylanilyticus T3, respectively (Figure1). The sequence variation may be associated with loop regions and give rise to the diverse substrate-specificity characteristic. The catalytic domain of AbfCelf comprised two key active sites, both Glu, which were conserved in all members of GH 51 family. On the basis of the results of sequence and structure alignment between AbfCelf and 1QW9, Glu173 is the acid/base, and Glu292 is the nucleophile. 3728

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry Figure 3. continued

6.0, sodium acetate buffer for a period of time. (c) Effect of pH on enzyme activity. Activity was measured at 40 °C for 15 min at the indicated pH values in 100 mM of each buffer. (d) Effect of pH on enzyme stability. The enzyme was incubated at 40 °C in pH 4.0−8.0 100 mM buffers for a period of time. After that, residual activity was assayed at optimum temperature and pH. In all panels, the highest activity was taken as 100%. Reaction substrate was 0.84 mM PNP-Abf.

oligosaccharides and PNP-glycosides. AbfCelf was able to hydrolyze PNP-Abf and arabino-oligosaccharides, whereas no activity was detected toward PNP-α-D-xylopyranoside and PNP-β-D-xylopyranoside. Homology Modeling. The structure of AbfCelf was built using the Swiss Model server with the chain B of 1QW9 as template. The sequence identity between template and target was 48.6%, and the QMEAN Z-score of the structure was −1.893, which was an evaluation parameter in homology modeling.31,32 RC plot analysis showed that out of 502 residues, 84.4% were in most favored region, 14.4% in additionally allowed region, and 0.9% in generously allowed region. Only 0.3% of residues were in disallowed regions, indicating that the conformation was reliable. As shown in Figure 4a, AbfCelf was organized into two domains: a characteristic (β/α)8 barrel (residues 19−382) and a 12-stranded β-sandwich with a jelly roll topology (residues 1− 18 and 383−502). In Figure 4b, the coloring approximates the actual electrostatic potential. The catalytic area was hydrophobic, where binding clefts and protein surfaces conferred the extensive range of specificities. Docking. After building the reliable structure of AbfCelf, substrate analysis can be carried out with the aid of the docking method. For the reason for lacking reliable tools (software), docking an enzyme with its macrosubstrate precisely is almost impossible today. Thus, this study examined only small oligosaccharides and PNP-glycosides in silico. ΔG obtained from Autodock was used as a rapid evaluation of binding free energy.29 Namely, the binding affinity between ligand and receptor can be expressed in terms of ΔG, which is given in Table 1. The negative number of ΔG means the positive binding affinity between the ligand and receptor. The more negative ΔG is, the larger is the binding constant.33 In contrast, when ΔG was a positive number (e.g., arabinotetraose), the ligand could not combine into the receptor, and thus the docking cannot be finished. As Table 1 shows, the ΔG of arabinotetraose was the only one positive number in the total tested substrates. The ΔG’s of other substrates and all products were negative numbers, among which arabinotriose, arabinobiose, and β-D-xylopyranose achieved the most negative numbers.



DISCUSSION The strain C. fimi was sequenced in 2013, and the gene Celf_3321 was marked as “AF domain-containing protein”.34 This work confirmed the gene segment encoding a GH 51 AF. Moreover, the AF from C. fimi was not reported before. Therefore, it is possible that this enzyme would be a novel AF with prospect activity that was worthy to be studied. GH 51 AFs including AbfCelf can remove both α-1,2- and α1,3-arabinofuranosyl moieties from arabinan and arabinoxylans. AbfCelf was also active on α-1,5-linked oligosaccharides, which is a characteristic of GH 43 AFs.13 The substrate specificity of

Figure 3. Biochemical properties of AbfCelf on pH and temperature. (a) Effect of temperature on enzyme activity. Reactions were conducted for 15 min in 100 mM, pH 6.0, sodium acetate buffer. (b) Effect of temperature on enzyme stability. The enzyme was incubated at temperatures ranging from 40 to 60 °C in 100 mM, pH 3729

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry Table 1. Activities of AbfCelf and Molecular ΔG Obtained by Using AutoDock 4.2

enzyme activity of AbfCelf on these pyranosidic synthetic substrates also cannot be detected. In these reactions, product release is the rate-limiting step. AbfCelf displayed specific activity as only 0.05 U/mg toward arabinotetraose, the biggest substrate among the six tested ones, while obtained a positive ΔG of this substrate. In this case, substrate binding is the rate-limiting step, due to the difficulty for the substrate to enter the binding crevice of AbfCelf during docking. Moreover, PNP-α-L-arabinofuranoside bound with AbfCelf easily and product α-L-arabinofuranoside released rapidly according to the ΔG in Table 1. The real AbfCelf activity on PNP-α-L-arabinofuranoside was coincident with this prediction. Additionally, AbfCelf exhibited a higher activity with arabinobiose than arabinotriose, which was in accord with the difference of ΔG between the substrate and product of docking experiment. Thus, ΔG, and the difference of ΔG between the substrate and product, could describe enzymatic reactions and suggest the reaction possibility. The structure of AbfCelf is shown in Figure 4. Its (β/α)8 barrel, which exists commonly in various GH families, participated in substrate binding and catalysis. The circularly permuted form of the catalytic (β/α)8 barrel varied in location, size, and shape. There is correlation of specificity differences with sequence variation in the β → α segments of the characteristic (β/α)8 barrel.36 The β-sandwich of AbfCelf was previously functioned as a domain for binding the carbohydrate molecules, but lost such function during evolution.25 The two catalytic residues (Glu173 and Glu292) were situated at Cterminal extensions of β-strands 4 and 7 in the (β/α)8 fold, which did superimpose perfectly with different GH 51 AFs. Other residues at β → α 4, such as Gln171, Asn172, Trp178, Glu181, and Tyr189, that might contain specificity-denoting characteristics, participated in the interaction with sugar ligands.37

AbfCelf is partly different from some previously wellcharacterized GH 51 AFs such as AbfATK4 from G. caldoxylolyticus TK4.16 The substrates of AFs include natural polymeric polysaccharides and chromogenic PNP-glycosides. The formers are sugar beet arabinan, debranched arabinan, linear 1,5-α-arabinan, arabinogalactan, gum arabic, arabinoxylan (wheat/rye), xylan (larchwood/beechwood/oat spelt), CMC, mannan, glucomannan, and Avicel, etc. Many relative enzymes are necessary for the complete degradation of these natural substrates. On the other hand, different synthetic substrates were diffusely used also, such as PNP-α/β-L-arabinofuranoside, PNP-α/β-L-arabinopyranoside, PNP-α/β-D-xylopyranoside, PNP-α-D-galactopyranoside, PNP-α/β-D-glucopyranoside, PNP-β-D-fucopyranoside, PNP-α-L-rhamnopyranoside, PNP-β-D-glucosaminide, PNP-α-D-mannoside/mannopyranoside, PNP-α/β-D-galactoside, and so on. These PNP-glycosides can be degraded by a single enzyme, and therefore, they are used to detect the activity of such enzyme. Consequently, the substrates cover 30 kinds of various PNP-glycosides and polysaccharides for AFs, and the large number of substrates leads to large amount of experimental determination in the analysis of substrate specificity. Thus, structure understanding and theoretical computation were utilized to analyze the substrate specificity of GH 51 AFs in this paper. Among many enzymatic reactions, product release or substrate binding is the rate-limiting step.35 The more negative ΔG of substrate implied stronger substrate binding affinity in the active center, while the more negative ΔG of product signified the product released more difficultly. PNP-β-Dxylopyranose and PNP-α-D-xylopyranose bound into AbfCelf with ΔG below −6 kcal/mol, while that value of the product was below −7 kcal/mol (Table 1). Apparently, AbfCelf had a higher binding affinity with the product than these substrates, which meant product would occupy the active site and hinder the next binding of a new substrate molecule. In reality, the 3730

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry

Figure 5. Interaction between ligands and key residues in the active site of AbfCelf. Overlapping of AbfCelf and the homologous structure (1QW9-B chain) in (wall-eyed) stereo. Residues are shown in stick model. PNP-Abf of AbfCelf by docking is in green, while that of template is magenta. Potential hydrogen bonds are shown as dotted blue lines; their length in angstroms is indicated. The α-helices and βsheets are shown in red and yellow, respectively. Loops and other secondary structures are shown in blue.

identity between template and AbfCelf made difficulty in completely overlapping of the crystal and docking conjugate. In addition, the differences in side chains might also lead to the environmental diversity between crystallization and docking. AbfCelf shared less hydrogen bonds based on the docking experiment, and such results matched the lower catalytic efficiency and affinity determined experimentally, indicating the conformation of ligand docking into the binding site of AFs can be simulated in a general credibility. Finally, the protein−ligand ΔG could manage to find putative ligands in silico and analyze the substrate specificity to some extent. Because L-arabinose residues are linked to form α-1,5-L-arabinan to which L-arabinofuranose units are attached, the lower protein−ligand ΔG of α-1,5-linked oligosaccharides exhibits potential hydrolytic activity against sugar beet arabinan, debranched arabinan, and linear arabinan.38 Additionally, Larabinosyl residues constitute side chains on the β-(1→4)linked xylose backbones in xylans and arabinoxylans.39 Accordingly, the ΔG of xylopyranoside gave a clue that PNPxylopyranoside, xylan, and substituted xylan could be the putative ligands for AbfCelf so that some bifunctional AFs are discovered. This mainly depends on the credibility of the method. In this study, AbfCelf was characterized and showed potent activity, 19.4 U/mg. Substrates were docked into the binding site to obtain the ΔG, indicating the reaction possibility. Then these prediction results were compared with experimental detected enzyme activity, which gave a clue of feasibility for substrate-specificity analysis in silico. After the present work, there is every possibility for us to screen virtually to select diverse bifunctional AFs with high active catalytic performance, thereby expanding the utilization and application in food industries. Also, before the characterization of an enzyme, the present approach to find putative substrates can save huge amounts of experimental labor, time, and cost. We hope to

Figure 4. Structure of AbfCelf. (a) The structure of AbfCelf created by Swiss Model. The α-helices and β-sheets are shown in red and yellow, respectively. Loops and other secondary structures are shown in green. The catalytic residues, Glu173 and Glu292, are shown in purple balls. (b) The solvent-accessible surface of AbfCelf. Positive and negative residues are shown in blue and red, respectively. The catalytic area is shown in purple.

The interaction between substrate and binding pocket of AbfCelf is shown in Figure 5. Some key residues were responsible for catalysis and substrate binding interactions like Glu27, Arg67, Asn172, Glu173, His242, Tyr244, Glu292, and Gln351, whose location and function are conserved in GH 51 AFs.37 The result was in good agreement with sequence alignment in Figure 1. AbfA crystal 1QW9 was used to test the rationality of docking method in simulating the interaction between enzyme and substrate. The conformation of the ligand with the lowest ΔG was overlapped with that of the 1QW9, which showed the main part of the substrate PNP-Abf in crystal and docking conjugate were partly coincided (Figure 5). Figure 5 also showed the hydrogen-bond network at the active site of AbfCelf. In the structure of covalent glycosyl-enzyme intermediate of AbfA, the arabinofuranose moiety was bound by a large number of hydrogen bonds, at least seven possible hydrogen bonds, which was also shown by Hövel et al.25 In the AbfCelf complex, the whole PNP-Abf can form six hydrogen bonds and therefore was less tightly bonded than AbfA complex. Interestingly, AbfA hydrolyzed substrate with higher catalytic affinity according to its kcat/Km of 134 s−1 mM−1, compared with that of AbfCelf, 82.4 s−1 mM−1. The sequence 3731

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry

(13) Remond, C.; Boukari, I.; Chambat, G.; O’Donohue, M. Action of a GH 51 α-L-arabinofuranosidase on wheat-derived arabinoxylans and arabino-xylooligosaccharides. Carbohyd. Polym. 2008, 72 (3), 424−430. (14) Zhang, G.; Mao, L.; Zhao, Y.; Xue, Y.; Ma, Y. Characterization of a thermostable xylanase from an alkaliphilic Bacillus sp. Biotechnol. Lett. 2010, 32 (12), 1915−1920. (15) Apiraksakorn, J.; Nitisinprasert, S.; Levin, R. E. Grass degrading β-1, 3−1, 4-D-glucanases from Bacillus subtilis GN156: Purification and characterization of glucanase J1 and pJ2 possessing extremely acidic pI. Appl. Biochem. Biotechnol. 2008, 149 (1), 53−66. (16) Canakci, S.; Belduz, A. O.; Saha, B. C.; Yasar, A.; Ayaz, F. A.; Yayli, N. Purification and characterization of a highly thermostable α-Larabinofuranosidase from Geobacillus caldoxylolyticus TK4. Appl. Microbiol. Biotechnol. 2007, 75 (4), 813−820. (17) Lee, S. F.; Forsberg, C. W. Purification and characterization of an α-L-arabinofuranosidase from Clostridium acetobutylicum ATCC 824. Can. J. Microbiol. 1987, 33 (11), 1011−1016. (18) Fritz, M.; Ravanal, M. C.; Braet, C. A family 51 α-Larabinofuranosidase from Penicillium purpurogenum: purification, properties and amino acid sequence. Mycol. Res. 2008, 112 (8), 933−942. (19) Debeche, T.; Cummings, N.; Connerton, I.; Debeire, P.; O’Donohue, M. J. Genetic and biochemical characterization of a highly thermostable α-L-arabinofuranosidase from Thermobacillus xylanilyticus. Appl. Environ. Microbiol. 2000, 66 (4), 1734−1736. (20) Zhang, L. J.; Tang, X.; Cui, D. B.; Yao, Z.; Gao, B.; Jiang, S.; Wei, D. Z. A method to rationally increase protein stability based on the charge−charge interaction, with application to lipase LipK107. Protein Sci. 2014, 23 (1), 110−116. (21) Gonnelli, G.; Rooman, M.; Dehouck, Y. Structure-based mutant stability predictions on proteins of unknown structure. J. Biotechnol. 2012, 161 (3), 287−293. (22) Brinkworth, R. I.; Breinl, R. A.; Kobe, B. Structural basis and prediction of substrate specificity in protein serine/threonine kinases. Proc. Natl. Acad. Sci. U.S.A. 2003, 100 (1), 74−79. (23) De Voss, J. J.; Sibbesen, O.; Zhang, Z.; Ortiz de Montellano, P. R. Substrate docking algorithms and prediction of the substrate specificity of cytochrome P450cam and its L244A mutant. J. Am. Chem. Soc. 1997, 119 (24), 5489−5498. (24) Garay-Malpartida, H. M.; Occhiucci, J. M.; Alves, J.; Belizário, J. E. CaSPredictor: a new computer-based tool for caspase substrate prediction. Bioinformatics 2005, 21 (suppl 1), 169−176. (25) Hövel, K.; Shallom, D.; Niefind, K.; Belakhov, V.; Shoham, G.; Baasov, T.; Schomburg, D. Crystal structure and snapshots along the reaction pathway of a family 51 α-L-arabinofuranosidase. EMBO J. 2003, 22 (19), 4922−4932. (26) Wang, S. H.; Yang, Y.; Zhang, J.; Sun, J.; Matsukawa, S.; Xie, J. L.; Wei, D. Z. Characterization of abnZ2 (yxiA1) and abnZ3 (yxiA3) in Paenibacillus polymyxa, encoding two novel endo-1, 5-α-Larabinanases. Bioresour. Bioprocess. 2014, 1 (1), 1−11. (27) Lineweaver, H.; Burk, D. The determination of enzyme dissociation constants. J. Am. Chem. Soc. 1934, 56 (3), 658−666. (28) Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J. M. PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 1993, 26 (2), 283−291. (29) Morris, G. M.; Huey, R.; Lindstrom, W.; Sanner, M. F.; Belew, R. K.; Goodsell, D. S.; Olson, A. J. AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J. Comput. Chem. 2009, 30, 2785−2791. (30) Degrassi, G.; Vindigni, A.; Venturi, V. A thermostable αarabinofuranosidase from xylanolytic Bacillus pumilus: purification and characterisation. J. Biotechnol. 2003, 101 (1), 69−79. (31) Arnold, K.; Bordoli, L.; Kopp, J.; Schwede, T. The SWISSMODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics 2006, 22 (2), 195−201. (32) Schwede, T.; Kopp, J.; Guex, N.; Peitsch, M. C. SWISSMODEL: an automated protein homology-modeling server. Nucleic Acids Res. 2003, 31 (13), 3381−3385.

develop better catalysts in this way for efficient production of Larabinose and pectin in the future, and AFs will have greater prospects to provide desirable food production properties.



ASSOCIATED CONTENT

S Supporting Information *

Standard curves of PNP and BSA and Lineweaver−Burk plot of AbfCelf. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Fax: +86-21-64252563. E-mail: [email protected]. Funding

This work was supported by the National Special Fund for State Key Laboratory of Bioreactor Engineering (2060204), partially supported by the National Natural Science Foundation of China (no. 31201296) and the Fundamental Research Funds for the Central Universities, People ’s Republic of China. Notes

The authors declare no competing financial interest.



REFERENCES

(1) Saha, B. C. Alpha-L-Arabinofuranosidases: biochemistry, molecular biology and application in biotechnology. Biotechnol. Adv. 2000, 18 (5), 403−423. (2) Matsuo, N.; Kaneko, S.; Kuno, A.; Kobayashi, H.; Kusakabe, I. Purification, characterization and gene cloning of two α-L-arabinofuranosidases from Streptomyces chartreusis GS901. Biochem. J. 2000, 346, 9−15. (3) Bonnin, E.; Dolo, E.; Le Goff, A.; Thibault, J. F. Characterisation of pectin subunits released by an optimized combination of enzymes. Carbohydr. Res. 2002, 337 (18), 1687−1696. (4) Chen, J.; Yang, R. J.; Chen, M.; Wang, S. H.; Li, P.; Xia, Y.; Zhou, L.; Xie, J. L.; Wei, D. Z. Production optimization and expression of pectin releasing enzyme from Aspergillus oryzae PO. Carbohydr. Polym. 2014, 101, 89−95. (5) Wang, S. H.; Yang, Y.; Yang, R. J.; Zhang, J.; Chen, M.; Matsukawa, S. G.; Xie, J. L.; Wei, D. Z. Cloning, characterization of a cold-adapted endo-1,5-α-L-arabinanase from Paenibacillus polymyxa and rational design for acidic applicability. J. Agric. Food Chem. 2014, 62 (33), 8460−8469. (6) Numan, M. T.; Bhosle, N. B. α-L-Arabinofuranosidases: the potential applications in biotechnology. J. Ind. Microbiol. Biotechnol. 2006, 33 (4), 247−260. (7) Gobbetti, M.; De Angelis, M.; Arnaut, P.; Tossut, P.; Corsetti, A.; Lavermicocca, P. Added pentosans in breadmaking: fermentations of derived pentoses by sourdough lactic acid bacteria. Food Microbiol. 1999, 16 (4), 409−418. (8) Voragen, A. G. J.; Rombouts, F. M.; Searle-van Leeuwen, M. F.; Schols, H. A.; Pilnik, W. The degradation of arabinans by endoarabinanase and arabinofuranosidases purified from Aspergillus niger. Food Hydrocolloids 1987, 1 (5), 423−437. (9) Henrissat, B.; Davies, G. J. Glycoside hydrolases and glycosyltransferases. Families, modules, and implications for genomics. Plant Physiol. 2000, 124 (4), 1515−1519. (10) Lombard, V.; Ramulu, H. G.; Drula, E.; Coutinho, P. M.; Henrissat, B. The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids Res. 2014, 42 (D1), D490−D495. (11) Suzuki, H.; Murakami, A.; Yoshida, K. I. Motif-guided identification of a glycoside hydrolase family 1 α-L-arabinofuranosidase in Bif idobacterium adolescentis. Biosci., Biotechnol., Biochem. 2013, 77 (8), 1709−1714. (12) Davies, G.; Henrissat, B. Structures and mechanisms of glycosyl hydrolases. Structure 1995, 3 (9), 853−859. 3732

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733

Article

Journal of Agricultural and Food Chemistry (33) Williams, D.; Stephens, E.; O’Brien, D.; Zhou, M. Understanding noncovalent interactions: ligand binding energy and catalytic efficiency from ligand-induced reductions in motion within receptors and enzymes. Angew. Chem. 2004, 43, 6596−6616. (34) Abt, B.; Foster, B.; Lapidus, A.; Clum, A.; Sun, H.; Pukall, R.; Klenk, H. Complete genome sequence of Cellulomonas f lavigena type strain (134T). Stand. Genomic Sci. 2010, 3 (1), 15. (35) Garcia-Viloca, M.; Gao, J.; Karplus, M.; Truhlar, D. G. How enzymes work: analysis by modern rate theory and computer simulations. Science 2004, 303 (5655), 186−195. (36) Gottschalk, T. E.; Tull, D.; Aghajari, N.; Haser, R.; Svensson, B. Specificity modulation of barley α-amylase through biased random mutagenesis involving a conserved tripeptide in β→α loop 7 of the catalytic (β/α) 8-barrel domain. Biochemistry 2001, 40 (43), 12844− 12854. (37) Sakon, J.; Adney, W. S.; Himmel, M. E.; Thomas, S. R.; Karplus, P. A. Crystal structure of thermostable family 5 endocellulase E1 from Acidothermus cellulolyticus in complex with cellotetraose. Biochemistry 1996, 35 (33), 10648−10660. (38) Beldman, G.; Schols, H. A.; Pitson, S. M.; Searle-van Leeuwen, M. J. F.; Voragen, A. G. J. Arabinans and arabinan degrading enzymes. Adv. Macromol. Carbohydr. Res. 1997, 1 (1), 64. (39) Kormelink, F. J.; Gruppen, H.; Viëtor, R. J.; Voragen, A. G. Mode of action of the xylan-degrading enzymes from Aspergillus awamori on alkali-extractable cereal arabinoxylans. Carbohydr. Res. 1993, 249 (2), 355−367.

3733

DOI: 10.1021/jf5059683 J. Agric. Food Chem. 2015, 63, 3725−3733