The 1.3 Å crystal structure of a novel endo‐β‐1,3‐glucanase of glycoside hydrolase family 16 from alkaliphilic Nocardiopsis sp. strain F96

Abstract

β-1,3-Glucan, a polymer of β-1,3-linked glucose, is the main constitute of botanical, fungal cell walls, and a major structural and storage polysaccharide of marine macroalga. The class of enzymes known as β-1,3-glucanases are well characterized in plants, viruses, and bacteria. Based on the hydrolysis reactions catalyzed by the glucanase, β-1,3-glucanases are classified into exo-β-1,3-glucanases (EC 3.2.1.58) and endo-β-1,3-glucanases (EC 3.2.1.6 and EC 3.2.1.39). The β-1,3-glucanases can play various physiological roles. In plants, β-1,3-glucanases have been implicated in the protection against fungal pathogens through their ability to hydrolyze β-1,3-glucan, a major cell-wall component, and in cell differentiation.1, 2 β-1,3-Glucanase expression in plant seeds plays important roles in the regulation of seed germination and dormancy, and in the defense against seed pathogens.3 In viruses, β-1,3-glucanases are involved in degrading the host cell wall either during virus release and/or are packaged in the virion particle and involved in virus entry.4 In bacteria, a metabolic function has been reported for endo-β-1,3-glucanases.5, 6 Endo-β-1,3-glucanases hydrolyze internal β-1,3-glucosyl linkages, while endo-β-1,3-1,4-glucanases only hydrolyze internal β-1,4-glucosyl linkages when the glucosyl residue itself is linked at the O-3 position. Endo-β-1,3-glucanases are also able to hydrolyze mixed linked β-1,3-1,4-glucans such as lichenan, but prefer β-1,3-glucans such as laminarin.5 Despite these functional differences, bacterial endo-β-1,3-glucanases share sequence similarity with endo-β-1,3-1,4-glucanases5 and belong to glycoside hydrolase family 16 (GHF16).7 Both types of GHF16 enzymes share 15–30% amino acid sequence homology. Microbial and plant β-glucanases belong to different family of glycoside hydrolases without any sequence similarity and unrelated three dimensional structures. The microbial enzymes belong to GHF16 with β-sandwich architecture, whereas the plant enzymes are classified as members of GHF17 and adopt a (β/α)8 TIM-barrel fold. These two β-glucanases which show the same substrate specificity and activity independently despite their structural differences give another example of functional convergent evolution.8 Beside these β-glucanases, there are numerous examples of enzymes that catalyze the same reactions being concocted independently. Probably the best-known example is that of the Ser/His/Asp(Glu) catalytic triad, which is found in at least five different protein folds that cannot easily be considered to be homologous.9, 10 On the basis of the amino acids sequence, endo-β-1,3-glucanase (BglF) from alkaliphilic Nocardiopsis sp. strain F96 exhibits the highest homology to those GHF16 endo-β-1,3-glucanases. In contrast with the other endo-β-1,3-glucanases, this enzyme showed more than eightfold greater activity toward a β-1,3-1,4-glucan rather than β-1,3-glucans with an optimum temperature of 70°C (at pH 6.0).11 These results suggested that BglF would be a novel endo-β-1,3-glucanase. Mutational analysis revealed that Glu123 and Glu128 should be the catalytic residues of BglF.11 While the crystal structures of two wild-type Bacillus endo-β-1,3-1,4-glucanases from B. macerans12 and B. licheniformis13 have been analyzed followed by structural studies on engineered hybrid and circularly permuted variants14-17 and also the crystal structure of a natural circularly permuted endo-β-1,3-1,4-glucanase from Fibrobacter succinogenes18 has been determined, the structure of a bacterial endo-β-1,3-glucanase has not yet been reported. Although the crystal structures of barley endo-β-1,3-glucanase and banana endo-β-1,3-glucanase have been reported, there is neither sequential nor tertiary structural homology between the plant and bacterial enzymes.19, 20 Here, we present the first crystal structure of an endo-β-1,3-glucanase of GHF16 from alkaliphilic Nocardiopsis sp. strain F96 at 1.3 Å resolution and compare it with the other homologous structures to address its substrate preference. The gene cloning, protein expression, purification, crystallization, and diffraction for native BglF have been previously reported.11, 21-23 Selenomethionine (SeMet)-labeled BglF was produced using a methionine auxotroph Escherichia coli strain B834(DE3) as host. The cells were cultured in SeMet core medium (Wako, Japan; contains L-alanine 0.5 g, L-arginine HCl 0.58 g, L-aspartic acid 0.4 g, L-cystine 30 mg, L-glutamic acid 0.67 g, L-glutamine 0.33 g, glycine 0.54 g, L-histidine 60 mg, L-isoleucine 0.23 g, L-leucine 0.23 g, L-lysine HCl 0.42 g, L-phenylalanine 0.13 g, L-proline 0.1 g, L-serine 2.08 g, L-threonine 0.23 g, L-tyrosine 0.17 g, L-valine 0.23 g, adenine 0.5 g, guanosine 0.67 g, thymine 0.17 g, uracil 0.5 g, sodium acetate 1.5 g, sodium succinate 2.06 g, ammonium chloride 0.75 g, and dipotassium hydrogen phosphate 10.5 g for 1 L of culture medium) supplemented with glucose 10 g/L, magnesium sulfate heptahydrate 0.25 g/L, iron (II) sulfate heptahydrate 4.2 mg/L, thiamine 0.5 mg/L, 50 mg/L L-selenomethionine, and ampicillin 50 mg/L. The cells were grown until the OD600 of the culture medium reached 0.7–0.8, then isopropyl-1-thio-β-D-galactopyranoside was added to the final concentration of 0.2 mM and the growth was continued for overnight at 25°C. We performed the purification of the SeMet-labeled BglF using the protocols as established for the native protein.23 Prior to crystallization, the enzyme activity was measured by determination of the reduced sugar amount that were released from the substrate by using the 3,5-dinitrosalicylic acid (DNS) method.24 The reaction mixtures, composed of 5 μL of enzyme solution (∼0.5 mg/mL) and 65 μL of 1.5% laminarin in 100 mM NaH2PO4-NaOH buffer (pH 6.0), were incubated at 37°C for 10 min. The reaction was stopped by the addition of 100 μL of DNS solution (1% DNS, 1% sodium hydroxide, and 0.05% sodium disulfide) and boiling at 100°C for 10 min. After boiling, 30 μL of 40% potassium sodium tartrate was then added to the solution and the mixture was cooled on ice bath until it reached room temperature. Absorbance at 570 nm was measured to estimate the enzyme activity. One unit of enzyme activity was defined as the amount of enzyme required for 1 μmol of reducing sugar released from the substrate per minute at 37°C. Crystals of native and SeMet-labeled BglF were obtained by the hanging-drop vapor-diffusion method at 293 K. Prior to crystallization, the purified protein was concentrated to 5 mg/mL in sample buffer [5 mM Tris-HCl (pH 7.5), 0.05 mM phenylmethylsulphonylfluoride and 2.5 mM dithiothreitol]. SeMet-labeled BglF was crystallized under similar conditions with the native BglF. On a cover slide, 1 μL of protein solution and 1 μL of reservoir [0.1M Tris-HCl (pH 8.0), 1.5M ammonium sulfate, 0.1M sodium chloride, and 1% (v/v) ethanol] were mixed and equilibrated against a 800-μL reservoir. Crystals appeared and reached a final size within a week. The crystals had a stacked plate form with 0.1 mm × 0.1 mm × 0.005 mm dimensions. A crystal was soaked in the cryo-protectant solution, consisting of 0.1M Tris-HCl (pH 8.0), 1.6M ammonium sulfate, 0.175M sodium chloride, 2% (v/v) ethanol, and 20% (v/v) glycerol, and then placed directly into a cold nitrogen-gas stream at 100 K. Single-wavelength anomalous diffraction (SAD) data were collected using synchrotron X-ray radiation source at SPring-8 (Hyogo, Japan) on the beamline BL26B1 using a Jupiter 210 detector. Diffraction data were recorded from the crystal at the peak wavelength of selenium absorption (λ = 0.97864 Å). A total of 360 images were collected in 1° oscillation with a crystal-detector distance of 150 mm. All diffraction data were obtained from a single crystal in the resolution range 40.49–1.60 Å and processed using the HKL2000 program package (DENZO and SCALEPACK25). The space group was determined to be primitive monoclinic P21 with unit cell parameters a = 34.61, b = 71.65, c = 40.48 Å, and β = 90.83°, containing one BglF molecule per asymmetric unit. The structure of BglF was determined by SAD method using the anomalous signals from SeMet-labeled protein. The six selenium sites in methionine residues 5, 108, 127, 157, 210, and 234 were identified and located in the asymmetric unit of the SeMet-labeled BglF crystal using SOLVE package26 at resolution range 2.5–30 Å. Five sites had reasonable occupancies (0.87–1.13) and high Patterson peak (15.9–18.8σ), while one site had low occupancy (0.42) and low Patterson peak height (6.5σ), probably because of a high temperature factor. The initial SAD phases had a mean figure of merit of 0.53 for data up to 2.5 Å resolutions, and of 0.61 after density modification using RESOLVE27 but RESOLVE was only able to build 11 residues without any side chains. By using ANALYZE_SOLVE routine in SOLVE, the phases were extended into 1.6 Å using already known Se sites without any refinement. At this point, the SAD phases had a mean figure of merit of 0.38 and of 0.71 after density modification using RESOLVE. The SAD phases were subjected into automatic model building, including side-chain docking and placement of solvent molecules using ARP/wARP.28 An initial model (227 out of 245 residues) was automatically traced with ARP/wARP. Further manual model building was done with the graphics program XtalView/Xfit29 package to build residues gap between residues 113–120/126–129/229–232. This model was used as a template for a molecular replacement search against the native data using MOLREP30 package. The result from MOLREP was further used as a template for automated model building using automated model building starting from existing model routine provided by ARP/wARP. Further visual inspection of electron density maps, manual rebuilding and refinement were carried out with XtalView/Xfit29 and REFMAC.31 The final native model comprises residues 7–243. The stereochemistry of the final model was verified with PROCHECK.32 Data collection, refinement, and structure quality statistics are shown in Table I. The coordinates have been deposited to the PDB with the accession number 2HYK. We carried out structure analysis using the following computer programs: T-Coffee33 server for multiple sequence alignment; SuperPose,34 a protein superposition server, for superimposition and calculation of root mean square deviations (RMSDs); ESPript35 version 2.2 for preparation of the multiple sequence alignment figure; and Pymol36 for the depiction of structures. The homologues used in the structural comparison were a hybrid Bacillus endo-β-1,3-1,4-glucanase H(A16-M) (B. macerans and B. amyloliquefaciens; PDB ID: 2AYH)14 and the complex structure with β-glucan tetrasaccharide (PDB ID: 1U0A),37 endo-β-1,3-1,4-glucanases from B. macerans (PDB ID: 1MAC)12 and B. licheniformis (PDB ID: 1GBG).13 The final overall structure contains one protein molecule, three glycerol molecules, two ethanol molecules, one calcium ion, and 219 water molecules in an asymmetric unit. The molecule of BglF is a single-domain polypeptide chain which has dimensions of ∼51 Å × 37 Å × 33 Å. It comprises 15 β-strands and three short 310-helices [Fig. 1(A)]. The overall fold of BglF can be defined as a classical sandwich-like β-jelly roll motif in which all the strands are connected by loops and three short 310-helices between β7/β8, β10/β11 and β13/β14. The sandwich is formed by the face-to-face packing of two antiparallel sheets containing seven and eight strands in the order of β2-β5-β14-β7-β8-β9-β10 and β1-β3-β4-β15-β6-β11-β12-β13, respectively. The two sheets enclose an extensive hydrophobic interior that is closed off by the numerous connections between the strands. Both β-sheets are twisted and bent, forming a convex and a concave side of the molecule. The glucan-binding site of BglF is located in a 32 Å long deep channel at the concave side. The deep extended binding cleft can accommodate six β-D-glucopyranosyl units covering subsites −4 to −1 at the nonreducing end and subsites +1 and +2 at the reducing end, with the scissile β-1,3 glycosidic bond positioned between subsites −1 and +1 (based on Davies et al., 1997)38 [Fig. 1(B)]. This arrangement of the subsites was assigned by the structural comparison with the complex structure of H(A16-M). At the bottom of the cleft mainly polar side chains are located, especially acidic amino acid residues. They are thought to interact with polar groups of the polysaccharide through hydrogen bonds formation and to position it for cleavage. Additionally, the upper and lower rims of the cleft are lined with aromatic side-chains containing residues which contribute in substrate binding and positioning of polysaccharide substrate by means of aromatic stacking interactions with the glucopyranose rings as have been observed in many carbohydrate binding proteins. A: Overall structure of BglF viewed along the substrate binding cleft from subsite −4. The protein is depicted as ribbon diagram; the first and second layers of β-sheets and 310 helices are colored green, orange and red, respectively, and labeled according to its appearance from N terminal. The putative catalytic residues and calcium ion are drawn as stick and ball, respectively. Zooming into the calcium binding site of BglF shows the detail of calcium coordination. The amino acid residues, calcium ion and water molecules are drawn as ball and stick. Water molecules and calcium ion are colored red and yellow, respectively. The (2Fo − Fc) Fourier map was calculated using final model phases contoured at 1.5σ above the mean density at 1.3 Å. B: Four superposed structures. BglF is shown in orange. Endo-β-1,3-1,4-glucanases from B. macerans, B. licheniformis, and hybrid Bacillus are shown in dark blue, green, and light blue, respectively. The unique loop regions of BglF are indicated with arrows heads. Loop 36–40 and loop 117–120 which are located near to subsite −2 are indicated by violet and brown arrows, respectively. Loop 146–150 and loop 195–198 are located on subsite +2 and behind subsite +2, respectively, as pointed by red and black arrows. The residues and the distances between two adjacent residues on the rim of the active clefts are shown and labeled with distances given in Å. A lichenan (a β-1,3-1,4-glucan) model was docked to show subsites location. Each glucose residue is labeled based on its subsite location. A calcium ion is bound with nearly perfect pentagonal-bipyramidal geometry on the convex face of the protein, coordinating to three backbone carbonyl oxygen atoms (Glu14, Gly58, and Asp237), a carbonyl side-chain oxygen (Oδ1) atom of Asp237 and three water molecules [Fig. 1(A)]. This calcium ion has been suggested to play a role in stabilizing the protein structure on the other GHF16 glycoside hydrolases.18, 39-41 The carbonyl oxygen atom of Gly58 and a water molecule are bound at the apex position, while the others are arranged in the pentagonal plane. The calcium ion is located on the opposite surface of the active site, with the distance from the catalytic residue Glu123 ∼25 Å. For further structural analysis, BglF structure was compared with the structures of H(A16-M) and its complex with β-glucan tetrasaccharide. The overall comparison shows that they share amino acid sequence identity ∼29%, similar overall topology (average rms difference for the Cα atoms of 170 residues between the two structures is 1.26 and 1.29 Å for the backbone atoms) and similar active site topology [Figs. 1(B), 2 and 3]. The catalytic center involves the putative nucleophile Glu123 and acid/base catalyst Glu128, and both residues are located on β8. Based on affinity labeling experiments on the H(A16-M) which showed the inhibitor, 3,4-epoxybutyl β-D-cellobioside, binds covalently to Glu105,42, 43 the respective residue in BglF, Glu123, is predicted to have similar function as nucleophile. The mutants of the putative catalytic residues, Glu123Gln and Glu128Gln, completely abolished BglF activity.11 The two catalytic residues reach into the groove of loops-like fingers and approach the oxygen atom of the scissile glycosidic bond. The side chains of Ile124, Ile126, and Met127 point into the hydrophobic core of the enzyme. The superimposition of BglF and H(A16-M) shows that three acidic residues of Glu123, Asp125, and Glu128 are well fitted with the respective catalytic residues of H(A16-M), despite of one residue insertion between Asp125 and Glu128 [Fig. 2(B)]. The conformations of the three side chains are fixed by hydrogen bonds. One reaches from Trp118 Nε1 to Glu123 Oε2 (2.8 Å) and one from Glu123 Oε1 to Asp125 Oδ1 (2.6 Å). Additionally, Asp125 has hydrogen bond between Asp125 Oδ2 and His142 Nδ1 (3.2 Å), while the respective residue Asp107 of H(A16-M) makes only one hydrogen bond with Glu105 Oε2 [Fig. 2(D)]. The carboxyl group of Glu128 which might function as the proton donor makes a hydrogen bond to Trp103 Nε1 (2.91 Å) [Fig. 2(D)]. This Trp103 is located on β7, whereas Glu109 of H(A16-M) makes hydrogen bond with Gln119 which is located on β9. Moreover, in H(A16-M), the glucose residue on subsite −1 (glucose −1) forms aromatic stacking interaction with Phe92, while the respective residue in BglF is Trp107 which might promotes stronger interaction. A: Comparison between subsites −4 to −1 of BglF and H(A16-M). Amino acid residues of BglF and H(A16-M) are drawn as sticks, colored in orange and blue, respectively, and labeled. The BglF structure are drawn as ribbon diagram and colored as in Figure 1. β-1,3-1,4 glucan tetrasaccharide from H(A16-M) complex are drawn as sticks, colored in yellow and labeled. Zooming into residue Trp118 on subsite −2 shows this residue forms multiple hydrophobic stacking interactions with Pro119-Pro144-Pro201 (purple colored). B: The detail comparison between subsites −4 to −1 of BglF and H(A16M) by rotating 45° anticlockwise of (A). C: Three glycerol molecules make extensive hydrogen bonds within them and with amino acid residues on subsite −2, −1, and +1. Those glycerol molecules are drawn as stick and labeled. D: The catalytic and auxiliary amino acids in BglF (above) and H(A16-M) (below) show the environment similarity around the catalytic residues. The amino acids residues are drawn as stick and labeled. In this figure water molecules are shown as red balls and hydrogen bonds are shown as dash lines with distances given in Å. Sequence alignment of GHF16 β-glucanases. Secondary structures are labeled according to its appearance from N terminal for BglF (above) and H(A16-M) (below). The alignment was generated using T-Coffee30 and then rendered using ESPript.32 Identical residues are shaded in black, and similar residues are shown in bold. The sequence number is based on BglF. Star marked residues indicate the catalytic pair. By these structural similarities, a reaction mechanism proposes conclusively Glu128 to be involved in the first reaction step as a general acid that protonating the oxygen at the scissile glycosidic bond. This mechanism is supported by the available biochemical data on homologous enzymes which have similar active-site geometry.12, 43, 44 The resulting intermediate is stabilized by the nucleophilic carboxyl group of Glu123 binding to the partially positively charged C1 atom of the cleaved bond in either ionic or covalent way. A water molecule activated by the Glu128 carboxylate attacks the C1 atom completing the reaction cycle under overall retention of configuration of the anomeric carbon and formally re-establishes the protonation state of the active site. This catalytic water molecule near to Glu128 is not observed probably because of the binding of glycerol molecule on subsite −1 (will be discussed in the next section) or it has low occupancy. The function of Asp125 remains unclear. The isosteric replacement of this Asp125 carboxylate with an amide function in Asp125Asn mutant (data not shown) as well as in Asp105Asn mutant of H(A16-M) leaves residual activity whereas the activity of Asp105Lys mutant of the later is no longer measurable.12 This seems to indicate that a conservation of shape is required at residue 105 in H(A16-M) or residue 125 in BglF, but not necessarily of function. Compared with the three GHF16 Bacillus endo-β-1,3-1,4-glucanases, BglF has three unique loops and one unique 310 helix (helix η3). Those loops are loop β2–β3, β7–η1, and β9–β10 which are located near to subsites −2, −2, and +2, respectively. The residues of 36–40 on loop β2–β3 and the residues of 117–120 on loop β7–η1 narrow the width of the active cleft on subsite −2 from 8.51 Å (distance between Phe30 Cε1 to Trp103 Cζ2) to 7.66 Å (distance between Gly38 Cα to Trp118 Cζ3) [Fig. 1(B)]. Similar finding was also observed on subsite +2 where the residues of 146–150 on loop β9–β10 narrows the width of active cleft on subsite +2. Despite the fact that loop β9–β10 has one glycine residue near to the end of the loop (Gly155) and two glycine residues on most front of the loop (Gly148–Gly149) which might introduce flexibility, this loop is rather fixed by hydrophobic interaction between Phe147 and Pro119 and by water mediated hydrogen bond between O atom of Glu150 and Nη2 atom of Arg198. The residues of 36–40 on loop β2–β3, the residues of 117–120 on loop β7–η1 and the residues of 146–150 on loop β9–β10 limit the accessibility of β-1,3-glucans, e.g., laminarin to the BglF active cleft because of its helical conformation45 and favor the binding and catalysis of β-1,3-1,4-glucans, e.g., lichenan which has elongated conformation. On subsite −2 of H(A16-M) complex structure, O2 atom of glucose −2 makes direct hydrogen bond with Oε2 atom of Glu63 and through a bridging water molecule with Oε1 atom of Glu63. The superimposition between BglF and H(A16-M) complex structures reveals that the respective residue in BglF is Arg76 which has longer side chain and too close interaction with glucose −2. Furthermore, Tyr94 of H(A16-M) which forms aromatic stacking interaction with glucose −2 is not conserved and changed to Leu109 in BglF. On subsite −3, the glucose residue in H(A16-M) makes hydrogen bond to Arg65 in which this residue is not conserved in BglF and replaced with Thr78. Alternatively, the residue Asp31 of BglF is predicted to make hydrogen bond with the glucose residue as its side chain position is near to the position of the side chain of Arg65 [Fig. 2(B)]. Trp118 is predicted to promote substrate binding and positioning on subsite −2 by means of aromatic stacking interactions with the glucose residue. This Trp118 has conformation with the aromatic ring plane facing perpendicular along the active site cleft, whereas the corresponding residue in H(A16-M), Trp103, makes angle about 30° with the cleft axis [Fig. 2(A,B)]. The conformation of Trp118 makes this residue able to make more effective aromatic stacking interaction with the glucose residue on this subsite. The conformational change in Trp118 is caused by three residues insertion (Pro119, Asp120, and Ser121) between residues 103 and 104 of H(A16-M). Such insertion forms additional unique loop on the upper side of subsite −2. Furthermore, Trp118 is involved in multiple hydrophobic stacking interactions with Pro119, Pro144, and Pro201, respectively [Fig. 2(A)]. This proline cluster promotes rigidity to Trp118 in particular and subsites −2 in general on increasing temperature. This interpretation was confirmed by the analysis of Trp118Ala mutant. The mutant almost abolished its activity (data not shown), and the fact suggests this residue might contribute to substrate binding. These structural differences between BglF and H(A16-M) suggest the existence of a new recognition scheme at subsite −2 and −3 as follows: To reduce close interaction with guanidium group of Arg76, the glucose −2 might adopt its alternate conformation with its HOCH2CH side-chain facing the exterior of BglF (180° rotation along its major axis). As consequence of such conformation, the glucopyranose ring will be lifted and makes hydrophobic stacking interaction with Trp118. Furthermore, the glucose −3 will also adopt its alternate conformation with HOCH2CH side-chain facing the interior of the binding cleft. The existence of Thr78 with shorter side-chain on this subsite instead of arginine residue as in H(A16-M) support this prediction. Another interesting feature of this BglF structure is the binding of three glycerol molecules on the active cleft. Those three glycerol molecules make extensive hydrogen bonds with each other and with amino acid residues on subsites −2, −1, and +1. Glycerol A which is located on subsite −1 has geometry that mimics glucose residue on that subsite in the complex structure of H(A-16M) with O3C3C2 and O2C2C1O1 resemble O6C6 C5 and C4C5 O5C1, respectively [Fig. 2(C)]. This glycerol A made water molecule-mediated hydrogen bonds between O1glycerol B O2, O2Glu41 Oε1, O2Arg76 Nη1, O2Asn39 Nδ2, O3Asn213 Nδ2 and direct hydrogen bond between O3Trp103 Nε1. The overall interactions between the glycerol molecules and the amino acid residues on subsites −2, −1, and +1 suggest the high affinity of those subsites. Having high affinity on those subsites is probably the way that this enzyme compensates its limited angle and narrow active site while maintaining its substrate specificity. A complete understanding of the substrate recognition of BglF and its substrate specificity requires additional structural and biochemical characterizations. The authors are grateful to the beamline staffs of SPring-8 BL41XU (Nobutaka Shimizu and Masahide Kawamoto) and BL26B1 (Masaki Yamamoto, Go Ueno).