Binding Thermodynamics of Reaction Intermediate Analogues as Potent Inhibitors of GH20 Exo-β-N-Acetylglucosaminidases

Piyanat Meekrathok1, Keith A. Stubbs2 and Wipa Suginta1,3*

1Biochemistry-Electrochemistry Research Unit, Institute of Science, Suranaree University of Technology, Nakhon Ratchasima 30000, Thailand

2School of Molecular Sciences, The University of Western Australia, 35 Stirling Highway, Crawley, WA, Australia, 6009

3Center of Excellence in Advanced Functional Materials, Suranaree University of Technology, Nakhon Ratchasima 30000, Thailand

Correspondence: Wipa Suginta, Biochemistry-Electrochemistry Research Group, Institute of Science, Suranaree University of Technology, Nakhon Ratchasima 30000, Thailand. E-mail: wipa@sut.ac.th

Running title: Potent inhibitors of GH20 -N-acetylglucosaminidases

Keywords: β-N-acetylglucosaminidase; chitin recycling; GH20 glycoside hydrolases; marine Vibrios; reaction intermediate analogues; isothermal titration microcalorimetry

Abstract

Exo--N-acetylglucosaminidases (GlcNAcases) are hydrolytic enzymes involved in the metabolism of chitin in bacteria and in eukaryotic glycosphingolipid metabolism, with genetic defects in human GlcNAcases (HexA and HexB) resulting in Tay-Sachs and Sandhoff diseases, respectively. Here, we determined the effects of three known inhibitors of exo--N-acetylglucosaminidases (PUGNAc, NHAcCAS and NHAcDNJ) on a GH20 exo--N-GlcNAcase (VhGlcNAcase) from the pathogenic bacterium Vibrio harveyi, in dose-response experiments. Molecular interactions between the inhibitors and the enzyme were investigated by isothermal calorimetry (ITC), and were confirmed using molecular docking. VhGlcNAcase was strongly inhibited by these compounds, with PUGNAc having the lowest IC50 value, of 1.2 μM. Molecular docking suggested that the inhibitors mimicked reaction intermediates, with enzyme-inhibitor interactions being similar to those of the enzyme with the substrate diNAG. The equilibrium dissociation constants (Kd) obtained from ITC were 0.19 μM for PUGNAc, 12.9 μM for NHAcCAS and 25.6 μM for NHAcDNJ, confirming that PUGNAc was the most potent inhibitor. The ITC data indicated that the binding of the inhibitors to the enzyme was driven by enthalpy. The negative heat capacity change (ΔCp) of -0.34 ± 0.05 kcal·mol−1·K−1 indicates that hydrophobic interactions make a substantial contribution to the molecular interactions between PUGNAc and the enzyme. Our results suggest that PUGNAc is a highly potent inhibitor, and suggest its usefulness as a scaffold for potential drugs targeting GlcNAcase-related metabolic diseases.

1. Introduction

Bacterial GH20 -N-acetylglucosaminidases (GlcNAcases) are exolytic enzymes that play a crucial role in the biomass recycling of chitin, a homopolysaccharide of β-1,4-linked N-acetylglucosamine (NAG) units [1-3]. Chitin is one of the most abundant biopolymers on earth, and is the primary structural component of fungal cell walls and the exoskeletons of crustaceans and insects. GlcNAcases that act on chitin cleave the glycosidic bond between adjacent GlcNAc units in the chitooligosaccharide chain, resulting in either retention or inversion at the anomeric centre of the resultant hemiacetal, relative to the starting material [4,5]. These enzymes are sub-divided into glycoside hydrolase families 3 (GH3), 20 (GH20) and 84 (GH84), each exhibiting unique substrate recognition. GH3 enzymes use a typical retaining mechanism involving the formation and breakdown of a covalent glycosyl-enzyme intermediate [6,7]. In contrast, GH20 and GH84 enzymes employ a substrate-assisted retaining mechanism [8-10], which involves the carbonyl oxygen of the C2-acetamido group acting as a nucleophile to displace the aglycone leaving group, with the formation of an oxazolinium intermediate.

Bacterial GlcNAcases and human β-hexosaminidases A and B (HexA and HexB) are notable members of GH20. Bacterial GlcNAcases in particular recognize a linear oligosaccharide, removing the N-acetyl-β-D-glucosamine/N-acetyl-β-D-galactosamine units from the non-reducing end of the oligosaccharide chain [11,12]. On the other hand, HexA and HexB hydrolyse β-GalNAc and β-GlcNAc residues from glycosphingolipids, such as the ganglioside GM2 [13]. Mutations in HexA or HexB lead to lysosomal accumulation of gangliosides, resulting in the neurodegenerative disorders known as Tay-Sachs and Sandhoff diseases, respectively [13,14]. For GH84, rigorous analysis has been conducted for human O-GlcNAcase, which removes β-GlcNAc from serine and threonine residues of O-GlcNAcylated proteins, this enzyme being implicated in many diseases including Alzheimer’s disease [15], diabetes [16,17] and cancer [18,19]. As a result of the biological importance of GlcNAcases in the modeling of glycoconjugates in cells, various chemical tools and the design of inhibitory small molecules [20-24] have been considered. 1,2-dideoxy-2ʹ-methyl-α-D-glucopyranoso-[2,1-d]-Δ2ʹ-thiazoline (NAG-thiazoline) [22], which mimics the geometry of the oxazolinium intermediate in the catalytic reaction sequence, has been characterized as a highly potent inhibitor of GlcNAcases from both GH20 and GH84 [9,22,25]. Recently, O-(2-acetamido-2-deoxy-D-glucopyranosylidene)-amino N-phenylcarbamate (PUGNAc) has also been shown to potently inhibit the GlcNAcases from all three families [25-28]. Iminosugars, such as 2-acetamino-1,2-dideoxynojirimycin (NHAcDNJ) [21] and 6-acetamido-6-deoxy-castanospermine (NHAcCAS) [20] have been also found to be the potent inhibitors of GlcNAcases [29]. Work on the use of this type of molecules as drugs targeting GlcNAcase-related metabolic diseases has resulted in NAG-thiazoline being suggested as a treatment for Tay-Sachs and Sandhoff diseases, by acting as a pharmacological chaperone that stabilizes the mutant form of HexA and HexB, resulting in an increase of the amount of HexA and HexB in lysosomes [24,30].

We have previously reported the recombinant expression and characterization of a novel GH20 exo-β-N-acetylglucosaminidase from V. harveyi (VhGlcNAcase) [1,31,32], the primary and opportunistic pathogen that causes luminous Vibriosis, which devastates commercial shrimp- and fish-farms worldwide [33]. The amino acid sequence of VhGlcNAcase is closest to that of the α-chain of HexA, with 30% sequence identity [34], and it is a member of GH20. The enzyme was identified in the periplasm of the bacterium and is thought to play an important role in the degradation of chitooligosaccharides and transport of degradation products through the outer membrane. Preliminary crystallographic data of VhGlcNAcase in the absence and the presence of substrate was previously reported [35] and kinetic modelling of the enzymic reaction suggests that VhGlcNAcase contains four GlcNAc binding subsites [1]. Here, we report that three reaction intermediate analogues act as highly potent inhibitors of VhGlcNAcase. Our study focuses on investigation of the thermodynamic parameters that underlie the molecular interactions between these compounds and VhGlcNAcase. The kinetic and thermodynamic data obtained from our study may help to elucidate the further development of rationally designed drugs based on potent inhibitors of the GH20 glycoside hydrolases.

2. Materials and methods

2.1. Recombinant expression and purification of VhGlcNAcase

Recombinant wild-type VhGlcNAcase was expressed in E. coli M15 (pREP) cells as a C-terminally His6-tagged polypeptide as described by Meekrathok and Suginta [31]. Briefly, cells expressing recombinant VhGlcNAcase were harvested by centrifugation at 4,500 ×g for 30 min and resuspended in lysis buffer containing 20 mM Tris-HCl buffer, pH 8.0, 150 mM NaCl, 1 mM phenylmethylsulphonyl fluoride (PMSF), 5% (v/v) glycerol and 1 mg mL-1 lysozyme, and then lysed on ice using a Sonopuls ultrasonic homogenizer with a 6-mm diameter probe. For purification, the crude enzyme obtained after centrifugation was immediately applied to a gravity-fed TALON® SuperflowTM metal affinity resin column (Clontech Laboratories, Inc., USA) operated at 4 °C, followed by the addition of 20 column volumes of equilibration buffer (20 mM Tris-HCl buffer, pH 8.0 containing 150 mM NaCl), and then 7 column volumes of equilibration buffer containing 5 mM imidazole. The protein was then eluted with 50 mM potassium phosphate buffer, pH 7.0 containing 150 mM imidazole. Fractions with GlcNAcase activity were then further purified on a HiPrep 26/60 Sephacryl S-300 prepacked column connected to an ӒKTAprime purification system (Amersham Bioscience, Piscataway, New Jersey, USA), using the running buffer of 50 mM potassium phosphate, pH 7.0. The eluted fractions were pooled and subjected to several rounds of Vivaspin-20 ultrafiltration membrane concentration (10 kDa molecular-weight cutoff, Vivascience AG, Hannover, Germany). The final protein concentration was determined by measuring absorbance at 280 nm [36].

2.2. GlcNAcase activity assay

The activity of the enzyme was determined in 96-well microtiter plates, using pNP-GlcNAc as substrate. PUGNAc, NHAcCAS, and NHAcDNJ were prepared by published methods [20,37,38]. The assay mixture (200 µL) contained 125 µM pNP-GlcNAc, 0.1 µg of VhGlcNAcase and 0, 1, 10 or 100 μM of the appropriate compound, dissolved in 100 mM phosphate buffer, pH 7.0. The reaction mixtures were incubated at 37 °C for 10 min with constant agitation in an Eppendorf ThermoMixer® Comfort (Eppendorf AG, Hamburg, Germany), and the reaction terminated by the addition of 100 µL of 3 M Na2CO3. The amount of p-nitrophenol (pNP) released was determined spectrophotometrically at 405 nm in a Benchmark PlusTM microplate spectrophotometer (Bio-Rad Laboratories, Hercules, California, USA). The molar concentrations of the liberated pNP were calculated from a calibration curve constructed with pNP standards of 0 to 20 nmol per well. The hydrolytic activity of the enzyme was defined as one unit of enzyme producing 1 nmol of pNP in 1 min at 37 °C.

2.3. Determination of dose-response curves

The half maximal inhibitory concentration (IC50) of each compound against VhGlcNAcase was determined from a dose-response curve using pNP-GlcNAc as substrate. A 200 µL reaction mixture was prepared in a 96-well microtiter plate containing 125 µM pNP-GlcNAc, 0.1 µg of VhGlcNAcase and varying concentrations of the compound (a ten-fold dilution series with a concentration range of 10-10 to 10-2 M) in 100 mM phosphate buffer, pH 7.0. The reaction mixture was incubated at 37 °C for 10 min with constant agitation, and terminated by the addition of 100 µl 3 M Na2CO3. The enzyme activity was estimated from the liberated pNP, quantitated as described previously. The IC50 value was obtained from a plot of the logarithm of inhibitor concentration versus fractional activity of the enzyme, in GraphPad Prism version 6.0. (GraphPad Software, California, USA).

2.4. Isothermal titration microcalorimetry (ITC)

ITC studies of binding of the compounds to VhGlcNAcase were carried out in triplicate, using the MicroCal-PEAQTM ITC system (Malvern Instruments Ltd.) at 25 °C with a stirring speed of 600 rpm. The ligand solution, containing 2.3 μL of 0.15 mM PUGNAc, 0.4 mM NHAcCAS or 0.4 mM NHAcDNJ, was injected into the 300-μl sample cell containing 10 μM purified VhGlcNAcase in 50 mm potassium phosphate buffer, pH 7.0. The control measurement was made by injecting the corresponding ligand into the sample cell containing only buffer. Each injection was repeated 18 times over 120-s intervals and the ITC data were collected and analyzed using the MicroCal-PEAQ ITC software. The ITC profile, obtained by titrating the ligand into VhGlcNAcase, was then subtracted from the profile of ligand to buffer, before the titration curve was fitted by a single-site binding model in the nonlinear least square algorithm. Direct measurements of binding affinity and thermodynamics in a single experiment allowed determination of the equilibrium binding association constant (Ka), reaction stoichiometry (n) and enthalpy change (ΔH°). The Gibb’s free energy change (ΔG°) and the entropy change (ΔS°) were calculated from the relationship ΔG° = ΔH° – TΔS° = -RT ln(Ka) where R is the gas constant (1.98 cal·K−1·mol−1) and T is the absolute temperature in kelvin. Values presented are means ± S.D. obtained from experiments carried out in triplicate.

Temperature-dependence experiments were carried out in triplicate using the MicroCal-PEAQTM ITC system as described above at 15, 20, 25, and 30 °C with a stirring speed of 600 rpm. PUGNAc (2.3 μL, 0.15 mM) was injected into the 300-μl sample cell containing 10 μM purified VhGlcNAcase in 50 mm potassium phosphate buffer, pH 7.0 as described previously. The heat capacity changes (ΔCp) of the binding reactions were obtained by examining the temperature-dependent change in reaction enthalpy: ΔG° = ΔH° – TΔS = -RT ln(Ka) and the ΔCp is equal to ∂ΔH°/∂T, the slope of the linear plot of the enthalpy change (ΔH°) versus temperature (T) [39]. Each parameter of the entropy change (ΔS°) was calculated as follows [39,40]: Mixing entropy change (ΔS°mix) = R ln (1/55.5) = −33 J·mol−1·K−1 = -8 cal·mol−1·K−1 = 0.0078 kcal·mol−1·K−1, which is a statistical correction that reflects mixing of solute and solvent molecules. Solvent entropy change (ΔS°solv) = ΔCp ln (298.15/385.15). Note that the entropy of solvation (ΔS°solv) is close to zero for proteins near 385 K; Cp can be correlated to the solvation entropy change (ΔS°solv) of the binding reaction at T = 25 °C (298 K). Conformational entropy change ΔS°conf = ΔS° − ΔS°mix − ΔS°solv. Finally, the reaction entropy change ΔS° = ΔS°solv + ΔS°mix + ΔS°conf.

2.5. Molecular Docking

All hydrogen atoms were added into the atomic coordinates of the VhGlcNAcase structure in complex with NAG (PDB id 6EZS) using the Protonation and Tautomers function in the GOLD 5.3 program [41]. The other configuration parameters were set up following the default values. The three-dimensional coordinates of the ligand, water and ion molecules were removed from the structure of the enzyme by the Protein Setup wizard in the program. The chemical structures of PUGNAc, NHAcCAS and NHAcDNJ were taken from the crystal structures of GlcNAcase from Paenibacillus sp. TS12 [29] deposited in the Protein Data Bank. Automatic GA parameter settings were used in all GOLD docking calculations, with 100 GA operations per ligand. The binding site was defined so as to include all amino acid residues within a radius of 7 Å from individual atoms of N-acetylglucosamine. The ChemPLP scoring function embedded in the GOLD package was applied in all docking calculations. This scoring function includes the contributions of hydrogen bonds, which are the predominant type found in GlcNAcases.

3. Results

3.1. Evaluation of the inhibitory effects of the compounds on VhGlcNAcase

We first examined the effects of the putative inhibitors on the hydrolytic activity of VhGlcNAcase. The chemical structures of PUGNAc, NHAcCAS and NHAcDNJ, are presented in Fig. 1. The residual activity of VhGlcNAcase was determined after the enzyme was exposed to different concentrations of each compound. As seen in Fig. 2, the activity of VhGlcNAcase, measured using the synthetic substrate pNP-GlcNAc, was suppressed in a concentration-dependent manner. PUGNAc showed the greatest effect, with the activity of the enzyme almost completely abolished at 100 μM. Similar results were obtained with the other two compounds, though the inhibitory effects were less pronounced, with residual activities of 23% and 35% being observed with 100 μM NHAcCAS and NHAcDNJ, respectively (Fig. 2). To quantify the inhibitory effects of these compounds, dose-response experiments were performed. The IC50 value for each compound was determined from the dose-response plot of the fractional residual activity of the enzyme (vo/vi) against the logarithm of the inhibitor concentration. The IC50 values for PUGNAc (Fig. 3A), NHAcCAS (Fig. 3B) and NHAcDNJ (Fig. 3C) were estimated to be 1.2 ± 0.04, 70.7 ± 2.7, 96.2 ± 4.2 μM, respectively, confirming that PUGNAc was the most potent inhibitor of VhGlcNAcase.

3.2. Molecular docking and interaction analysis

Molecular docking was used to examine specific interactions of the natural substrate chitobiose (diNAG) with the inhibitors, using the crystal structure of VhGlcNAcase in complex with NAG (PDB id 6EZS; unpublished data) as a template. Docking diNAG into the active site of VhGlcNAcase gave a Piecewise Linear Potential (PLP) docking score of 71.8 ± 2.1 [41], indicating that the substrate fitted well into the substrate-binding pocket of the enzyme, with the substrate located within the active site (Fig. 4). The pyranose ring of the NAG in the -1 subsite adopts a 4E/boat conformation whereas the NAG in the +1 subsite adopts a 1C4/chair conformation, which is similar to what was observed with diNAG in complex with the chitobiase from Serratia marcescens (SmChb, PDB id: 1QBB) [42]. In Fig. 4 (upper left panel), the polar and charged residues R274, Q398, Y530, D532, E584 and W546 are located within a distance of 3.8-Å. These residues make hydrogen bonds with various hydroxyl groups of NAG in the -1 subsite (assigned as -1NAG). In Fig. 4 (lower panel), five aromatic residues, W487, W505, Y530, W546 and W582, form the hydrophobic wall of the substrate-binding pocket of the enzyme, with W582 stacked against the pyranose ring of -1NAG, whereas the side chain of W546 makes a hydrogen bond with 6-OH of -1NAG and also stacks against the planar face of NAG in the +1 subsite (assigned as +1NAG). All the inhibitors fitted well within subsite -1, with no steric clash against the surface of VhGlcNAcase (Fig. 4, lower panels). PUGNAc showed the highest PLP fit score of 62.7 ± 1.2, followed by NHAcCAS (score = 53.1 ± 2.1), and NHAcDNJ (score = 46.9 ± 0.2). PUGNAc, for which the enzyme had the greatest affinity, based on activity measurements, was found to form a H-bond network with the binding residues, similar to that of diNAG. In addition, three key aromatic residues that bind to diNAG, Y530, W546 and W582, were also found to interact with PUGNAc. On the other hand, fewer interactions were observed with NHAcCAS and NHAcDNJ. Notably, the stacking interaction of W546 and the phenyl carbamate moiety of PUGNAc is missing in the other two inhibitors. The interactions between the substrate/inhibitor and the enzyme are summarized in Table 1.

3.3. Binding thermodynamics of VhGlcNAcase with reaction intermediate analogues

To gain further insight into the modes of binding of the inhibitors with VhGlcNAcase, ITC was performed to determine the equilibrium dissociation binding constant (Kd), stoichiometry (n), enthalpy change (ΔH°), entropy change (ΔS°) and Gibb’s free energy change (ΔG°). The heat release profiles of the wild-type enzyme (Fig. 5, upper panels) were measured during titration with different concentrations of each inhibitor (Fig. 5A for PUGNAc, 5B for NHAcCAS and 5C for NHAcDNJ). The individual troughs of the thermogram were integrated, yielding the enthalpy change as kcal per mol of injectant plotted against the molar ratio, inhibitor:VhGlcNAcase (Fig. 5A-C, lower panels). This analysis gave Kd values for PUGNAc, NHAcCAS and NHAcDNJ of 0.19, 12.9, and 25.6 μM, respectively (Table 2), an order of values that agreed with those of the IC50 values determined using dose-response experiments. Theoretical curve fitting of single-site binding gave a stoichiometry (n) of 0.9 for PUGNAc and 1.0 for NHAcCAS and NHAcDNJ (Table 2), confirming one molecule of inhibitor bound per molecule VhGlcNAcase.

Thermodynamic parameters obtained from the ITC experiments, including the enthalpy change (ΔH°), the entropy change (−TΔS°) and the free energy change (ΔG°), were further analyzed (inset, Fig. 5, lower panels). Plots of the free energy change (ΔG°) as the sum of negative ΔH° and positive −TΔS° terms demonstrate how the enthalpy and the entropy changes contribute to binding of the inhibitors by the enzyme. The results which are similar for all three compounds, indicate that the bindings are exergonic (-ΔG°) and are driven by enthalpy (Table 2), with ΔG° values ranging from -6.3 to -9.2 kcal·mol−1.

3.4. Temperature effects on inhibitor-enzyme binding

Since PUGNAc was the most potent inhibitor of VhGlcNAcase identified in this study, we performed further thermodynamic analysis of this compound. ITC experiments were carried out at different temperatures (Fig. 6A-D, upper panels), allowing heat capacity changes (Cp) to be calculated from the secondary plots of binding isotherms (Fig. 6A-D, lower panels). The equilibrium binding constants (Kd) for PUGNAc at 15, 20, 25, and 30 oC were 0.11, 0.13, 0.19, and 0.13 μM, respectively (Table 3). The theoretical fit with a single-site binding model gave the stoichiometry (n) close to 1.0 over the entire range of the tested temperatures, confirming a one-to-one binding mode. The heat capacity change (Cp) linked to hydration of the system was obtained from the slope of a linear plot of the enthalpy change (ΔH°) versus temperature (Fig. 7A). The Cp of PUGNAc was calculated to be -0.34 ± 0.05 kcal·mol−1·K−1. The reaction entropy change (ΔS°) for the binding of PUGNAc by VhGlcNAcase reveals a large increase in the solvent-entropy change (ΔS°solv) and a large unfavorable conformational entropy change (ΔS°conf) in the enzyme (Fig. 7B). The results indicate that PUGNAc binds to VhGlcNAcase rigidly and the binding results in a large favorable ΔS°solv, compensating a large unfavorable ΔS°conf.

4. Discussion

The exo-β-N-acetylglucosaminidase (VhGlcNAcase) from V. harveyi is an exolytic chitin-degrading enzyme that belongs to family 20 of the glycoside hydrolases [1,31]. VhGlcNAcase plays a crucial role in the recycling of chitin biomass by the marine Vibrio, as it helps to completely degrade chitin degradation products, generated by secreted chitinases [43,44] and transported through outer membrane of the bacterium through chitoporin [45,46], to GlcNAc monomers that are readily metabolized further in the chitin catabolic pathway and finally serve as nitrogen and carbon sources for the cells.

VhGlcNAcase uses a substrate-assisted catalytic mechanism, involving the formation of an oxazolinium intermediate that is then broken down to give a hemiacetal, retaining the anomeric conformation of the substrate. This mechanism requires the amino acids D437 and E438 for catalysis [31]. An understanding of the reaction mechanism and the development of inhibitors of GlcNAcases has led to efforts to develop these types of molecules as drugs targeting GlcNAcase-related metabolic diseases. For example, NAG-thiazoline has been suggested as a treatment for Tay-Sachs and Sandhoff diseases, acting as a pharmacological chaperone that acts to stabilize the mutant form of HexA and HexB, resulting in an increased amount of HexA and HexB in the lysosome [24,30]. The compounds studied here have been shown to lower the enzymatic activity of GlcNAcases in general, as demonstrated previously with GH20 and GH84 GlcNAcases [21,25,29,47], but NHAcDNJ, for example, has also been used therapeutically and reported to increase the impaired GlcNAcase activity in adult Tay-Sachs and infantile Sandhoff cell lines [48]. Since the compounds studied here have been employed as potent inhibitors of GH20 enzymes, in this study we carried out a series of experiments to examine the effects of the three compounds on the activity of VhGlcNAcase. Our results showed that all compounds strongly inhibited VhGlcNAcase activity at concentrations in the low μM range. PUGNAc was found to be the most potent inhibitor of VhGlcNAcase with an IC50 value of 1.2 μM, whereas NHAcCAS and NHAcDNJ showed moderate effects. The inhibitory effect of PUGNAc on VhGlcNAcase was similar to that on human HexA [29], while its C4-epimer, Gal-PUGNAc was found to be more effective than PUGNAc against Paenibacillus sp. TS12 β-Hex1 [29] and human HexB [29,49].

Our molecular docking analysis indicates that the substrate-binding pocket of VhGlcNAcase can accommodate a short chain of chitooligosaccharides, agreeing with the previously reported kinetic data, which showed that VhGlcNAcase preferred a chitooligosaccharide chain of 2-4 GlcNAc units. The active sites of human HexA and HexB also have a small pocket in which the C2-acetamido group of GalNAc-isofagomine nestles between three tryptophan residues, W405, W424 and W489 (equivalent to W487, W505 and W582 in VhGlcNAcase) [50,51]. In the VhGlcNAcase structure modelled with diNAG, five aromatic residues, W487, W505, W546, W582 and Y530, form the wall in the active site pocket, which is stabilized by hydrogen bonding interactions. W582 was found to stack directly with the plane of the pyranose ring of the -1NAG moiety, and these interactions are expected to contribute to binding. All aromatic residues are completely conserved in other GH20 enzymes and play an important role in substrate binding [9,51]. The pyranose ring of PUGNAc appears to be in a 4H3 conformation, similar to that observed for this compound with other GH20 bacterial chitinolytic β-N-acetyl-D-hexosaminidases [52], and thought to be the conformation of the transition state for enzymes that use substrate-assisted catalysis [53]. PUGNAc forms hydrogen bonds with polar and charged residues at both -1 and +1 subsites. The pyranose ring of PUGNAc makes stacking interactions with W582 and the extended phenylcarbamate moiety stacks against W546 in the +1 subsite to facilitate binding (Fig. 4). Additionally, a trigonal sp2-hybridization at the pseudoanomeric position (C1) potentially enhances the binding, through partial mimicry of the transition state of these types of enzymes [53]. This is presumably why PUGNAc shows the strongest binding and is the most potent inhibitor of VhGlcNAcase. On the other hand, the piperidine rings of NHAcCAS and NHAcDNJ can only make stacking interactions with W582 and the amine nitrogen of both analogues does not interact with other residues in the active site.

The enthalpic changes produced by increasing concentrations of the inhibitors were measured using ITC, allowing binding affinities and thermodynamic parameters [54,55] for VhGlcNAcase to be determined. The study revealed that PUGNAc had the highest affinity for VhGlcNAcase, supporting the results of the inhibition experiments. From the ITC experiments, VhGlcNAcase binds to all compounds through enthalpy-driven reactions, indicating that electrostatic interactions dominate the binding. It was also found that the stoichiometry (n) of binding of the inhibitor to VhGlcNAcase was 1:1 for all compounds tested. Large increases in enthalpy with increasing temperature are often compensated by an entropy decrease, resulting in only a small change in free energy, the so called ‘entropy-enthalpy compensation’. This thermodynamic compensation is a natural consequence of finite Cp values and in general can take place due to multiple weak interactions and limited free energy windows [56]. The effect is particularly apparent in aqueous systems, in which non-covalent interactions dominate binding. In a GH19 chitinase, Ohnuma et al. [57] found Cp for binding of (GlcNAc)6 to BcChi-A-E61A to be -0.11 ± 0.008 kcal·mol−1·K−1, suggesting that an aromatic residue (W103) of BcChi-A may contribute to binding through a hydrophobic interaction with (GlcNAc)6. Regarding the binding of PUGNAc to VhGlcNAcase, the large negative value of Cp (-0.34 ± 0.05 kcal·mol−1·K−1) reflects stronger hydrophobic interactions and a larger accessible area of nonpolar surface than in BcChi-A, with W582 and W546 stacking against the aromatic rings of the bound inhibitor. The entropy change (ΔS°) of the reaction with PUGNAc reveals a large favourable increase in the solution (ΔS°solv), which compensates for a large unfavorable conformational entropy change (ΔS°conf) (Fig. 7B). It is likely that a large change in ΔS°solv results from the greater dehydration of the enzyme at the binding interface.

In conclusion, kinetic and thermodynamic characterization of the interaction of VhGlcNAcase with some known inhibitors of exo-β-N-acetylglucosaminidases suggested that PUGNAc is the most potent inhibitor for VhGlcNAcase, and the data are well supported by number of interactions observed by docking simulation. The other compounds tested, NHAcCAS and NHAcDNJ, showed moderate inhibitory effects on the enzyme. The binding affinities obtained from ITC analysis suggested that VhGlcNAcase bound to the compounds with Kd values in the nM - µM range and in a 1:1 ratio. Analysis of the thermodynamic parameters indicates that binding of all compounds to VhGlcNAcase is driven mainly by enthalpy, with the degree of inhibition depending on the functional moiety and the chain length of the compounds.

Conflicts of interest

All authors declare no conflict.

Acknowledgements

This research was supported by Suranaree University of Technology (SUT) and by Office of the Higher Education Commission under the National Research University (NRU) Project of Thailand (FtR.33/2559). WS was funded by the Thailand Research Fund and Suranaree University of Technology through a Basic Research Grant (Grant no. BRG578001) and an SUT grant (SUT1-102-60-24-12). The authors gratefully acknowledge the Biochemistry-Electrochemistry Research Unit and Biochemistry Laboratory, and the Centre for Scientific and Technological Equipment, Suranaree University of Technology for providing the facilities for carrying out this research and thank Dr. Kiattawee Choowongkomon, Department of Biochemistry, Faculty of Science, Kasetsart University, Bangkok, Thailand for use of the software GOLD 5.3. KAS also thanks the Australian Research Council for funding (FT100100291).

Author contributions

WS was the grant holder and supervised the entire project. PM designed and performed all the experiments. WS also provided reagents and materials used in this study. KAS synthesized the inhibitors tested. PM and WS analyzed the data and wrote the paper with support from KAS.

References

[1] W. Suginta, D. Chuenark, M. Mizuhara, T. Fukamizo, Novel β-N-acetylglucosaminidases from Vibrio harveyi 650: cloning, expression, enzymatic properties, and subsite identification, BMC Biochem. 11 (2010) 40.

[2] N.N. Thi, W.A. Offen, F. Shareck, G.J. Davies, N. Doucet, Structure and activity of the Streptomyces coelicolor A3(2) β-N-acetylhexosaminidase provides further insight into GH20 family catalysis and inhibition, Biochemistry 53 (2014) 1789-1800.

[3] J. Zhou, Z. Song, R. Zhang, C. Chen, Q. Wu, J. Li, X. Tang, B. Xu, J. Ding, N. Han, Z. Huang, A Shinella β-N-acetylglucosaminidase of glycoside hydrolase family 20 displays novel biochemical and molecular characteristics, Extremophiles 21 (2017) 699-709.

[4] G. Davies, B. Henrissat, Structures and mechanisms of glycosyl hydrolases, Structure 3 (1995) 853-859.

[5] J. Gebler, N.R. Gilkes, M. Claeyssens, D.B. Wilson, P. Beguin, W.W. Wakarchuk, D.G. Kilburn, R.C. Miller, Jr., R.A. Warren, S.G. Withers, Stereoselective hydrolysis catalyzed by related β-1,4-glucanases and β-1,4-xylanases, J. Biol. Chem. 267 (1992) 12559-12561.

[6] D.J. Vocadlo, C. Mayer, S. He, S.G. Withers, Mechanism of action and identification of Asp242 as the catalytic nucleophile of Vibrio furnisii N-acetyl-β-D-glucosaminidase using 2-acetamido-2-deoxy-5-fluoro-α-L-idopyranosyl fluoride, Biochemistry 39 (2000) 117-126.

[7] D.J. Vocadlo, S.G. Withers, Detailed comparative analysis of the catalytic mechanisms of β-N-acetylglucosaminidases from families 3 and 20 of glycoside hydrolases, Biochemistry 44 (2005) 12809-12818.

[8] S. Drouillard, S. Armand, G.J. Davies, C.E. Vorgias, B. Henrissat, Serratia marcescens chitobiase is a retaining glycosidase utilizing substrate acetamido group participation, Biochem. J. 328 (1997) 945-949.

[9] B.L. Mark, D.J. Vocadlo, S. Knapp, B.L. Triggs-Raine, S.G. Withers, M.N. James, Crystallographic evidence for substrate-assisted catalysis in a bacterial β-hexosaminidase, J. Biol. Chem. 276 (2001) 10330-10337.

[10] F.V. Rao, H.C. Dorfmueller, F. Villa, M. Allwood, I.M. Eggleston, D.M.F. van Aalten, Structural insights into the mechanism and inhibition of eukaryotic O‐GlcNAc hydrolysis, EMBO J. 25 (2006) 1569-1578.

[11] P.W. Robbins, K. Overbye, C. Albright, B. Benfield, J. Pero, Cloning and high-level expression of chitinase-encoding gene of Streptomyces plicatus, Gene 111 (1992) 69-76.

[12] K. Sandhoff, T. Kolter, Processing of sphingolipid activator proteins and the topology of lysosomal digestion, Acta Biochim. Pol. 45 (1998) 373-384.

[13] R.A. Gravel, J.T.R. Clarke, M.M. Kaback, D. Mahuran, K. Sandhoff, K. Suzuki, The GM2 gangliosidoses, in: C.R. Scriver, A.L. Beaudet, W.S. Sly, D. Valle (Eds.), The Metabolic Basis of Inherited Disease, McGraw-Hill, New York, 1995, pp. 1807-1839.

[14] R. Myerowitz, Tay-Sachs disease-causing mutations and neutral polymorphisms in the Hex A gene, Hum. Mutat. 9 (1997) 195-208.

[15] S.A. Yuzwa, D.J. Vocadlo, O-GlcNAc and neurodegeneration: biochemical mechanisms and potential roles in Alzheimer’s disease and beyond, Chem. Soc. Rev. 43 (2014) 6839-6858.

[16] J. Ma, G.W. Hart, Protein O-GlcNAcylation in diabetes and diabetic complications, Expert Rev. Proteomics 10 (2013) 365-380.

[17] K. Vaidyanathan, L. Wells, Multiple tissue-specific roles for the O-GlcNAc post-translational modification in the induction of and complications arising from type II diabetes, J. Biol. Chem. 289 (2014) 34466-34471.

[18] C.M. Ferrer, V.L. Sodi, M.J. Reginato, O-GlcNAcylation in cancer biology: linking metabolism and signaling, J. Mol. Biol. 428 (2016) 3282-3294.

[19] J.P. Singh, K. Zhang, J. Wu, X. Yang, O-GlcNAc signaling in cancer metabolism and epigenetics, Cancer Lett. 356 (2015) 244-250.

[20] R.H. Furneaux, G.J. Gainsford, J.M. Manson, P.C. Tyler, The chemistry of castanospermine, part I: synthetic modifications at C-6, Tetrahedron 50 (1994) 2131-2160.

[21] E. Kappes, G. Legler, Synthesis and inhibitory properties of 2-acetamido-2-deoxynojirimycin (2-acetamido-5-amino-2,5-dideoxy-D-glucopyranose, 1) and 2-acetamido-1,2-dideoxynojirimycin (2-acetamido-1,5-imino-1,2,5-trideoxy-D-glucitol, 2), J. Carbohyd. Chem. 8 (1989) 371-388.

[22] S. Knapp, D. Vocadlo, Z. Gao, B. Kirk, J. Lou, S.G. Withers, NAG-thiazoline, an N-acetyl-β-hexosaminidase inhibitor that implicates acetamido participation, J. Am. Chem. Soc. 118 (1996) 6804-6805.

[23] K.A. Stubbs, N. Zhang, D.J. Vocadlo, A divergent synthesis of 2-acyl derivatives of PUGNAc yields selective inhibitors of O-GlcNAcase, Org. Biomol. Chem. 4 (2006) 839-845.

[24] M.B. Tropak, S.P. Reid, M. Guiral, S.G. Withers, D. Mahuran, Pharmacological enhancement of β-hexosaminidase activity in fibroblasts from adult Tay-Sachs and Sandhoff patients, J. Biol. Chem. 279 (2004) 13478-13487.

[25] M.S. Macauley, G.E. Whitworth, A.W. Debowski, D. Chin, D.J. Vocadlo, O-GlcNAcase uses substrate-assisted catalysis: kinetic analysis and development of highly selective mechanism-inspired inhibitors, J. Biol. Chem. 280 (2005) 25313-25322.

[26] K.A. Stubbs, M. Balcewich, B.L. Mark, D.J. Vocadlo, Small molecule inhibitors of a glycoside hydrolase attenuate inducible AmpC-mediated β-lactam resistance, J. Biol. Chem. 282 (2007) 21382-21391.

[27] R.J. Dennis, E.J. Taylor, M.S. Macauley, K.A. Stubbs, J.P. Turkenburg, S.J. Hart, G.N. Black, D.J. Vocadlo, G.J. Davies, Structure and mechanism of a bacterial β-glucosaminidase having O-GlcNAcase activity, Nat. Struct. Mol. Biol. 13 (2006) 365-371.

[28] D.L. Dong, G.W. Hart, Purification and characterization of an O-GlcNAc selective N-acetyl-β-D-glucosaminidase from rat spleen cytosol, J. Biol. Chem. 269 (1994) 19321-19330.

[29] T. Sumida, K.A. Stubbs, M. Ito, S. Yokoyama, Gaining insight into the inhibition of glycoside hydrolase family 20 exo-β-N-acetylhexosaminidases using a structural approach, Org. Biomol. Chem. 10 (2012) 2607-2612.

[30] K.S. Bateman, M.M. Cherney, D.J. Mahuran, M. Tropak, M.N. James, Crystal structure of β-hexosaminidase B in complex with pyrimethamine, a potential pharmacological chaperone, J. Med. Chem. 54 (2011) 1421-1429.

[31] P. Meekrathok, W. Suginta, Probing the catalytic mechanism of Vibrio harveyi GH20 β-N-acetylglucosaminidase by chemical rescue, PLoS One 11 (2016) e0149228.

[32] P. Sirimontree, T. Fukamizo, W. Suginta, Azide anions inhibit GH-18 endochitinase and GH-20 exo β-N-acetylglucosaminidase from the marine bacterium Vibrio harveyi, J. Biochem. 159 (2016) 191-200.

[33] B. Austin, X.H. Zhang, Vibrio harveyi: a significant pathogen of marine vertebrates and invertebrates, Lett. Appl. Microbiol. 43 (2006) 119-124.

[34] M.J. Lemieux, B.L. Mark, M.M. Cherney, S.G. Withers, D.J. Mahuran, M.N. James, Crystallographic structure of human β-hexosaminidase A: interpretation of Tay-Sachs mutations and loss of GM2 ganglioside hydrolysis, J. Mol. Biol. 359 (2006) 913-929.

[35] P. Meekrathok, M. Burger, A.T. Porfetye, I.R. Vetter, W. Suginta, Expression, purification, crystallization and preliminary crystallographic analysis of a GH20 β-N-acetylglucosaminidase from the marine bacterium Vibrio harveyi, Acta Crystallogr. F 71 (2015) 427-433.

[36] S.C. Gill, P.H. von Hippel, Calculation of protein extinction coefficients from amino acid sequence data, Anal. Biochem. 182 (1989) 319-326.

[37] E.D. Goddard-Borger, K.A. Stubbs, An improved route to PUGNAc and its galacto-configured congener, J. Org. Chem. 75 (2010) 3931-3934.

[38] K.A. Stubbs, J.P. Bacik, G.E. Perley-Robertson, G.E. Whitworth, T.M. Gloster, D.J. Vocadlo, B.L. Mark, The development of selective inhibitors of NagZ: increased susceptibility of Gram-negative bacteria to β-lactams, ChemBioChem 14 (2013) 1973-1981.

[39] H. Zakariassen, M. Sorlie, Heat capacity changes in heme protein-ligand interactions, Thermochim. Acta 464 (2007) 24-28.

[40] K.P. Murphy, D. Xie, K.S. Thompson, L.M. Amzel, E. Freire, Entropy in biological binding processes: estimation of translational entropy loss, Proteins 18 (1994) 63-67.

[41] M.L. Verdonk, J.C. Cole, M.J. Hartshorn, C.W. Murray, R.D. Taylor, Improved protein-ligand docking using GOLD, Proteins 52 (2003) 609-623.

[42] I. Tews, A. Perrakis, A. Oppenheim, Z. Dauter, K.S. Wilson, C.E. Vorgias, Bacterial chitobiase structure provides insight into catalytic mechanism and the basis of Tay-Sachs disease, Nat. Struct. Biol. 3 (1996) 638-648.

[43] W. Suginta, A. Vongsuwan, C. Songsiriritthigul, H. Prinz, P. Estibeiro, R.R. Duncan, J. Svasti, L.A. Fothergill-Gilmore, An endochitinase A from Vibrio carchariae: cloning, expression, mass and sequence analyses, and chitin hydrolysis, Arch. Biochem. Biophys. 424 (2004) 171-180.

[44] W. Suginta, A. Vongsuwan, C. Songsiriritthigul, J. Svasti, H. Prinz, Enzymatic properties of wild-type and active site mutants of chitinase A from Vibrio carchariae, as revealed by HPLC-MS, FEBS J. 272 (2005) 3376-3386.

[45] W. Suginta, W. Chumjan, K.R. Mahendran, P. Janning, A. Schulte, M. Winterhalter, Molecular uptake of chitooligosaccharides through chitoporin from the marine bacterium Vibrio harveyi, PLoS One 8 (2013) e55126.

[46] W. Suginta, W. Chumjan, K.R. Mahendran, A. Schulte, M. Winterhalter, Chitoporin from Vibrio harveyi, a channel with exceptional sugar specificity, J. Biol. Chem. 288 (2013) 11038-11046.

[47] M.S. Macauley, Y. He, T.M. Gloster, K.A. Stubbs, G.J. Davies, D.J. Vocadlo, Inhibition of O-GlcNAcase using a potent and cell-permeable inhibitor does not induce insulin resistance in 3T3-L1 adipocytes, Chem. Biol. 17 (2010) 937-948.

[48] A.J. Steiner, G. Schitter, A.E. Stutz, T.M. Wrodnigg, C.A. Tarling, S.G. Withers, D.J. Mahuran, M.B. Tropak, 2-Acetamino-1,2-dideoxynojirimycin-lysine hybrids as hexosaminidase inhibitors, Tetrahedron Asymmetry 20 (2009) 832-835.

[49] K.A. Stubbs, M.S. Macauley, D.J. Vocadlo, A selective inhibitor Gal-PUGNAc of human lysosomal β-hexosaminidases modulates levels of the ganglioside GM2 in neuroblastoma cells, Angew. Chem. Int. Ed. 48 (2009) 1300-1303.

[50] T. Maier, N. Strater, C.G. Schuette, R. Klingenstein, K. Sandhoff, W. Saenger, The X-ray crystal structure of human β-hexosaminidase B provides new insights into Sandhoff disease, J. Mol. Biol. 328 (2003) 669-681.

[51] B.L. Mark, D.J. Mahuran, M.M. Cherney, D. Zhao, S. Knapp, M.N. James, Crystal structure of human β-hexosaminidase B: understanding the molecular basis of Sandhoff and Tay-Sachs disease, J. Mol. Biol. 327 (2003) 1093-1109.

[52] T. Liu, H. Zhang, F. Liu, L. Chen, X. Shen, Q. Yang, Active-pocket size differentiating insectile from bacterial chitinolytic β-N-acetyl-D-hexosaminidases, Biochem. J. 438 (2011) 467-474.

[53] G.E. Whitworth, M.S. Macauley, K.A. Stubbs, R.J. Dennis, E.J. Taylor, G.J. Davies, I.R. Greig, D.J. Vocadlo, Analysis of PUGNAc and NAG-thiazoline as transition state analogues for human O-GlcNAcase: mechanistic and structural insights into inhibitor selectivity and transition state poise, J. Am. Chem. Soc. 129 (2007) 635-644.

[54] M.L. Doyle, Characterization of binding interactions by isothermal titration calorimetry, Curr. Opin. Biotechnol. 8 (1997) 31-35.

[55] G.A. Holdgate, W.H. Ward, Measurements of binding thermodynamics in drug discovery, Drug Discov. Today 10 (2005) 1543-1550.

[56] A. Cooper, C.M. Johnson, J.H. Lakey, M. Nöllmann, Heat does not come in different colours: entropy-enthalpy compensation, free energy windows, quantum confinement, pressure perturbation calorimetry, solvation and the multiple causes of heat capacity effects in biomolecular interactions, Biophys. Chem. 93 (2001) 215-230.

[57] T. Ohnuma, M. Sorlie, T. Fukuda, N. Kawamoto, T. Taira, T. Fukamizo, Chitin oligosaccharide binding to a family GH19 chitinase from the moss Bryum coronatum, FEBS J. 278 (2011) 3991-4001.

Figure legends

Fig. 1. Chemical structures of PUGNAc, NHAcCAS and NHAcDNJ used in this study and the structure of the presumed intermediate in GlcNAcase substrate-assisted catalytic mechanisms.

Fig. 2. Relative activity of WT VhGlcNAcase against pNP-GlcNAc in the absence and presence of various concentrations of inhibitors, in 100 mM sodium phosphate buffer, pH 7.0. The values shown are mean ± S.D. obtained from experiments carried out in triplicate.

Fig. 3. Dose-response plot of the fractional activity of WT VhGlcNAcase as a function of various concentrations of inhibitors. The value of IC50 for each inhibitor was determined from this graph with the error bars representing the standard deviation (S.D.) from triplicate experiments.

Fig. 4. Analysis of VhGlcNAcase interaction with the natural substrate and inhibitors by the use of molecular docking. Details of the hydrogen-bonding network (upper panels) and hydrophobic interactions (lower panels) show the positions of the regularly surface-exposed residues in the active site of diNAG, PUGNAc, NHAcCAS and NHAcDNJ complexed with VhGlcNAcase.

Fig. 5. Inhibitor binding to WT VhGlcNAcase. Microcalorimetric titrations for the binding of PUGNAc (A), NHAcCAS (B) and NHAcDNJ (C) to VhGlcNAcase and its control traces for ligand to buffer (upper panels) and thermographic binding isotherms with theoretical fits (lower panels) obtained from ITC experiments. Thermodynamic parameter values of analogue binding (inset) were plotted as bar graphs with means ± S.D. obtained from experiments carried out in triplicate.

Fig. 6. Microcalorimetric titrations and control traces for ligand to buffer (upper panels) and thermographic binding isotherms with theoretical fits (one set of sites) (lower panels) for the binding of PUGNAc to VhGlcNAcase, obtained from ITC experiments at 15 °C - 30 °C (A-D).

Fig. 7. ΔCp, ΔSsolv, and ΔSconf. PUGNAc binding to VhGlcNAcase was tested at 15, 20, 25 and 30 °C. (A) The thermodynamic parameters ΔG° (black), ΔH° (blue) and -TΔS° (green): values were obtained as described in Table 3 and plotted against the temperature. (B) The entropy change (ΔS°) of VhGlcNAcase-PUGNAc reaction. ΔS°solv and ΔS°conf were calculated according to the equations quoted in the Materials and Methods.

Table 1 Summary of the interactions of the natural substrate and inhibitors observed in the substrate binding pocket of VhGlcNAcase.

Subsite

Ligand

diNAG

PUGNAc

NHAcCAS

NHAcDNJ

R274, Q398, W487,

R274, Q398, D532,

R274, Q398, Y530,

R274, Q398, D532,

-1

W505, Y530, D532,

W582, E584

W582, E584

W582, E584

W582, E584

+1

W546

W546

W546

Normal-and-underlined: hydrogen bonding residues

Bold: hydrophobic interaction residues

Bold-and-underlined: hydrogen bonding and hydrophobic interaction residues

Table 2 Binding parameters obtained from ITC.

Inhibitor

Isothermal microcalorimetric (ITC) parameter

IC50

Kd

ΔG°

ΔH°

ΔS°

-TΔS°

n

(μM)

(μM)

(kcal/mol)

(kcal/mol)

(kcal/mol/K)

(kcal/mol)

PUGNAc

1.2 ± 0.04

0.19 ± 0.04

-9.18 ± 0.12

-9.44 ± 0.09

-0.001 ± 0.0

0.26 ± 0.03

0.9 ± 0.1

NHAcCAS

70.7 ± 2.7

12.9 ± 0.87

-6.67 ± 0.04

-7.15 ± 0.18

-0.002 ± 0.001

0.48 ± 0.20

1.0 ± 0.0

NHAcDNJ

96.2 ± 4.2

25.6 ± 0.95

-6.27 ± 0.02

-7.82 ± 0.73

-0.005 ± 0.002

1.55 ± 0.72

1.0 ± 0.1

Table 3 Effects of temperature on the binding of PUGNAc to VhGlcNAcase.

Temperature

(°C)

Isothermal microcalorimetric (ITC) parameter

Kd

ΔG°

ΔH°

ΔS°

-TΔS°

n

(μM)

(kcal/mol)

(kcal/mol)

(kcal/mol/K)

(kcal/mol)

15

0.11 ± 0.04

-9.20 ± 0.23

-6.19 ± 0.22

0.012 ± 0.002

-3.01 ± 0.38

0.9 ± 0.1

20

0.13 ± 0.04

-9.23 ± 0.17

-9.71 ± 0.31

-0.002 ± 0.00

0.47 ± 0.18

0.7 ± 0.1

25

0.19 ± 0.02

-9.17 ± 0.07

-9.27 ± 0.25

-0.0003 ± 0.00

0.10 ± 0.21

0.9 ± 0.0

30

0.13 ± 0.02

-9.55 ± 0.09

-12.07 ± 0.38

-0.008 ± 0.001

2.55 ± 0.45

0.9 ± 0.1

Fig. 1.

2

Fig. 2.

Fig. 3.

Fig. 4.

Fig. 5.

Fig. 6.

Fig. 7.