If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To whom correspondence may be addressed: Children's Hospital Oakland Research Inst., 5700 Martin Luther King, Jr. Way, Oakland, CA 94609. Tel.: 510-450-7932; Fax: 510-450-7914;
* This work was supported in part by National Institutes of Health Grant AI44079 (to M. J. J.) and by the Studienstiftung des Deutschen Volkes (to A. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Hyaluronidases are enzymes that degrade hyaluronan, an important component of the extracellular matrix. The mammalian hyaluronidases are considered to be involved in many (patho)physiological processes like fertilization, tumor growth, and metastasis. Bacterial hyaluronidases, also termed hyaluronate lyases, contribute to the spreading of microorganisms in tissues. Such roles for hyaluronidases suggest that inhibitors could be useful pharmacological tools. Potent and selective inhibitors are not known to date, although l-ascorbic acid has been reported to be a weak inhibitor of Streptococcus pneumoniae hyaluronate lyase (SpnHL). The x-ray structure of SpnHL complexed with l-ascorbic acid has been elucidated suggesting that additional hydrophobic interactions might increase inhibitory activity. Here we show that l-ascorbic acid 6-hexadecanoate (Vcpal) is a potent inhibitor of both streptococcal and bovine testicular hyaluronidase (BTH). Vcpal showed strong inhibition of Streptococcus agalactiae hyaluronate lyase with an IC50 of 4 μm and weaker inhibition of SpnHL and BTH with IC50 values of 100 and 56 μm, respectively. To date, Vcpal has proved to be one of the most potent inhibitors of hyaluronidase. We also determined the x-ray structure of the SpnHL-Vcpal complex and confirmed the hypothesis that additional hydrophobic interactions with Phe-343, His-399, and Thr-400 in the active site led to increased inhibition. A homology structural model of BTH was also generated to suggest binding modes of Vcpal to this hyaluronidase. The long alkyl chain seemed to interact with an extended, hydrophobic channel formed by mostly conserved amino acids Ala-84, Leu-91, Tyr-93, Tyr-220, and Leu-344 in BTH.
The glycosaminoglycan hyaluronan (hyaluronic acid; HA),
The abbreviations used are: HA, hyaluronic acid; BTH, bovine testicular hyaluronidase; BSA, bovine serum albumin; BVH, bee venom hyaluronidase; IC50, 50% inhibition of equiactive enzyme concentrations; SagHL, S. agalactiae hyaluronate lyase; SpnHL, S. pneumoniae hyaluronate lyase; Vc, l-ascorbic acid; Vcpal, l-ascorbic acid 6-hexadecanoate.
consisting of repeating disaccharide units of (β,1–4)-d-glucuronic acid (β,1–3)-N-acetyl-d-glucosamine, is a major component of the extracellular matrix of animal tissues. It is present in significant quantities in the skin (dermis and epidermis), brain, and central nervous system (
). It has been identified in essentially all vertebrates and in the capsule of some bacteria. HA is the common preferred substrate of a class of enzymes termed hyaluronidases. The enzymes from bacterial sources are hyaluronate lyases (EC 4.2.2.1) (
). The main degradation products are usually unsaturated disaccharides or tetrasaccharides, which might be utilized by bacteria as an additional carbon source (
). The degradation of HA in the host extracellular matrix decreases the viscoelasticity of the ground substance leading to increased spreading of bacteria and the associated toxins in the tissue (
The hyaluronidases of eukaryotes are present in many tissues and organs, e.g. the liver, kidney, testes, etc. as well as in the venom of bees, wasps, etc. and the salivary glands of leeches. The latter enzymes act as endo-N-acetylglucosaminidases (hyaluronate glycanohydrolases, EC 3.2.1.35 and EC 3.2.1.36) (
). Bovine testicular hyaluronidase (BTH) and the distantly related bee venom hyaluronidase (BVH) are well known representatives of these eukaryotic enzymes, but structural information is only available for the bee venom hyaluronidase (
), and three of them have been expressed in different cell lines, but no larger amounts of the enzymes have been produced for enzymological characterization and structural determination due to difficult isolation and/or purification. Furthermore, in contrast to the bacterial enzymes, the human enzymes, the PH-20 protein, Hyal-1 (
), and Hyal-2, probably exist in several isoforms. They seem to play crucial roles in physiological and pathophysiological processes like fertilization (
). But, due in part to the lack of potent and selective inhibitors, most of these functions are far from clear. Selective hyaluronidase inhibitors are needed to study the relationships between the enzyme activity and the physiological and pathophysiological effects of hyaluronidases. Moreover hyaluronidase inhibitors could be useful as drugs, e.g. in the treatment of arthroses or, combined with antibiotics, in antibacterial therapy of hyaluronate lyase producing bacteria, e.g. Streptococcus pneumoniae, a major human Gram-positive pathogen.
). For bacterial hyaluronate lyases, l-ascorbic acid (Vc) (Fig. 1) has been described as a competitive inhibitor of the S. pneumoniae enzyme (SpnHL) with an IC50 value of ∼6 mm (
). Therefore, no potent and selective hyaluronidase inhibitors are known to date. Recently, however, the structural basis of the SpnHL-Vc interaction has been elucidated by means of x-ray crystallography (
). This structure shows binding of Vc in the active site of the enzyme, and among all the interactions between Vc and the enzyme the importance of hydrophobic interactions with Trp-292 and Tyr-408 is evident. This in turn has suggested that Vc derivatives with increased hydrophobicity might have higher inhibitory activity. To test this hypothesis, l-ascorbic acid 6-hexadecanoate (Vcpal), a highly effective antioxidant (
We therefore investigated the inhibitory effect of Vcpal on the enzymatic activity of two bacterial hyaluronate lyases from S. pneumoniae and S. agalactiae strain 4755 (SagHL) and BTH as the representative of mammalian hyaluronidases. Additionally the structure of SpnHL, co-crystallized with Vcpal, was determined at 1.65-Å resolution providing an opportunity to examine the protein-Vcpal interactions at the molecular level. A homology model of BTH
Coordinates of the BTH homology model are available from the authors upon request.
based on the crystal structure of bee venom hyaluronidase in complex with a HA tetrasaccharide fragment was constructed, and potential modes of binding of Vcpal were predicted by flexible docking. The results should set the stage for the design and development of more potent and selective hyaluronidase inhibitors.
EXPERIMENTAL PROCEDURES
Materials—Hyaluronan (hyaluronic acid) from Streptococcus zooepidemicus was purchased from Aqua Biochem (Dessau, Germany). Bovine serum albumin (BSA) was obtained from Serva (Heidelberg, Germany). The investigated hyaluronidases were enzyme preparations from different sources. Lyophilized hyaluronidase from bovine testis (Neopermease®) (200,000 IU; 4 mg plus 25 mg of gelatin/vial according to the declaration of the supplier) was a gift from Sanabo (Vienna, Austria). Stabilized hyaluronate lyase, i.e. 200,000 IU (0.572 mg from S. agalactiae strain 4755 plus 2.2 mg of BSA and 37 mg of Tris-HCl/vial according to the declaration of the supplier) of lyophilized hyaluronate lyase, was kindly provided by id-Pharma (Jena, Germany). S. pneumoniae hyaluronate lyase (
). The enzyme was concentrated to 5 mg/ml in 10 mm Tris-HCl buffer, 150 mm NaCl, 1 mm dithiothreitol, pH 7.4 using centrifugal spin devices with 50-kDa molecular mass cut-off (Millipore) and used for the inhibition studies and for the production of crystals with the inhibitor. The enzyme concentration was determined photometrically at 280 nm using a calculated molar absorption coefficient of 127,000 liters·mol-1·cm-1 (
). l-Ascorbic acid 6-palmitate was purchased from Sigma. All other chemicals were of analytical grade and were obtained either from Merck, Fisher Scientific, or Sigma.
Determination of Inhibitory Activity on Hyaluronidases—The inhibitory effect of Vc and Vcpal on the hyaluronate lyases from S. pneumoniae, S. agalactiae strain 4755, and hyaluronidase from bovine testis was measured using a turbidimetric assay (
). Enzyme activity was quantified by determining the turbidity caused by the residual high molecular mass substrate (molecular mass > 6–8 kDa) precipitated with cetyltrimethylammonium bromide. The incubation mixture contained 120 μl of citrate-phosphate buffer (solution A: 0.1 m Na2HPO4, 0.1 m NaCl, solution B: 0.1 m citric acid, 0.1 m NaCl; solution A and B were mixed in the appropriate proportions to reach pH 5.0), 30 μl of BSA solution (0.2 mg/ml in water), 30 μl of HA substrate solution (2 mg/ml in water), 50 μl of H2O, 10 μl of Me2SO, and 30 μl of enzyme solution. To account for the different enzymatic mechanisms of bacterial and bovine hyaluronidases the assays were normalized by using equiactive concentrations of 54 ng of bovine testicular hyaluronidase or 2.9 ng of hyaluronate lyase from S. agalactiae strain 4755 or 1.7 ng of hyaluronate lyase from S. pneumoniae in 30 μl of BSA solution. To determine the enzyme activities in the presence of the test compounds, instead of 10 μl of Me2SO 10 μl of varying concentrations of inhibitor were used resulting in a final concentration range from 1.1 to 300 μm for Vcpal and from 1 to 20 mm for Vc, respectively. The final Me2SO concentration was 3.8% (v/v).
After incubation of the assay mixture for 30 min at 37 °C, 720 μl of a 2.5% (w/v) cetyltrimethylammonium bromide solution (2.5 g of cetyltrimethylammonium bromide dissolved in 100 ml of 0.5 m sodium hydroxide solution, pH 12.5) were added to precipitate the residual high molecular weight substrate and to stop the enzyme reaction. This mixture was again incubated at 25 °C for 20 min, and the turbidity of each sample was determined at 600 nm with a Uvikon 930 UV spectrophotometer (Kontron, Eching, Germany). Experiments were performed in quadruplicate. The turbidity of the sample without inhibitor was taken as reference for 100% enzyme activity, whereas the turbidity of the sample without enzyme (30 μl of BSA solution were used instead) was taken as reference for 0% enzyme activity. The activities were plotted against the logarithm of the inhibitor concentration, and IC50 values were calculated by curve fitting of the experimental data with Sigma Plot 8.0 (SPSS Inc., Chicago, IL).
Crystallization of the Complex—For crystallization experiments, the Vcpal was dissolved in 5% Me2SO (v/v), 10 mm Tris-HCl buffer, 150 mm NaCl, 1 mm dithiothreitol, pH 7.4 to the final concentration of 50 mm. The hanging drop vapor diffusion (
) using Linbro culture plates (Hampton Research Inc.) at room temperature was used to grow the crystals of the enzyme-inhibitor complex. Equal volumes of protein and reservoir solution (1 μl each) and various amounts of inhibitor solutions (0.1, 0.5, and 1.0 μl) were mixed and equilibrated against 1 ml of the reservoir solution. The reservoir solution contained 60–65% saturated ammonium sulfate, 0.2 m NaCl, 2% dioxane, and 0.1 m sodium citrate buffer (pH 6.0) as reported for the native enzyme crystallization (
X-ray Diffraction—The crystals of the enzyme-inhibitor complex were cryoprotected using 30% xylitol (w/v), 3.5 m ammonium sulfate, and 0.1 m sodium citrate buffer (pH 6.0) as reported for the native crystals (
) and frozen in liquid nitrogen. Standard fiber loops (Hampton Research Inc.) of a suitable size were used to pick up and mount the frozen crystals under a nitrogen flow at -180 °C. The x-ray diffraction data for the enzyme-inhibitor complex was collected using rotation (oscillation) photography and a Quantum 4u CCD detector. The crystallographic setup of beamline 5.0.1 of the Berkeley Center for Structural Biology, Advanced Light Source, Lawrence Berkeley National Laboratory was used. The collected data were analyzed, indexed, integrated, and scaled using the HKL2000 software package (
), calculated from 5% of reflections set aside at the outset, was used to monitor the progress of refinement. Water molecules were placed into 3 σ positive peaks in maps when density was also evident in maps and suitable hydrogen bonding partners were available. The ligand Vcpal was modeled into the catalytic site according to the strong, immediately evident difference density. Sulfate and xylitol molecules, deriving from the crystallization and cryocooling solutions, respectively, were modeled into suitably shaped regions of electron density. Final statistics for the model are shown in Table II. Programs of the CCP4 package (
Construction of BTH Model—All calculations were done on a Silicon Graphics OCTANE R10000 work station. For the homology modeling of BTH the crystal structures of bee venom hyaluronidase were used as templates (Protein Data Bank codes 1fcq, 1fcu, and 1fcv). Since the sequence identity between BTH and bee venom hyaluronidase is only 32%, a multiple sequence alignment including the human PH-20 protein was performed to get more reliable results. The amino acid sequence of BTH (
) was aligned with the sequences of bee venom hyaluronidase extracted from the crystal structures and of the human PH-20 protein (Swiss-Prot accession number p38567) using the program ClustalW (
) and the SYBYL module PROTABLE. Then the model was protonated using SYBYL 6.8 (Tripos Inc., St. Louis, MO) and energetically minimized with the AMBER all-atom force field (
) with AMBER all-atom charges (distance-dependent dielectricity constant, 4) up to a root mean square gradient of 0.1 kcal/mol/Å (Powell conjugate gradient). Surfaces and lipophilic potentials of the model were calculated and visualized by the program MOLCAD contained within SYBYL 6.8.
Flexible Docking of l-Ascorbic Acid 6-Hexadecanoate Using FlexX— The three-dimensional structure of Vcpal was generated with the SYBYL SKETCH module and energetically minimized using Tripos Force Field with Gasteiger-Hueckel charges (distance-dependent dielectric constant, 1) up to a root mean square gradient of 0.05 kcal/mol/Å (Powell conjugate gradient). The binding pocket of BTH was defined by the amino acids within a sphere of 4 Å around Tyr-220 and Trp-341. All docking calculations were performed with FlexX, Version 1.12 (
). Previous versions of FlexX tended to overestimate affinity of hydrogen-bonded ligands and performed poorly with relatively lipophilic binding sites (
). Due to the amphiphilic substrate hyaluronan the active site of BTH consists of many polar, mostly ionic amino acids but also contains some hydrophobic residues. Therefore we used the recently developed scoring function ScreenScore that incorporates hydrophobic (
) and ScreenScore as a fitness and scoring function for base placement and flexible incremental docking, standard parameters were applied. The results were analyzed and visualized with FlexV, Version 1.6, and SYBYL 6.8.
RESULTS AND DISCUSSION
Comparison of Inhibitory Activities of l-Ascorbic Acid and l-Ascorbic Acid 6-Hexadecanoate on Different Hyaluronidases—Recently the inhibitory effects of Vc on SpnHL and BTH have been investigated. In contrast to BTH, Vc inhibits the SpnHL with an IC50 value of around 6 mm under the experimental conditions used (
). In our normalized assays with equiactive concentrations the bacterial enzymes were weakly inhibited by Vc with IC50 values of 6 mm for SagHL and 32 mm for SpnHL, respectively, whereas the bovine enzyme was not affected up to 100 mm. Therefore, Vc proved to be a weak but selective inhibitor of streptococcal hyaluronate lyases. Under physiological conditions Vc is suggested to trigger a natural, low level defense mechanism in the human organism against pneumococcal invasion. That Vc weakly inhibits bacterial hyaluronate lyase (
) indicates for the first time that such effects are possibly at least in part related to this enzyme, which is also described as a “spreading factor” facilitating bacterial spread in the host tissues.
The crystal structure of the S. pneumoniae hyaluronate lyase-Vc complex has been determined (
) and suggests which variations of this compound might lead to stronger inhibitors of the enzyme. One possibility is the introduction of lipophilic chains in the 6-position of Vc corresponding to the presence and topology of hydrophobic residues in the catalytic cleft (
). Hydrophobic interactions also play an important role for the substrate HA since it has been shown to form a 2-fold helical structure with an extensive hydrophobic patch of about eight CH units in aqueous solutions. Hyaluronan combines hydrophilic regions with hydrophobic patches, both characteristic of lipids (
). As a consequence of these suggestions, the effects of the more hydrophobic compound Vcpal and of hexadecanoic acid on the bacterial hyaluronate lyases from Streptococcus species and on the bovine testicular hyaluronidase were determined.
In agreement with the inhibitory effects that have been observed for various fatty acids on S. dysgalactiae hyaluronidase (
), hexadecanoic acid showed an inhibition of the activity of SagHL by 80% at maximal test concentration of 295 μm (Table I), whereas Vcpal strongly inhibited SagHL with an IC50 of 4 μm (Fig. 2). Equiactive concentrations of SpnHL and bovine testicular hyaluronidase revealed IC50 values of 100 and 56 μm, respectively (Table I). Thus, Vcpal is up to 1500 times more active than Vc (Table I) and proved to be among the most potent inhibitors of bacterial and bovine hyaluronidases described to date.
Table IInhibition of hyaluronidase activity IC50 values of SpnHL, SagHL, and BTH for Vc, Vcpal, and hexadecanoic acid determined as described under “Experimental procedures” are reported.
Fig. 2Inhibitory activities of increasing concentrations of Vcpal on SpnHL (▴), SagHL (•), and BTH (▪). The final enzyme concentrations were 6.22 ng/ml, 10.6 ng/ml, and 0.2 μg/ml for SpnHL, SagHL, and BTH, respectively. Each point represents the mean value of four independent measurements ± S.E. as described under “Experimental Procedures.” c, concentration.
Vcpal exhibited about 25-fold higher inhibitory activity on SagHL compared with SpnHL under our experimental conditions, but the factor for Vc was only 6-fold. In contrast to Vc, overall selectivity of Vcpal for the bacterial enzymes versus BTH could not be observed. The markedly stronger inhibition of all investigated enzymes by Vcpal compared with Vc clearly shows that the long alkyl chain increases the affinity of the molecule by additional hydrophobic interactions with the enzyme.
The Binding Mode of l-Ascorbic Acid Hexadecanoate within the Active Site of S. pneumoniae Hyaluronate Lyase—X-ray crystallography of the crystallized SpnHL-Vcpal complex was used for determination of the binding mode. The SpnHL part of the complex structure contains an N-terminal α-domain and a C-terminal β-domain connected by a 10-residue linker and has only slight changes when compared with the native SpnHL (
) crystal structures. The active site of the enzyme is situated in the predominant cleft between the α- and the β-domains. The catalytic mechanism as proposed (
). Catalysis occurs at the narrowest part of the catalytic cleft and involves residues Asn-349, His-399, and Tyr-408. In addition, a hydrophobic patch consisting of Trp-291, Trp-292, and Phe-343 contributes to precise positioning of the substrate.
Measured electron density at the catalytic site allowed for the satisfactory localization and the reliable placement of Vcpal in the active site of the enzyme where the compound was found to bind as shown in Fig. 3A. In the final x-ray structure, additional difference electron density was evident at the catalytic site, suggestive of the existence of alternative modes of inhibitor binding. However, despite repeated attempts, only the conformation shown was well supported by the density. The likely existence of other binding modes was also apparent in the somewhat elevated B-factors for the inhibitor in the final structure (Table II). Surprisingly the initial assumption that the Vc portion would have an intact ring proved wrong. No support for a lactone ring was evident from electron density, although the Vc moiety of the inhibitor binds to the same region as intact Vc (
). Instead it seems as if the Vc portion of the inhibitor binds in a ring-opened form (Fig. 3A). It is known that oxidation of Vc gives l-dehydroascorbic acid, which is rapidly degraded under physiological conditions (pH 7.4) via l-diketogulonate to yield l-erythrulose and oxalate (
). Possibly Vcpal is modified in a similar manner, but only the hydrolyzed lactone moiety was modeled into the electron density (Fig. 3A). Presumably ring opening occurs only in the course of co-crystallization due to the long time needed for this process but not during the enzymatic assay.
Fig. 3A, the SpnHL binding site for l-ascorbic acid 6-hexadecanoate. Hydrogen bonds are represented as dotted lines. Shown is the final sigmaA-weighted omit map electron density calculated to a resolution of 1.65 Å when the inhibitor was excluded from the model and contoured at 2.0 σ. B, schematic diagram of interactions of l-ascorbic acid 6-hexadecanoate with SpnHL. Hydrogen bonds are represented as dotted lines. Other residues shown form hydrophobic interactions with the inhibitor. The figure was made with LIGPLOT (
). C, binding mode of hexasaccharide substrate. The coordinates of the hexasaccharide substrate originate from its complex with SpnHL (Protein Data Bank code 1loh (
)). D, binding mode of l-ascorbic acid 6-hexadecanoate. The cryoprotectant xylitol molecule close to the inhibitor (see text) is also shown (bottom right corner).
As shown in Fig. 3 the l-ascorbic acid part of Vcpal binds within the catalytic site explaining the inhibitory activity. Although the presumed existence of ligand disorder necessitated a conservative approach to water-fitting, water molecules are present in the catalytic site, some mediating protein-ligand interactions. For example, water molecules numbered 162 and 328 bridge the interaction between inhibitor O-2 and protein residues Asn-349 and Asp-352. Fig. 3B also illustrates that the majority of protein-inhibitor interactions are hydrophobic, the only exceptions being hydrogen bonds between the hydroxy groups at C-4 and C-5 and Asn-290 and others from carboxylate to Tyr-408 and Arg-462. The hydrophobic face of the Vc portion lies flat on the side chain of Trp-292. Such an arrangement is commonly observed in complexes of carbohydrate-binding proteins with their ligands (
). The palmitoyl moiety fits in a mainly hydrophobic surface crevice. The aliphatic chain is well defined by density with the exception of the three terminal carbon atoms for which density did not permit modeling. The palmitoyl group forms hydrophobic interactions with Trp-291 and Phe-343, both contributing to the hydrophobic patch along with His-399 and Thr-400. None of the interactions is sufficiently close and geometrically suitable to be considered as a weak hydrogen bond involving methylene carbons of the palmitoyl group. Fig. 3, C and D, shows a comparison of the binding mode of Vcpal with that of a substrate-based hexasaccharide (Protein Data Bank code 1loh (
)). The binding sites overlap, but the surface cleft that accommodates the palmitoyl group is not directly involved in substrate binding. Interestingly one of the three well ordered cryoprotectant xylitol molecules bound to SpnHL fits close to the end of the palmitoyl moiety. The distance between the penultimate resolved carbon of the palmitoyl group and atom O-5 of xylitol is only 4.0 Å (Fig. 3D). This position of the bound xylitol suggests that the affinity of the present inhibitor might be enhanced through addition of matching groups at the aliphatic end of the palmitoyl moiety.
Homology Model of Bovine Testicular Hyaluronidase—Due to the low but significant amino acid sequence identity between insect and mammalian hyaluronidases the crystal structure of BVH (
) has provided the first possibility to construct reliable comparative models of mammalian hyaluronidases like BTH, human PH-20 protein, and Hyal 1–4. To predict the potential binding mode of Vcpal at the active site of BTH, a homology model based on crystal structures of bee venom hyaluronidase (
) was constructed. In the case of less than 40% identity the correct sequence alignment is the most important factor affecting the quality of the resulting models (
). Given that the sequence identity between BTH and bee venom hyaluronidase is just 32% an improvement could be obtained by constructing a multiple, instead of a pairwise, alignment of BTH, human PH-20 protein, and the BVH (Fig. 4) using ClustalW (
), by the SYBYL module PROTABLE, and by comparison of the Ramachandran plot with that of the template (Protein Data Bank code 1fcq) and was completed by addition of hydrogen atoms and by energy minimization.
Fig. 4Sequence alignment of the bovine testicular hyaluronidase, human PH-20 (hPH) protein, and bee venom hyaluronidase as the basis of homology modeling. α-Helices and β-strands are differently shaded. Additionally amino acids residues within the active site of BVH are highlighted in bold type, and sequence conserved residues are marked with asterisks.
Due to the homology modeling approach, the BTH model and the crystal structure of bee venom hyaluronidase closely superimpose with a root mean square deviation of 1.5 Å. Similar to BVH, the model has an open (β/α)7 barrel structure with a large groove perpendicular to the barrel axis. The groove is mainly shaped by loops connecting the β-strands. The putative HA binding site is rather open and formed by several residues of the groove that are mostly conserved between mammalian and bee hyaluronidase (
). As observed for the BVH crystal structure the HA binding site of BTH is dominated by several hydrophobic amino acids. Fig. 5A illustrates the proposed binding mode of a HA disaccharide within the active site of the BTH model. By convention, the sugar residue subsites are labeled from -n to +n with -n at the non-reducing end and +n at the reducing end of the substrate. Cleavage occurs between the -1 and +1 subsites (
). The subsite -1 for the N-acetylglucosamine residue forms a small pocket defined by the tyrosines Tyr-220, Tyr-265, and Tyr-305 (not shown) as well as the tryptophan Trp-341 (Tyr-184, Tyr-227, Tyr-268, and Trp-301 in BVH). The neighboring subsite -2 for glucuronic acid is comprised of Tyr-93 at the bottom as well as by Trp-341 and Leu-344 on the two sides (Tyr-55, Trp-301, and Ser-304 in BVH). The backbone NH group of Leu-344 and the hydroxy group of Ser-343 (Ser-303 in BVH) probably interact with the carboxylic group of d-glucuronic acid. This structural analysis based on the sequence alignment indicates that the residues forming the active site of hyaluronidases are highly conserved (Fig. 4). Thus, appropriate homology models will be most reliable in regions where competitive inhibitors bind.
Fig. 5A, proposed binding mode of HA disaccharide within the active site of the BTH model. Subsite -1 for N-acetylglucosamine (GlcNAc) and subsite -2 for d-glucuronic acid (GlcA) are shown (see text for details). The HA disaccharide is modeled into the binding site of BTH model according to its binding mode found in the crystal structure of bee venom hyaluronidase (Protein Data Bank code 1fcv (
)). Shown is the proposed binding mode of Vcpal within the active site of the BTH model with only Asp-147 protonated (B) and within the same active site with Asp-147 and Glu-149 protonated (C). Additionally the HA fragment (tetrasaccharide) is shown (its carbon atoms are colored in yellow) as modeled into the binding site of the BTH model according to its binding mode found in the crystal structure of bee venom hyaluronidase (Protein Data Bank code 1fcv (
)). The protein surface is represented by a Connolly surface colored by lipophilicity (the warmer the color, the more lipophilic the amino acids). Both catalytically active residues are shown with their protonation state used for the different calculations. For demonstration purposes the Connolly surface of both catalytically active residues was omitted.
) indicated two catalytic acidic amino acids corresponding to Asp-111 and Glu-113 for BVH (Asp-147 and Glu-149 for BTH) that were indeed shown to be involved in catalysis (
)). Due to the high amino acid homology in the HA binding region, the topology of all amino acids involved in catalysis is probably very similar to that observed for BVH. Therefore, the proposed mechanism of HA degradation is likely to be a double displacement substrate-assisted mechanism as suggested for BVH (
Potential Binding Mode of l-Ascorbic Acid 6-Hexadecanoate to Bovine Testicular Hyaluronidase—The homology model of BTH was used for flexible ligand docking to predict the binding mode of Vcpal. To locate the binding pocket, the BTH model was superimposed with the crystal structure of BVH in complex with a HA tetrasaccharide fragment. The active site of BTH was defined by the set of amino acids lying within a sphere of 4 Å around residues Tyr-220 and Trp-341 (BVH Tyr-184 and Trp-301), both close to the reducing end of the tetrasaccharide, and the amino acids Asp-147 and Glu-149 (BVH Asp-111 and Glu-113) involved in catalysis. In the crystal structure of BVH the aspartate and glutamate side chains are in close proximity, independent of whether an acidic or a neutral pH was used for crystallization (
) Glu-113 in BVH (equivalent to Glu-149 in BTH) was proposed to act as the proton donor, while the nucleophile is the acetamido oxygen of the HA substrate probably forming a covalent oxazolinium intermediate. In the next step this intermediate is hydrolyzed by a water molecule resulting in the observed retention of the configuration at the anomeric carbon atom (
), the protonation of Glu-113 in BVH and Glu-149 in BTH, respectively, should be simultaneous with substrate binding and the displacement of water molecules from the active site. However, it cannot be ruled out that Glu-149 remains unprotonated in the inhibitor-bound form. Therefore, the δ-oxygen of Asp-147 was protonated in all flexible docking calculations, whereas Glu-149 was treated in both forms.
Each of the FlexX calculations was analyzed by visual inspection of the docked conformations (309 target-bound conformations for both residues protonated and 237 for only Asp-147 protonated). Each docking series resulted in a preferred binding mode of compound Vcpal comprising about 70% of all proposed positions. In both cases, the Vc moiety acts as an anchor and was placed in such a way that an optimal hydrogen bonding pattern was achieved. In the case of both residues protonated (mode 1), Vc interacts with the backbone oxygen atoms of Tyr-263 and Tyr-265 and with the side chain oxygen of Tyr-279. In the other case (mode 2), the hydrogen bonding interaction is formed with both amino acids relevant for catalysis, Asp-147 and Glu-149. Thus, the specific binding mode seems to depend on the protonation state of the catalytic residues as depicted in Fig. 5, B and C. If Asp-147 and Glu-149 are both protonated the long alkyl chain of Vcpal occupies the region where the tetrasaccharide was found in the crystal structure of bee venom hyaluronidase (mode 1, Fig. 5B). In the other case the palmitoyl moiety projects to a channel-like region that probably belongs to the binding site of the polymeric HA before being degraded since it is directly attached to the site of the HA tetrasaccharide fragment (mode 2, Fig. 5C).
Both of the proposed binding modes are consistent with the competitive binding mechanism of Vcpal toward BTH. Nevertheless, with respect to potential hydrophobic interactions, binding mode 1 seems to be favored. Depicting the hydrophobic surrounding of Vcpal by means of hydrophobicity-colored Connolly surfaces shows that in this binding mode the long alkyl chain favorably interacts with an extended, strongly hydrophobic channel (brown color), which is formed by the mostly conserved amino acids Ala-84, Leu-91, Tyr-93, Tyr-220, and Leu-344 in BTH (Phe-46, Ile-53, Tyr-55, Tyr-184, and Ser-304 in BVH). In binding mode 2 the alkyl chain occupies a less hydrophobic region (green color) (Fig. 5, B and C).
Comparison of Binding Modes of l-Ascorbic Acid and l-Ascorbic Acid 6-Hexadecanoate to Hyaluronidases—The binding of Vc to SpnHL is mainly caused by hydrophobic interactions with the side chain of Trp-292 since the indole moiety is stacked with the five-member ring of Vc. This structural arrangement helps Vc to bind in the narrowest part of the binding site comprising the amino acids Trp-292, Arg-462, and Arg-466 (
). In the SpnHL-Vcpal structure, the Vcpal moiety of the inhibitor is detected in the same, narrow part of the active site but in a different conformation. This alternative binding mode is probably forced by the long aliphatic side chain, the binding of which is induced by strong hydrophobic interactions with residues of the hydrophobic patch. The ring-opened form in the crystal structure is not a prerequisite of that mode since the intact Vc moiety may be docked in a similar position without changing interactions of the palmitoyl chain.
Multiple binding modes for the Vc moiety may also be possible in the case of BTH as suggested by the docking studies for Vcpal; the Vc moiety can be placed in two different but adjacent parts of the active site of BTH (Fig. 5, B and C). This result probably follows from the topology of the HA binding site of BTH that is far more open than that of the streptococcal enzymes. Interactions of Vc itself may be too weak and/or immobilization and orientation effects cannot sufficiently act on the small molecule within this wide crevice so that additional hydrophobic substituents are needed to impart tight binding on BTH. In conclusion, the observed selectivity of Vc for bacterial hyaluronate lyases versus BTH seems to be due to a closer fit in a narrow binding pocket.
Conclusions—Hyaluronidases are considered to be involved in many (patho)physiological processes like fertilization and microbial spread, but most roles of the enzymes are still unknown. Inhibitors of these enzymes could be useful as pharmacological tools to unravel the precise functions of hyaluronidases. In this respect the present investigations on Vcpal are a first step toward the discovery of low molecular weight hyaluronidase inhibitors. Vcpal showed strong inhibition of S. agalactiae hyaluronate lyase (IC50 4 μm), whereas S. pneumoniae hyaluronate lyase and bovine testicular hyaluronidase were inhibited with IC50 values of only 100 and 56 μm, respectively. To date, Vcpal has proved to be one of the most potent inhibitors of hyaluronidases.
The elucidation of the x-ray structure of the SpnHL-Vcpal complex confirmed the hypothesis that hydrophobic interactions of the hexadecanoic moiety with Phe-343, His-399, and Thr-400 in the active site essentially contribute to the strong binding. Homology modeling of BTH and docking approaches suggest that Vcpal may interact with this enzyme in two alternative binding modes. In the more probable mode, the long alkyl chain seems to interact with an extended, hydrophobic channel formed by the mostly conserved amino acids Ala-84, Leu-91, Tyr-93, Tyr-220, and Leu-344 in BTH.
The knowledge of the binding modes of Vcpal on hyaluronidases provides several options for the design of more potent inhibitors. The combination of a small hydrophilic molecule like Vc and a larger lipophilic residue like the hexadecanoic moiety appears to be an appropriate strategy. Therefore, future work should be directed to mainly hydrophilic but more potent inhibitors than Vc serving as anchor in the center of the binding site. Such compounds might be suggested by structure-based ligand design methods. Hits inhibiting hyaluronidases should then be substituted in appropriate positions by larger lipophilic substituents, which might increase inhibitory activity as shown for hexadecanoic acid. Some of these approaches are the subjects of ongoing work.
Acknowledgments
The technical assistance of Susanne Bollwein is gratefully acknowledged. The diffraction data were collected at the Berkeley Center for Structural Biology, Advanced Light Source Lawrence Berkeley National Laboratory using beamline 5.0.1.
The atomic coordinates and structure factors (code 1W3Y) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).
72535-4&id=gr1.jpg){kind=link}
72535-4&id=gr2.jpg){kind=link}
72535-4&id=gr3.jpg){kind=link}
72535-4&id=gr4.jpg){kind=link}
72535-4&id=gr5.jpg){kind=link}