Novel inhibitors of anthrax edema factor

Several pathogenic bacteria produce adenylyl cyclase toxins, such as the edema factor (EF) of Bacillus anthracis. These disturb cellular metabolism by catalyzing production of excessive amounts of the regulatory molecule cAMP. Here, a structure-based method, where a 3D-pharmacophore that fit the active site of EF was constructed from fragments, was used to identify non-nucleotide inhibitors of EF. A library of small molecule fragments was docked to the EF-active site in existing crystal structures, and those with the highest HINT scores were assembled into a 3D-pharmacophore. About 10,000 compounds, from over 2.7 million compounds in the ZINC database, had a similar molecular framework. These were ranked according to their docking scores, using methodology that was shown to achieve maximum accuracy (i.e., how well the docked position matched the experimentally determined site for ATP analogues in crystal structures of the complex). Finally, 19 diverse compounds with the best AutoDock binding/docking scores were assayed in a cell-based assay for their ability to reduce cAMP secretion induced by EF. Four of the test compounds, from different structural groups, inhibited in the low micromolar range. One of these has a core structure common to phosphatase inhibitors previously identified by high-throughput assays of a diversity library. Thus, the fragment-based pharmacophore identified a small number of diverse compounds for assay, and greatly enhanced the selection process of advanced lead compounds for combinatorial design.     

Structural basis for the interaction of Bordetella pertussis adenylyl cyclase toxin with calmodulin

CyaA is crucial for colonization by Bordetella pertussis, the etiologic agent of whooping cough. Here we report crystal structures of the adenylyl cyclase domain (ACD) of CyaA with the C-terminal domain of calmodulin. Four discrete regions of CyaA bind calcium-loaded calmodulin with a large buried contact surface. Of those, a tryptophan residue (W242) at an alpha-helix of CyaA makes extensive contacts with the calcium-induced, hydrophobic pocket of calmodulin. Mutagenic analyses show that all four regions of CyaA contribute to calmodulin binding and the calmodulin-induced conformational change of CyaA is crucial for catalytic activation. A crystal structure of CyaA-calmodulin with adefovir diphosphate, the metabolite of an approved antiviral drug, reveals the location of catalytic site of CyaA and how adefovir diphosphate tightly binds CyaA. The ACD of CyaA shares a similar structure and mechanism of activation with anthrax edema factor (EF). However, the interactions of CyaA with calmodulin completely diverge from those of EF. This provides molecular details of how two structurally homologous bacterial toxins evolved divergently to bind calmodulin, an evolutionarily conserved calcium sensor.     

ATP conformations and ion binding modes in the active site of anthrax edema factor: A computational analysis

The Edema Factor (EF), one of the virulence factors of anthrax, is an adenylyl cyclase that promotes the overproduction of cyclic‐AMP (cAMP) from ATP, and therefore perturbs cell signaling. Crystallographic structures of EF bound to ATP analogs and reaction products, cyclic‐AMP, and Pyrophosphate (PPi), revealed different substrate conformations and catalytic‐cation binding modes, one or two cations being observed in the active site. To shed light into the biological significance of these crystallographic structures, the energetics, geometry, and dynamics of the active site are analyzed using molecular dynamics simulations. The ATP conformation observed in the one‐metal‐ion structure allows stronger interactions with the catalytic ion, and ATP is more restrained than in the structure containing two Mg²⁺ ions. Therefore, we propose that the conformation observed in the one‐ion crystal structure is a more probable starting point for the reaction. The simulations also suggest that a C3′‐endo sugar pucker facilitates nucleophilic attack. Additionally, the two‐cation binding mode restrains the mobility of the reaction products, and thus their tendency to dissociate. Proteins 2009. © 2009 Wiley‐Liss, Inc.     

Structure-based inhibitor discovery against adenylyl cyclase toxins from pathogenic bacteria that cause anthrax and whooping cough

Edema factor (EF) and CyaA are adenylyl cyclase toxins secreted by pathogenic bacteria that cause anthrax and whooping cough, respectively. Using the structure of the catalytic site of EF, we screened a data base of commercially available, small molecular weight chemicals for those that could specifically inhibit adenylyl cyclase activity of EF. From 24 compounds tested, we have identified one quinazoline compound, ethyl 5-aminopyrazolo[1,5-a]quinazoline-3-carboxylate, that specifically inhibits adenylyl cyclase activity of EF and CyaA with approximately 20 microm Ki. This compound neither affects the activity of host resident adenylyl cyclases type I, II, and V nor exhibits promiscuous inhibition. The compound is a competitive inhibitor, consistent with the prediction that it binds to the adenine portion of the ATP binding site on EF. EF is activated by the host calcium sensor, calmodulin. Surface plasmon resonance spectroscopic analysis shows that this compound does not affect the binding of calmodulin to EF. This compound is dissimilar from a previously described, non-nucleoside inhibitor of host adenylyl cyclase. It may serve as a lead to design antitoxins to address the role of adenylyl cyclase toxins in bacterial pathogenesis and to fight against anthrax and whooping cough.     

Physiological calcium concentrations regulate calmodulin binding and catalysis of adenylyl cyclase exotoxins

Edema factor (EF) and CyaA are calmodulin (CaM)-activated adenylyl cyclase exotoxins involved in the pathogenesis of anthrax and whooping cough, respectively. Using spectroscopic, enzyme kinetic and surface plasmon resonance spectroscopy analyses, we show that low Ca(2+) concentrations increase the affinity of CaM for EF and CyaA causing their activation, but higher Ca(2+) concentrations directly inhibit catalysis. Both events occur in a physiologically relevant range of Ca(2+) concentrations. Despite the similarity in Ca(2+) sensitivity, EF and CyaA have substantial differences in CaM binding and activation. CyaA has 100-fold higher affinity for CaM than EF. CaM has N- and C-terminal globular domains, each binding two Ca(2+) ions. CyaA can be fully activated by CaM mutants with one defective C-terminal Ca(2+)-binding site or by either terminal domain of CaM while EF cannot. EF consists of a catalytic core and a helical domain, and both are required for CaM activation of EF. Mutations that decrease the interaction of the helical domain with the catalytic core create an enzyme with higher sensitivity to Ca(2+)-CaM activation. However, CyaA is fully activated by CaM without the domain corresponding to the helical domain of EF.     

A novel mechanism for adenylyl cyclase inhibition from the crystal structure of its complex with catechol estrogen

Catechol estrogens are steroid metabolites that elicit physiological responses through binding to a variety of cellular targets. We show here that catechol estrogens directly inhibit soluble adenylyl cyclases and the abundant trans-membrane adenylyl cyclases. Catechol estrogen inhibition is non-competitive with respect to the substrate ATP, and we solved the crystal structure of a catechol estrogen bound to a soluble adenylyl cyclase from Spirulina platensis in complex with a substrate analog. The catechol estrogen is bound to a newly identified, conserved hydrophobic patch near the active center but distinct from the ATP-binding cleft. Inhibitor binding leads to a chelating interaction between the catechol estrogen hydroxyl groups and the catalytic magnesium ion, distorting the active site and trapping the enzyme substrate complex in a non-productive conformation. This novel inhibition mechanism likely applies to other adenylyl cyclase inhibitors, and the identified ligand-binding site has important implications for the development of specific adenylyl cyclase inhibitors.     

Effects of initial settings on computational protein–ligand docking accuracies for several docking programs

In this study, the influences of initial settings, i.e. initial conformations, configurations and docking parameters, on docking results were investigated. The conformations used in the study were generated by the CAMDAS program. After the conformational search calculations, five structures were selected from the conformer groups according to their conformation energies and root mean square deviations against crystal structures; for example, the lowest energy conformer, as well as the closest and farthest conformers to the crystal structure, was retrieved. Several docking parameter settings were used (default, high speed, generating 50 poses). In this study, docking calculations were conducted using the GOLD, eHiTS, AutoDock, AutoDock vina, FRED and DOCK programs. The success rates of GOLD, eHiTS and FRED were better than those of AutoDock, AutoDock vina and DOCK. The docking results using the farthest conformations were worse than those obtained using other conformations, indicating that some conformation search for the ligand molecule should be performed before the docking calculations.     

Docking studies on monoamine oxidase-B inhibitors: estimation of inhibition constants (K(i)) of a series of experimentally tested compounds

Monoamine oxidase (EC1.4.3.4; MAO) is a mitochondrial outer membrane flavoenzyme that catalyzes the oxidation of biogenic amines. It has two distinct isozymic forms designated MAO-A and MAO-B, each displaying different substrate and inhibitor specificities. They are the well-known targets for antidepressant and neuroprotective drugs. Elucidation of the X-ray crystallographic structure of MAO-B has opened the way for molecular modeling studies. A series of experimentally tested (1-10) model compounds has been docked computationally to the active site of the MAO-B enzyme. The AutoDock 3.0.5 program was employed to perform automated molecular docking. The free energies of binding (DeltaG) and inhibition constants (K(i)) of the docked compounds were calculated by the Lamarckian Genetic Algorithm (LGA) of AutoDock 3.0.5. Excellent to good correlations between the calculated and experimental K(i) values were obtained.     

MD simulations of anthrax edema factor: calmodulin complexes with mutations in the edema factor "switch a" region and docking of 3'-deoxy ATP into the adenylyl cyclase active site of wild-type and mutant edema factor variants

Bacillus anthracis, a spore-forming infectious bacterium, produces an exotoxin, called the edema factor (EF), that functions in part by disrupting internal signalling pathways. When complexed with human host cell calmodulin (CaM), EF becomes an active adenylyl cyclase, producing the internal signal substance cyclic-AMP in an uncontrolled fashion. Recently, the crystal structures for uncomplexed EF and EF:CaM complexes in the presence and absence of a substrate analog (3'-deoxy-ATP), were reported. EF mutational studies have implicated a number of residues important in CaM binding and/or in the generation of the adenylyl cyclase active site, formed by the movements of the EF switch A, B and C regions upon CaM binding. Here we report on the results of molecular dynamics (MD) simulations on two EF:CaM complexes, one containing wild-type EF and the other containing EF in which a cluster of residues in the switch A region (L523, K525, Q526 and V529) have been mutated to alanine. The switch A mutations cause a large increase in the flexibility of the switch C region, the rupture of a number of EF-CaM interactions, an expansion of the carboxyl-terminal domain of CaM, and a change in the Ca(2+) ion binding abilities of the CaM that is in complex with EF. The results indicate the importance of the mutated switch A residues in maintaining a compact EF:CaM complex that appears to be a prerequisite for the generation of a fully-functional adenylyl cyclase active site. The effects of mutating key residues (K346, K353, H577, E588, D590 and N639) in the active site region of EF (to alanine) on the ability of EF to bind the 3'-deoxy-ATP substrate analog were also examined. Active-site residue substitutions at positions 583 (N583A) and 577 (H577A) were found to be particularly disruptive for the placement of the adenine ring moiety into the position found in the x-ray crystal structure of the ligand-protein complex.     

MD Simulations of Anthrax Edema Factor: Calmodulin Complexes with Mutations in the Edema Factor “Switch A” Region and Docking of 3′-deoxy ATP into the Adenylyl Cyclase Active Site of Wild-Type and Mutant Edema Factor Variants

Abstract

Bacillus anthracis, a spore-forming infectious bacterium, produces an exotoxin, called the edema factor (EF), that functions in part by disrupting internal signalling pathways. When complexed with human host cell calmodulin (CaM), EF becomes an active adenylyl cyclase, producing the internal signal substance cyclic-AMP in an uncontrolled fashion. Recently, the crystal structures for uncomplexed EF and EF:CaM complexes in the presence and absence of a substrate analog (3′-deoxy-ATP), were reported. EF mutational studies have implicated a number of residues important in CaM binding and/or in the generation of the adenylyl cyclase active site, formed by the movements of the EF switch A, B and C regions upon CaM binding. Here we report on the results of molecular dynamics (MD) simulations on two EF:CaM complexes, one containing wild-type EF and the other containing EF in which a cluster of residues in the switch A region (L523, K525, Q526 and V529) have been mutated to alanine. The switch A mutations cause a large increase in the flexibility of the switch C region, the rupture of a number of EF-CaM interactions, an expansion of the car-boxyl-terminal domain of CaM, and a change in the Ca²⁺ ion binding abilities of the CaM that is in complex with EF. The results indicate the importance of the mutated switch A residues in maintaining a compact EF:CaM complex that appears to be a prerequisite for the generation of a fully-functional adenylyl cyclase active site. The effects of mutating key residues (K346, K353, H577, E588, D590 and N639) in the active site region of EF (to alanine) on the ability of EF to bind the 3′-deoxy-ATP substrate analog were also examined. Active-site residue substitutions at positions 583 (N583A) and 577 (H577A) were found to be particularly distruptive for the placement of the adenine ring moiety into the position found in the x-ray crystal structure of the ligand-protein complex.     

Structural evidence for a ligand coordination switch in liver alcohol dehydrogenase

The use of substrate analogues as inhibitors provides a way to understand and manipulate enzyme function. Here we report two 1 A resolution crystal structures of liver alcohol dehydrogenase in complex with NADH and two inhibitors: dimethyl sulfoxide and isobutyramide. Both structures present a dynamic state of inhibition. In the dimethyl sulfoxide complex structure, the inhibitor is caught in transition on its way to the active site using a flash-freezing protocol and a cadmium-substituted enzyme. One inhibitor molecule is partly located in the first and partly in the second coordination sphere of the active site metal. A hydroxide ion bound to the active site metal lies close to the pyridine ring of NADH, which is puckered in a twisted boat conformation. The cadmium ion is coordinated by both the hydroxide ion and the inhibitor molecule, providing structural evidence of a coordination switch at the active site metal ion. The structure of the isobutyramide complex reveals the partial formation of an adduct between the isobutyramide inhibitor and NADH. It provides evidence of the contribution of a shift from the keto to the enol tautomer during aldehyde reduction. The different positions of the inhibitors further refine the knowledge of the dynamics of the enzyme mechanism and explain how the crowded active site can facilitate the presence of a substrate and a metal-bound hydroxide ion.     

Bicarbonate activation of adenylyl cyclase via promotion of catalytic active site closure and metal recruitment

In an evolutionarily conserved signaling pathway, 'soluble' adenylyl cyclases (sACs) synthesize the ubiquitous second messenger cyclic adenosine 3',5'-monophosphate (cAMP) in response to bicarbonate and calcium signals. Here, we present crystal structures of a cyanobacterial sAC enzyme in complex with ATP analogs, calcium and bicarbonate, which represent distinct catalytic states of the enzyme. The structures reveal that calcium occupies the first ion-binding site and directly mediates nucleotide binding. The single ion-occupied, nucleotide-bound state defines a novel, open adenylyl cyclase state. In contrast, bicarbonate increases the catalytic rate by inducing marked active site closure and recruiting a second, catalytic ion. The phosphates of the bound substrate analogs are rearranged, which would facilitate product formation and release. The mechanisms of calcium and bicarbonate sensing define a reaction pathway involving active site closure and metal recruitment that may be universal for class III cyclases.     

Ligand-induced conformational changes: improved predictions of ligand binding conformations and affinities

A computational docking strategy using multiple conformations of the target protein is discussed and evaluated. A series of low molecular weight, competitive, nonpeptide protein tyrosine phosphatase inhibitors are considered for which the x-ray crystallographic structures in complex with protein tyrosine phosphatase 1B (PTP1B) are known. To obtain a quantitative measure of the impact of conformational changes induced by the inhibitors, these were docked to the active site region of various structures of PTP1B using the docking program FlexX. Firstly, the inhibitors were docked to a PTP1B crystal structure cocrystallized with a hexapeptide. The estimated binding energies for various docking modes as well as the RMS differences between the docked compounds and the crystallographic structure were calculated. In this scenario the estimated binding energies were not predictive inasmuch as docking modes with low estimated binding energies corresponded to relatively large RMS differences when aligned with the corresponding crystal structure. Secondly, the inhibitors were docked to their parent protein structures in which they were cocrystallized. In this case, there was a good correlation between low predicted binding energy and a correct docking mode. Thirdly, to improve the predictability of the docking procedure in the general case, where only a single target protein structure is known, we evaluate an approach which takes possible protein side-chain conformational changes into account. Here, side chains exposed to the active site were considered in their allowed rotamer conformations and protein models containing all possible combinations of side-chain rotamers were generated. To evaluate which of these modeled active sites is the most likely binding site conformation for a certain inhibitor, the inhibitors were docked against all active site models. The receptor rotamer model corresponding to the lowest estimated binding energy is taken as the top candidate. Using this protocol, correct inhibitor binding modes could successfully be discriminated from proposed incorrect binding modes. Moreover, the ranking of the estimated ligand binding energies was in good agreement with experimentally observed binding affinities.     

Crystal Structure of Dicamba Monooxygenase: A Rieske Nonheme Oxygenase that Catalyzes Oxidative Demethylation

Dicamba (3,6-dichloro-2-methoxybenzoic acid) is a widely used herbicide that is efficiently degraded by soil microbes. These microbes use a novel Rieske nonheme oxygenase, dicamba monooxygenase (DMO), to catalyze the oxidative demethylation of dicamba to 3,6-dichlorosalicylic acid (DCSA) and formaldehyde. We have determined the crystal structures of DMO in the free state, bound to its substrate dicamba, and bound to the product DCSA at 2.10-1.75 Å resolution. The structures show that the DMO active site uses a combination of extensive hydrogen bonding and steric interactions to correctly orient chlorinated, ortho-substituted benzoic-acid-like substrates for catalysis. Unlike other Rieske aromatic oxygenases, DMO oxygenates the exocyclic methyl group, rather than the aromatic ring, of its substrate. This first crystal structure of a Rieske demethylase shows that the Rieske oxygenase structural scaffold can be co-opted to perform varied types of reactions on xenobiotic substrates.     

Docking simulation with a purine nucleoside specific homology model of deoxycytidine kinase, a target enzyme for anticancer and antiviral therapy

5'-Phosphorylation, catalyzed by human deoxycytidine kinase (dCK), is a crucial step in the metabolic activation of anticancer and antiviral nucleoside antimetabolites, such as cytarabine (AraC), gemcitabine, cladribine (CdA), and lamivudine. Recently, crystal structures of dCK (dCKc) with various pyrimidine nucleosides as substrates have been reported. However, there is no crystal structure of dCK with a bound purine nucleoside, although purines are good substrates for dCK. We have developed a model of dCK (dCKm) specific for purine nucleosides based on the crystal structure of purine nucleoside bound deoxyguanosine kinase (dGKc) as the template. dCKm is essential for computer aided molecular design (CAMD) of novel anticancer and antiviral drugs that are based on purine nucleosides since these did not bind to dCKc in our docking experiments. The active site of dCKm was larger than that of dCKc and the amino acid (aa) residues of dCKm and dCKc, in particular Y86, Q97, D133, R104, R128, and E197, were not in identical positions. Comparative docking simulations of deoxycytidine (dC), cytidine (Cyd), AraC, CdA, deoxyadenosine (dA), and deoxyguanosine (dG) with dCKm and dCKc were carried out using the FlexX docking program. Only dC (pyrimidine nucleoside) docked into the active site of dCKc but not the purine nucleosides dG and dA. As expected, the active site of dCKm appeared to be more adapted to bind purine nucleosides than the pyrimidine nucleosides. While water molecules were essential for docking experiments using dCKc, the absence of water molecules in dCKm did not affect the ability to correctly dock various purine nucleosides.     

Methodology and problems of protein-ligand docking: case study of dihydroorotate dehydrogenase,thymidine kinase,and phosphodiesterase 4

The docking methodology was applied to three different therapeutically interesting enzymes: human dihydroorotate dehydrogenase (DHODH), Herpes simplex virus type I thymidine kinase (HSV1 TK) and human phosphodiesterase 4 (PDE4). Programs FlexX, AutoDock and DOCK where used. The three targets represent three distinct cases. For DHODH and HSV1 TK, the binding modes of substrate and inhibitors within the active site are known, while the binding orientation of cAMP within PDE4 has been solely hypothesized. Active site of DHODH is mainly hydrophobic and the binding mode of the inhibitor brequinar was used as a template for evaluating the docking strategies. The presence of cofactors revealed to be crucial for the definition of the docking site. The HSV1 TK active site is small and polar and contains crystal water molecules and ATP. Docking of thymidine and aciclovir (ACV) within the active site was analyzed by keeping or removing water molecules. It showed the crucial role of water in predicting the binding of pyrimidines and purines. The crystal structure of PDE4 contains magnesium and zinc cations as well as catalytic water molecule but no ligand. Several docking experiments of cAMP and rolipram were performed and the results showed clear-cut dependence between the ligand orientation and the presence of metals in the active site. All three cases show specific problems of the docking methodology, depending on the character of the active site.     

Automated docking of monosaccharide substrates and analogues and methyl α-Acarviosinide in the glucoamylase active site

Glucoamylase is an important industrial glucohydrolase with a large specificity range. To investigate its interaction with the monosaccharides D-glucose, D-mannose, and D-galactose and with the substrate analogues 1-deoxynojirimycin, D-glucono-1,5-lactone, and methyl αacarviosinide, MM3(92)-optimized structures were docked into its active site using AutoDock 2.1. The results were compared to structures of glucoamylase complexes obtained by protein crystallography. Charged forms of some substrate analogues were also docked to assess the degree of protonation possessed by glucoamylase inhibitors. Many forms of methyl αa-carviosinide were conformationally mapped by using MM3(92), characterizing the conformational pH dependence found for the acarbose family of glucosidase inhibitors. Their significant conformers, representing the most common states of the inhibitor, were used as initial structures for docking. This constitutes a new approach for the exploration of binding modes of carbohydrate chains. Docking results differ slightly from x-ray crystallographic data, the difference being of the order of the crystallographic error. The estimated energetic interactions, even though agreeing in some cases with experimental binding kinetics, are only qualitative due to the large approximations made by AudoDock force field. © 1997 Wiley-Liss, Inc.     

L-aspartase from Escherichia coli: substrate specificity and role of divalent metal ions

The enzyme L-aspartase from Escherichia coli has an absolute specificity for its amino acid substrate. An examination of a wide range of structural analogues of L-aspartic acid did not uncover any alternate substrates for this enzyme. A large number of competitive inhibitors of the enzyme have been characterized, with inhibition constants ranging over 2 orders of magnitude. A divalent metal ion is required for enzyme activity above pH 7, and this requirement is met by many transition and alkali earth metals. The binding stoichiometry has been established to be one metal ion bound per subunit. Paramagnetic relaxation studies have shown that the divalent metal ion binds at the recently discovered activator site on L-aspartase and not at the enzyme active site. Enzyme activators are bound within 5 A of the enzyme-bound divalent metal ion. The activator site is remote from the active site of the enzyme, since the relaxation of inhibitors that bind at the active site is not affected by paramagnetic metal ions bound at the activator site.     

Phosphate-bound structure of an organophosphate-degrading enzyme from Agrobacterium radiobacter

OpdA is a binuclear metalloenzyme that can hydrolyze organophosphate pesticides and nerve agents. In this study the crystal structure of the complex between OpdA and phosphate has been determined to 2.20 Å resolution. The structure shows the phosphate bound in a tripodal mode to the metal ions whereby two of the oxygen atoms of PO₄ are terminally bound to each metal ion and a third oxygen bridges the two metal ions, thus displacing the μOH in the active site. In silico modelling demonstrates that the phosphate moiety of a reaction product, e.g. diethyl phosphate, may bind in the same orientation, positioning the diethyl groups neatly into the substrate binding pocket close to the metal center. Thus, similar to the binuclear metallohydrolases urease and purple acid phosphatase the tripodal arrangement of PO₄ is interpreted in terms of a role of the μOH as a reaction nucleophile.     

Key substrate recognition residues in the active site of a plant cytochrome P450, CYP73A1. Homology guided site-directed mutagenesis.

CYP73 enzymes are highly conserved cytochromes P450 in plant species that catalyse the regiospecific 4-hydroxylation of cinnamic acid to form precursors of lignin and many other phenolic compounds. A CYP73A1 homology model based on P450 experimentally solved structures was used to identify active site residues likely to govern substrate binding and regio-specific catalysis. The functional significance of these residues was assessed using site-directed mutagenesis. Active site modelling predicted that N302 and I371 form a hydrogen bond and hydrophobic contacts with the anionic site or aromatic ring of the substrate. Modification of these residues led to a drastic decrease in substrate binding and metabolism without major perturbation of protein structure. Changes to residue K484, which is located too far in the active site model to form a direct contact with cinnamic acid in the oxidized enzyme, did not influence initial substrate binding. However, the K484M substitution led to a 50% loss in catalytic activity. K484 may affect positioning of the substrate in the reduced enzyme during the catalytic cycle, or product release. Catalytic analysis of the mutants with structural analogues of cinnamic acid, in particular indole-2-carboxylic acid that can be hydroxylated with different regioselectivities, supports the involvement of N302, I371 and K484 in substrate docking and orientation.