Novel neuraminidase inhibitors: identification ... - Future Science

4 downloads 0 Views 1MB Size Report
Novel neuraminidase inhibitors: identification, biological evaluation and investigations of the binding mode. Background: The pathogenicity of influenza A and B ...
Research Article Special Focus: Computational Chemistry For reprint orders, please contact [email protected]

Novel neuraminidase inhibitors: identification, biological evaluation and investigations of the binding mode Background: The pathogenicity of influenza A and B viruses depends on the function of influenza neuraminidase (NA). Emerging resistant influenza A viruses of subtype H1N1 increasingly challenge the effectiveness of established NA inhibitors. Recent computational studies have indicated several weak points of NA that can be exploited for rational inhibitor design to conquer this imminent threat, such as the opening of the binding pocket due to the flexibility of the 150-, 245- and 430-loops. Methods: We employed shape-focused virtual screening based on a recently discovered lead compound, katsumadain A, to identify novel promising compounds with significant inhibitory efficacy on NA and resistance-breaking capacity on oseltamivir-resistant strains. A potential binding mode of these compounds was derived employing ligand-based techniques and protein–ligand docking using representative protein conformations selected from molecular dynamics simulations. Results: Five novel compounds were identified by virtual screening. Their IC50 values, determined in chemiluminescence-based NA inhibition assays, are in the range of 0.18–17 µM. In particular, artocarpin exhibits high affinity toward three H1N1 oseltamivir-sensitive influenza A viruses. It also inhibits the NA of an oseltamivir-resistant H1N1 isolate.

Influenza particularly puts the young, the elderly and patients with chronic diseases at high risk [1,2,101] . People in these high-risk categories can often contract pneumonia, which can be fatal in serious cases. Seasonal influenza causes an estimated death toll of 250,000–500,000 people a year [101] . Among the three different influenza serotypes, influenza A and B viruses are critical for infectious diseases, while influenza C viruses play only a minor role in humans. Influenza A viruses are the most virulent human pathogens, being responsible for several major influenza pandemics with strains of high transmission rates. Moreover, the high virulence and lethality of H5N1 and the imminent risk of virus reassortment and mutation raise strong concerns about human health [1] . The 2009 flu pandemic was caused by a new H1N1 influenza A virus (H1N1v) that emerged by reassortment of avian, human, classical as well as European swine viruses [2] . It was characterized by a high transmission rate and low virulence [3] . Although this pandemic developed less critically than anticipated, these recent events highlight the urgent need for novel antiviral therapeutics [4,5] . Influenza virus neuraminidase (NA) has become the most established target for flu treatment. This viral enzyme enables the transport of the virus through mucus. Most importantly,

it facilitates the detachment of mature viruses from the cell surface and suppresses their selfaggregation by specifically cleaving off sialic acid from the cell-surface glycoprotein and the envelope of progeny viruses, respectively  [6] . Inhibitors of influenza NA prevent the release of progeny viruses and, hence, prevent the spread and replication of the virus. Computational approaches have played a key role in the rational drug design and optimization of NA inhibitors (NAIs). Zanamivir (Relenza®) was the first approved NAI [7,8] . The highly successful optimization campaign was driven by employing the software GRID  [9] to extend protein–ligand interactions by introducing a basic functional group in the conserved pocket of the binding site, which forms a salt bridge with Glu-119. A drawback is the insufficient oral absorption of zanamivir, which has been optimized in oseltamivir (Tamiflu®), the first orally bioavailable NAI on the market [10] . Today, oseltamivir is considered the first choice for the treatment of influenza infections. In addition, it can be applied for prophylaxes. Recently, the cyclopentane derivative, peramivir  [11] , has been approved for emergency use in hospitalized patients suffering from H1N1 influenza. However, the potential of the narrow portfolio of anti-influenza

10.4155/FMC.10.292 © 2011 Future Science Ltd

Future Med. Chem. (2011) 3(4), 437–450

Johannes Kirchmair1, Judith M Rollinger2 , Klaus R Liedl1, Nora Seidel3, Andi Krumbholz3 & Michaela Schmidtke†3 1 Institute of Theoretical Chemistry and Center for Molecular Biosciences, University of Innsbruck, Austria 2 Institute of Pharmacy/Pharmacognosy and Center for Molecular Biosciences Innsbruck, University of Innsbruck, Austria 3 Institute of Virology and Antiviral Therapy, Friedrich Schiller University, Hans-Knoell-Straße 2, 07745 Jena, Germany † Author for correspondence: Tel.: +49 364 1939 5723 Fax: +49 364 1939 5702 E-mail: michaela.schmidtke@ unijena.de

ISSN 1756-8919

437

Research Article | Kirchmair, Rollinger, Liedl, Seidel, Krumbholz & Schmidtke Key Terms influenza: Influenza is a major infectious disease responsible for the 250,000 - 500,000 deaths per year. The ability of influenza viruses to rapidly alternate in viral genetics is particularly problematic for prophylaxis and treatment with vaccines and drugs.

shape-focused virtual screening: Shape-focused

virtual screening is a particularly powerful approach to identify novel bioactive compounds. While it considers shape as the major characteristic to describe similarities among molecules, chemical (pharmacophore-like) features are also taken into account.

438

drugs is limited and threatened by emerging drug resistances. Before the flu season of 2007/2008, antiviral resistance to NAIs had rarely been reported [12–14] . In that season, however, the situation changed for the worse, when oseltamivir-resistant H1N1 viruses carrying the amino acid substitution H274Y spontaneously emerged and spread globally  [15–17] . In 2009 the latter were mostly replaced by the influenza A virus H1N1v that is NAI susceptible [18] . The fast changes in oseltamivir susceptibility of worldwide circulating viruses underline an urgent need in new compounds breaking resistance. Among the most common amino acid substitutions leading to oseltamivir resistance of NA are E119V, R152K, D198N, H274Y, E276D, R292K and N294S [19] . Several weak points of NA have recently been revealed, which have not, or only rudimentarily, been exploited for rational inhibitor design, so far. For example, x-ray crystallographic studies have indicated a high degree of conformational flexibility of parts of influenza NA [20] , which may allow distinct, innovative scaffolds to be accommodated within the binding site. NSC 89853, for example, has recently been identified as a bulky NAI that is only accommodated by an open x-ray structure conformation of the NA binding site [21] . Observations on conformational flexibility of influenza NA in x-ray structures stimulated further computational investigations employing molecular dynamics (MD) simulation techniques. Amaro et al. identified extended conformational shifts of the 150- and the 430-loops that were not anticipated from the x-ray structures [22, 23] . These conformations could be systematically exploited for the design of NAIs with alternative binding modes. Several research groups followed this idea and suggested novel inhibitor leads targeting open NA conformations. However, the merit of these studies is limited by the exploration of compounds closely related to oseltamivir [24–26] or the lack of experimental validation of the bioactivity of the proposed inhibitors [27] . Recently, we have identified compounds of NAIs from seed extract of Alpinia katsumadai and have elucidated their potential binding mode [28] . The diarylheptanoid, katsumadain A (1; Table 1) was identified as the most promising lead structure. It inhibited the NA of the human influenza virus A/PR/8/34 of subtype H1N1 as well as of H1N1 swine influenza viruses, with IC50 values between 0.9 and 1.64 µM. Its activity against an oseltamivir-resistant H1N1 isolate carrying the Future Med. Chem. (2011) 3(4)

amino acid substitution H274Y was markedly lower. Employing protein–ligand docking techniques we found that katsumadain A is unlikely to bind to the active site of the target in a conformation of NA that has been observed in x-ray experiments already, due to its extended shape. Considering the flexible 150- and 430-loop regions of NA, extensive MD simulations were performed to derive representative conformations of influenza NA. Major conformational differences among those selected frames were reflected in distinct conformations in the area of the flexible loops described in the work of Amaro et al., which also includes the 245-loop. Overall, we were able to identify one particular frame that obtained high-docking scores for the placement of katsumadain A, favorable interaction patterns between the ligand and the protein and very good surface complementarity. In the current study, the proposed binding mode of katsumadain A was taken as a starting point for shape-focused virtual screening in order to identify novel, biologically active chemical entities. Five compounds selected from the NCI database for biological testing (a molecular library provided by the US NIH Developmental Therapeutics Program, containing more than 200,000 small organic molecules  [29] ), were evaluated for their NA-inhibiting potential. All demonstrated strong activities against three oseltamivir-susceptible H1N1 strains. Their inhibitory potential against the oseltamivirresistant H1N1 strain is approximately 2–20times lower and one of the compounds, artocarpin (4) (a twice isoprenylated flavone present in different species of the genus Artocarpus), also exhibits high activity on an oseltamivirresistant H1N1 isolate. The binding properties of these bioactive compounds are investigated using protein–ligand docking techniques on x-ray structural models as well as representative MD frames. Experimental setup Preparations for the MD simulations, trajectory clustering, shape-focused virtual screening and protein–ligand docking were processed on an Intel Core 2 Quad Q6600 workstation with 4  ×  2.4  GHz, 2  ×  4  MB L2-Cache and 4  GB  R AM, running OpenSuse  11.2. The production MD was run on 16  cores of an AMD Opteron cluster, 2 × AMD Opteron 2382, 4 × 2.6 GHz each, 4 × 512 kB L2-Cache, 6 MB L3-Cache and 4 GB RAM. „„Hardware

future science group

Novel neuraminidase inhibitors „„Shape-focused

virtual screening The NCI database was downloaded directly from the NCI website [102] and prepared using Pipeline Pilot Student Edition (version 6.1.5; Accelrys, CA, USA). Unwanted molecules were filtered using the Organic Filter and Bad Valence Filter nodes. In accordance to the protein setup, the compounds were ionized at pH 6.5 using the Ionize Molecule at pH component. The default acid/base sites collection and the default partial least squares regression models were applied. Compounds were filtered for 150 ≤ molecular weight ≤ 600 and AlogP ≤ 6 [30] . Conformational ensembles were calculated for the refined compound collection using OMEGA (version 2.3.2; OpenEye, NM, USA) [31] and stored in the OpenEye binary data format (.oeb.gz). ROCS (version 3.0.0; OpenEye [32] was employed for similarity-based screening of the NCI database. Katsumadain A, our most promising hit, identified from A. katsumadai, was used as a template for shape-focused virtual screening. A representative query conformation was determined using OMEGA to create a single conformer per molecule (using OMEGA’s maxconf flag). ROCS screening was performed using the default settings. The hits were ranked employing the ComboScore, a scoring function that assesses the goodness of the alignment between the query and the candidate molecules. ComboScore puts exactly equal weights on both of its components, a shape-based scoring function (represented by the ShapeTanimoto score) and a function considering pharmacophore-like chemical pattern matching (represented by the ScaledColor score) [32] . Each component can yield values from zero to one, where zero stands for absolute dissimilarity and one denotes identity. ComboScore is the sum of ShapeTanimoto and ScaledColor and, hence, can yield values from zero to two, where higher values indicate a closer similarity. „„Alignment

studies The Flexible Ligand Alignment algorithm implemented in MOE (version 2009.10 ; Chemical Computing Group CCG, Montreal, QC, Canada) was employed for ligand-based alignment. The conformations, as obtained for the hits during ROCS screening, were used as seed structures.

„„MD

simulations Amber 10 [33] was employed for MD simulations using PDB (Protein Data Bank) entry open conformation, apo structure (2hty [20] ; chain B) as future science group

| Research Article

a structural basis, following the approach of Amaro et al. [22,23] . The PDB2PQR server [34, 35] was used to prepare the system and to protonate the target according to pH 6.5. Crystal waters were preserved 5 Å around the protein using VMD [36] . A TIP3P water box was placed 10 Å around the protein. Rectangular periodic boundary conditions and the particle mesh Ewald approach were applied for calculating electrostatics (cutoff 10 Å). An initial minimization, limited to solvent molecules, was performed for 500  steepest descent minimization cycles and 500 conjugate gradient cycles. This was followed by a minimization of the full system for 2500 cycles using the steepest descent algorithm and 2500 cycles using the conjugate gradient algorithm. The system was gradually warmed to 300°K over the initial 50 ps of the simulation by applying weak restraints to the protein. An equilibration step of 100 ps under constant pressure was performed, allowing the water molecules to relax. The production MD was processed with a 2-fs time step and the SHAKE algorithm keeping all bond lengths involving hydrogens fixed. A 20-ns section of the trajectory was submitted to frame clustering. To obtain ten representative protein conformations for ligand docking, the trajectory was clustered considering the conformational flexibility (in terms of the backbone root mean square deviation) of all residues of the first shell building the binding concavity of NA (residues 114–119, 134–140, 145–152, 156, 178–180, 222–227, 244–246, 276–277, 292, 294, 347–350, 371, 403–406, 423, 425– 432 and 437– 441). Amber  10 includes 11 different algorithms that allow for clustering trajectories. Shao et al provided an excellent in-depth ana­lysis of the performance of these different algorithms, their advantages and disadvantages [37] . On the basis of their observations, we selected the average linkage algorithm for clustering our trajectory (using default settings). „„Protein–ligand

docking The genetic docking algorithm, GOLD (version  4.1; Cambridge Crystallographic Data Centre, Cambridge, UK) [38] , was employed for protein–ligand docking. The 3D seed structures of the ligands were generated from scratch using OMEGA. Besides the deactivation of ‘early termination’, program defaults were used for docking and GOLDScore was employed for pose ranking. www.future-science.com

439

Research Article | Kirchmair, Rollinger, Liedl, Seidel, Krumbholz & Schmidtke „„Visualization

Molecular structures were visualized using VROCS (version 3.0.0; OpenEye), VIDA (version 4.0.0; OpenEye) and PyMol (version 1.0; DeLano Scientific, CA, USA). „„Compound

preparations Oseltamivir carboxylate (GS4071) was kindly provided by F Hoffmann-La Roche AG (Basel, Switzerland; lot no. RO0640802–002) and zanamivir (GG167) by GlaxoSmithKline (Uxbridge, UK). Compound stocks were prepared in water and stored at 4°C. Stock solutions of compounds 1–6 were prepared in DMSO. All five tested small organic molecules were ordered from the NCI database, USA. Their purity was checked using TLC and LC–MS and revealed to be >95%.

„„Cells

& viruses Madin–Darby canine kidney (MDCK) cells (Friedrich-Loeffler Institute, Riems, Germany) were grown in Eagle minimum essential medium supplemented with fetal bovine serum 10%, penicillin 100  U/ml and streptomycin 100  U/ml. Influenza viruses  A/Jena/5555/09 and A/Jena/5528/09 were isolated in MDCK cells from nasal swabs obtained from flu patients. Stocks of these clinical H1N1 isolates, of the H1N1 influenza virus  A/Puerto Rico/8/34 (A/PR/8/34; Institute of Virology, Philipps University Marburg, Germany), and the oseltamivir-resistant human H1N1 isolate A/342/09 (Robert Koch Institute, Berlin, Germany) were propagated in MDCK cells in serum-free medium formulated with trypsin 2 µg/ml and bicarbonate 1.2 mM, aliquoted and stored at -80°C until use.

„„Chemiluminescence-based

NA inhibition assay Neuraminidase activity and enzyme inhibition were determined with the commercially available NA-Star® kit (Tropix, Applied Biosystems, Darmstadt, Germany) that utilizes a 1,2-dioxetane sialic acid derivative for the substrate, as described in detail recently [28] . To evaluate the concentration required to reduce NA enzyme activity by 50%, at least six serial threefold compound dilutions (dilution in NA-Star assay buffer) were tested a minimum of three times. Katsumadain  A was included as control compound on each test plate. At first the enzyme  inhibition assay was performed with influenza virus A/PR/8/34. Compounds (3–6) 440

Future Med. Chem. (2011) 3(4)

with IC50 lower than 10 µM were then tested against two recent H1N1 isolates as well as an oseltamivir-resistant isolate. „„Determination

of cytotoxicity The cytotoxicity of test compounds was determined on 2-day-old MDCK confluent cell monolayers grown in 96-well plates as described previously [39] . Briefly, after removal of cell culture medium, serial twofold dilutions of compounds were added to the cells for 72 h (37°C, 5% CO2 and each concentration in triplicate in cell culture medium). Then, the cells were fixed and stained with a crystal violet formalin solution. After dye extraction, the optical density of individual wells was quantified spectrophotometrically at 550/630 nm with a microplate reader. Cell viability of individual compound-treated wells was evaluated as the percentage of the mean value of optical density resulting from six mock-treated cell controls, which was set 100%. The 50% cytotoxic concentration (CC50 ) was defined as the compound concentration reducing the viability of untreated cell cultures by 50%. Results & discussion virtual screening The NCI database was screened for potential inhibitors of influenza NA employing the shapefocused screening engine ROCS for similaritybased prioritization of candidate molecules. The screening was based on the previously identified NA-inhibiting natural compound katsumadain A (1) isolated and identified from A. katsumadai [28] . The basic concept of shape-based methods is the generation of a complementary image of the binding site by considering the shape of a ligand (referred to as a template or query). Molecules complying with this mapping are expected to be potentially active on the particular target addressed by the template. ROCS assesses the volume overlap between two molecules by Gaussians that are parameterized according to the hardsphere volume of heavy atoms. It considers shape and chemical functionality (encoded by the socalled ‘color force field’) for hit ranking. The ComboScore function puts exactly equal weights on its both components, the ShapeTanimoto and the ScaledColor score, which we found to obtain robust predictive power [40] . The 3D conformation of the query compound, katsumadain A (1) , was calculated using OMEGA, as exemplified in the ‘Methods’ section (depiction provided in Table  1). Recent

„„ Similarity-based

future science group

Novel neuraminidase inhibitors

A/Jena/5258/09 A/Jena/5555/09 A/342/09 A/PR/8/34

10

100

Inhibition (%)

1

110 100 90 80 70 60 50 40 30 20 10 0 0.01

0.1

1

10

100

Inhibition (%)

0.1

110 100 90 80 70 60 50 40 30 20 10 0 0.01

0.1

1

10

100

Inhibition (%)

110 100 90 80 70 60 50 40 30 20 10 0 0.01

110 100 90 80 70 60 50 40 30 20 10 0 0.01

0.1

1

10

100

Inhibition (%)

Inhibition (%)

A

110 100 90 80 70 60 50 40 30 20 10 0 0.01

0.1

1 Concentration (µg/ml)

10

100

B

C

D

E

| Research Article

Figure 1. Dose-dependent inhibition of the neuraminidase of the oseltamivir-sensitive H1N1 influenza viruses A/PR8/34, A/Jena/5528/09 and A/5555/09 as well as the oseltamivir-resistant A/340/09 by compounds 1, 3–6 (A–E) in chemiluminescence-based neuraminidase inhibition assay. Mean values and standard deviations from at least three independently performed assays are shown.

investigations have demonstrated the robustness of ROCS regarding the query conformation. The screening engine obtains high accuracy with any reasonable low-energy conformation as long as it is within the sampling space of the conformational model generator. Use of an future science group

experimentally determined conformation (such as an x-ray protein–ligand complex) does not necessarily lead to superior performance [32,40] . The NCI structural database was pre-processed employing Pipeline Pilot and converted to a multiconformational library using OMEGA. The www.future-science.com

441

442

Future Med. Chem. (2011) 3(4)

Artocarpin

4

241010

MW

1182–34–9

7608–44–8

436.5

358.3

516.5

252239–83–1 476.6

CAS‡

102049 1244–78–6

91529

NCI†

HO

O

OH

Structure

O

O

O O

OH

O

O

O

HO O (R )

O

O

O

O

O OH

HO

OH

OH

(R )

O

O

O

O

OH

OH

OH

ROCS overlay with katsumadain A§

1.057/0.44/0.613

0.991/0.491/0.499

1.072/0.637/0.435

ComboScore/ ShapeScore/ScaledColor score with compound (1) ¶

Compound identifier for the NCI database. ‡ CAS registration number. § Chemical features of katsumadain A (1) considered for shape-focused virtual screening: hydrogen-bond acceptors are depicted as red spheres and aromatic moieties as green discs, respectively. Carbons are indicated in blue/green color, oxygens in red color. ¶ Values obtained by ComboScore and its components, ShapeScore and ScaledColor score. MW: Molecular weight.



Quercetin 5,7,3’,4’-tetramethyl ether

1,4-dicaffeoylquinic acid

2

3

Katsumadain A

1

Compound Name

Table 1. Compounds active on influenza neuraminidase identified by shape-focused screening.

Research Article | Kirchmair, Rollinger, Liedl, Seidel, Krumbholz & Schmidtke

future science group

future science group

Compound identifier for the NCI database. CAS registration number. Chemical features of katsumadain A (1) considered for shape-focused virtual screening: hydrogen-bond acceptors are depicted as red spheres and aromatic moieties as green discs, respectively. Carbons are indicated in blue/green color, oxygens in red color. ¶ Values obtained by ComboScore and its components, ShapeScore and ScaledColor score. MW: Molecular weight. †

§

294.3 604847 35486–73–8 1-(5-hydroxy-2,2dimethyl-2H-1benzopyran-6-yl)-2phenyl-ethanone 6

www.future-science.com



OH

O

552.5 363258 49619–87–6 4’-O-methylochnaflavone 5

CAS‡

MW NCI†

HO

OH

Structure

O

O

O

O

O

O

OH

O

OH

ROCS overlay with katsumadain A§

1.040/0.501/0.539

0.976/0.600/0.376

| Research Article

Compound Name

Hits of significantly lower molecular size than the query compound katsumadain A (1) . In the first case (2 and 5), low ComboScore values are mainly caused by a lack of agreement between chemical interaction patterns, which is reflected by the low ScaledColor score and can be observed in the alignment depictions presented in Table 1. The shape alignment seems reasonable, while the chemical interaction patterns show obvious differences. In the second case (3, 4 and  6) , low ComboScore values are a result of partial matching. These candidate molecules cover only a fraction of katsumadain A. This fraction, however, includes the ligand core and, hence, all but one of the hydrogen-bond donor and acceptor features present in katsumadain A. As discussed in detail in ‘Computational studies on the putative binding Mode’ section, the alignment of the bioactive compounds with the query structure, as proposed by ROCS, does not n

Table 1. Compounds active on influenza neuraminidase identified by shape-focused screening (cont.).

actual ROCS screening process was run employing default settings and hit ranking was performed with ComboScore. The top-ranked compounds yielded ComboScore values of approximately 1.2. Considering the range of the ComboScore, this indicates a moderate degree of similarity between katsumadain A and the candidate molecules. This is expected, since katsumadain  A is a natural product and the NCI compound collection is not primarily focused on compounds from natural sources. Among the chemical structures derived from the shape-focused screening of the model compound katsumadain  A, we tried to select chemically diverse compounds for evaluation of their NA-inhibiting potential. Some synthetic candidates identified during the screening were out of stock, so that we finally obtained six natural product-like compounds, which were then checked for purity. Since one candidate could not fulfil the purity criteria (see Compound preparations), five compounds (2–6) were submitted to experimental evaluation. As illustrated in Table 1 and reflected by their fairly low ComboScore values (ComboScore 193.61§ >279.33 § 19.43 >180.83 § 21.78

PR/8/34

Jena/5528/09 Jena/5555/09

342/09

2.14 17.0 1.09 0.18 2.05 1.33

1.76 Not determined 0.39 0.23 3.54 2.04

4.55 Not determined 14.83 0.55 40.65 6.35

1.72 Not determined 0.98  0.30 2.01 2.21

† The CC50 were calculated from the mean dose–response curve of at least two independent assays each with three parallels per compound concentration. ‡ The 50% NA-inhibitory concentrations were calculated from the mean dose–response curve of at least three independent assays each with two parallels per compound concentration. § Maximum tested concentration; 100 µg/ml. CC50: 50% cytoxic concentration; MDCK: Madin–Darby canine kidney; NA: Neuraminidase.

necessarily represent the binding mode we expect at the target site. The reason for the screening success seems to be the remarkable strength of ROCS to identify related chemical patterns among molecules in a global and fuzzy manner, which can be particularly useful for finding moderately related bioactive compounds and for scaffold hopping. In this particular case, the method was able to recognize the ligand core of katsumadain A as a central feature for bioactivity and most of the identified entities share a related ligand core. „„NAI

Key Term Flavonoids: Common plant metabolites that show a broad bioactivity spectra. They have repeatedly been associated with anti-viral activity..

444

identified by virtual screening Representatives of the chemical class of caffeoylquinic acids are widely spread plant metabolites. Some of them have been reported to be endowed with antiviral activities [41,42] ; however, they have never been described before as NAIs. The selected candidate, 1,4-dicaffeoylquinic acid  (2) , has mainly been identified as a secondary plant metabolite (i.e., organic compounds that do not directly affect plant growth, development or reproduction but play important roles in protection against pests, coloring, as plant hormones and attractants) in species of the Asteraceae family, for example, in artichokes [43] , arnica flowers [44] or echinacea [45] . The synthetic 2H chromen derivative (6) has never been associated with any activity, but is closely related to the chalcon lonchocarpine isolated from different Leguminosae species, for example, from the genus Lonchocarpis. Further selected candidates (3–5) belong to the flavonoids (substituted phenyl benzopyranes). These widespread plant metabolites show reams of intriguing bioactivities that are of substantial importance for human healthcare Future Med. Chem. (2011) 3(4)

in terms of nutrition and herbal remedies. Many studies already reported the NA-inhibiting activities of different types of flavonoids [46–51] . Based on the NA-in vitro screening of 25 flavonoids, Liu et al. could demonstrate a decreasing NA potency from aurones, flavon(ol)es, isoflavones, flavanon(ol)es to flavan(ol)es [46] . They also stressed the importance of a 4´-OH and 7-OH group, the keto group in C4 and the C2–3 functionality. Derived from a recently performed quantitative structure–activity relationship study based on published NA-inhibiting activities of different flavonoids, Mercader et al. analyzed those 2D and 3D descriptors being of relevance for a NA-inhibiting activity [52] . They underline the importance of electric charges, masses and polarizabilities of the atoms present in the substituted phenyl-benzopyranes. A total of 18 polyphenols (chalcons, flavonoids and coumarines) – some of them substituted with one or two isoprene groups in position C6 or C6 and C3´– have been described recently from Glycyrrhiza uralensis [48] . Among all the tested isolates, the five-membered ring compounds demonstrated the most promising NA inhibition. The most active NAI was the C6 isoprenylated five-ring coumarin glycyrol. Our selection was independent of all these findings and exclusively considered the information of the shape-focused virtual screening. Based on these data, we selected three flavonoids for biological evaluation. Quercetin 5,7,3´,4´-tetramethyl ether (3) is a widely distributed plant secondary metabolite. After selection of our hit candidates, this compound was identified by another group as an anti-influenza constituent of elderberry, preventing H1N1 infection in vitro  [53] . In contrast to future science group

Novel neuraminidase inhibitors our computational approach, Roschek et al. performed a direct analysis in real time-TOF-MS ana­lysis hyphenated with a direct binding assay in the search for the antiviral principle in the complex natural mixture. The authors demonstrated a direct binding of compound 3 to H1N1 virus particles preventing the viruses for entering host cells, thus halting H1N1 infection in vitro. Artocarpin (4) is known as a constituent of different species of the genus Artocarpus (belonging to the mulberry family), for example Artocarpus heterophyllus [54] . This tree is native to Southeast Asia; however, it is cultivated throughout the tropics due to its edible fruits (Jackfruit). Artocarpin is a twice isoprenylated flavone (C3 and C6) and is not known for any anti-influenza or NA inhibiting activity, so far. Previously, Miki et  al. screened naturally occurring biflavonoids for the inhibitory activity against influenza NA [55] . They used the best NA-inhibiting biflavonoid, ginkgetin with an IC50 against H1N1 NA (A/PR/8/34) of 55 µg/ml and against H3N2 NA (A/Guizhou/54/89) of 9.8 µg/ml, to build sialic acid conjugates. Some of these caused significant survival effects in the influenza virus-infected mice. Our hit list contained 4´-O-methylochnaflavone (5) , which we selected for a putative NA-inhibiting candidate. It is a rare naturally occurring biflavonoid isolated from the genus Ochna [56] and from Japanese honeysuckle, Lonicera japonica  [57] , which previously showed a moderate HIV-integrase-inhibiting effect [58] . Whereas compound 5 is composed of an apigenin and a 4´methylluteolin moiety linked via the hydroxyl groups of the respective B-rings, the apigenin moiety in ginkgetin is connected in position 8 via a C–C linkage to the C3´ position of 4´,7-dimethylapigenin. „„Cytotoxicity

& NA inhibitory activity The results from cytotoxicity studies revealed a good compatibility of the test compounds (Table 2) . Moreover, all five natural compounds selected by virtual screening inhibited the NA

| Research Article

of influenza virus A/PR/8/34 at noncytotoxic concentrations in the micromolar dose range. The IC50 of four of them (3–6) was even lower than 10 µM. Their positive activity was further confirmed in NA-inhibitory tests with the two clinical H1N1v isolates, A/Jena/5528/09 and A/Jena/5555/09. The dose–response curves obtained with the three oseltamivir-sensitive viruses and compounds 3–6, as well as katsumadain A, compare well (Figure 1) . Artocarpin (4) exhibited the best activity. In comparison to the control compound katsumadain A, its IC50 was approximately tenfold lower (180–300  nM). This natural compound also showed the strongest activity against the oseltamivir-resistant clinical isolate A/342/09, whereas the IC50 was approximately two- to three-times higher in comparison to the drug-susceptible strains. In all tests, oseltamivir and zanamivir were included as additional controls. The IC50 for individual test viruses are summarized in Table 3. They confirm the NAI susceptibility of influenza viruses A/PR/8/34, A/Jena/5528/09 and A/Jena/5555/09 as well as the oseltamivir resistance of the isolate A/342/09. „„Computational

studies on the putative binding mode Considering topological aspects, one may be tempted to follow an alignment of katsumadain A and these inhibitors as exemplified in F igure  2A . The overall superposition seems sound in terms of shape; however, from a pharmacophoric perspective, one will recognize the apparent lack in chemical feature complementarity. Assuming that the heterocyclic ligand core is the key for enzyme inhibition, a crucial role of the heteroatoms present in this area can be anticipated. An adapted superposition that satisfies this pharmacophore idea was derived by guided superposition (Figure 2B) . Only a minor rotation of the katsumadain A molecule is sufficient to establish a superposition of the three hydrogen-bond acceptors of the ligand cores, whereas the overall alignment does not suffer

Table 3. Susceptibility of test virus against the control compounds oseltamivir and zanamivir. NAI

IC50 (nM) determined for influenza virus A PR/8/34

Oseltamivir carboxylate 0.23 Zanamivir 0.15

Jena/5528/09

Jena/5555/09

342/09

0.12 0.03

0.13 0.04

>30 0.39

The 50% neuraminidase-inhibitory concentrations were calculated from the mean dose–response curve of at least three independent assays each with two parallels per compound concentration. NA: Neuraminidase.

future science group

www.future-science.com

445

Research Article | Kirchmair, Rollinger, Liedl, Seidel, Krumbholz & Schmidtke from this manipulation. From the ligand-based perspective we, therefore, propose a pharmacophore including three hydrogen-bond acceptors and one aromatic feature in the ligand core, as well as one aromatic and several lipophilic features in the terminal regions of the ligands (Figure 2C) .

A

B

C

C D

Figure 2. Studies of the potential binding mode of identified neuraminidase inhibitors to neuraminidase. (A) Proposed, intuitive ligandbased alignment of the flavonole-derived inhibitors ( 3 turquoise, 4 grey and 5 orange) with katsumadain A (1, green). (B) Adapted ligand-based alignment based on the chemical feature-based pharmacophore concept. (C) Ligand-based common-feature pharmacophore derived from the adapted overlay. Hydrogenbond acceptors are indicated with red arrows, the aromatic area is indicated as a blue circle and the hydrophobic area is indicated in a brown circle. (D) Targetbased alignment of katsumadain A (1, green) and 4 (grey) in the active site of neuraminidase, derived by protein–ligand docking on a representative MD frame. The overlay of the ligand-derived pharmacophore with the structure-derived putative binding mode demonstrates the remarkable agreement of both binding models.

446

Future Med. Chem. (2011) 3(4)

In order to derive the putative ligand binding mode of the most promising inhibitor (4) from a structure-based point of view, we employed a docking protocol similar to the one reported for the ana­lysis of the binding mode of katsumadain A, where this compound was unlikely to be found accommodated by NA in a conformation observed in x-ray experiments with small druglike molecules. This was the reason for us to commence MD simulations in order to analyze the time-dependent behavior of the apo-crystal structure of NA (see ‘Methods’ section). Ten representative protein conformations were selected from the 20-ns trajectory-employing clustering techniques. Katsumadain A was subsequently docked to the ten protein conformations using GOLD. Docking results for one particular frame (the representative frame of cluster 4, which represents 23% of the collected MD frames and is related to the x-ray open conformation, yet showing an extended binding pocket) indicated a plausible binding mode for katsumadain A for several different reasons. All poses of katsumadain A for this particular frame are almost identical. The poses obtain the highest GOLDScore values and feature an extended hydrogen-bonding network and optimum surface complementarity. Moreover, the proposed binding mode is in agreement with experimentally observed structure–activity relationship on protein mutants. Katsumadain A shows a considerabe lower activity against the oseltamivir-resistant human H1NI isolate. According to the German Reference Laboratory for influenza viruses at the Robert Koch Institute Berlin, which provided us with the oseltamivirresistant virus as control for antiviral testing, it is known that this virus carries the H274T mutation that imparts oseltamivir resistance. The conformational shift of Glu-276 induced by this mutation causes a reduction of the binding site in this key interaction area of katsumadain A and, hence, the repulsion of the ligand. In a docking simulation using the mutated protein structure, we were unable to dock the ligand in a comparable docking pose into the binding pocket. For more details see elsewhere [28] . In the current work, ten iterative protein– ligand docking cycles were processed for 4 on the crystal structures 2hu0 (H5N1, open conformation, holo structure), 2hu4 (H5N1, closed conformation, holo structure) and 2hty (H5N1, open conformation, apo structure) [20] and each of the ten protein frames. The most likely binding mode was identified considering the GOLDSCORE ranking as well as the chemical feature and future science group

Novel neuraminidase inhibitors shape complementarity of the docking poses with respect to the protein environment. Another good indicator for the reliability of a predicted pose is the degree of geometric variation between the

| Research Article

docking poses of a ligand for a particular protein conformation. In case the generated ligand poses are consistent, the proposed binding geometry can be considered more likely.

A

B

Glu 276

245 loop

Arg 292 Arg 371 Arg 224

Tyr 406

Arg 118 430 loop

C

150 loop

D

Figure 3. Potential binding mode of artocarpin (4) to influenza neuraminidase. (A) Overlay of influenza neuraminidase monomers: open, apo crystal structure (red), closed holo crystal structure (blue), representative frame 3 of the molecular dynamics simulation (green) and 4 docked (carbons yellow, oxygens red). Frame 3 shows a more open conformation of the 245-loop, which facilitates the accommodation of the dihydroxyphenyl moiety. (B) 4 forms hydrogen bond interactions with Arg-118, Arg-224, Glu-276, Arg-292 and Arg-371, which are indicated with blue dashed lines. The lines not only indicate bonds to binding partners of this frame but also to the other protein structures, which illustrates, in part, the flexibility of the residues involved. Also, hydrophobic contacts with the side chain of Arg-224 and Tyr-406 seem to play an important role for affinity. (C) The overlay of 4 docked to frame 3 with the binding mode identified for katsumadain A (1, violet; see [28] for more detail) stresses the pharmacophoric relation of both inhibitors. Intriguingly, the structurally derived binding modes are in agreement with the pharmacophore model derived in the ligand-based approach, which is a further indication for the plausibility of the presented binding model. (D) The visualization of the protein surface of frame 3 (grey surface) and 4 (mesh) documents compatible shapes of these binding partners.

future science group

www.future-science.com

447

Research Article | Kirchmair, Rollinger, Liedl, Seidel, Krumbholz & Schmidtke The best overlay among the proposed docking solutions for 4 was found with frame 3 (Figure 3) , a medium-populated cluster of protein conformations, accounting for 7% of the frames collected during the production MD simulations. At the same time, docking poses obtain the maximum GOLDScore (56.58) among all poses of all frames. Besides, poses that are in accordance with the ones for frame 3 are also present with all other representative frames. Intriguingly, this binding mode could not be observed for the investigated x-ray structures. Considering the chemical feature and shape complementarity, we identified this binding mode as the most likely one. Strikingly, this binding mode is in agreement with the pharmacophore model derived earlier from the active compounds (Figure 2D) . Conclusion The shape-focused virtual screening engine ROCS was employed for the identification of novel inhibitors of influenza NA. All five compounds selected from the virtual screening hit list for biological evaluation turned out to exhibit significant inhibitory activity on the viral protein of oseltamivir-susceptible influenza A viruses. Four of the bioactive compounds are flavonoids – a class of widely distributed natural products that have been repeatedly associated with antiviral and NA-inhibiting activity. Artocarpin (4) is the most interesting compound identified in the current work, inhibiting the NA of oseltamivirsusceptible, as well as resistant H1N1 influenza A viruses at nanomolar concentrations. A ligand-derived pharmacophore model indicated the potential key role of three hydrogen-bond acceptors and hydrophobic/aromatic interactions in ligand binding. Independently from that, protein–ligand docking confirmed this initial model. The structure-based studies on the potential binding mode of flavonoids indicate, once more, that influenza NA is a

highly flexible protein and that the x-ray structures of this enzyme available today may not be adequate to explain the binding of bulky or rigid chemical entities, such as the ones identified in this work [22,23,27,59,60] . Future perspective Shape-focused virtual screening has proven to be a powerful approach for identifying novel synthetic and also natural product-like enzyme inhibitors. Combining this rapid and accurate ligand-based approach with a downstream, thorough docking algorithm, accounting for shape and flexibility of the ligand binding site, is expected to be particularly useful for identifying novel bioactive compounds. In future work we will be focusing on such a multistep approach to investigate synthetic compounds with inhibitory activity on influenza NA. Acknowledgement The authors thank Ulrike Grienke (Institute of Pharmacy/ Pharmacognosy, University of Innsbruck, Austria) for the TLC and LC–MS measurements of compounds 1–6, Birgit Jahn (Institute of Virology and Antiviral Therapy, Jena University Hospital, Germany) for technical assistance and Bernhard Randolf and Hannes Wallnöfer (Institute Institute of Theoretical Chemistry and Center for Molecular Biosciences, University of Innsbruck, Austria) for valuable advice concerning the molecular dynamics simulations and IT support.

Financial & competing interests disclosure The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert t­estimony, grants or patents received or pending, or royalties. No writing assistance was utilized in the production of this manuscript.

Executive summary „„

A total of five novel inhibitors of influenza neuraminidase (NA) have been identified in a successful virtual screening campaign employing shape-focused similarity metrics. Four of them are flavonoids, a class of plant metabolites that were repeatedly identified as exhibiting interesting antiviral and also NA-inhibiting activity.

„„

The most active compound identified in this work, artocarpin (4), inhibits the NA of oseltamivir-susceptible and also exhibits activity on oseltamivir-resistant H1N1 influenza A viruses at nanomolar concentrations.

„„

A ligand-based common-feature pharmacophore model indicates the central role of hydrogen bonding functionalities in ligand binding. Protein–ligand docking of the flavonoids on representative protein conformations selected from molecular dynamic simulations allowed us to derive a structure-based binding model, which confirms the prior developed ligand-based binding hypothesis.

„„

The insights gained from this modeling study and the structure–activity relationship data will be exploited for the development of novel inhibitors of influenza NA with improved resistance profiles.

448

Future Med. Chem. (2011) 3(4)

future science group

Novel neuraminidase inhibitors Bibliography

11

Papers of special note have been highlighted as: n of interest nn of considerable interest 1

2

nn

3

4

Webster RG, Walker EJ. Influenza – the world is teetering on the edge of a pandemic that could kill a large fraction of the human population. Am. Sci. 91, 122–129 (2003). Smith GJ, Vijaykrishna D, Bahl J et al. Origins and evolutionary genomics of the 2009 swine-origin H1N1 influenza A epidemic. Nature 459, 1122–1125 (2009).

historical perspectives and lessons learned. Antiviral Res. 71, 372–378 (2006). 14

Bauer K, Richter M, Wutzler P, Schmidtke M. Different neuraminidase inhibitor susceptibilities of human H1N1, H1N2, and H3N2 influenza A viruses isolated in Germany from 2001 to 2005/2006. Antiviral Res. 82, 34–41 (2009).

n

Investigates the dynamics of the loops involved in protein dynamics in more detail and points out its potential implications for drug development.

16

Moscona A. Global transmission of oseltamivir-resistant influenza. N. Engl. J. Med. 360, 953–956 (2009).

17

Weinstock DM, Zuccotti G. The evolution of influenza resistance and treatment. JAMA  Assoc. 301, 1066–1069 (2009).

18

Itoh Y, Shinya K, Kiso M et al. In vitro and in vivo characterization of new swine-origin H1N1 influenza viruses. Nature 460, 1021–1025 (2009).

26 Park JW, Jo WH. Computational design of

Investigates the characteristics of the 2009 H1N1 influenza A virus and its susceptibility to approved NAIs.

27 Cheng LS, Amaro RE, Xu D, Li WW,

Gives a comprehensive overview on influenza neuraminidase (NA) as a target in drug discovery.

7

von Itzstein M. The war against influenza: discovery and development of sialidase inhibitors. Nat. Rev. Drug Discov. 6, 967–974 (2007).

8

von Itzstein M, Wu WY, Kok GB et al. Rational design of potent sialidase-based inhibitors of influenza-virus replication. Nature 363, 418–423 (1993).

nn

McCammon JA. Characterizing loop dynamics and ligand recognition in human- and avian-type influenza neuraminidases via generalized born molecular dynamics and end-point free energy calculations. J. Am. Chem. Soc. 131, 4702–4709 (2009).

13 Hayden FG. Antivirals for influenza:

Das K, Aramini JM, Ma LC, Krug RM, Arnold E. Structures of influenza A proteins and insights into antiviral drug targets. Nat. Struct. Mol. Biol. 17, 530–538 (2010).

Gong JZ, Xu WF, Zhang J. Structure and functions of influenza virus neuraminidase. Curr. Med. Chem. 14, 113–122 (2007).

10

23 Amaro RE, Cheng XL, Ivanov I, Xu D,

15

6

9

inhibitors for influenza. N. Engl. J. Med. 353, 1363–1373 (2005).

Observes and analyzes wide-open conformations of the NA active site that have not been observed experimentally so far, employing molecular dynamics simulations.

Dawood FS, Jain S, Finelli L et al. Emergence of a novel swine-origin influenza A (H1N1) virus in humans. N. Engl. J. Med. 360, 2605–2615 (2009).

Neumann G, Noda T, Kawaoka Y. Emergence and pandemic potential of swine-origin H1N1 influenza virus. Nature 459, 931–939  (2009).

nn

nn

12 Moscona A. Drug therapy – neuraminidase

Provides a fascinating view on the emergence of the 2009 H1N1 influenza A epidemic.

5

n

Sidwell RW, Smee DF. Peramivir (BCX-1812, RWJ-270201): potential new therapy for influenza. Expert Opin. Invest. Drugs 11, 859–869 (2002).

| Research Article

Describes the development of the first potent NA inhibitor (NAI) zanamivir, and demonstrates the key role of computational methods for rational drug design. Goodford PJ. A computational procedure for determining energetically favorable binding sites on biologically important macromolecules. J. Med. Chem. 28, 849–857 (1985). Kim CU, Lew W, Williams MA. Influenza neuraminidase inhibitors possessing a novel hydrophobic interaction in the enzyme active site: design, synthesis, and structural ana­lysis of carbocyclic sialic acid analogues with potent anti-influenza activity. J. Am. Chem. Soc. 119, 681–690 (1997). Provides exciting insights to the design of the first orally bioavailable NAI, oseltamivir.

future science group

nn

19

Meijer A, Lackenby A, Hungnes O et al. Oseltamivir-resistant influenza virus A (H1N1), Europe, 2007–2008 season. Emerg. Infect. Dis. 15, 552–560 (2009).

Baz M, Abed Y, Simon P, Hamelin ME, Boivin G. Effect of the neuraminidase mutation H274Y conferring resistance to oseltamivir on the replicative capacity and virulence of old and recent human influenza A (H1N1) viruses. J. Infect. Dis. 201, 740–745 (2010).

20 Russell RJ, Haire LF, Stevens DJ et al. The

structure of H5N1 avian influenza neuraminidase suggests new opportunities for drug design. Nature 443, 45–49 (2006). n

21

Milestone in the development of antiinfluenza drugs, indicating novel opportunities to target influenza NA from x-ray structures. An JH, Lee DCW, Law AHY et al. A novel small-molecule inhibitor of the avian influenza H5N1 virus determined through computational screening against the neuraminidase. J. Med. Chem. 52, 2667–2672 (2009).

22 Amaro RE, Minh DDL, Cheng LS et al.

Remarkable loop flexibility in avian influenza N1 and its implications for antiviral drug design. J. Am. Chem. Soc. 129, 7764–7765 (2007).

www.future-science.com

24 Durrant JD, McCammon JA. Potential

drug-like inhibitors of group 1 influenza neuraminidase identified through computeraided drug design. Comput. Biol. Chem. 34, 97–105 (2010). 25 Li Y, Zhou BC, Wang RX. Rational design of

tamiflu derivatives targeting at the open conformation of neuraminidase subtype 1. J. Mol. Graph. Model. 28, 203–219 (2009). novel, high-affinity neuraminidase inhibitors for H5N1 avian influenza virus. Eur. J. Med. Chem. 45, 536–541 (2010). Arzberger PW, McCammon JA. Ensemblebased virtual screening reveals potential novel antiviral compounds for avian influenza neuraminidase. J. Med. Chem. 51, 3878–3894 (2008). 28 Grienke U, Schmidtke M, Kirchmair J et al.

Antiviral potential and molecular insight into neuraminidase inhibiting diarylheptanoids from Alpinia katsumadai. J. Med. Chem. 53, 778–786 (2010). 29 Milne GWA, Nicklaus MC, Driscoll JS,

Wang SM, Zaharevitz D. National cancer institute drug information-system 3D database. J. Chem. Inf. Comput. Sci. 34, 1219–1224 (1994). 30 Ghose AK, Pritchett A, Crippen GM. Atomic

physicochemical parameters for three dimensional structure directed quantitative structure-activity relationships III: modeling hydrophobic interactions. J. Comp. Chem. 9, 80–90 (1988). 31

Hawkins PCD, Skillman AG, Warren GL, Ellingson BA, Stahl MT. Conformer generation with OMEGA: algorithm and validation using high quality structures from the protein databank and cambridge structural database. J. Chem. Inf. Model. 50, 572–584 (2010).

449

Research Article | Kirchmair, Rollinger, Liedl, Seidel, Krumbholz & Schmidtke 32 Hawkins PCD, Skillman AG, Nicholls A.

43 Gil-Izquierdo A, Conesa MA, Ferreres F,

Comparison of shape-matching and docking as virtual screening tools. J. Med. Chem. 50, 74–82 (2007).

Gil MI. Influence of modified atmosphere packaging on quality, vitamin C and phenolic content of artichokes (Cynara scolymus l.). Eur. Food Res. Technol. 215, 21–27 (2002).

33 Case DA, Cheatham TE, Darden T et al. The

Amber biomolecular simulation programs. J. Comp. Chem. 26, 1668–1688 (2005).

35

Dolinsky TJ, Nielsen JE, McCammon JA, Baker NA. PDB2PQR: an automated pipeline for the setup of poisson-boltzmann electrostatics calculations. Nucleic Acids Res. 32, W665–W667 (2004).

hydroxycinnamoylquinic acids of Arnica flowers and burdock roots using a standardized LC-DAD–ESI-MS profiling method. J. Agric. Food Chem. 56(21), 10105–10114 (2008). 45

Du G-H. Anti-influenza virus activities of flavonoids from the medicinal plant Elsholtzia rugulosa. Planta Med. 74, 847–851 (2008). 47 Ryu YB, Curtis-Long MJ, Lee JW et al.

Structural characteristics of flavanones and flavones from Cudrania tricuspidata for neuraminidase inhibition. Bioorg. Med. Chem. Lett. 19, 4912–4915 (2009).

37 Shao JY, Tanner SW, Thompson N,

Cheatham TE. Clustering molecular dynamics trajectories: 1. Characterizing the performance of different clustering algorithms. J. Chem. Theory Comput. 3, 2312–2334 (2007).

48 Ryu YB, Kim JH, Park S-J et al. Inhibition of

neuraminidase activity by polyphenol compounds isolated from the roots of Glycyrrhiza uralensis. Bioorg. Med. Chem. Lett. 20, 971–974 (2010).

38 Jones G, Willett P, Glen RC, Leach AR,

Taylor  R. Development and validation of a genetic algorithm for flexible docking. J. Mol. Biol. 267, 727–748 (1997).

49 Jeong HJ, Ryu YB, Park S-J et al.

Neuraminidase inhibitory activities of flavonols isolated from Rhodiola rosea roots and their in vitro anti-influenza viral activities. Bioorg. Med. Chem. 17, 6816–6823 (2009).

39 Schmidtke M, Schnittler U, Jahn B,

Dahse HM, Stelzner A. A rapid assay for evaluation of antiviral activity against coxsackie virus B3, influenza virus A, and herpes simplex virus type 1. J. Virol. Methods 95, 133–143 (2001).

50 Nagai T, Miyaichi Y, Tomimori T, Suzuki Y,

Yamada H. Inhibition of influenza virus sialidase and anti-influenza virus activity by plant flavonoids. Chem. Pharm. Bull. 38, 1329–1332 (1990).

40 Kirchmair J, Distinto S, Markt P et al. How

to optimize shape-based virtual screening: choosing the right query and including chemical information. J. Chem. Inf. Model. 49, 678–692 (2009).

51

41 Hamauzu Y, Yasui H, Inno T, Kume C,

Omanyuda M. Phenolic profile, antioxidant property, and anti-influenza viral activity of chinese quince (Pseudocydonia sinensis schneid.), quince (Cydonia oblonga Mill.) and apple (Malus domestica Mill.) fruits. J. Agric. Food Chem. 53, 928–934 (2005).

Bauer R, Khan IA, Wagner H. TLC and HPLC ana­lysis of Echinacea pallida and E. Angustifolia roots. Planta Med. 54, 426– 430 (1988).

46 Liu A-L, Liu B, Qin H-L, Lee SM, Wang Y-T,

36 Humphrey W, Dalke A, Schulten K. VMD:

visual molecular dynamics. J. Mol. Graphics 14, 33–38, (1996).

Alberte RS. Elderberry flavonoids bind to and prevent H1N1 infection in vitro. Phytochemistry 70, 1255–1261 (2009). 54 Lin C.N, Lu C.M, Huang P-L. Flavonoids

from Artocarpus heterophyllus. Phytochemistry 39, 1447–1451 (1995).

44 Lin LZ, Harnly JM. Identification of

34 Dolinsky TJ, Czodrowski P, Li H et al.

PDB2PQR: expanding and upgrading automated preparation of biomolecular structures for molecular simulations. Nucleic Acids Res. 35, W522–W525 (2007).

53 Roschek B, Fink RC, McMichael MD, Li D,

52

Nagai T, Miyaichi Y, Tomimori T, Yamada H. Inhibition of mouse liver sialidase by plant flavonoids. Biochem. Biophys. Res. Commun. 163, 25–31 (1989). Mercader AG, Pomilio AB. QSAR study of flavonoids and biflavonoids as influenza H1N1 virus neuraminidase inhibitors. Eur. J. Med. Chem. 45, 1724–1730 (2010).

55

Miki K, Nagai T, Suzuki K et al. Antiinfluenza virus activity of biflavonoids. Bioorg. Med. Chem. Lett. 17, 772–775 (2007).

56 Okigawa M, Kawano N, Aqil M, Rahman W.

Ochnaflavone and its derivatives: a new series of diflavonyl ethers from Ochna squarrosa linn. J. Chem. Soc. Perkin Trans. 1, 580–583 (1976). 57 Son KH, Park JO, Chung KC et al.

Flavonoids from the aerial parts of Lonicera japonica. Arch. Pharmacal. Res. 15, 365–370 (1992). 58 Hong H, Neamati N, Winslow HE et al.

Identification of HIV-1 integrase inhibitors based on a four-point pharmacophore. Antiviral Chem. Chemother. 9, 461–472 (1998). 59 Landon MR, Amaro RE, Baron, R et al.

Novel druggable hot spots in avian influenza neuraminidase H5N1 revealed by computational solvent mapping of a reduced and representative receptor ensemble. Chem. Biol. Drug Des. 71, 106–116 (2008) 60 Durrant JD, McCammon JA. Potential

drug-like inhibitors of group 1 influenza neuraminidase identified through computeraided drug design. Comput. Biol. Chem. 34, 97–105 (2010). „„Website 101 WHO. Influenza (seasonal) fact sheet N°211

www.who.int/mediacentre/factsheets/ fs211/en Accessed on 03 November 2010. 102 NCI website

http://dtp.nci.nih.gov Accessed 24 August 2009.

42 Shi, S, Huang K, Zhang Y, Zhao Y, Du Q.

Purification and identification of antiviral components from laggera pterodonta by high-speed counter-current chromatography. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci. 859, 119–124 (2007).

450

Future Med. Chem. (2011) 3(4)

future science group