Integrated in Silico Tools for Exploiting the Natural Products’ Bioactivity

• Abstract
• Introduction
• Methods to Access the NPs’ Bioactivity
• Virtual Screening Concepts
• Structure-based approaches
• Ligand-based approaches
• Neural networks
• Filtering Experiments - Virtual Libraries
• Integrated Approaches
• Limits and Expectations
• Conclusions
• Acknowledgements
• References

Abstract

Whereas computational methods for molecular design are well established in medicinal chemistry research, their application in the field of natural products is still not exhaustively explored. This article gives a short introduction into both the potential for the application of computer-assisted approaches, such as pharmacophore modelling, virtual screening, docking, and neural networking to efficiently access the bioactive metabolites, and the requirements and limitations related to this specific field. The challenge is which selection criteria and/or multiple filtering tools to apply for a target-oriented isolation of potentially bioactive secondary metabolites. Application examples are provided where in silico tools and classical methods used by natural product scientists are used in an effort to maximize their efficacy in drug discovery. Thus, integrated computer-assisted strategies may help to process the huge amount of available structural and biological information in a reasonably short time for a straightforward search of bioactive natural products.

Introduction

‘Diseases are as old as life is which is exposed to physical, chemical, biological, and psychic stimuli’ [1]. Humankind was accordingly engaged from earliest times in attempts to find reliable remedies which were exclusively found in natural products (NPs) derived from plants, animals and minerals. The 19^th and 20^th century development of potent synthetic drug substances stimulated the chemists’ research and offer us today a vast palette of synthetic drugs. Particularly in industrialized societies, it is justified to ask about the present day impact of naturally derived remedies, as either phytotherapeutics or single molecular entities. The answer is simple: statistics show that NPs are still the most favoured source of new drugs for clinical use. This was demonstrated recently by Newman and co-authors, who circumstantiate that in the total drug launches of small molecules from 1981 to 2002 only 39 % can be classified as truly synthetic in origin [2], [3]. The reason behind is that the NPs’ diversity in chemical space is highly superior to that exhibited by compounds originating from combinatorial chemistry. NPs show a wide range of pharmacological activities because their biosynthesis is a result of diverse stimuli from numerous enzymes [4].

Despite the great impact of naturally derived remedies, major pharmaceutical companies scaled down their engagement in NPs’ drug discovery. Obviously, it became a commercially unprofitable endeavour: the proverbial search for the needle in the haystack. In the last 15 years high-throughput screening and combinatorial chemistry were supposed to be promising strategies to accelerate the search for new drug leads; however, results were disappointing [5]. Huge quantities of data generated by these methods together with the bulk of already published data about structural and biological information of pharmacological targets and their ligands require computational (i. e., in silico) approaches to handle the almost overwhelming amount of data in a reasonably short time. This ‘data-mining’ permits us to extract knowledge from a large set of data in order to make predictions of new events [6]. This is of crucial importance in the field of drug discovery from nature, because (i) NPs have definitely a great impact as drug substances because of their astonishing chemical and biological diversity, (ii) they represent an almost infinite source of chemical entities, and (iii) millions of already known data (biological and structural information available from the literature) can be used to extract knowledge by applying computational methods.

This report aims to give a short overview of different in silico techniques and integrated computational approaches dedicated to a rationalized drug discovery process of NPs.

Methods to Access the NPs’ Bioactivity

In principle, there are two completely diverse ways to discover a NPs’ bioactivity. A simplified scheme is shown in Fig. [1]. When starting from a holistic level, a whole extract of a plant, fungus, micro-organism etc. is biologically investigated, usually by considering potential selection criteria for a more focused process, e. g., hints from folk medicine, bio-rational or phylogenetic criteria. In the field of plants, it has been estimated that only 5 to 15 % of the approximately 250,000 described higher plant species have ever been tested for some type of biological activity [2]. In order to ascertain a proposed or assumed pharmacological effect, target validation accompanied by the choice of a suitable bioassay (binding assay or cellular test) is a major challenge in the early drug discovery process and indispensable before proceeding with a bioguided characterization of the extract under investigation. In this process active fractions are further separated and retested in an iterative way until the active ingredients are isolated.

When searching for active NPs on a molecular base, a set of reliable structural information is mandatory to start with. Up to today, approximately 170,000 NPs are known and chemically identified [7]. Although only a minority of secondary metabolites has ever been tested on any target, information on activities (hits and non-hits) is continuously increasing and is all-too often lying idle.

The correlation between a compound’s biological activity and its structure is the basis for deriving a structure-activity relationship (SAR). By a set of such data pairs, the human brain is trained to find the inherent relationship and thus to obtain knowledge on the structural requirements for an increased target affinity and/or selectivity. For more diverse and manifold datasets the human brain is overextended and asks for computational help. This may be coped by using an artificial neural network to derive a quantitative SAR (QSAR) and to predict structural features for an activity from a large database.

By spatial arrangement of essential features of a set of diverse ligands (hits), a ligand-based pharmacophore model is generated, which is not only a helpful tool for rationalizing a QSAR, but also for the virtual screening of large databases.

Our basic structural knowledge on pharmacological targets is continuously increasing thanks to accelerating technological achievements in genomics and structure elucidation. 3D structures are useful in defining topographies of the complementary surfaces of ligands and can thus be exploited to identify virtual hits by docking the NPs’ conformers into the binding site. The best requirement for any predictive tool is, however, provided by the 3D structure of a ligand-target complex, where information is available for both the ligand’s binding conformation and the exact binding site of the target. Adapted from this information, the structure-based pharmacophore model is an efficient filtering tool for in silico screening. More detailed information about these data-mining tools will be offered in the following section.

Virtual Screening Concepts

In the early drug discovery process, virtual screening technologies have largely enhanced the impact of computational chemistry and nowadays chemoinformatics plays a predominant role in drug research [8]. The goal of applying such methods is to mine more or less large compound databases in silico and to select a limited number of candidates for their potential of having the desired pharmacological effect. The pharmacophore concept has proven to be extremely successful in tackling this task. The prior use of pharmacophore models in the biological screening of NPs is an efficient procedure since it eliminates quickly molecules that do not possess the required features, thus leading to a dramatic enrichment when compared to a purely random screening experiment. In this way, less experiments direct the search for bioactive metabolites towards a more focused experiment.

By definition, a pharmacophore model represents the spatial arrangement of chemical groups or so-called ‘features’ in a molecule that are known or thought to determine its activity. The ‘features’ are located relative to each other in coordinate space as points surrounded by a sphere of tolerance. This represents the region in space that should be occupied by a certain chemical functionality capable of the feature’s kind of interaction, which is based on electrostatic interactions, H-bonding or hydrophobic interactions. In organic molecules, different structural motifs can express the similar chemical behaviour and therefore the same biological effect. In silico techniques are roughly divided into two groups: structure-based approaches and ligand-based approaches [9], [10].

Structure-based approaches

In this group knowledge about the target’s 3D structure and its binding site is essential to perform a protein-ligand high throughput docking, which turned out to be a valuable structure-based virtual screening method [11], [12]. For the generation of a structure-based 3D pharmacophore model the conformation of the ligand in the protein’s binding site is additionally required. This information is provided by numerous co-crystal structures that are stored in the Brookhaven Protein Data Bank (PDB) [13]. A new software tool has recently been described for the successful generation of such chemical feature-based models. The software LigandScout [14], [15] is a programme for ligand interpretation and data mining in the PDB. The performance of this programme allows an automatic detection and classification of protein-ligand interactions into hydrogen bonds, charge transfer, and lipophilic regions. The information implemented in a structure-based pharmacophore pattern (i. e., 3D coordinates of interaction points) is ideally used as data-mining tool [16].

The structure-based approach benefits not only from the accurate geometry of ligand-target interactions by XRA or NMR experiments but, in turn, also from the disclosed ligand’s conformation relevant for the activity. Successful application examples within NPs have recently been published in the area of acetylcholinesterase inhibitory plant constituents [17] and X-linked inhibitors of apoptosis (XIAP) [18]. In the latter study, a high-resolution 3D structure of the XIAP BIR3 domain complexed with the N-terminal end of the Smac/Diablo protein [19], an endogenous ligand of the respective XIAP binding pocket, was used as starting point to virtually screen an in-house 3D-NP database. Embelin from the Japanese Ardisia herb emerged as virtual small molecule weight hit, which was found to be a fairly potent inhibitor of XIAP using a fluorescence polarization binding assay. In Fig. [2], embelin is shown in an energetically preferred conformation docked into the well-defined binding pocket of the BIR3 domain (AutoDock program [20]).

Ligand-based approaches

In many cases no experimental information on the target’s protein structure has become available, not to mention the biological conformation of the ligand. Nevertheless, several lead discovery projects have already reached a well-advanced stage, because the predominant source of knowledge about activities is still the hit or lead information, which can be sufficient to create a ligand-based pharmacophore model. Thereby, the underlying hypothesis is purely obtained from ligand information using algorithms aimed at defining the common ensemble of steric and electrostatic features of different compounds which are necessary for their interaction with a specific biological target [21]. This approach may provide essential information for the early selection of promising drug candidates, allowing the perception and understanding of key interactions between a receptor and a ligand on a generalized level. Several successful applications within this subject have been performed using the Catalyst programme [22], one of the leading software packages in chemical feature-based pharmacophore modelling. A study recently published has demonstrated the power of these methods applied to pharmacophore modelling of sigma-1 ligands [23]. Therein, some reliable pharmacophore models could be extracted solely from ligand information. In Fig. [3] A, potent sigma-1 receptor ligands are structurally aligned to derive distinct common features. Their 3D arrangement in combination with a spatial restriction (not shown in Fig. [3]) was then used for the generation of a pharmacophore model, which was able to retrieve compounds with high affinity values, among them also NPs, like solanidine (Fig. [3] B).

Neural networks

Artificial neural network simulations are based on collections of mathematical models that are interconnected and organized in different layers. They are analogous to an adaptive human learning process and usually trained with learning sets applying one or more molecular descriptors in order to form clusters that enable us to distinguish between different objects and their properties. The resulting models are then applied to make predictions on test sets, until the validated models may be used to derive a QSAR of chemically related structures or to mine larger datasets. One may distinguish between supervised and unsupervised learning methods as discussed in detail by Zupan and Gasteiger [6], [24].

Neural network models are powerful data-mining tools to unravel the complexity of both the relationships between structural information and biological activity, and the high amount of biological information already published for thousands of compounds. This is especially true in the field of NPs, where enormous amounts of data have to be processed and at the same time complex relationships can be studied. In 1982, a self-organizing neural network was introduced by Kohonen [25] where objects from a multidimensional space are projected into a lower-dimensional one. This is usually a 2D plane representing the conserved similarity relationship of the objects. Recently an application example within the field of NPs was published by Wagner et al. [26]. The authors used a dataset of 103 structurally diverse sesquiterpene lactones and their NF-κB inhibitory activity to derive an SAR. By the application of multiple 3D structure representations as descriptors, a single model was achieved which provided detailed information on the structural influence of the investigated biological activity.

A comprehensive survey for computer-aided molecular selection of NPs applying integrated computational 2D or 3D QSAR methods is described by Bernard and co-authors [27].

Filtering Experiments - Virtual Libraries

The advent of structure databases has provided a basis for the development and feasibility of automatic methods in the search for new lead structures. Conceptually, all the virtual screening concepts presented above have their origins in synthetic chemistry. Their application, however, is just as well adaptable to NPs’ chemistry. Prior to the in silico filtering experiment, a 3D structure database requires an efficient generation of reasonable, energetically minimized conformations assumed to meet approximately those conformations that might be of biological relevance [28]. The underlying algorithms for 3D structure generation and conformation analysis are implemented in commercial software tools, e. g., in CORINA [6] or the Catalyst programme [22].

In the field of NPs only few commercial 3D databases are available, e. g., the Traditional Chinese Medicinal Database (TCMD) [29] or the Dictionary of Natural Product Database (DNP) launched by Chapman & Hall [7]. Moreover, a number of non-commercial in-house created databases have been used by different groups for their virtual screening studies on NPs, e. g., a marine natural product database (MNDP) [30], a database based on the ‘de materia medica’ of Pedanius Dioscurides (DIOS) [31] or a natural product database (NPD) [17], [31], [32]. The library of the National Cancer Institute (NCI) contains more than half a million compounds from both synthetic and natural origin that have been collected and tested by the NCI since 1955. About half of the synthetic compounds, which represent the large majority of the samples, may be used for free and are thus in the public domain (”Open NCI Database”) [33]. An interesting property prediction approach to the more than 250,000 compounds contained in this open database was provided by Poroikov and co-authors [34]. By use of the programme PASS (prediction of activity spectra for substances) an in silico tool for complex searches of 565 different types of activities is provided, e. g., in the case of antineoplastic effects, the authors could demonstrate a substantial data set enrichment over random selection by the use of PASS-predicted probabilities.

Integrated Approaches

The search for bioactive NPs is a complex and multi-disciplinary challenge. Despite the power of described in silico filtering tools (indication for putative ligands, enrichment of the dataset) the results may be too vague for an NP researcher, because the disposability of a metabolite for pharmacological testing is usually a procedure consuming much effort and time. Thus, the computer-aided molecular selection is best combined with further discovery methods, labelled as integrated or in combo approaches, for increasing the probability of discovering a hit. A set of different in silico methodologies was previously applied by Cherkasov and co-authors [35] to aid in the discovery of non-steroidal ligands for human sex hormone binding globulin from natural sources. Therein, a rigorously cross-validated neural network-based QSAR model identified 105 prospective compounds from a structure collection of 23,836 commercial natural substances. This stringent QSAR ranking was combined with docking studies and a pharmacophore-aided database search. The integrated computational methods resulted in a convincing predictive tool which identified a set of 29 structurally diverse NPs, of which every fourth compound was able to inhibit the target protein in a micromolar range.

Some strategies which combine in silico tools and classical methods for activity exploitation are schematized in Fig. [4] and examples will be outlined here: As soon as a sensitive data-mining tool has been developed and has proved itself by more or less selectively finding the active compounds within a test set, it can be applied for screening a 3D multi-conformational database (Fig. [4] A). The subsequent procedure consists of the evaluation of the virtual hits considering physico-chemical properties, toxicity and pharmacokinetics. In this stage additional virtual filtering tools for the profiling of ADME parameters [36] might have an invaluable impact on aiding a refined selection of compounds. Then, a sensible choice of natural materials known to contain the focused metabolites and worth investigating in detail is a crucial step which requires a comprehensive study in the literature considering the hit(s) content in the natural source, its availability and maybe hints from ethnopharmacology. As further selection criteria, it is advisable to perform a preliminary assay with those crude extracts and fractions assumed to contain the promising metabolite(s). Those fractions that scored well are then subjected to a bioguided process. This strategy was successfully embarked in the search for cyclooxygenase inhibiting metabolites from Morus alba [32].

A different integrated procedure is schematized in Fig. [4] B. Applying this approach, the selection of the natural material is not guided by virtual prediction, but a number of extracts is roughly screened with a bioassay to identify the active ones. A similar strategy is to collect information about the traditional application of natural preparations in the field of the focused pharmacological target. A 3D database is then generated consisting of all the metabolites known from the literature to be included in that/those extract(s) that came off well or in the traditionally used preparation, respectively. The resulting focused database is virtually screened on an established pharmacophore model of the aiming target. The putative hits may then be identified by modern analytical tools like LC-MS or LC-NMR to isolate them in a target-oriented way for pharmacological testing. This strategy is especially helpful for intricate pharmacological assays, which would turn a bioguided fractionation into an unrealistic endeavour.

Such a combined approach to rationalize a phytochemical lead discovery was performed by Bernard and co-workers [37]. Starting with an in vitro screening of traditionally used anti-inflammatory plant extracts on phospholipase A₂, a focused structural database was generated and virtually screened on a established ligand-based pharmacophore model for human non-pancreatic phospholipase A₂ [38]. The combination of experimental data with database exploitation and molecular modelling resulted in the efficient identification of betulin and betulinic acid as extract ingredients with anti-phospholipase A₂ in vitro effects.

A similar strategy was performed for a rationalized discovery of non-alkaloid AChE inhibitors from chicory root in our group [39]. Within a database of 47 known metabolites from Cichorium intybus, the pharmacophore modelling and docking experiments revealed five sesquiterpene lactones as prospective anticholinesterase structures. Experimental data confirmed two of them as active AChE inhibiting plant constituents.

Limits and Expectations

• As can be seen in Fig. [1], all the computational approaches are based on structural biology, either of the ligand, the target or both. Thus, applicants of these techniques will be completely lost if the structural and biomolecular knowledge is not reliable enough. In this sense it is the basic research engaged with conscientious structure elucidation of target proteins and secondary metabolites that at all enables a sensible and appropriate use of in silico techniques.

• The compilation of today’s known secondary metabolites is far from being complete and will never reach an approximate completeness because of the infinite source of nature. The NP databases used for virtual screens are, however, limited to known structures and are accordingly restricted to already determined and published entities, in contrast to truly virtual libraries produced by computational design.

• For the development of a fully functional neural network it is of pivotal importance to feed and train the net not only with hits, but as well with non-hits. Unfortunately, these are all too often not published.

• The significance of virtual predictions has to be subjected to an impartial opinion. Thus, the validation of a generated model is one of the crucial steps. Even after the proof of concept, one has to be aware that in silico techniques are just filtering tools that at best might enrich the pool with potential candidates and give an indication of the ligand-target interaction.

• Beside serendipitous findings, in silico results may just be as accurate as the information that was initially fed into the computer. To avoid a distorted view on virtually obtained predictions, special caution will be demanded if two or more highly active structures are used to generate a ligand-based hypothesis without being aware of their unique binding mechanism.

• From the target’s side it is advantageous to select proteins with at least partly hydrophobic and cavity-shaped binding sites to ensure distinct overall interactions with a potential ligand. In contrast, cell surface receptors and large protein-protein interactions with relatively flat and sprawled binding domains usually provide a risky and unreliable hypothesis and in turn shady virtual hits.

Conclusions

There is a constant need for new drug leads in the battle against a number of difficult to treat diseases. Thanks to advances in genomics and structural biology, our cumulative knowledge of potential therapeutic targets for exploration continues to grow. On the other hand, there is a steady rise in potential small molecular weight ligands for these targets, based on the power of combinatorial chemistry, advances in synthetic chemistry and - of paramount importance in the field of NPs - the continuous structure elucidation of new chemical entities from the planet’s untapped molecular diversity. Because NPs have been the single most productive source of drug leads, much effort is spent on the discovery of their bioactivities. To process the huge amount of structural and biological data from the pool of both ligands and targets, in silico techniques such as virtual screening and neural network analyses can have a substantial impact on the discovery success rate. These techniques, however, must not be used exclusively as activity-predicting tools, since the results provide merely an indication for a putative activity: it is mainly by the creation of interfaces between computational tools and well-established methods from NP scientists, e. g., bioguided fractionation, on-line analytical activity profiling, ethnopharmacological screening, that a reasonable standard of success can be achieved. Early results indicate that these integrated computational approaches are efficient and trendsetting strategies to exploit nature’s multitude of bioactivities, and thus have the potential to accelerate the drug discovery process of NPs.

Acknowledgements

The authors thank Dr. Christian Laggner (Institute of Pharmacy/Computer Aided Molecular Design Group, University of Innsbruck, Austria) for providing Fig. [3].

References

• 1 Schmitz R. Geschichte der Pharmazie. Vol. 1, Von den Anfängen bis zum Ausgang des Mittelalters. Eschborn; Govi-Verlag 1998: p 3
  Search in Google Scholar
• 2 Cragg G M, Newman D J. Biodiversity: A continuing source of novel drug leads.  Pure Appl Chem. 2005;  77 7-24
  CrossrefPubMedSearch in Google Scholar
• 3 Newman D J, Cragg G M, Snader K M. Natural products as sources of new drugs over the period 1981 - 2002.  J Nat Prod. 2003;  66 1022-37
  CrossrefPubMedSearch in Google Scholar
• 4 Hadacek F. Secondary metabolites as plant traits: current assessment and future perspectives.  CRC Crit Rev Plant Sci. 2002;  21 273-322
  PubMedSearch in Google Scholar
• 5 Lahana R. How many leads from HTS?.  Drug Discov Today. 1999;  4 447-8
  CrossrefPubMedSearch in Google Scholar
• 6 Gasteiger J, Teckentrup A, Terfloth L, Spycher S. Neural networks as data mining tools in drug design.  J Phys Org Chem. 2003;  16 232-45
  CrossrefPubMedSearch in Google Scholar
• 7 Chapman & Hall/CRC Press L LC. Dictionary of natural products. Available at http://www.chemnetbase.com. Accessed February 2006
  Search in Google Scholar
• 8 Xu J, Hagler A. Chemoinformatics and drug discovery.  Molecules. 2002;  7 566-600
  PubMedSearch in Google Scholar
• 9 Güner O F. Pharmacophore perception, development, and use in drug design. IUL Biotechnology Series. La Jolla; International University Line 2000
  Search in Google Scholar
• 10 Böhm H -J, Schneider G, Kubinyi H, Mannhold R, Timmerman H. Virtual screening for bioactive molecules. Vol. 10, In: Mannhold R, Kubinyi H, Timmermann H, editors Methods and principles in medicinal chemistry. New York; Wiley 2000
  Search in Google Scholar
• 11 Abagyan R, Totrov M. High-throughput docking for lead generation.  Curr Opin Chem Biol. 2001;  5 375-82
  CrossrefPubMedSearch in Google Scholar
• 12 Schneider G, Böhm H J. Virtual screening and fast automated docking methods.  Drug Discov Today. 2002;  7 64-70
  PubMedSearch in Google Scholar
• 13 Berman H, Westbrook J, Feng Z, Gilliland G, Bhat T, Weissig H. et al . The protein data bank.  Nucleic Acids Res. 2000;  28 235-42
  CrossrefPubMedSearch in Google Scholar
• 14 Wolber G, Langer T. LigandScout: 3D pharmacophores derived from protein-bound ligands and their use as virtual screening filters.  J Chem Inf Model. 2005;  45 160-9
  CrossrefPubMedSearch in Google Scholar
• 15 Inte:Ligand G mbH. Vienna, Austria. LigandScout: 3D pharmacophore modelling tool. Available at http://www.inteligand.com. Accessed February 2006
  Search in Google Scholar
• 16 Congreve M, Murray C W, Blundell T L. Structural biology and drug discovery.  Drug Discov Today. 2005;  10 895-907
  CrossrefPubMedSearch in Google Scholar
• 17 Rollinger J M, Hornick A, Langer T, Stuppner H, Prast H. Acetylcholinesterase inhibitory activity of scopolin and scopoletin discovered by virtual screening of natural products.  J Med Chem. 2004;  47 6248-54
  CrossrefPubMedSearch in Google Scholar
• 18 Nikolovska-Coleska Z, Xu L, Hu Z, Tomita Y, Li P, Roller P P. et al . Discovery of embelin as a cell-permeable, small-molecular weight inhibitor of XIAP through structure-based computational screening of a traditional herbal medicine three-dimensional structure database.  J Med Chem. 2004;  47 2430-40
  CrossrefPubMedSearch in Google Scholar
• 19 Wu G, Chai J, Suber T L, Wu J W, Du C, Wang X. et al . Structural basis of IAP recognition by Smac/DIABLO.  Nature. 2000;  408 1008-12
  PubMedSearch in Google Scholar
• 20 Morris G M, Goodsell D S, Huey R, Olson A J. Distributed automated docking of flexible ligands to proteins: parallel applications of AutoDock 2.4  J Comput Aided Mol Des. 1996;  10 293-304
  CrossrefPubMedSearch in Google Scholar
• 21 Langer T, Krovat E M. Chemical feature-based pharmacophores and virtual library screening for discovery of new leads.  Curr Opin Drug Discov Dev. 2003;  6 370-6
  PubMedSearch in Google Scholar
• 22 Accelrys I nc., San D iego. CA, USA. Catalyst: programme for modelling, database management, and querying for drug discovery. Available at http://www.accelrys.com. Accessed January 2006
  Search in Google Scholar
• 23 Laggner C, Schieferer C, Fiechtner B, Poles G, Hoffmann R D, Glossmann H. et al . Feature based pharmacophore models for sigma1 receptor, ERG2 and EBP.  J Med Chem. 2005;  48 4754 -64
  CrossrefPubMedSearch in Google Scholar
• 24 Zupan J, Gasteiger J. Neural networks in chemistry and drug design, 2nd edition. Weinheim; Wiley-VCH 1999
  Search in Google Scholar
• 25 Kohonen T. Self-organized formation of topologically correct feature maps.  Biol Cybern. 1982;  43 59-69
  CrossrefPubMedSearch in Google Scholar
• 26 Wagner S, Hofmann A, Siedle B, Terfloth L, Merfort I, Gasteiger J. Development of a structural model for NF-κB inhibition of sesquiterpene lactones using self-organizing neural networks.  J Med Chem. 2006;  49 241-52
  PubMedSearch in Google Scholar
• 27 Bernard P, Berton J Y, Chrétien J R. Computer-aided molecular selection and design of natural bioactive molecules.  Curr Opin Drug Discov Dev. 1999;  2 213-23
  PubMedSearch in Google Scholar
• 28 Kirchmair J, Laggner C, Wolber G, Langer T. Comparative analysis of protein-bound ligand conformations with respect to catalyst's conformational space subsampling algorithms.  J Chem Inf Model. 2005;  45 422-30
  CrossrefPubMedSearch in Google Scholar
• 29 Lu A, Liu B, Liu H, Zhou J, Xie G. A traditional Chinese medicine plant-compound database aid its application for searching.  Int Electron J Mol Des. 2004;  3 672-83
  PubMedSearch in Google Scholar
• 30 Lei J, Zhou J. A marine natural product database.  J Chem Inf Comput Sci. 2002;  42 742-8
  CrossrefPubMedSearch in Google Scholar
• 31 Rollinger J M, Haupt S, Stuppner H, Langer T. Combining ethnopharmacology and virtual screening for lead structure discovery: COX-inhibitors as application example.  J Chem Inf Comput Sci. 2004;  44 480-8
  CrossrefPubMedSearch in Google Scholar
• 32 Rollinger J M, Bodensieck A, Seger C, Ellmerer E P, Bauer R, Langer T. et al . Discovering COX-inhibiting constituents of Morus root bark: activity-guided versus computer-aided methods.  Planta Med. 2005;  71 399-405
  Thieme ConnectPubMedSearch in Google Scholar
• 33 Development Therapeutics Program N CI/NIH. Open NCI database. Available at http://dtp.nci.nih.gov.accesdistant.sorbonne-universite.fr/webdata.html. Accessed in March 2006
  Search in Google Scholar
• 34 Poroikov V V, Filimonov D M, Ihlenfeldt W -D, Gloriozova T A, Lagunin A A, Borodina Y V. et al . PASS biological activity predictions in the enhanced open NCI database browser.  J Chem Inf Comput Sci. 2003;  43 228-36
  CrossrefPubMedSearch in Google Scholar
• 35 Cherkasov A, Shi Z, Fallahi M, Hammond G L. Successful in silico discovery of novel nonsteroidal ligands for human sex hormone binding globulin.  J Med Chem. 2005;  48 3203-13
  CrossrefPubMedSearch in Google Scholar
• 36 Van de Waterbeemd H. High-throughput and in silico techniques in drug metabolism and pharmacokinetics.  Curr Opin Drug Discov Dev. 2002;  5 33-43
  PubMedSearch in Google Scholar
• 37 Bernard P, Scior T, Didier B, Hibert M, Berthon J -Y. Ethnopharmacology and bioinformatic combination for leads discovery: application to phospholipase A2 inhibitors.  Phytochemistry. 2001;  58 865-74
  CrossrefPubMedSearch in Google Scholar
• 38 Bernard P, Pintore M, Berthon J -Y, Chretien J R. A molecular modeling and 3D QSAR study of a large series of indole inhibitors of human non-pancreatic secretory phospholipase A₂ .  Eur J Med Chem. 2001;  36 1-19
  CrossrefPubMedSearch in Google Scholar
• 39 Rollinger J M, Mock P, Zidorn C, Ellmerer E P, Langer T, Stuppner H. Application of the in combo screening approach for the discovery of non-alkaloid acetylcholinesterase inhibitors from Cichorium intybus .  Curr Drug Discov Technol. 2005;  2 185-93. Corrections: Curr Drug Discov Technol 2006; in press
  CrossrefPubMedSearch in Google Scholar

Mag. pharm. Dr. Judith Maria Rollinger

Institute of Pharmacy/Pharmacognosy

Leopold-Franzens-Universität Innsbruck

Innrain 52c

Josef-Moeller Haus

6020 Innsbruck

Austria

Phone: +43-512-507-5308

Fax: +43-512-507-2939

Email: judith.rollinger@uibk.ac.at

References

• 1 Schmitz R. Geschichte der Pharmazie. Vol. 1, Von den Anfängen bis zum Ausgang des Mittelalters. Eschborn; Govi-Verlag 1998: p 3
  Search in Google Scholar
• 2 Cragg G M, Newman D J. Biodiversity: A continuing source of novel drug leads.  Pure Appl Chem. 2005;  77 7-24
  CrossrefPubMedSearch in Google Scholar
• 3 Newman D J, Cragg G M, Snader K M. Natural products as sources of new drugs over the period 1981 - 2002.  J Nat Prod. 2003;  66 1022-37
  CrossrefPubMedSearch in Google Scholar
• 4 Hadacek F. Secondary metabolites as plant traits: current assessment and future perspectives.  CRC Crit Rev Plant Sci. 2002;  21 273-322
  PubMedSearch in Google Scholar
• 5 Lahana R. How many leads from HTS?.  Drug Discov Today. 1999;  4 447-8
  CrossrefPubMedSearch in Google Scholar
• 6 Gasteiger J, Teckentrup A, Terfloth L, Spycher S. Neural networks as data mining tools in drug design.  J Phys Org Chem. 2003;  16 232-45
  CrossrefPubMedSearch in Google Scholar
• 7 Chapman & Hall/CRC Press L LC. Dictionary of natural products. Available at http://www.chemnetbase.com. Accessed February 2006
  Search in Google Scholar
• 8 Xu J, Hagler A. Chemoinformatics and drug discovery.  Molecules. 2002;  7 566-600
  PubMedSearch in Google Scholar
• 9 Güner O F. Pharmacophore perception, development, and use in drug design. IUL Biotechnology Series. La Jolla; International University Line 2000
  Search in Google Scholar
• 10 Böhm H -J, Schneider G, Kubinyi H, Mannhold R, Timmerman H. Virtual screening for bioactive molecules. Vol. 10, In: Mannhold R, Kubinyi H, Timmermann H, editors Methods and principles in medicinal chemistry. New York; Wiley 2000
  Search in Google Scholar
• 11 Abagyan R, Totrov M. High-throughput docking for lead generation.  Curr Opin Chem Biol. 2001;  5 375-82
  CrossrefPubMedSearch in Google Scholar
• 12 Schneider G, Böhm H J. Virtual screening and fast automated docking methods.  Drug Discov Today. 2002;  7 64-70
  PubMedSearch in Google Scholar
• 13 Berman H, Westbrook J, Feng Z, Gilliland G, Bhat T, Weissig H. et al . The protein data bank.  Nucleic Acids Res. 2000;  28 235-42
  CrossrefPubMedSearch in Google Scholar
• 14 Wolber G, Langer T. LigandScout: 3D pharmacophores derived from protein-bound ligands and their use as virtual screening filters.  J Chem Inf Model. 2005;  45 160-9
  CrossrefPubMedSearch in Google Scholar
• 15 Inte:Ligand G mbH. Vienna, Austria. LigandScout: 3D pharmacophore modelling tool. Available at http://www.inteligand.com. Accessed February 2006
  Search in Google Scholar
• 16 Congreve M, Murray C W, Blundell T L. Structural biology and drug discovery.  Drug Discov Today. 2005;  10 895-907
  CrossrefPubMedSearch in Google Scholar
• 17 Rollinger J M, Hornick A, Langer T, Stuppner H, Prast H. Acetylcholinesterase inhibitory activity of scopolin and scopoletin discovered by virtual screening of natural products.  J Med Chem. 2004;  47 6248-54
  CrossrefPubMedSearch in Google Scholar
• 18 Nikolovska-Coleska Z, Xu L, Hu Z, Tomita Y, Li P, Roller P P. et al . Discovery of embelin as a cell-permeable, small-molecular weight inhibitor of XIAP through structure-based computational screening of a traditional herbal medicine three-dimensional structure database.  J Med Chem. 2004;  47 2430-40
  CrossrefPubMedSearch in Google Scholar
• 19 Wu G, Chai J, Suber T L, Wu J W, Du C, Wang X. et al . Structural basis of IAP recognition by Smac/DIABLO.  Nature. 2000;  408 1008-12
  PubMedSearch in Google Scholar
• 20 Morris G M, Goodsell D S, Huey R, Olson A J. Distributed automated docking of flexible ligands to proteins: parallel applications of AutoDock 2.4  J Comput Aided Mol Des. 1996;  10 293-304
  CrossrefPubMedSearch in Google Scholar
• 21 Langer T, Krovat E M. Chemical feature-based pharmacophores and virtual library screening for discovery of new leads.  Curr Opin Drug Discov Dev. 2003;  6 370-6
  PubMedSearch in Google Scholar
• 22 Accelrys I nc., San D iego. CA, USA. Catalyst: programme for modelling, database management, and querying for drug discovery. Available at http://www.accelrys.com. Accessed January 2006
  Search in Google Scholar
• 23 Laggner C, Schieferer C, Fiechtner B, Poles G, Hoffmann R D, Glossmann H. et al . Feature based pharmacophore models for sigma1 receptor, ERG2 and EBP.  J Med Chem. 2005;  48 4754 -64
  CrossrefPubMedSearch in Google Scholar
• 24 Zupan J, Gasteiger J. Neural networks in chemistry and drug design, 2nd edition. Weinheim; Wiley-VCH 1999
  Search in Google Scholar
• 25 Kohonen T. Self-organized formation of topologically correct feature maps.  Biol Cybern. 1982;  43 59-69
  CrossrefPubMedSearch in Google Scholar
• 26 Wagner S, Hofmann A, Siedle B, Terfloth L, Merfort I, Gasteiger J. Development of a structural model for NF-κB inhibition of sesquiterpene lactones using self-organizing neural networks.  J Med Chem. 2006;  49 241-52
  PubMedSearch in Google Scholar
• 27 Bernard P, Berton J Y, Chrétien J R. Computer-aided molecular selection and design of natural bioactive molecules.  Curr Opin Drug Discov Dev. 1999;  2 213-23
  PubMedSearch in Google Scholar
• 28 Kirchmair J, Laggner C, Wolber G, Langer T. Comparative analysis of protein-bound ligand conformations with respect to catalyst's conformational space subsampling algorithms.  J Chem Inf Model. 2005;  45 422-30
  CrossrefPubMedSearch in Google Scholar
• 29 Lu A, Liu B, Liu H, Zhou J, Xie G. A traditional Chinese medicine plant-compound database aid its application for searching.  Int Electron J Mol Des. 2004;  3 672-83
  PubMedSearch in Google Scholar
• 30 Lei J, Zhou J. A marine natural product database.  J Chem Inf Comput Sci. 2002;  42 742-8
  CrossrefPubMedSearch in Google Scholar
• 31 Rollinger J M, Haupt S, Stuppner H, Langer T. Combining ethnopharmacology and virtual screening for lead structure discovery: COX-inhibitors as application example.  J Chem Inf Comput Sci. 2004;  44 480-8
  CrossrefPubMedSearch in Google Scholar
• 32 Rollinger J M, Bodensieck A, Seger C, Ellmerer E P, Bauer R, Langer T. et al . Discovering COX-inhibiting constituents of Morus root bark: activity-guided versus computer-aided methods.  Planta Med. 2005;  71 399-405
  Thieme ConnectPubMedSearch in Google Scholar
• 33 Development Therapeutics Program N CI/NIH. Open NCI database. Available at http://dtp.nci.nih.gov.accesdistant.sorbonne-universite.fr/webdata.html. Accessed in March 2006
  Search in Google Scholar
• 34 Poroikov V V, Filimonov D M, Ihlenfeldt W -D, Gloriozova T A, Lagunin A A, Borodina Y V. et al . PASS biological activity predictions in the enhanced open NCI database browser.  J Chem Inf Comput Sci. 2003;  43 228-36
  CrossrefPubMedSearch in Google Scholar
• 35 Cherkasov A, Shi Z, Fallahi M, Hammond G L. Successful in silico discovery of novel nonsteroidal ligands for human sex hormone binding globulin.  J Med Chem. 2005;  48 3203-13
  CrossrefPubMedSearch in Google Scholar
• 36 Van de Waterbeemd H. High-throughput and in silico techniques in drug metabolism and pharmacokinetics.  Curr Opin Drug Discov Dev. 2002;  5 33-43
  PubMedSearch in Google Scholar
• 37 Bernard P, Scior T, Didier B, Hibert M, Berthon J -Y. Ethnopharmacology and bioinformatic combination for leads discovery: application to phospholipase A2 inhibitors.  Phytochemistry. 2001;  58 865-74
  CrossrefPubMedSearch in Google Scholar
• 38 Bernard P, Pintore M, Berthon J -Y, Chretien J R. A molecular modeling and 3D QSAR study of a large series of indole inhibitors of human non-pancreatic secretory phospholipase A₂ .  Eur J Med Chem. 2001;  36 1-19
  CrossrefPubMedSearch in Google Scholar
• 39 Rollinger J M, Mock P, Zidorn C, Ellmerer E P, Langer T, Stuppner H. Application of the in combo screening approach for the discovery of non-alkaloid acetylcholinesterase inhibitors from Cichorium intybus .  Curr Drug Discov Technol. 2005;  2 185-93. Corrections: Curr Drug Discov Technol 2006; in press
  CrossrefPubMedSearch in Google Scholar

Mag. pharm. Dr. Judith Maria Rollinger

Institute of Pharmacy/Pharmacognosy

Leopold-Franzens-Universität Innsbruck

Innrain 52c

Josef-Moeller Haus

6020 Innsbruck

Austria

Phone: +43-512-507-5308

Fax: +43-512-507-2939

Email: judith.rollinger@uibk.ac.at