Chapter 9 Discovering new lead compounds in pharmaceutical research and development
Traditional remedies invariably involve crude plant extracts containing multiple chemical constituents, which vary in potency from highly active (e.g. Digitalis leaf) to very weak (e.g. cinnamon bark). By contrast, orthodox medicine relies heavily on single (or a very small number of) chemically well-characterized active ingredients exhibiting selective activities at, in many cases, well-characterized biological targets. These medicines are generally very potent and many exhibit fairly narrow windows between an effective and a toxic dose. Orthodox medicines are formulated into doses that are carefully standardized for bioavailability.
Compounds derived from higher plants continue to feature among the most widely used orthodox medicines we have today (Martindale, see Further reading). These include analgesic agents (e.g. morphine, codeine and the non-steroidal anti-inflammatory drugs based originally on the structure of salicin), antimalarial treatments (e.g. quinine), antitumour drugs (e.g. vincristine and taxol) and asthma therapies (e.g. cromoglycate). Other plant-derived compounds are currently being evaluated in pharmaceutical development, an example of which is artemisinin, an extract of the sweet wormwood plant (Artemisia annua), which is being assessed in combination with chlorproguanil and dapsone as a new antimalarial treatment.
In some cases, natural materials continue to be the only viable commercial source of the active compound. For example, GlaxoSmithKline harvests up to 10 000 metric tons dry weight of poppy capsule per year to provide a source of opiate alkaloids.
‘High-throughput screening’ (HTS) is a major strategy for the discovery of new lead chemicals in the pharmaceutical industry. HTS uses miniaturized assay formats, usually microtitre plates in which, for example, 384 or 1536 different samples can be assayed, in volumes of less than 50 and 5 μl, respectively in one run. Using sophisticated automation equipment, typically, hundreds of thousands of samples are screened against each biological target of interest every day: the final numbers for each usually being dictated by the overall cost of the assay, which can vary from < 1 p per well to > 20 p per well. Screening collection sizes range from 400 000 to over 4 million.
HTS is often portrayed, by people who know little about it, as an activity requiring very little intellectual input. The reality is that HTS is a complex process that demands an understanding of the role of specific biological targets in disease progression; the development of bioassays capable of discovering modulators of the target; the design, miniaturization and automation of bioassays (which are automation friendly); an understanding of the macro- and micro-structure of the biological target so that the sample selection strategy is optimized; the engineering of custom-built robots capable of storage, retrieval and bioassay of millions of samples per annum and the development of software systems that can enable scientists involved to make sense of the mass of data that emerges.
Ideal biological assays for screening are those that enable identification of compounds acting on specific biological targets, involve a minimum number of reagent addition steps, perform reliably and predictably, are easily amenable to miniaturization and automation, and involve low-cost ingredients and detection technology. Biochemical targets of interest in pharmaceutical lead discovery range from enzymes to receptors (nuclear and transmembrane) to ion channels and, in the case of infectious disease, to whole microorganism cells.
An example of a biological assay that has the characteristics needed in a good screen is the squalene synthase enzyme assay, which wasdeveloped to look for inhibitors of squalene synthesis, a potential target for the identification of novel cholesterol lowering agents (Tait, 1992). Using either [1-14C] isopentenyl diphosphate as a precursor for squalene or [2-14C] farnesyl diphosphate as a direct substrate of squalene synthase, the production of radiolabelled squalene is determined after adsorption of assay mixtures onto silica gel thin-layer chromatography sheets and selective elution of the diphosphate precursors into a solution of sodium dodecyl sulfate at alkaline pH. The use of [2-14C] farnesyl diphosphate, and of an endogenous oxygen consumption system (ascorbate/ascorbate oxidase) to prevent further metabolism of squalene, allows the method to be applied as a dedicated assay for squalene synthase activity. The assay can be readily operated in microtitre plate format, which allows 96 or 384 samples to be screened per plate. It can be deployed either in a quantitative, low-throughput mode or in a qualitative, high-throughput mode, which has proved to be resistant to interference by compounds other than selective inhibitors.
The endogenous neuropeptide bradykinin (BK) is implicated in the mediation of various types of pain in the mammalian CNS. Antagonism of bradykinin to its receptors is a potential target for the development of new analgesic agents. An assay has been devised to detect compounds that antagonize binding of radiolabelled bradykinin to BKII receptors expressed in Chinese hamster ovary (CHO) cells (Sampson et al., 2000). Compounds under test are added to the wells of microtitre plates to which CHO cells have adhered. After incubation with radiolabelled bradykinin, the excess labelled ligand is removed by washing. The plates are then counted in a scintillation counter so as to assess binding of labelled bradykinin to the receptors expressed on the surfaces of the cells. This particular screen suffers interference from compounds that possess cytotoxicity through a variety of mechanisms. It is therefore essential to run follow-up control assays against other cell types to distinguish false positives.
In the infectious disease arena, it is still common to run high-throughput, whole-cell antifungal or antibacterial assays to detect samples that inhibit growth of the designated strain, e.g. Candida albicans, Staphylococcus aureus. Optical density or colour changes using a redox indicator are the most frequently used assay technologies. Assays in this therapeutic area may be mechanism based. For example, a C. albicans cell-free translation system using polyurethane as a synthetic template, has been established to search for compounds that inhibit fungal protein synthesis (Kinsman et al., 1998).
Screening plant extracts for antitumour activity involves assays against a wide variety of cancer cell lines and mechanism-based in vitro targets, which have been documented extensively (Pezzuto, 1997). Among the most frequently used mechanism-based assays are those assessing activity against the biological targets of existing antitumour drugs, such as topoisomerases I and II, collagenase, tubulin binding and stabilization, endocrine hormone synthesis and androgen and oestrogen receptor binding. Mechanism-based assays often require sophisticated or expensive reagents: an assay for activity of DNA ligase I involves incubating plant extracts with recombinant human DNA ligase I cDNA and its radiolabelled substrate, and measuring uptake 5′-32P labelled phosphomonoesters into alkaline-phosphatase-resistant diesters (Tan et al., 1996).
During the 1980s and early 1990s, natural product samples were the mainstay of HTS programmes within the pharmaceutical industry, due at least in part to the lack of availability of large numbers of synthetically derived chemicals. Over recent years, this situation has changed dramatically. The highly competitive arena of drug discovery provides pharmaceutical companies with a clear incentive to be first to discover and patent new lead molecules. Thus, a range of technologies has evolved to facilitate ever-increasing numbers of samples to be rapidly generated and evaluated.
Most large pharmaceutical houses have built up a compound bank containing hundreds of thousands of chemical compounds, which reflect the chemistries of earlier medicines developed by the company. Chemical diversity in these collections can be supplemented by acquisition of new compound types from the growing ranks of specialist compound vendors. Computational modelling techniques are utilized to generate sets of specific interest for given biological targets. Methods are available for electronically filtering out ‘undesirable’ compounds and techniques such as pharmacophore analysis, two- or three-dimensional structure searching or chemical clustering can be used to derive sets of the required size. Ready access to these compound collections is facilitated by the use of robotic storage and retrieval facilities, which can present the samples in formats appropriate for HTS bioassays.
Combinatorial chemistry techniques are widely applied in the drug-discovery process, especially for generating large numbers of compounds for lead discovery and in the optimization of lead compounds. Using robotic systems, tens of thousands of compounds can be synthesized from a small number of reagents in a few days. To date, however, it appears that the most successful of these compound libraries, in terms of yielding interesting bioactive molecules, have utilized focused chemistry based on structural knowledge of the biological target and the pharmacophoric features required to affect it.
To supplement the chemical diversity of the compound banks and the chemically focused combinatorial libraries, a number of pharmaceutical companies continue to screen natural extracts. Historically, large collections of microbial organisms (notably fungi and filamentous bacteria) were built up from a diversity of environmental niches and emphasis was placed on the development of a range of fermentation conditions capable of eliciting the microbes to produce a variety of secondary metabolites. The extracts generated for HTS from this source can be reproduced on demand, should further studies on bioactivity of interest be required. In particular, industry found microbial fermentations to be a prolific source of antibiotics. More recently, from the same source, valuable immunosuppressant drugs and lipid-lowering agents have been added to the medicine chest.
Plant samples also feature in the HTS programmes of a small number of pharmaceutical discovery organizations. The feasibility of using plants in a drug-discovery programme depends on ensuring that effective procurement strategies are in place to source both the primary material and additional supplies should these be required.
Advances in screening technology have increased the throughput capacity of an average HTS from tens of thousands of samples to hundreds of thousands of samples over the last decade. Even so, the availability of so many samples for HTS means that choices might need to be made about the most appropriate sub-set of samples for each particular target. The sample selection strategy may then be ‘diversity-based’, i.e. samples are chosen to represent as wide a spectrum of chemical diversity as possible, or ‘focused’, i.e. the samples represent specific chemical types only.
Both strategies are likely to play a role in a pharmaceutical company’s methodology. Diversity screening may yield an unexpected interaction between a compound and a biological target, although the question of what constitutes ‘representative’ chemical diversity in the vast area of potential chemical space remains unanswered. Focused screening requires a large amount of prior information about a target, and this might not always be available. A combination of both approaches may be adopted. Computational methodologies for hit identification are continuously being developed. For example, compound databases enabling three-dimensional chemical structure searching are often used. If there are known ligands for a target, these can be used to construct a pharmacophore, which can then be utilized to search further chemical databases and select molecules with desired features. Chemical clustering can be used to derive sets of the required size. In the case of combinatorial chemistry-derived libraries, targeted sets can be generated with desired chemical properties, by using appropriately selected chemical building blocks. Natural products offer a potentially infinite source of chemical diversity unmatched by synthetic or combinatorially derived compound collections (Strohl, 2000), thus making them a desirable tool for diversity based screening. If a focused strategy is adopted, however, it is necessary to develop different techniques for natural product sample selection in order that the most appropriate samples are accessed for relevant targets.
The United Nations Convention on Biological Diversity (CBD) of 1992 has, to date, 191 parties, including 168 signatories (see its website http://www.cbd.int and the references therein). The key objectives of the Convention are to ensure the conservation of biological diversity, the sustainable use of natural resources and to implement fair and equitable sharing of benefits. Within the framework of the Convention are the concepts of the sovereignty of states over genetic resources and their obligation to facilitate access. The contracting parties are expected to establish measures for benefit sharing in the event of commercial utilization. This involves collaboration between the collector, the source country and the commercial partner. It is now normal practice to draw up a legal agreement to cover these issues and many companies have issued policy statements relating to this area.
As an example, a statement on GlaxoSmithKline’s website (http://www.gsk.com) describes how a pharmaceutical company addresses such issues. The policy recognizes the importance of matters considered at Rio and subsequent meetings of the Congress of the Parties and goes on to state that GlaxoSmithKline will collaborate only with organizations that can demonstrate both the expertise and the authority to supply natural materials. Only relatively small quantities of plant material are collected, from sustainable sources. GlaxoSmithKline supports the CBD’s role in providing a framework for the conservation of biological diversity and the sustainable use of its components and the CBD objective ‘to provide fair and equitable sharing of the benefits arising from the use of genetic resources’. GlaxoSmithKline further supports the approach laid down in the CBD and in the Bonn Guidelines of leaving it to national governments to determine the conditions under which access to genetic resources should be given and for the parties concerned mutually to agree on the benefits to be shared. Agreements will cover such matters as the permitted use of the resources and the nature and timing of any benefits that are to be shared. This approach allows national governments the flexibility to determine what rules will best serve their national interests and allows the stakeholders involved to reach agreement appropriate to each particular case.
Before a decision is made on what natural materials will be evaluated in a given screen, it is essential to gather some information on whether a target is indeed appropriate for input of natural product extracts. For example, if the biological target is very highly tractable, if there are significant time constraints and cost of goods issues, and if data suggest it is likely to be relatively straightforward to obtain synthetically derived, small-molecule lead compounds then it might be inappropriate to screen natural products against that target. However, if a target is of a class where it is difficult to find small molecule hits, e.g. involving protein–protein interactions, or if there is a strong precedent or rationale for natural-product-derived actives, then natural product input should be considered. The latter may be exemplified by, for example, the antimicrobial area, where the track record of drug discovery from microbial sources is beyond dispute. The same rationale would apply to the superb track record of plant species in yielding analgesic medicines.
If the target appears to be suitable for natural product input, the sample selection for the screen needs to be considered. Various strategies can be adopted, depending on the extent of the available natural materials collections and on the capacity and ‘robustness’ of the target itself.
Some companies have access to large and diverse natural materials collections. Such collections are likely to include samples acquired to add diversity to the potential collection (and inherent in the desire for taxonomic diversity is the assumption that this will be reflected in chemical diversity of extracts subsequently generated). Most collections will also include samples particularly selected for various reasons, e.g. a microbial producer of a given compound or a plant used ethnomedically for a given condition. These large collections probably still only reflect a fraction of the world’s potential biodiversity. It has been estimated that only around 70 000 fungal species are known, out of an estimated 1 500 000 (see the UN CBD website http://www.cbd.int and the references therein). Further, it has been estimated that only about 1% of microbial biodiversity actually comprises ‘culturable’ organisms (Amman et al., 1995). Thus, a huge number of strains may not be amenable to conventional isolation and cultivation methodologies, and many groups are now working on applying cloning techniques to harness the potential chemical diversity of these organisms.
A diversity-based approach requires acquisition of pre-selected taxonomic groups. The strategy may utilize the assumption that taxonomic diversity will inherently be reflected in the chemical diversity of the extracts subsequently prepared and screened. Various techniques can be employed to analyse the taxonomic spread of a plant collection and then make efforts to fill gaps so that the collection more completely reflects available diversity.
A more focused approach depends on having prior knowledge about selected samples, which might suggest that they contain particular chemical classes of interest or that they possess desirable biological properties. This strategy can be considered under two headings ‘chemical targeting’ and ‘biological targeting’.
This utilizes natural materials as sources of specific compounds of interest to a particular disease area, or as sources of chemical classes deemed to have suitable pharmacophores. In this way, chemical types that are under-represented in an existing sample collection can be identified. Plant-derived chemicals can provide an effective means of filling any ‘gaps’, thereby enhancing the overall diversity of available chemistry. It may not always be necessary to expend resource taking this process to full isolation and structure elucidation. In some cases, a set of plants can be selected on the basis that they are reported to produce a general chemical class of interest and appropriate crude or semi-purified extracts can be prepared in order to enrich the extract with the desired chemical types.
This adopts what may be thought of as a disease-driven process. Plant samples can be selected for biological evaluation using some type of information associated with them that suggests their relevance for evaluation in a given therapeutic target. Perhaps the most striking observations available are ethnobotanical reports of traditional medicinal uses of plants. A number of orthodox medicines available commercially today were discovered by following leads provided from indigenous knowledge (Cox, 1994).
Various approaches are adopted in assimilating the information needed to select plants of particular relevance for a given disease target. Some research groups rely on developing a network of ethnobotanists, who work closely with indigenous colleagues and traditional doctors in various countries. The outcome of this approach is a low number of plant samples identified for evaluation in the laboratory and a great deal of information on their use. Some pharmaceutical companies prefer only to use information that is already in the public domain. Various journals and books hold a significant amount of information relating to the ethnobotanic uses of various plants; for example, there are many publications describing the properties of plants used in Chinese Traditional Medicine (e.g. Chang and But, 1986). Perhaps the most time-effective way of searching for information relating to reported biological or chemical properties of plants is to use electronic data stored in a range of databases. One key example of this is the NAPRALERT (Natural Products Alert) database (Loub et al., 1985). This system was initiated and is maintained at the University of Illinois at Chicago. It contains a wealth of information in the form of a huge number of references relating to reports of biological activity in the scientific literature, ethnobotanical reports and phytochemical data.
For chemical information, databases such as the Chapman and Hall Dictionary of Natural Products (2000) can be useful tools. This database contains information on well over 100 000 natural products, often including the species from which the compound originates. Searches can be carried out on the basis of chemical structure, sub-structure, structural similarity, presence of particular functional groups, etc. in order to build a set of organisms reported to produce these compounds. Alternatively, the database can be searched on the basis of species or genus, so as to build a list of compounds reported to derive from particular organism groups. This can be particularly useful in the process of de-replication and compound identification (see later). In addition, searches can be carried out on a range of other fields, such as molecular weight, which are also useful in compound identification following identification of a molecular ion by mass spectrometry analysis, reported uses of the natural product, literature references, log P, etc. The database comprises not just plant metabolites but compounds from all natural sources, and although the data are not complete, it can represent a valuable starting point by which to build chemically focused sets of samples for screening.
In the case of preparing samples that might be expected to contain specific chemical entities, phytochemical procedures reported in the literature can be used to generate semi-purified extracts, or extracts enriched in the compound or chemical class of interest. In the case of plants with an ethnomedical use, it might be possible to prepare extracts using methodology as recommended for use in traditional medicine.
These ‘targeted’ approaches are likely to involve smaller numbers of plant samples than a high throughput, random screening programme. However, the actual numbers of plants selected and screened can be tailored by the scientist, by making the selection criteria more or less stringent. For example, a selection process may result in a small number of plants reported to produce specific compounds of interest, but if the assay is capable of screening much larger numbers of samples, the set can be extended to include all those plants reported to produce metabolites of the much broader chemical class. Similarly, a set can be extended taxonomically to an optimum size by including plants of related species or genera, on the basis that they might also produce related chemistry. Making the selection criteria slightly more ‘fuzzy’ in this manner also allows a greater role for the element of luck—always important in the drug discovery process!
Preparation methods are tailored towards the type of natural material being processed and the strategy for analysis being undertaken. For plant samples, the standard approach is to acquire and store dried plant material. In rare cases, for example if an ethnomedical report dictates use of fresh material, then this may be undertaken, but would not be the norm. Although a very small amount of plant material—less than half a gram—may provide sufficient extract to allow testing in many hundreds of bioassays, it is only sensible to collect a larger amount of material. For natural product samples to remain competitive sources of lead compounds, it is necessary to be able to very rapidly follow-up any active extracts. For this reason, collection of a few hundred grams of dry material is more typical. Plant material is finely ground using techniques ranging from pestle and mortar to industrial grinding apparatus, as appropriate.
Microbial strain collections are maintained either as freeze-dried cultures or are preserved by low-temperature storage. Required strains are generally revived by inoculation onto agar and/or growth media, and then are cultivated on a medium—or, more often, a range of media—that have been developed with the aim of promoting secondary metabolite production. Factors like incubation times and temperature, media composition and agitation rates can all have a significant effect on growth and metabolite synthesis. The next step is to generate an extract for analysis in a range of bioactivity screens. Often, there will be no prior knowledge of which chemical types may be present in the samples, and which will be active in any given bioassay. Techniques will therefore be aimed at solubilizing as wide a range of compounds as possible, typically using an alcoholic solvent, such as methanol or ethanol, or an aqueous alcoholic solvent. Extraction may take the form of a cold infusion or it may involve hot-solvent extraction using a Soxhlet apparatus (Silva et al., 1998). If the natural source material is under investigation because it has been reported to contain a desired compound or chemical class, then a bespoke extraction procedure will be adopted, probably using a literature report. If the target molecule is less specific, for example if the aim is to access any alkaloid molecules that might be present, then alkaloid-enriched extracts can be generated by following a method such as acid–base partitioning.
An alternative approach might favour the development of fractionation methods, so that several fractions originating from a single sample are generated prior to biological testing. Although more labour intensive, this can have the benefit of eliminating many of the problems that can be seen when testing crude extracts in some bioassays, e.g. frequent actives caused by commonly occurring interfering compounds such as tannins, saponins, flavonoids, etc. Such actives can take a significant time to de-replicate amd eliminate and, in some circumstances, it may be desirable to reduce this resource by spending time before biological testing to do some semi-purification of extracts. Techniques used can include solvent partitioning or chromatographic fractionation.
Biological analysis is likely to yield a number of active extracts, as defined by showing a certain level of activity in a given screen. A process then needs to begin that will either lead to a full identification of a fully characterized, active compound or to a partial identification of activity to the level of a family of known compounds.
Before commencing full bioassay-guided fractionation of the active samples it is necessary to review the tolerance of a given assay to crude or semi-purified extracts of natural materials. The aim of evaluating such samples in a biological assay is to identify compounds that interact with a particular biological target, e.g. an enzyme or receptor. However, in practice, most assays utilize a measurable system, such as colour, light or radioactivity, enabling a high throughput of samples. This leads to the possibility of detection of non-specific interactions, which are particularly problematic when investigating multi-component, uncharacterized extracts as opposed to single chemical entities. Examples of natural products that generate such effects are detergent-like compounds, which disrupt cell membranes, polyphenolics, which form complexes with a wide array of proteins, antioxidants and ultraviolet (UV) quenchers. It is vital to be able to detect, and eliminate, such actives as rapidly as possible (VanMiddlesworth and Cannell, 1998). This process can be speeded up by testing a standard set of known interfering compounds and extracts in an assay prior to the full screen. This provides data to indicate the tolerance of the assay to such samples, and can at times lead to a decision not to proceed with testing of crude extracts against that target. The physicochemical properties of a compound can give useful clues as to its identity. The most commonly used properties include high-performance liquid chromatography (HPLC) retention time and UV spectra data that are readily acquired through standard analytical techniques. By comparing these data with those of known compounds, it may be possible to characterize the components of a mixture without the need for full isolation—or at least, it may be possible to narrow the possibilities. If a library of such data from a relatively large number of natural products is used, this can be very effective in identifying those compounds that are present in more than one organism. If full isolation of the active component is warranted, various methods may be adopted. There is no single, best isolation technique, nor is there any single, correct method for any given compound. Most separation methods involve some form of chromatography—typically preparative HPLC. An isolation method would normally involve solvent partitioning, followed by a crude chromatography step such as a silica column or counter-current chromatography with a relatively small number of fractions based on polarity, followed by final purification through a high-resolution separation step such as HPLC. At each stage of the purification, the active compound is tracked by bioassay of the fractions. The only sure way to identify the structure of a bioactive metabolite is to demonstrate activity using the isolated compound and then to determine its structure by nuclear magnetic resonance (NMR) and mass spectrometry (MS). Until a compound is isolated, it is also impossible to determine its concentration and, hence, its potency in a given assay. It is also prudent to check that the concentration of the metabolite in the extract and the activity in the unpurified extract tally, so that a minor but active component does not go unaccounted for. Secondary testing will then be undertaken on the active compound to determine the mechanism and the selectivity of action and eventually, to evaluate in vivo activity. The chemical structure of the compound will also be evaluated to give some indication of the classical ‘drug-like’ qualities of the molecule. These are to some extent subjective but include consideration of whether the compound is of sufficiently low molecular weight to allow ready chemical modification, and not prevent drug uptake; of whether the compound is likely to be stable with respect to oral uptake, whether it has sites that may be suitable for modification, whether it is likely to have a suitable log P, and so on. In fact, some classes of natural product are not suitable as drug candidates by these criteria, as they are too big and complex or possess unsuitable redox properties. If the molecule is deemed to be of interest, related metabolites from the sample, or related species, may be accessed for structure activity determination. Even if the structure of the metabolite is not novel, this does not preclude it becoming a lead molecule, particularly if the mode of action is novel.
The discovery of a series of novel and highly potent euphane triterpenes illustrates the potential of plant extracts to generate useful chemical leads in a high-throughput screening programme.
During a ‘random’ screening programme to search for novel inhibitors of human thrombin, which were capable of blocking the formation of blood clots and hence could be of value in treating and preventing deep vein thrombosis, some 150 000 samples, including synthetic compounds and bacterial, fungal and plant extracts, were evaluated. Methanolic extracts of Lantana camara (Verbenaceae) leaves, obtained from a UK garden, were found to display potent activity.
Large-scale extraction and bioassay-guided chromatographic fractionation led to the identification of a series of novel compounds, which were characterized by NMR and MS as 5,5-trans-fused cyclic lactone-containing euphane triterpenes (O’Neill et al., 1998). The compounds showed ICso values of the order of 50 nm against thrombin.
After the initial activity was detected, literature searches on Lantana camara revealed this plant species to be reported to be toxic to grazing animals, which, on ingestion of the leaves, develop hepatotoxicity and photosensitization (Sharma and Sharma, 1989). These toxic effects have been attributed to the lantadenes, a series of pentacyclic triterpenes. A further study of haematological changes in sheep following Lantana poisoning demonstrated a significant increase in blood coagulation time and prothrombin time, with an associated decrease in blood sedimentation rate, total plasma protein and fibrinogen (Uppal and Paul, 1982). This observation might be associated with the thrombin inhibitory translactone-containing euphane triterpenes described above.
The biological activity of these compounds has been reported in detail (Weir et al., 1998). Their mechanism of action as inhibitors of blood clotting is via acylation of the active site Ser 195 residue of thrombin. This acylating activity has been found to be generic against other serine protease enzymes and this finding forms the basis for exploitation in drug discovery.
Amman RL, Ludwig W, Schleifer KH. Phylogenetic identification and in situ detection of individual microbial cells without cultivation. Microbiological Reviews. 1995;59:143-169.
Chang H-M, But P-H. Pharmacology and applications of Chinese materia medica. Singapore: World Scientific Publishing Co, 1986.
Cox P. The ethnobotanical approach to drug discovery: strengths and limitations. In: Ethnobotany and the search for new drugs. Chichester: Wiley; 1994:25-41. (Ciba Foundation Symposium 185)
Dictionary of natural products. 5th edn. Chapman and Hall/CRC Press, 2000.
Kinsman OS, Chalk PA, Jackson HC, et al. Isolation and characterisation of an antifungal antibiotic (GR 135402) with protein synthesis inhibition. Journal of Antibiotics. 1998;51(I):41-49.
Loub WD, Farnsworth NR, Soejarto DD, Quinn ML. NAPRALERT: Computer handling of natural product research data. Journal of Chemical Information and Computing Sciences. 1985;25:99-103.
Martindale, the complete drug reference. Internet version. http://www.csi.micromedex.com.
O’Neill MJ, Lewis JA, Noble HM, et al. Isolation of translactone containing triterpenes with thrombin inhibitory activity from the leaves of Lantana camara. Journal of Natural Products. 1998;61(11):1328-1331.
Pezzuto JM. Plant derived anticancer agents. Biochemical Pharmacology. 1997;53(2):121-133. and references therein
Sampson JH, Phillipson JD, Bowery N, et al. Ethnomedically selected plants as sources of potential analgesic compounds: indications of in vitro biological activity in receptor binding assays. Phytotherapy Research. 2000;14:24-29.
Sharma OMP, Sharma PD. Natural products of the Lantana plant—the present and prospects. Journal of Scientific Industrial Research. 1989;48:471-478.
Silva GL, Lee I-S, Kinghorn AD. Special problems with the extraction of plants. In: Cannell RJP, editor. Natural products isolation. Totowa, NJ: Humana Press; 1998:343-363.
Strohl WR. The role of natural products in a modern drug discovery program. Drug Discovery Today. 2000;5(2):39-41.
Tait RM. Development of a radiometric spot-wash assay for squalene synthase. Analytical Biochemistry. 1992;203(2):310-316.
Tan GT, Lee S, Lee I-S, et al. Natural product inhibitors of human DNA ligase I. Biochemical Journal. 1996;314:993-1000.
Uppal RP, Paul BS. Haematological changes in experimental Lantana poisoning in sheep. Indian Veterinary Journal. 1982:18-24.
VanMiddlesworth FW, Cannell RJP. Dereplication and partial identification of natural products. In: Cannell RJP, editor. Natural products isolation. Totowa, NJ: Humana Press; 1998:343-363.
Weir MP, Bethell SS, Cleasby A, et al. Novel natural product 5,5-trans-lactone inhibitors of human a-thrombin: mechanism of action and structural studies. Biochemistry. 1998;37:6645-6657.
* Disclaimer: the information in this chapter represents the authors’ personal views. In no way does it represent the views of GlaxoSmithKline