《天然药物化学》课程参考文献（天然药物研究与开发）Natural Products as Leads to Potential Drugs - An Old Process or the New Hope for Drug Discovery.pdf

Natural Products as Leads to Potential Drugs: An Old Process or the New Hope for Drug Discovery? David J. Newman† Natural Products Branch, DeVelopmental Therapeutics Program, DCTD, National Cancer InstitutesFrederick, P.O. Box B, Frederick, Maryland 21702 ReceiVed April 5, 2007 I. Introduction From approximately the early 1980s, the “influence of natural products” upon drug discovery in all therapeutic areas apparently has been on the wane because of the advent of combinatorial chemistry technology and the “associated expectation” that these techniques would be the future source of massive numbers of novel skeletons and drug leads/new chemical entities (NCEa ) where the intellectual property aspects would be very simple. As a result, natural product work in the pharmaceutical industry, except for less than a handful of large pharmaceutical companies, effectively ceased from the end of the 1980s. What has now transpired (cf. evidence shown in Newman and Cragg, 20071 and Figures 1 and 2 below showing the continued influence of natural products as leads to or sources of drugs over the past 26 years (1981–2006)) is that, to date, there has only been one de novo combinatorial NCE approved anywhere in the world by the U.S. Food and Drug Administration (FDA) or its equivalent in other nations for any human disease, and that is the kinase inhibitor sorafenib (1, Chart 1), which was approved by the FDA in late 2005 for renal carcinoma. However, the techniques of combinatorial chemistry have revolutionized the deVelopment of active chemical leads where currently, instead of medicinal chemists making derivatives from scratch, a procedure is used whereby syntheses are based on combinatorial processes so that modifications can be made in an iterative fashion. An example of such a process would be the methods underlying the ultimate synthesis of the antibiotic linezolid (Zyvox, 2) by the Pharmacia (now Pfizer) chemists starting from the base molecules developed in the late 1980s by DuPont Pharmaceutical, who reported the underlying antibiotic activity and mechanism of action of this novel class of molecules, the oxazolidinones.2–6 Although the early (late 1980s to late 1990s) combinatorial chemical literature is replete with examples of libraries containing hundreds of thousands to millions of new compounds, as stated rather aptly by Lipinski in the early 2000s, if the early libraries had been disposed of, the productivity of pharmaceutical drug discovery would have materially improved in the prior decade.7 However, in the late 1990s, synthetic chemists realized that the combinatorial libraries that had been synthesized up to that time (with the exception of those based on intrinsically bioactive compounds such as nucleosides, peptides, and to some extent carbohydrates) lacked the “complexity” normally associated with bioactive natural products, items such as multiple chiral centers, heterocyclic substituents, and polycyclic structures. Although chemists had probably accepted that as a “basic rule”, natural products were different from synthetic compounds; the 1999 analysis by Henkel et al.8 was perhaps the first of the † Contactinformation.Telephone:+301.846.5387.Facsimile:+301.846.6178. E-mail: dn22a@nih.gov. The views expressed in this review are those of the author and are not necessarily the position of the U.S. Government. a Abbreviations: -AST-IV, -arylsulfotransferase-IV; BIOS, biologyoriented synthesis; Cdks, cyclin dependent kinases; DOS, diversity-oriented synthesis; E-FABP, epidermal fatty acid binding protein; EGFR, epidermal growth factor receptor; FKBP-12, FK binding proteins; FXR agonists, farnesoid X receptor agonist; GPCR, G-protein-coupled receptor; hERG, human ether-a-go-go-related-gene K+ channel; HIF-1R, hypoxia-inducible factor-1R; IGF1R, insulin-like growth factor 1 receptor; LTA4H, leukotriene A4 hydrolase/aminopeptidase; mEST, murine estrogen sulfotransferase; M6p-IGFR2, insulin-like growth factor II/mannose 6-phosphate receptor; NCE, new chemical entity; NMDA, N-methyl-D-aspartate; NodH sulfotransferase (gene product from Rhizobium NodH); PAPS, 3′-phosphoadenosine 5′-phosphosulfate; PFT, protein fold topology; PI3K, phosphoinositol-3- kinase; PKC, protein kinase C, exists in multiple isoforms; PSSC, protein structure similarity clustering; SCONP, structural classification of natural products; SEA, similarity ensemble approach; Shp-2, Src homology 2 domain containing tyrosine phosphatase 2; TOR, target of rapamycin; VEGFR-2, vascular endothelial growth-factor receptor-2. Figure 1. Source of small molecule drugs, 1981–2006: major categories, N ) 983 (in percentages). Codes are as in ref 1. Major categories are as follows: “N”, natural product; “ND”, derived from a natural product and usually a semisynthetic modification; “S”, totally synthetic drug often found by random screening/modification of an existing agent; “S*”, made by total synthesis, but the pharmacophore is/was from a natural product. Figure 2. Sources of small molecule drugs, 1981–2006: all categories, N ) 983 (in percentages). Codes are as in ref 1. Major categories are as follows: “N”, natural product; “ND”, derived from a natural product and usually a semisynthetic modification; “S”, totally synthetic drug often found by random screening/modification of an existing agent; “S*”, made by total synthesis, but the pharmacophore is/was from a natural product. The subcategory is as follows: “NM”, natural product mimic. J. Med. Chem. 2008, 51, 2589–2599 2589 10.1021/jm0704090 This article not subject to U.S. Copyright. Published 2008 by the American Chemical Society Published on Web 04/05/2008 Downloaded by CKRN CNSLP MASTER on August 8, 2009 Published on April 5, 2008 on http://pubs.acs.org | doi: 10.1021/jm0704090

modern treatments demonstrating the intrinsic structural differences between synthetic libraries, in this case those of Bayer AG, and the natural product structures shown in the Chapman and Hall Dictionary of Natural Products, made available to the nonspecialists in the field. Thus, the concept of diversity-oriented synthesis (DOS) has now come into vogue, where compounds resembling natural products in terms of their complexity (as defined above), or that are based on natural product topologies, have been made by a significant number of synthetic chemists. The “absolute origin of the term” is a trifle difficult to discern. Certainly it was used in Schreiber’s group9,10 in the late 1990s, and the concept was used by Nicolaou in the same time frame as exemplified by the reports on the benzopyran libraries.11–13 DOS-sourced molecules have been or are being tested in a large number and variety of biological screens in order to determine their role(s) as leads to novel drug entities and/or biological probes. To give an idea of the vast amount of work reported in the literature in the time frame from 2000 to date, over 300 articles have been published in the chemical literature, with 60 plus being reviews, when the term “diversity oriented synthesis” is used as the search parameter in the “Web of Science”. II. Discussion of Specific Topics II.1. Syntheses around Privileged Structures, aka the Intrinsic Differences of Natural Products. The concept of “privileged structures”, among which I personally include natural products with known bioactivities and would also extend this definition to the majority of secondary metabolites, was first suggested by Evans et al. in relation to the benzdiazepines14 and then extended to natural product structures such as indoles and other partial structural motifs by Mason et al. and references therein.15 One can argue fairly successfully that peptides are possibly the best recognized of privileged structures, though purine and pyrimidine bases and their corresponding nucleosides may well be a very close second. Following on from this comment, one of the seminal reviews in the history of use of peptidomimetics derived from natural products is that by Wiley and Rich.16 This paper gives an excellent history of what might be considered to be “precombinatorial discoVeries from natural product scaffolds” and should be read by those who wish to see where a significant number of earlier leads in a large variety of biological screens have come from. The current use of natural product-based privileged structures as leads to novel bioactive compounds is, as mentioned above, probably best demonstrated by the work of Nicolaou’s group formally reported in a series of papers in Journal of the American Chemical Society in 2000.11–13 By use as the base structure, a benzopyran or a partially reduced benzopyran (from data showing over 10 000 structures with these base structures in the natural products literature), a series of iterative molecules based on combinatorial syntheses were tested against a wide series of biological assays. To date, four distinct, previously unrecognized biological activities have been reported from this relatively small series of compounds. These currently include an inhibitor (3) of NADH/ubiquinone oxidoreductase with cytostatic activity against specific cell Chart 1 2590 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 9 MiniperspectiVe Downloaded by CKRN CNSLP MASTER on August 8, 2009 Published on April 5, 2008 on http://pubs.acs.org | doi: 10.1021/jm0704090

lines,17 a compound (4) with antibacterial activity against methicillin-resistant Staphylococcus aureus, 18 nonsteroidal FXR agonists (5a–c) which have helped define the interactions within this receptor for the first time,19,20 and an inhibitor (6) of hypoxia-inducible factor-1R (HIF-1R).21 These are probably the most diverse activities yet shown from a single base natural product structure, and it will be very interesting to see how many more biological results will be reported from these series in due course. The potential for such synthetic strategies is further exempli- fied by another review from the same group that, although covering the benzopyrans, expands the natural products to polysaccharides, the eleutherobin/sarcodictyin derivatives, glycopeptide antibiotics such as vancomycin and epothilones.22 All of these can be considered to be part of the collection of privileged structures. Both earlier and then in a similar time frame to the work above, synthetic and natural products chemists were investigating the potential of the simple analogues of adenine, 6-dimethylaminoadenine (7), and from a Castanea sp., isopentenyl adenine (8), as inhibitors of the mitotic histone Hi kinase (better known as cyclin-dependent kinase 1/cyclin B).23,24 Further investigation with other purine-based compounds showed that the plant secondary metabolite olomucine (9), originally isolated from the cotyledons of the radish and that had been synthesized in 1986 by Parker et al.,25 inhibited cyclin dependent kinases (Cdks) with IC50 values in the low micromolar range. This finding disproved the then existing dogma that no “specific” kinase inhibitors could be found for ATP-binding sites because they would be swamped by the normal cellular levels of ATP (which are in the 1-5 mM range). Further development of the olomucine structure led to roscovitine (10), with an IC50 against Cdks of 450 nM and then from a focused combinatorial library, the purvalanols (11a,b) with IC50 values in the 4–40 nM range.26 Currently, roscovitine (10) is in phase II clinical trials for cancer under the names CYC202 and seliciclib, with a recent full publication giving details of the phase I trial.27 Since there are basic similarities between the enzymic mechanisms of kinases and sulfotransferases, both performing a transfer reaction of anionic groups and both binding adenosinebased substrates, with kinases using ATP as the phosphoryl donor and sulfotransferases using 3′-phosphoadenosine 5′- phosphosulfate (PAPS), it was a logical extension of this work to look at the interactions of purine scaffold libraries utilized for roscovitine and olomucine with suitable target enzymes. Thus, by use of inositol 1,4,5-trisphosphate 3-kinase (IP3K) as the target of a suitable library, compound 12 was found with an IC50 value of 10.2 µM, compared to >100 µM against Cdk1/ cyclin B.28 Variations in the same library had been tested earlier against the NodH sulfotransferase from Rhizobium melioti giving a new compound (13, Chart 2) with equipotent activity (IC50 ≈ 20 µM) against this enzyme and Cdk2/cyclin A.29 Following from these studies, the use of a similar library protocol, but with a slight variation in the substituents, yielded two further nanomolar purine based inhibitors, one (14) demonstrating activity against a murine estrogen sulfotransferase (mEST) but with low activities against kinase targets while another compound in the same series (15) demonstrated a Ki of 96 nM against -arylsulfotransferase-IV (-AST-IV),30 with no other inhibitory effects against sulfotransferases/kinases at 10 µM. The value of this natural product-based approach can be seen in the comments on inhibitor discovery with sulfatases in a 2004 review by Rath et al.31 where, using synthetic inhibitors not based on natural products, the IC50 values are in the 50+ µM range for Est and NodH in comparison to the 1000 times more potent inhibitors from the purine libraries. Thus, by utilizing a simple “biologically active motif” and then using the techniques of combinatorial chemistry to produce focused libraries, Meijer and Schultz demonstrated very effectively that potent inhibitors of a variety of relatively closely related enzymes could be devised. Contemporaneously with these results, Waldmann’s group at the Max Planck Institute in Dortmund, Germany, began to publish some very interesting data on their work with solidphase syntheses around indolactam V (16), the core structure of the teleocidins (an example of which, teleocidin B, is shown in Chart 2 (17)). Indolactam V (16) was a known PKC modulator of both PKCR and PKCδ, whereas one of the 31 analogues made (18) only modulated PKCδ. 32–34 Also referred to in the above review by Breinbauer et al.34 is the work on the syntheses around the marine-sourced Cdc25 phosphatase inhibitor dysidiolide (19) and the only known inhibitors from nature of the her/neu tyrosine kinase, the nakijiquinones A-D (20a–d), also isolated from marine sources. The influence of these two very small focused libraries will be dealt with in section II.2 below. II.2. Syntheses around Privileged Structures Leading to Current Clinical Candidates and Approved Drugs. In addition to the compounds referred to above in section II.1 and those discussed below in section III, there are currently agents in clinical use and in advanced clinical trials that further exemplify the utility of “compounds based on natural product structures be they partially or totally synthetic”. Perhaps the best examplar in the current crop of novel antitumor agents in clinical trials would be the work done by chemists at the Eisai Research Institute utilizing the original work by Kishi’s group on the synthesis of halichondrin B (21), which led after 200 modifications/molecules to the compound currently known as E7389 (eribulin, 22), which is a novel tubulin inhibitor now in phase III clinical trials against breast cancer in the U.S. and the EU.35 Though totally synthetic, the basic halichondrin scaffold can be seen quite clearly on comparison of the two compounds. A further series of examples are those based on modification of the well-known natural product rapamycin (23) (effectively by modification at one site), which has led to three clinical drugs (sirolimus 24, everolimus 25, and temsirolimus (CCI-779) 26) and to one in phase II clinical trials (deforolimus (A23573) 27) (Chart 3). In all cases, modifications were made in the one area, the C-43 alcoholic hydroxyl group that avoids both the FKBP- 12 and the target of rapamycin (TOR) binding sites, since modifications in other areas would negate the basic biological activity of this molecule.36 For a further discussion on modification of molecules from a similar biosynthetic source (polyketide immunophilin ligands) one should consult the recent review by Koehn.37 III. Inter- and Intramolecular Interactions of Compounds and Proteins III.1. Introduction. The interactions of small molecules with proteins leading to a change in tertiary structure, and in the case of complexes of protein subunits, quaternary structure, were probably first recognized as a result of the 1960s X-ray crystallographic work of the Perutz group on the hemoglobin molecule on binding of oxygen, which was reported in 1970.38 The shift in the contact faces between the heterodimeric R MiniperspectiVe Journal of Medicinal Chemistry, 2008, Vol. 51, No. 9 2591 Downloaded by CKRN CNSLP MASTER on August 8, 2009 Published on April 5, 2008 on http://pubs.acs.org | doi: 10.1021/jm0704090

hemoglobin dimers that occurred on initial binding of oxygen and the increase in binding strength/rate of addition of further oxygen molecules to the individual heme units were shown to be directly related to the interaction between the oxygen and the heme prosthetic group, with the subsequent structural perturbation(s) transmitted through the whole complex. For many years, substrate interactions at the “active/binding site” of an enzyme were thought to be (relatively) specific to a given compound/enzyme/receptor set, though there were some examples of where a given ligand might interact with different “enzymes/proteins”. Such an example is the interaction of the opioid methadone with both the µ-opioid receptor, a G-proteincoupled receptor (GPCR), and the N-methyl-D-aspartate receptor (NMDA),39 with the possibility that both may be necessary for the full action of the drug.40 Similarly, the classical privileged structures, the benzodiazepines, bind to ion channels which is their usually recognized mode of action, but also affect mitochondrial proteins.41 Then from a potentially significant medical aspect, two compounds with formally quite dissimilar activities, astemizole, an H1 receptor inhibitor (28), and cisapride, a 5-HT4 agonist (29), may both lead to unexpected cardiac events because of their unexpected inhibition of the hERG ion channel.42 There are two groups of investigators (one in Germany and the other in Australia) who over the past few years have used such “unexpected protein-ligand interactions” as the basis to discover and develop drug candidates and biological probes based on natural products or on synthesized compounds based on such molecules. Each will be discussed in turn, and very recently, a third group (in the U.S.) has used the ligand (or pharmacophore) to investigate/discover relationships among proteins but have not limited themselves to the use of only natural products or their related structures. As will become apparent, the third iteration is reminiscent of a melding of the first two processes. III.2. Waldmann’s Methods. The earliest reported and currently best described process was based on the realization by Waldmann that “natural products were biologically validated starting points for library design”.34 In a series of elegant combinatorial syntheses around natural product structures, he Chart 2 2592 Journal of Medicinal Chemistry, 2008, Vol. 51, No. 9 MiniperspectiVe Downloaded by CKRN CNSLP MASTER on August 8, 2009 Published on April 5, 2008 on http://pubs.acs.org | doi: 10.1021/jm0704090

and his group have demonstrated that agents with increased biological activities, and in a significant number of cases, activities against targets that had not previously had inhibitors identified, could be produced relatively easily. In the initial series of experiments as mentioned above, the structures built around the marine-derived putative phosphatase inhibitor dysidiolide 19 led to activity against a variety of cancer tumor lines and activity against phosphatase Cdc25c, though the in vitro and in vivo activities did not track completely.43 Nakijiquinone C 20c was demonstrated to be an inhibitor of epidermal growth factor receptor (EGFR: her2/neu is a protooncogene from this class of receptors), in addition to having activity against c-ErbB2 and PKC, and cytotoxic to L1210 and KB cell lines.44 Waldmann’s group tested the compounds that they had synthesized, where they had made sequential changes to the terpene moiety and the aminoacids while keeping the quinone moiety relatively constant (specific details given in the original paper)45 against her2/neu and also against a suite of tyrosine kinases.46 From this initial library, the Waldmann group discovered not only inhibitors of the vascular endothelial growth-factor receptor-2 (VEGFR-2) (30) and insulin-like growth factor 1 receptor (IGF1R), albeit at micromolar levels, but also four inhibitors of the Tie-2 kinase, (31-34) a protein involved in angiogenesis and for which, at the beginning of their studies, no inhibitors were known. However, during the studies, the first natural product inhibitor (35) was reported from the plant Acacia aulacocarpa. 47 This report was followed by three related papers reporting synthetic pyrrolopyrimidines48–50 and substituted pyrazolopyrimidines51 demonstrating this activity. Using his data, Waldmann applied the following logic. Proteins may be regarded as biomolecules built up from individual building blocks (domains) that are parts of the overall protein that fold “independently” from the rest of the structure to give compact arrangements of secondary structures such as R-helices, -sheets, and -turns.46 These individual domains are interconnected by relatively short peptide linkers. Chart 3 MiniperspectiVe Journal of Medicinal Chemistry, 2008, Vol. 51, No. 9 2593 Downloaded by CKRN CNSLP MASTER on August 8, 2009 Published on April 5, 2008 on http://pubs.acs.org | doi: 10.1021/jm0704090