Sunteți pe pagina 1din 15

REVIEWS

THE EVOLVING ROLE OF NATURAL


PRODUCTS IN DRUG DISCOVERY
Frank E. Koehn and Guy T. Carter
Abstract | Natural products and their derivatives have historically been invaluable as a source
of therapeutic agents. However, in the past decade, research into natural products in the
pharmaceutical industry has declined, owing to issues such as the lack of compatibility of
traditional natural-product extract libraries with high-throughput screening. However, as
discussed in this review, recent technological advances that help to address these issues,
coupled with unrealized expectations from current lead-generation strategies, have led to a
renewed interest in natural products in drug discovery.

PHARMACOPHORE
Chemical substances derived from animals, plants and small-molecule natural-product patents granted in
The ensemble of steric and microbes have been used to treat human disease since the years 1984 2003. The chart shows a period of
electronic features that is the dawn of medicine. The investigation of natural increasing patent activity through the 1980s, a flatten-
necessary to ensure optimal products as source of novel human therapeutics ing or even slight decline from 1990 to 1999, and a
interactions with a specific
biological target structure
reached its peak in the Western pharmaceutical industry pickup of activity between 2000 2003. In the context
and to trigger (or to block) its in the period 1970 1980, which resulted in a pharma- of worldwide pharmaceutical R&D spending, which
biological response. ceutical landscape heavily influenced by non-synthetic essentially tripled from roughly US $10 billion to US
molecules. Of the 877 small-molecule New Chemical $30 billion over the same period, the overall trend in
NEW MOLECULAR ENTITY Entities (NCEs) introduced between 1981 and 2002, the 1990s can be viewed as a relative decline. What are the
(NME). A medication containing
an active ingredient that has not roughly half (49%) were natural products, semi-synthetic underlying reasons for the past decline, and what are
been previously approved for natural product analogues or synthetic compounds the future prospects for natural-product research in
marketing in any form. based on natural-product PHARMACOPHORES1. drug discovery?
Despite this success, pharmaceutical research into The decreased emphasis in the pharmaceutical
COMBINATORIAL CHEMISTRY
The generation of large
natural products has experienced a slow decline during industry on the discovery of natural products during
collections, or , of the past two decades. Although these trends seemed the past decade can be attributed to a number of factors:
compounds by synthesizing obvious to investigators working in the field, their first, the introduction of high-throughput screening
combinations of a set of downstream effects are somewhat difficult to measure (HTS) against defined molecular targets, which promp-
smaller chemical structures.
precisely, given the long product-development cycles ted many companies to move from natural products
encountered in the pharmaceutical industry. The extract libraries towards so-called y syn-
lengthy delay usually ten years or more between thetic chemical libraries; second, the development of
WyethResearch, the initial discovery of a potential therapeutic agent and COMBINATORIAL CHEMISTRY, which at first offered the
401 North Middletown subsequent market launch of a NEW MOLECULAR ENTITY prospect of simpler, more drug-like screening libraries
Road, PearlRiver,
New York 10965, USA.
(NME) means that agents reaching the market today are of wide chemical diversity; third, advances in molec-
Correspondence to F.E.K typically the products of discovery research programmes ular biology, cellular biology and genomics, which
& G.T.C initiated at least a decade ago. increased the n um ber of molecular targets a nd
e-mails: However, it is possible to gain a reasonable picture prompted shorter drug discovery timelines; fourth, a
KOEHNF@wyeth.com; of natural-product pharmaceutical discovery research declining emphasis among major pharmaceutical
CARTERG@wyeth.com
doi:10.1038/nrd1657
by examining the general trends found in patent sta- companies on infectious disease therapy, a tradi-
Published online 24 February 2005 tistics. FIGURE 1 shows worldwide pharmaceutical tional area of strength for natural products 2 ; and last,

206 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

400

Total worldwide natural-product patents


Original natural-product patents

300

Number of small-molecule
patents 200

100

0
1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Figure 1 | Worldwide pharmaceutical natural-product patents. Total natural-products patents are shown in gold. The data are
for all worldwide grants for patents claiming composition-of-matter or use of small-molecule natural products as pharmaceuticals.
A single natural product can give rise to several patents based on filings in multiple countries, or for multiple indications. Original natural-
product patents are shown in orange. The data are for all first-time grants of patents claiming novel composition of small-molecule
natural products as pharmaceuticals. These statistics were derived by systematic search of the Derwent World Patents Index.

possible uncertainties with regard to collection of bio- representative combinatorial, synthetic and natural-
materials as a result of the 1992 Rio Convention on product compound libraries on the basis of molecular
Biological Diversity3. diversity and - properties such as molecu-
The underlying reasons for these industry trends are lar mass, number of chiral centres, molecular flexibility
as much commercial as they are scientific, particularly in as measured by number of rotatable bonds and ring
the case of research into infectious disease4. As a result of topology, distribution of heavy atoms, and LIPINSKI-type9
these factors, y drug discovery environment calls for descriptors. Other investigators have differentiated
rapid screening, hit identification and hit-to-lead devel- natural products, trade drugs or other synthetic molecu-
opment. In this environment, traditional resource- lar libraries on the basis of scaffold architecture and
intensive natural-product programmes that are based pharmacophoric properties 10 , or other molecular
on extract-library screening, bioassay-guided isolation, descriptors11. These studies reveal that natural products
structure elucidation and subsequent production scale- typically have a greater number of chiral centres and
up face a distinct competitive disadvantage when com- increased steric complexity than either synthetic drugs
pared with approaches that utilize defined synthetic or combinatorial libraries8,12. Although drug and com-
DRUG-LIKE
Sharing certain characteristics
chemical libraries. However, emerging trends, coupled binatorial molecules tend to contain a significantly
with other molecules that act as with unrealized expectations from current R&D strate- higher number of nitrogen-, sulphur- and halogen-
drugs. The set of characteristics gies, are prompting a renewed interest in natural prod- containing groups, natural products bear a higher
size, shape and solubility in ucts as a source of chemical diversity and lead generation. number of oxygen atoms 8,11 . Multivariate statistical
water and organic solvents
analysis of molecular descriptors shows that natural
varies depending on who is
evaluating the molecules. Natural products, chemical diversity and HTS products differ significantly from synthetic drugs and
Characteristics of chemical diversity found in natural combinatorial libraries in the ratio of aromatic ring
IPIN I -OF- I products and synthetic libraries. Current thinking in atoms to total heavy atoms (lower in natural products),
analysis of the World the generation of drug leads embodies the concept of number of solvated hydrogen-bond donors and accep-
Drug Index led to the -of-
,which identifies several key
achieving high molecular diversity within the bound- tors (higher in natural products) and by greater molec-
properties that should be aries of reasonable DRUG-LIKE properties 5,6 . It has long ular rigidity. Natural-product libraries also have a
considered for small molecules been recognized that natural-product structures have broader distribution of molecular properties such as
that are intended to be orally the characteristics of high chemical diversity, biochemical molecular mass, octanol water partition coefficient
administered. These properties
specificity and other molecular properties that make and diversity of ring systems compared with synthetic
are: molecular mass <500 Da,
number of hydrogen-bond them favourable as lead structures for drug discovery, and combinatorial counterparts8,10,11. Indeed, less than
donors <5; number of and which serve to differentiate them from libraries of one-fifth of the ring systems found in natural products
hydrogen-bond acceptors <10; synthetic and combinatorial compounds 7 . Various are represented in current trade drugs. A perhaps unex-
calculated octanol water investigators have worked to measure by means of pected finding is that of Schneider and Lee who
partition coefficient (an
indication of the ability of a
computational chemistry those desirable chemical revealed that the fraction of natural product structures
molecules to cross biological features that distinguish natural products from other with two or more -of- 9 violations is quite low

membranes) <5. sources of drug leads. Feher and Schmid 8 examined (approximately 10%) and equal to that of trade drugs11.

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 0 7


REVIEWS

Recent advances in genomics and structural biology


O OH during the past 5 10 years are painting a clearer picture
H
O N
N of the diversity of proteins targeted by natural-product
O MeO H
H molecules. It has been shown that the number of
O unique protein architectures (or folds) in nature is
O
H OMe
H MeO much smaller than the number of protein families pre-
O
O OMe dicted by sequence similarity16,17. Structural motifs of
OMe
domains, the functional building blocks of proteins, are
OMe
Lovastatin Reserpine frequently conserved even when there is a low degree of
Target: HMG Co-A reductase Target: neurotransmitter vesicles
Cholesterol-lowering agent Antihypertensive/sedative sequence similarity. These units are combined in
modular fashion to fine-tune the function of full pro-
HO OH
teins18,19. This indicates that proteins populate the total
HO O FOLD SPACE in a highly non-uniform fashion, such that
O
NH localized concentrations of receptive binding are
N
H distributed onto clusters of superfolds20. Furthermore,
H2N N O
O HN OH
N the same protein-binding target or fold can serve differ-
O O ing functional roles in a number of higher organisms21.
HO NH O On the basis of this concept, a guiding principle has
O
HO
H
N
N emerged that natural products, by virtue of their molec-
O
O
OH
ular evolution to preferentially bind to these folds, are
O
OH O O validated starting points for screening-library design22.
O
HO3SO
H H This was the basis for work in the 1990s that effectively
O utilized molecular scaffolds based on yohimbine 23 ,
H HO

Artemisinin Micafungin
paclitaxel24 and vancomycin25,26 for SOLID-PHASE SYNTHESIS
Target: hemin Target: (1,3) Glucan synthase of focused libraries. Recent investigations have success-
Antimalarial Antifungal fully produced high-quality screening-type libraries by
Figure 2 | Examples of natural-product drugs. Natural products have become effective
solid-phase synthesis based on privileged natural-
drugs in a wide variety of therapeutic indications, as illustrated by the compounds shown, which product benzopyran scaffolds27,28 and focused libraries
modulate a diverse range of targets. HMG Co-A, 3-hydroxy-3-methylglutaryl coenzyme A. by parallel synthesis to find selective inhibitors of
receptor tyrosine kinases29.
Aside from binding in the active sites of enzymes
and acting as inhibitors of catalysis, the propensity of
Suitability of natural products for modulating bio- natural products to bind to a variety of protein
chemical reactions and proteinprotein interactions. domains and folding motifs is borne out by biological
Contemporary drug discovery is based in large part on activity of another type the capacity to modulate or
the screening of small molecules for their ability to inhibit protein protein interactions. As a result, these
bind or otherwise inhibit specific macromolecular molecules are effective modulators of cellular processes
(usually protein) targets. Given the virtually infinite such as the immune response, signal transduction,
number of small-molecule structures that can be gen- mitosis and apoptosis. An enzyme inhibitor must bind
erated, it is imperative that a molecular screening to the active site at the interior of the protein, in a man-
library covers a significant portion of chemical diversity ner which involves relatively few interactions. By con-
space, but is also favourably biased trast, a small-molecule disruptor of a protein protein
and - 13. Some of the first large
interaction must bind to the protein surface, where
combinatorial libraries, in some instances containing in tight binding involves a larger surface area with multiple
excess of one million compounds, were synthesized points of interaction. By virtue of their increased size,
only to find disappointingly low hit rates, or no hits at and natural selective pressure to bind gene products,
all. These libraries were designed more on the basis of natural products are effective scaffolds for this function30.
chemical accessibility and maximum achievable size Moreover, similar natural-product scaffolds can be
than on biologically relevant chemical diversity or structurally fine-tuned by nature to modulate protein
properties 14 . This is consistent with the concept that a protein interactions selectively, on the basis of a single
FOLD SPACE fruitful area in which to search for drug leads is the starting protein interaction. Perhaps the best known
The total repertoire of three- area of diversity space in which the chemical scaffolds example of this is the case of FK50631, rapamycin32 and
dimensional protein structures embody characteristics that promote binding to multi- ascomycin33,a family of closely related polyketide natural
or architectures.
ple target proteins so-called privileged structures 15 . products derived from soil actinomycetes that show
SOLID-PHASE SYNTHESIS
Natural products can be viewed as a population of profoundly different cellular effects via binding of the
Synthesis of compounds on the privileged structures selected by evolutionary pres- FK506-binding protein immunophilin proteins and
solid surface of an insoluble sures to interact with a wide variety of proteins and modulation of the protein protein interactions
resin support, which allows
other biological targets for specific purposes, a view involved in the signal transduction pathways of T-cell
them to be readily separated
(by filtration or centrifugation)
supported by the fact that natural products have activation and growth. All three of these compounds
from excess reagents, soluble become effective drugs in a wide variety of therapeutic FK506 (tacrolimus (Protopic; Fujisawa)), rapamy-
reaction by-products or solvents. indications (FIG. 2). cin (sirolimus (Rapamune; Wyeth)) and ascomycin

208 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

(pimecrolimus (Elidel; Novartis)) are now in clinical high-throughput screening (HTS). The natural-product
use for different therapeutic indications (FK506 and library itself might be composed of samples that are
rapamycin for liver and kidney transplantation, respec- either themselves mixtures such as crude extracts
tively, and ascomycin for atopic dermatitis) 34 . Natural (10 100s of components), semi-pure mixtures (5 10
products of this class have also recently shown potential compounds) or, alternatively, single purified natural
in the treatment of neurodegenerative disease35. products. In the case of pure libraries, the hit-detection
A different theme is evident when we examine the process is the same as that for synthetic pure libraries. In
large family of natural products that affect mitosis by the first case, however, heterogeneity of the natural-
modulating the protein protein interactions involved in product library samples adds two additional levels of
tubulin polymerization (FIG. 3). In this case, there is complexity to the screening process. The first of these is
broad diversity in the structure and mechanism of that once a response for the sample is detected by HTS,
tubulin-binding molecules found in plants, animals and one or more rounds of chemical purification and bio-
microorganisms 36 . Agents such as the plant-derived logical assay might be necessary for identifying and iso-
taxane paclitaxel and the marine-sponge-derived disco- lating the active component in the mixture. This is
dermolide 37 bind to microtubules and enhance their described in greater detail later in this article.
polymerization. The vinca alkaloids, along with the The second additional hurdle is that the complexity
macrocyclic polyether spongiastatin 38 and the recently of crude or semi-pure natural-product libraries, and
described marine tripeptide hemiasterlin39, bind to the the chemical nature of many of the components found
tubulin dimer vinca domain and inhibit microtubule in them, often challenges the robustness of HTS tech-
polymerization. Others, such as colchicine, podophyllo- nology. Before the advent of biotechnology, the diffi-
toxin, steganacin and combretastatin, disrupt micro- culty of obtaining purified protein targets directly
tubule assembly by binding to a site on soluble tubulin from tissues necessitated that much of the compound
distinct from that of the vinca alkaloids40. At low con- screening was performed using cellular in vitro or even
centrations, the actions of different agents can converge whole-animal systems. When applied to crude or other-
to a similar mechanism affecting tubulin dynamics40. wise partially purified natural-product libraries, these
platforms have a relatively good capacity to detect
Natural-product libraries and high-throughput screening. active components. Examples can be found in the
The exercise of identifying natural-product (and also antimicrobial in vitro and in vivo assays used in the 1970s
synthetic) molecules as potential drug leads most often and 1980s, which effectively served to detect many
involves automated testing of large collections (libraries) novel antibiotics. Advances in genomics and molecular
of compounds for activity as inhibitors or activators of biology have now made it possible to obtain relatively
a specific biological target, such as a cell-surface recep- large quantities of protein targets for high-throughput,
tor or enzyme. This process is commonly known as cell-free assay systems for directly detecting the catalytic
inhibition or target binding 41 . These developments
have greatly expanded the number of targets amenable
to HTS, but at the same time have introduced technical
complexities into the screening of natural-product
O libraries. These complications reduced the success rates
O
O O OH for lead discovery from natural-product libraries in the
NH
O early and mid-1990s, but technical improvements during
O the past few years have circumvented the problems in
S
O
HO O many HTS platforms.
HO HO N OH
O O An illustrative example is provided by the case of pro-
O O tein kinases, enzymes that catalyse the phosphorylation
O O
by ATP of protein tyrosine or serine/threonine residues,
OH
and which have a central role in cellular signalling path-
Paclitaxel Epothilone A ways that control the activation, growth and proliferation
Taxus breviffolia Sorangium cellulosum
of cells. As drug targets, protein kinases are superseded in
importance only by the G-protein-coupled receptors42.
O
MeO Using expressed and purified kinases, numerous HTS
NH technologies have been developed for detecting specific
O O MeO kinase substrate (ATP) inhibitors. Initially, assays were
OMe based on measuring the transfer of radioactive phos-
OH OH O O
O
phate ( 32 P or 33 P) from ATP to a protein or synthetic
OH H2N OMe peptide substrate a robust, but low-throughput, tech-
Discodermolide Colchicine nique. Newer HTS kinase assay platforms are based on
Discodermia dissoluta Colchiicum autumnale
fluorescence intensity, time-resolved fluorescence or flu-
Figure 3 | Structural themes in natural products that target tubulin proteinprotein orescence polarization 43 . Early complications in using
interactions. Microtubules, dynamic assemblies of the protein tubulin, are essential components these assays for screening natural-product libraries
of cells involved in mitosis. The natural products shown here inhibit cellular proliferation by came from two sources. First, for detecting competitive
disrupting tubulin polymerization and de-polymerization by different binding mechanisms. substrate inhibitors, it is best to screen at a compound

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 0 9


REVIEWS

concentration close to the Km of the enzyme for ATP, highly abundant components can exhibit inhibition as a
typically micromolar. This poses a challenge with mix- result of nonspecific binding, perturbation of the assay
ture natural-product libraries in which the relative con- pH or other physical properties. Second, natural-product
centrations of individual components in a sample are samples can contain compounds that either fluoresce or
not known precisely, and might vary by two orders of absorb at excitation or emission wavelengths of the
magnitude or more. Consequently, at any given assay fluorophore (typically fluorescein), or by light scatter-
dose, the concentration levels of trace components ing of insoluble components, which therefore affect
might not be high enough for detection, whereas the the readout of the assay44. This liability is shared with

Natural lipophilic isovaleryl C2C4 ester analogues on SAR from natural esters is used to design
disaccharide show greater potency; C4 position is optimal optimized stable adamantyl-modified dissacharide

OH b OH OH c
OH OH O O O
HO HO HO O
O
O OH O O OH HO
O O O
O O
O OH
O O O O O OH O O OH O
Core Core Core O O OH
HO OH HO OH HO OH Core
MIC (Staphyloccocus aureus): MIC (Staphyloccocus aureus): MIC (Staphyloccocus aureus): HO OH
8 g per ml 48 g per ml 4 g per ml

a f
OH
OH OH NH OH O
HO NH
OH OH OH O
HN NH O OH HO
OH
O H O HN NH
OH O OH
OH O O H O
H OH O H OH O
HN N O OH
OH O O
H OH O H OH
NH HN HO OH HN N O
HN
O H NH HN HO OH
HN
NH HN O H MIC (Staphyloccocus aureus):
O e NH HN 0.251 g per ml
HO NH HN O
Core HO
Adamantyl substitution on
O NH HN
disaccharide and cyclohexyl
O O are combined to produce
O semisynthetic analogue
Mannopeptimycin Cyclohexyl group in natural
MIC (Staphyloccocus aureus): 64 g per ml with optimized potency and
analogues is more potent in vivo activity
d OR1
OR4 OH NH OH
R2O
OH
HN NH
O OH
O H O
OH
H OH O H
OH O O OH R1 = nC6H11CO, R2, R3 and R4 = H MIC = 12 g per ml
HN N
O R2 = nC6H11CO, R1, R3 and R4 = H MIC = 48 g per ml
HO OH
HN NH HN
R3 = nC 6H11 CO, R1, R2 and R4 = H MIC > 128 g per ml
O H
R4 = nC6H11CO, R1, R2 and R3 = H MIC > 64 g per ml
NH HN
O
R3O NH HN Shotgun approach yields esters at four primary alcohol
O positions. Only substitutions at R1 and R2 retain potency,
O therefore modifications focused on the disaccharide.
Figure 4 | Using natural-product structureactivity relationships and shotgun synthesis to optimize biological activity:
the example of mannopeptimycins. The mannopeptimycins were originally isolated in the late 1950s as a complex of novel
antibiotic glycopeptides that show activity predominantly against Gram-positive bacteria51,52. However, the structural complexity,
along with the lack of broad-spectrum activity, reduced the urgency to further develop the antibiotic. The need to find new antibiotics
with specific activity against resistant strains, and advances in methods for purifying and characterizing the components of natural
product complexes led to re-examination of the mannopeptimycins in the 1990s. The mannopeptimycin complex consists of five
separate members, designated mannopeptimycin . The molecules consist of a cyclic hexapeptide core that contains two
stereoisomers of the amino acid, -amino--[4-(2-iminoimidazolidinyl)]--hydroxypropionic acid, which is glycosylated with
mannose mono and disaccharide moieties that in some cases bear isovalerate esters. a | The major component of the complex,
mannopeptimycin , is not esterified and has poor antibacterial activity. Initial structureactivity relationship studies, shown in the
figure, focused primarily on modifications to the disaccharide. Natural isovaleryl ester regio-isomers provided the initial direction for
SAR experiments, leading to c, a stable disaccharidic modification. d | Random (shotgun) esterification of primary alcohols quickly
yields SAR information about disaccharide positions necessary for potency. e | Cyclohexyl group in structural analogues were found
to be more potent. f | Combining optimal modifications to the core and disaccharide leads to a highly potent, optimized semi-
synthetic analogue. MIC, minimum inhibitory concentration; SAR, structureactivity relationship.

210 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

Processing
Initial bioactive sample experiments

Bioassay-guided
Bioactive concentrate
fractionation

Structure
Identification Pure bioactive compound determination
of compounds

Evaluation of 'natural SAR'


'Shotgun' synthesis of derivatives
Known compounds Novel bioactive compound Scale-up purification

Medicinal
Potential lead
chemistry

Development track candidate

Amount of compound required 110 g 110 mg 110 g 100 g

Figure 5 | Chemical process for natural product discovery. The natural product is extracted from the source, concentrated,
fractionated and purified yielding essentially a single biologically active compound. Identification of known compounds, thereby
avoiding replication of previous efforts, has been greatly aided by directly coupled HPLC-mass spectrometer (LC-MS) systems
and natural-product databases57. De novo structure determination of compounds that are novel has been revolutionized by
advances in spectroscopic techniques, particularly in high-resolution nuclear magnetic resonance technologies. Although the
determination of complex structures is technically challenging, it is no longer a major impasse in the drug discovery process.
In those cases in which the biological activity profile meets criteria for potency and selectivity, preliminary structureactivity
relationship (SAR) studies are conducted and the purification process is scaled up. Once the feasibility of modulating biological
response through synthetic modification is established, the hit is declared a lead and proceeds onward for additional
optimization by traditional medicinal chemistry.

synthetic screening libraries, but the issue is exacerbated HTS analysis of natural-product libraries using mass
with natural products because the presence of these spectrometry for example, electrospray ionization
interference compounds might not be fully character- Fourier transform ion cyclotron resonance mass
ized. However, it has now been shown that in kinase, spectrometry (ESI-FTICR) screening against RNA
protease and phosphatase assays, fluorescent-compound targets48 . These and other developments make it likely
and light-scattering interferences can be overcome by that future lead generation through the screening of
increasing the fluorophore concentration in the assay, by natural products need not be limited by HTS technology.
using red-shifted wavelength dyes45 or by the newer
technique of LIFETIME DISCRIMINATED POLARIZATION44. Structureactivity and structureproperty relation-
An effective strategy that helps to alleviate this ships of natural products. One of the distinguishing
interference, as well as shorten the time needed to features (or perhaps historical drawbacks) of natural
eventually isolate the active principle, is the imple- products is their frequent occurrence as suites or com-
mentation of purified or fractionated samples for plexes of structurally related analogues. The biological
screening from the original crude extract. There are significance of this expressed molecular diversity is
many variations of this theme and each offers advan- unclear, particularly when the suite contains members
tages of expediency or purity46,47. The samples produced (often major) that seem to lack biological activity or
by this fractionation approach are less complex mix- function. Essentially, why would an organism expend
tures for use in screening. The result of this procedure the resources needed to synthesize many analogues of
is that the final resolution of the active components is a molecule for which there is but a single purpose? One
simplified, typically only requiring one additional purifi- possible answer to this question is furnished by the
LIFETIME DISCRIMINATED cation step. Of potentially greater significance is that y , which is based on the proposi-
POLARIZATION interferences are reduced by the simplification of the tion that biological activity is a rare property for any
A method of reducing test- mixtures and the relative concentration of minor molecule to possess, and that there might be a selective
compound interference in
fluorescence-based screening
components is increased, thereby enhancing the advantage to an organism if it can generate and retain
by rejection of signals from opportunity to uncover novel biologically active metabo- chemical diversity at a low cost49. The existence of con-
short-lifetime sources. lites. Significant successes have also been achieved in the geners of a natural product series might therefore be

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 1 1


REVIEWS

(HPLC), nuclear magnetic resonance (NMR) a nd


Box 1 | Nuclear magnetic resonance spectroscopy experiments
mass spectrometry with which to purify and charac-
Nuclear magnetic resonance (NMR) spectroscopy experiments that are used to elucidate terize the components. The mannopeptimycins are a
the molecular structures of natural products are based on the application of pulse complex of novel antibiotic glycopeptides produced
sequences precisely timed radio-frequency and magnetic-field gradient pulses (usually by the Streptomyces hygroscopicus strain LL-AC98,
on the microsecond and millisecond timescale) that are designed to excite the atomic and have activity against methicillin-resistant staphylo-
nuclei of molecules and thereby produce diagnostic signals that can be analysed to cocci and vancomycin-resistant enterococci. The
determine the connectivity of the 1H, 13C and 15N nuclei of the molecule. More than mannopeptimycins inhibit cell-wall biosynthesis
1,000 of these pulse sequences have been developed, each designed to ascertain a different
through a unique mode of action, by binding to the
type of physical information about the molecule under study. Due to the many types and
membrane- bound cell-wall precursor lipid II (C 35 -
variations of pulse sequences, a system of acronyms has evolved in the field of NMR.
MurNAc-peptide-GlcNAc) in a different manner to
Listed here are a few representatives of the commonly used pulse-sequence acronyms a
reader might encounter in the literature of natural-product structure determination.
other lipid II binders 53 . They also bind lipoteichoic
acid in a way that serves to concentrate the antibiotic
COSY (COrrelated SpectroscopY) and TOCSY (TOtal Correlation SpectroscopY) on the cell surface.
Provides molecular connectivity information based on protonproton interactions
The mannopeptimycin complex as described 51
through covalent bonds and results in connectivity maps of proton-bearing carbon atoms.
consists of five separate members, designated manno-
NOESY (Nuclear Overhauser Enhancement SpectroscopY) peptimycin . The molecules consist of a cyclic hexa-
Provides molecular connectivity information based on protonproton through-space
peptide core that contains two stereoisomers of an
interactions. Also provides information about proton inter-atomic distances and
unprecedented amino acid, -amino--[4-(2-imi-
molecular conformation.
noimidazolidinyl)]--hydroxypropionic acid (Aiha),
HSQC (Heteronuclear Single Quantum Coherence Spectroscopy) which is glycosylated with mannose mono- and disac-
Provides connectivity information relating specific carbon atoms and the protons
charide moieties that in some cases bear isovalerate esters.
bound to them.
The major component of the complex, mannopepti-
HMBC (Heteronuclear Multiple Bond Correlation Spectroscopy) mycin is not esterified and has poor antibacterial
Provides long-range connectivity between protons and carbon atoms separated by two
activity. Strikingly, the presence and location of the iso-
to four covalent bonds.
valeryl group is crucial for the retention of antibacterial
INADEQUATE (Incredible Natural Abundance DoublE QUAntum TransferExperiment)
activity. This limited but important initial SAR was
Provides direct connectivity maps of covalently bound carbon atoms.
instrumental in designing a synthesis programme that
furnished analogues with improved antibacterial
potency and a favourable therapeutic window54 56 (FIG. 4).
the consequence of an need to generate its own
chemical diversity to optimize the activity of its secondary Technologies for natural-product lead discovery
metabolites, essentially doing its own structure activity The typical process of discovering natural-product hits,
relationship (SAR) optimization. and their progression towards development, is depicted
In any event, due to limitations in separation tech- in FIG. 5. In this generic scheme, the natural product is
nology, and methods for structure determination and extracted from the source, concentrated, fractionated
characterization, the occurrence of natural products and purified, yielding essentially a single biologically
as a complex has historically presented a challenge to active compound. Historically, this process has suffered
their development as drugs. Indeed, important natural- most often from three major hurdles. The first is the
product therapeutics such as ivermectin50 were devel- rapid identification of known compounds (dereplica-
oped and marketed in the past as complexes because it tion), to avoid the duplication of effort. This step has
was not possible to purify the individual components been greatly facilitated by the availability of reliable
at a sufficient scale. However, recent technological directly coupled HPLC-mass spectrometer (LC-MS)
advances have allowed natural-product investigators systems, and the general availability of natural-product
to unravel theses complexes, and exploit the chemical databases 57 . The pivotal development responsible for
diversity presented by them to obtain analogues with the success of LC-MS has been the introduction of effi-
improved potency and drug-like properties. An illus- cient and general methods for producing ions from the
tration of this principle is found in the mannopepti- effluent of HPLC separations. The most general of these
mycin family of antibiotics 51 (FIG. 4). The complex was methods, known as electrospray ionization (ESI) and
originally isolated in the late 1950s, when it was atmospheric pressure ionization (API), can generate the
shown to have potent activity predominantly against ions essential for mass spectrometric analysis for greater
Gram-positive bacteria 52 . At that time the chemical than 90% of analytes, ranging from amino acids to pro-
complexity of the complex, along with the lack of teins and nucleic acids. The correlation of both molecular
broad-spectrum activity (at that time a drawback for mass and UV absorption data with known compounds
antibiotics), reduced the urgency to further develop by database searching is ordinarily sufficient to classify
the antibiotic. sets of compounds 58 .
Re-examination of the complex in the late 1990s The second major hurdle in the process the
was prompted by the need to find new antibiotics de novo structure determination of compounds that
with specific activity against resistant strains of are NMEs is an area that has been revolutionized
Gram-negative organisms, and the availability of new by many advances in spectroscopic techniques, partic-
advances in high-performance liquid chromatography ularly in high-resolution NMR technologies. Of the

212 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

many NMR advances, those of particular importance 1H- 1 H and 1 H-[ 13 C, 15 N, 31 P] correlations and 1 H- 1 H
to natural-product structure determination fall into dipolar (through space) molecular connectivity data
one of two main areas: multidimensional pulse methods that essentially map out the structure of the com-
and sensitivity improvements. From its inception, and pound (BOX 1). In the area of sensitivity, stronger mag-
particularly since the advent of two-dimensional netic fields provided by superconducting magnets 61 ,
NMR methods 59 , high-resolution NMR spectroscopy cryogenic electronics62 and micro-probe technologies63
has seen continuous development and expansion of have dramatically lowered the amount of material
the array of experimental methods available to elucidate needed for structural analysis, to less than a milligram.
chemical structures 60 . New experiments, particularly The combination of cryogenic probe electronics with
multidimensional ones, provide scalar (through bond) correlation spectroscopy enables the development of

Box 2 | Synthesis of natural products to overcome supply problems: discodermolide


(+)-Discodermolide, an
antitumour polyketide from the O O
Caribbean sponge Discodermia
OH OH O O
dissoluta, was first isolated and
characterized in 1990 by OH H2N
Gunasekera and co-workers on (+)-Discodermolide
the basis of its immuno-
suppressive and cytotoxic
properties68. Subsequent studies
MeO
showed that discodermolide Me
possessed potent antitumour O N
OMe
activity as a result of induction of
O
tubulin polymerization and
microtubule stabilization in a Common precursor
manner similar to taxol, but with adapted from Smith
synthesis
tenfold greater potency102.
Moreover, discodermolide is
highly potent against P-
I
glycoprotein (Pgp)-expressing MeO
OMe
multiple-drug resistant (MDR) O O
OTBDMS
37 O N
tumour cells , and has improved Me
O
aqueous solubility compared with OTBDMS
O
OTBDMS
existing agents. These attributes
qualified the compound as a MeO
strong candidate for development Chemistry adapted
from Marshall synthesis
as an anticancer agent. However,
the low isolation yield of (+)-
discodermolide (~14 mg per kg),
combined with the limited natural
supply of the producing O
organism, made it necessary to
H OTBDMS O
pursue chemical synthesis as a O
means to supply the compound H 2N
OTBDMS
for furtherdevelopment.
(+)-Discodermolide representsa Chemistry adapted
daunting target for large-scale from Paterson synthesis
chemical synthesis, by virtue of its
13 stereogenic centres and high
degree of chemically sensitive (+)-Discodermolide
functionality. Following the initial
report of its isolation, the compound quickly became a target for several academic groups, and several successful syntheses
were reported. The first was achieved by Schreiber et al., who synthesized both enantiomers103. Schreibers work enabled
the assignment of the absolute stereochemistry and provided valuable structureactivity relationship (SAR) data. Later
syntheses by Smith104, Paterson105 and others succeeded in improving efficiency, yield and economy. Finally, by blending
the common precursor methodologies developed in the gram-scale synthesis reported by Smith104 with chemistry
developed in syntheses of the Paterson105 and Marshall groups106, the team at Novartis succeeded in synthesizing 60 g of
(+)-discodermolide107 111. The remarkable story of (+)-discodermolide from the initial isolation of a novel, complex
and rare natural product, to the successful production of a clinical drug substance is testimony to the power of
combining modern natural-product, synthetic and process chemistry to overcome supply problems.

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 1 3


REVIEWS

Test LC or
fractions LC/MS
+ +
Active? Pure?
Fractionate
(HPLC, partition,
HP20 and so on) + + Pure compound:
Crude extract Fractions Active? Pure? Determine structure
(NMR, MS, UV,
IR and so on)
+
Active? Pure?

Figure 6 | Generic scheme for bioassay-guided fractionation. Several cycles of fractionation are usually needed to obtain a
pure compound. Multiple active components can illuminate structureactivity relationships. HP20, a solid-phase adsorber; HPLC,
high-performance liquid chromatography; IR, infrared spectroscopy; LC/MS, liquid chromatography/mass spectrometry; NMR,
nuclear magnetic resonance; UV,ultraviolet.

still more powerful experiments, such as correlation NMR. Bioassay of the various products indicated that
experiments for low-abundance 13C and 15 N nuclei 64, only those esters formed on the terminal mannose of
which are unattainable with conventional hardware. the disaccharide retained potency (FIG. 4). The know-
Today, highly complex molecular structures, such as ledge gained through understanding the natural SAR
those of maitotoxin65 or lomaiviticin66, are largely deter- and the shotgun approach provided an early foundation
mined with these NMR techniques and knowledge of on which the overall synthetic strategy was developed.
the new molecular formula. Once the feasibility of modulating biological response
Determination of the molecular formula is crucial through synthetic modification is established, the hit is
to the process and is typically done by high-resolution declared a lead and proceeds onwards for additional
mass spectrometry on microgram quantities of material. optimization by traditional medicinal chemistry.
One of the most powerful of these techniques is The major bottleneck that continues to affect natural
Fourier transform ion cyclotron resonance mass spec- product drug discovery is the isolation and purification
trometry (FT-ICR/MS), which is capable of measuring of the active principles from an exceptionally complex
molecular mass with exceptional accuracy 67 . Com- matrix. Often the target compounds represent much
bining the tools of high-resolution mass spectrometry less than 1% by weight of the crude extract, and the
with two-dimensional NMR spectroscopy allows struc- approach remains highly experimental. Although
ture determination to be carried out on sub-milligram advances in separation technology, such as HPLC,
or milligram amounts of a compound in a matter of supercritical fluid chromatography (SFC) and capillary
hours or days, rather than weeks or months. Although electrophoresis (CE) have had a major impact on
the determination of complex structures is technically resolving power, often the purification step in the
challenging, it is no longer a major impasse in the process is rate-limiting. The challenge is twofold: first,
drug discovery process. one must correlate the biological signal of interest with
In those cases in which the biological activity pro- the effector compounds; second, one must then devise
file meets criteria for potency and selectivity, prelimi- preparative separation methods to yield sufficient quan-
nary SAR studies are conducted and the process is tities of the pure material. This latter point impinges on
scaled up. Nature often provides the first clues to SAR one of the fundamental concerns in natural-product
in the form of biosynthetic congeners, as seen in the drug discovery, often referred to as the y .
case of the mannopeptimycins (described in the pre- Unlike simpler synthetic compounds, natural products
vious section (FIG. 4)). A second avenue for exploring SAR can be limited in supply owing to sourcing limitations
in an expeditious manner is the approach to or the impracticality of synthesis. As indicated at the
chemical derivatization. Again using the example of bottom of FIG. 5, material requirements rapidly escalate
mannopeptimycins, key functional groups required as one approaches the medicinal chemistry and devel-
for antibiotic activity were identified by allowing the opment phases. The supply issue is particularly crucial
parent compound to react with nonspecific derivatizing for source organisms, such as marine invertebrates (for
agents such as anhydrides. In this approach, all of the example, the discodermolide producer 68), that have not
14 primary and secondary hydroxyl groups might be been productively cultured in the laboratory, but also
expected to form ester derivatives. By restricting the stoi- arise with plant products, such as paclitaxel. Microbial
chiometry of the reagents and monitoring the reactions products, as well as some plant-derived agents, are
by HPLC, conditions were found in which a mixture of amenable to culturing on production scale, and syn-
mainly mono-esterified components was produced 54 . thetic methodologies continue to be developed for the
The various isomers were subjected to preparative large-scale synthesis of highly complex products.
HPLC purification and the regio-chemistry of the newly Perhaps the most compelling recent examples are those
formed ester groups was defined by two-dimensional of discodermolide, described in BOX 2, and E7389, an

214 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

Select Separate Characterize

Natural
product PDA MS MSn
mixture

NMR

Effluent UV-visible spectra Mw

Identification

Fragment ions Substructures

Structure determination
Figure 7 | Affinity-based identification system for natural products. Affinity selection is performed on the first column of this
high-performance liquid chromatography-based system. Selected compounds are resolved into single components on the second
column, which are then interrogated by a series of detectors. A combination of UV spectra and molecular mass leads to the
identification of known compounds, whereas more extensive structural information is obtained for unknowns using tandem MS and
NMR. MS, mass spectrometry; NMR, nuclear magnetic resonance; PDA, photodiode array; UV, ultraviolet.

analogue of halichondrin B69,70. Both of these highly Similar correlations have been determined for com-
complex molecules have been synthesized in multi- pounds that bind to structured RNA targets that are
gram quantities sufficient for clinical trials. Successes amenable to FT-ICR/MS analysis. Using a method
such as these make the compelling case that virtually dubbed multitarget affinity/specificity screening (MASS),
no crucially important compound is beyond reach for Hofstadler was able to detect specific binding of the
clinical evaluation. aminoglycoside antibiotic paromomycin in chromato-
The general paradigm for bioassay-guided purifi- graphic fractions derived from cultures of Streptomyces
cation is shown in FIG. 6. As can be seen in the diagram, rimosus subspecies paromomycinus72. Non-covalent
progression depends on how many y of fraction- adducts of paromomycins with a synthetic RNA
ation and bioassay are required. In those cases for oligomer containing the Escherichia coli A-site (site of
which the bioassay turnaround time is lengthy, the action of aminoglycoside antibiotics) were observed in
time for a single round of fractionation can be pro- the mass spectrometer. The molecular masses of the
longed. In the early stages of such discovery research, aminoglycosides were determined by the differences in
time is crucial if NMEs are to progress into the next mass between the adducts and the free RNA. By analogy,
phase, and delays owing to bioassay turnaround can be molecular masses of unknown binders present in extracts
a practical limitation. of other antibiotic producers could be determined in the
For this reason, a number of innovative approaches same manner.
have been advanced to create - bioassays that Frontal affinity chromatography (FAC)73,74 combined
utilize affinities of the desired natural ligands with targets with mass spectrometry has also been used to simplify the
of interest. One such approach uses continuous-flow deconvolution of activities in natural-product extracts75.
enzymatic reactions that are capable of providing real- In FAC, the target is immobilized on a column and the
time read outs of inhibition of enzymatic activity71. In mixture is continuously infused through the system. The
this system, the effluent from HPLC is split in two compounds with the greatest affinity for the target will
streams, one to the enzyme assay and the other in par- have the longest times. Mass-spectro-
allel to a mass spectrometer. Correlating peaks detected metric monitoring of the effluent provides the character-
in the enzyme assay with the corresponding mass ization of the retained compounds, as they sequentially
spectra provides data that are characteristic of the elute in inverse order of their binding affinity.
bioactive compound. This correlation can be suffi- An idealized system that couples on-line affinity
cient to identify known compounds by database selection, separation and identification/structure
searching on the basis of molecular mass, and will at determination is depicted in FIG. 7. All of the compo-
least provide between the biological test and a nents of this system are connected in a flow path that is
physicochemical parameter. Such linkage allows the driven by an HPLC pump. The selected compounds
physicochemical characteristics (mass spectrometric are resolved in the separation step, which is shown in
data, HPLC retention time, UV absorbance and so on) FIG. 7 as an HPLC column that yields single compo-
to be substituted for bioassay data for subsequent nents or ,which are then subjected to UV-visible
cycles of fractionation. absorption spectroscopy, mass spectrometry and NMR

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 1 5


REVIEWS

Box 3 | Improving natural products: the example of bryostatin


In 1988, Wender et al. proposed a simplified
model of bryostatin that accounted for its HO
H3CO OAc
potent inhibition of the binding of phorbol
esters to protein kinase C, as well as the O O O O O O
natural ligand 1,2-diacylglycerol. Although a
O O
direct correlation of this interaction has not 1 OH HO
OH HO
yet been linked to antitumour activity, this O O O O

pharmacophoric model was used to probe


19
26
the essential structural features required for O 20 OH O R OH
OCH OCH3
chemotherapeutic efficacy. In the ensuing O
3
C7H15 O
15 years, several simplified analogues were O O
produced on the basis of the model that
Bryostatin 1 Analogue X R = CH3
showed enhanced potency. Two regions have Analogue Y R = H
been defined: one, referred to as the
recognition domain, consists of a particular Spacer domain Recognition domain
spatial arrangement of oxygen atoms at C1,
C19 and C26. The other domain is the spacer domain that holds the three oxygen atoms in the proper orientation.
Wenders primary focus has been on simplifying the latter piece; the first successful candidates have simplified A and B
rings, as shown in both X and Y.In the next phase of the simplification process, the fatty acyl residue at C20 was
investigated, and it was found that a simple saturated octanoate residue compared favourably, and finally simply
shortening the carbon chain by one unit at C26 converting the secondary alcohol function common to Bryostatin 1, and
X, to a primary alcohol in Y,provided enhanced potency over the natural product.

spectroscopy for characterization. In theory, one pass less sensitive than mass spectrometry) has been relieved
of a particular extract though such a system would to a great degree by the development of cryogenic flow
provide identification of the compounds responsible probes 83 and micro-coil (nanolitre volume) NMR
for the binding activity, including compounds of probes84. These, in conjunction with the introduction of
unknown structure that would be determined by on- on-line solid-phase extraction peak trapping, have
line MS/MS and NMR spectroscopy76,77. In practice, it greatly improved the sensitivity and flexibility range of
is usually more expedient to conduct various aspects of the technique. Using LC-UV-SPE-NMR-MS, it is now
this process independently. This is particularly true for possible to perform automated analysis of natural-
complete structure determination by NMR spectroscopy, product extracts in which the individual components
in which the availability of greater amounts of material are present in 10 50 g quantities77.
(milligrams versus micrograms) will greatly reduce Continuing developments in biosynthetic tech-
data-acquisition times. nologies offer great promise for the discovery a nd
Advances in NMR, MS and HPLC technology have development of natural-product-derived pharmaceu-
made hyphenated LC-NMR, and now LC-NMR-MS, ticals. Genetic methods for the creation and expres-
practical options for analysis and structure determina- sion of novel metabolites are now routinely used in
tion of complex natural-product mixtures, in many drug discovery programmes. The development of
instances circumventing the need for the traditional metabolic engineering and its potential applications to
approach of mixture fractionation and isolation of indi- drug discovery has recently been reviewed by Khosla
vidual components before structure elucidation by MS, and Keasling 85, and will not be further addressed in
and NMR. Although LC-NMR was introduced almost this article.
20 years ago78 , its application to the direct analysis of
natural products has been hampered by problems of Leveraging the properties of natural products
chromatographic peak diffusion, sensitivity and sup- A strategy that has been successfully used in recent
pression of unwanted background solvent signals79 . years follows from experiments aimed at total synthesis
The development of peak storage units served to elimi- of natural products. Many academic groups consider
nate peak diffusion problems encountered with earlier natural products ideal targets for testing their syn-
stopped-flow or on-flow methods, and improved back- thetic methodology, and many remarkable achieve-
ground solvent suppression using pulsed magnetic-field ments have been documented 86 ,87 . In the course of
gradients led to the first effective applications of LC- these heroic efforts it is often possible to define the
NMR to structure analysis of plant natural products crucial structural elements required for biological
without the need for chromatographic isolation 80,81 . activity. In this way, potent and selective products can
Direct coupling of electrospray mass spectrometry for be derived with fewer synthetic steps and at reason-
LC-NMR-MS has proven to be an effective combination able cost. A recent example that actually provided a
for the characterization of plant glycosides that are resis- simpler product with enhanced potency is provided
tant to analysis by LC-NMR alone82. Finally, the inher- by the work of Wender (BOX 3) on the bryostatin series
ent low sensitivity of NMR (approximately 1,000-fold of anticancer compounds88 91.

216 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

Box 4 | Methods for exploiting natural-product structures


CN
Systematic chemical mutagenesis O
Systematic variation of the individual amino-acid residues (side O
N
chains) allows the pinpointing of structural features essential for
O NH CH3 O
biological activity. An elegant application of a combination of NH
such processes was used to probe the crucial features of HUN- CH3 O
N O
7293, a naturally occurring cyclic depsipeptide that is a potent N
NH CH3
and selective inhibitor of cell-adhesion molecule expression. In O
N
an approach referred to as systematic chemical mutagenesis OCH3
(see figure), Boger and co-workers explored the effect on
biological activity (both potency and selectivity) of simplifying
HUN-7293
each of the seven residues of HUN-7293, including removal of
N-methyl groups93. These experiments provided a greater
understanding of the structural requirements for maintaining
specific biological activity. O

Synthetic mimetics O O HN Cl
Understanding the binding interactions of the natural product
and the target can lead to a model for synthetic mimetics. An O N O OCH3
H
example of this approach to mimetics of the cryptophycin
antitumour agents utilized an azepine scaffold to which residues Cryptophycin
were attached that resembled the overall geometry in the natural
product94.A synthetic strategy was developed that allowed
N HN
compounds such as the one shown in the figure to be prepared in N
reasonable overall yield. In this process, the stereochemical O O
arrangement of the side chains and side-chain composition were
studied to optimize the biological response. Azepine mimetic

Peptides are modular structures joined together by the binding interactions of the natural product and the
amide (peptide) bonds. Small peptides, which often target can lead to a model for synthetic mimetics. An
contain non-protein amino acids, are typically assem- example of this approach to mimetics of the crypto-
bled by multi-enzyme complexes referred to as non- phycin antitumour agents started with an azepine scaf-
ribosomal peptide synthases, and in many instances fold to which residues were attached that resembled
these compounds are cyclized. These peptides, which the overall geometry in the natural product 94 .A syn-
represent a major class of biologically active natural thetic strategy was developed that allowed the azepine
products, are particularly amenable to parallel syn- mimetic compound shown in BOX 4 to be prepared,
thetic methodologies owing to the repetitive nature and with a reasonable overall yield. In this process, the
of the bond-forming process. Systematic variation of stereochemical arrangement of the side chains and
the individual amino-acid residues (side chains) side-chain composition were studied to optimize the
therefore allows the pinpointing of structural features biological response.
essential for biological activity. An elegant application An elegant example of overcoming the lack of selec-
of a combination of such processes was used to probe tivity observed with certain highly potent cytotoxic
the crucial features of HUN-7293, a naturally occur- agents in the treatment of cancer is through the mech-
ring cyclic depsipeptide that is a potent and selective anism of monoclonal antibody delivery through conju-
inhibitor of cell-adhesion molecule expression 92 . In gate formation. Calicheamicin (FIG. 8) is a highly potent
an approach referred to as systematic chemical muta- enediyne DNA-damaging agent produced by Micro-
genesis (BOX 4), Boger and co-workers explored the monospora echinospora95. Although it is 100 1,000
effect on biological activity (both potency and selec- times more potent than conventionally used thera-
tivity) of simplifying each of the seven residues of peutic agents, calicheamicin itself lacks the therapeutic
HUN-7293, including removal of N-methyl groups 93. index required for systemic administration. The
These experiments provided a greater understanding potency of calicheamicin has been harnessed by
of the structural requirements for maintaining specific immunoconjugation a chemical modification fol-
biological activity. lowed by linkage to a monoclonal antibody that
In some cases, the secondary metabolite has been specifically binds to the CD33 cell-surface antigen
shown to have exquisite potency for a particular target, present on acute myeloid leukaemia cells. After inter-
but is not practical for use as a therapeutic agent, owing nalization into lysosomal vesicles, gemtuzumab
to various liabilities for example, the cost of goods (a ozogamicin (Mylotarg; Wyeth) is engineered to release
supply issue), metabolic liabilities, pharmaceutical calicheamicin, which migrates to the nucleus, cleaves
properties and so on. In these instances, understanding DNA and results in cell death96,97.

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 1 7


REVIEWS

HO focused library based on nakijiquinone C29 . Earlier


O
O efforts in the 1990s effectively utilized several natural
CH3SSS N
CH3 O CH3 H OCH3 products scaffolds including yohimbine 23 , paclitaxel24
I O CH3 and vancomycin 25,26 for solid-phase synthesis of
S O O
N HO O H focused libraries.
O OCH3 OH H
O The diversity-oriented approach seeks to leverage
H3C OCH3
HO
O
N O the privileged structural motifs of natural-product
H3CO OH CH3CH2 H scaffolds to synthesize combinatorial libraries capable
H3CO
of binding a wide range of biological targets98. Initial
Calicheamicin 1 investigations based on the diversity-oriented principles
have successfully produced high-quality screening
libraries by solid-phase synthesis based on natural-
a Chemical modifications
product benzopyran scaffolds28,29. Later refinements
b Antibody conjugation have made it possible to rapidly synthesize diversity-
oriented small-molecule microarray libraries to pro-
duce molecules that bind Hap3p,a subunit of the yeast
Hap2/3/4/5p transcription-factor complex 99 and HIV
protease100. It has also been shown that it is possible to
synthesize diversity-oriented natural-product-based
O AcBut linker-hydrazone libraries by chemical recombination of complex frag-
function cleaves under
(HN intracellular conditions
ments obtained by chemical degradation of diverse
n = 23 O bioactive natural products101.
Humanized IgG4 O H3C CH3
HO
anti-CD33 O O Conclusions
monoclonal antibody NHN S
S N
The remarkable chemical diversity encompassed by
CH3 O H3C CH3 OCH3
H natural products continues to be of relevance to drug
I O CH3
S O O discovery. Although y drug discovery engine
NHO O H
O OCH3 OH H operates at an accelerated pace compared with the
O
H3C
O OCH3 era in which natural products were pre-eminent
HO N O sources of drug leads, numerous approaches have
H3CO CH3CH2 H
OH H3CO been developed to capture their intrinsic value.
Calicheamicin Crucial breakthroughs in separation and structure-
Figure 8 | Harnessing highly toxic natural products for cancer therapy. One way to
determination technologies have lowered the hurdles
overcome the lack of selectivity observed with certain highly potent cytotoxic natural inherent in screening mixtures of structurally com-
products with potential for the treatment of cancer is to use monoclonal antibody delivery plex molecules. A greater understanding of the
through conjugate formation. Calicheamicin is a highly potent anticancer agent but lacks the exquisite specificity ingrained in secondary metabo-
therapeutic index necessary for systemic administration. The potency of calicheamicin has lites through the evolutionary process has focused
been harnessed by chemical modification followed by linkage to a monoclonal antibody that attention on their roles as mediators of protein protein
specifically binds to the unique CD33 cell-surface antigen present on acute myeloid leukaemia
cells. Ig, immunoglobulin.
interactions in vital cellular processes, and advances in
synthetic chemistry have revolutionized the processes
of material supply and the modulation of biological
activity through structural modifications. It seems
Natural products as building blocks for molecular that no compound is too complex to be recreated in
libraries. Instead of viewing natural products as a the laboratory.
stand-alone approach distinct from combinatorial Furthermore, our ability to model the binding of
synthesis, it is now much more effective to adopt these evolved, privileged structures with their targets
strategies that combine both approaches. In principle, enables the design of simpler mimetics that have
there seems to be a number of strategies through superior properties. Efforts to expand the impact of
which the unique molecular diversity of natural prod- natural chemical diversity on the drug discovery
ucts can be leveraged in the design of combinatorial process follow two main chemistry-driven paths. One
libraries. The target-oriented or focused-library seeks to simplify crude mixtures, as well as enhance
approach seeks to elaborate structural modifications the impact of minor components in assays, through the
onto an existing bioactive natural-product scaffold in creation of fractionated natural-product libraries.
a parallel, systematic fashion in order to improve its The other approach uses the power of combinatorial
inherent biological activity or drug-like properties. synthesis to amplify the structural context in which
This can be performed either by semi-synthetic modi- the unique features of natural products are expressed.
fication of the parent molecule, or by fully synthetic The confluence of these technologies with advances in
methods. This strategy was put into practice by genomics, metabolic engineering85 and chemical synthe-
Waldmann et al., who recently developed potent, sis offer exciting new possibilities to exploit the remark-
selective inhibitors against the TIE2 receptor tyrosine able chemical diversity of in
kinase by parallel synthesis of a small (74 member) the quest for new drugs.

218 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc


REVIEWS

1. Newman, D. J., Cragg, G. M. & Snader, K. M. Natural 26. Nicolaou, K. C., Winssinger, N. Hughes, R., Smethurst, C. & An analysis of why organisms produce natural
products as a source of new drugs over the period Cho, S. Y. New selenium-based safety-catch linkers: solid- products chemical diversity.
19812002. J. Nat. Prod. 66, 10021037 (2003). phase semisynthesis of vancomycin. Angew. Chem. Int. Ed. 50. Chabala, J. C. et al. Ivermectin, a new broad-spectrum
A detailed analysis and description of current 39, 10841088 (2000). antiparasitic agent. J. Med. Chem. 23, 11341136 (1980).
natural-product-derived therapeutic agents. 27. Nicolaou, K. C., Pfefferkorn, J. A., Roecker, A. J., Cao, G.-Q. 51. He, H. et al. Mannopeptimycins, novel antibacterial
2. Projan, S. J. Infectious diseases in the 21st century: & Barluenga, S. Natural product-like combinatorial libraries glycopeptides from Streptomyces hygroscopicus, LL-
increasing threats, fewer new treatments and a premium on based on privileged structures 1. General principles and AC98. J. Am. Chem. Soc. 124, 97299736 (2002).
prevention. Cur. Opin. Pharmacol. 3, 457458 (2003). solid phase synthesis of benzopyrans. J. Am. Chem. Soc. 52. De Voe, S. E. & Kunstmann, M. P. Antibiotic AC98 and
3. Kirsop, B. E. The convention on biological diversity: some 122, 99399953 (2000). production. US Patent 3,495,004 (1970).
implications for microbiology and microbial collections. A demonstration of the natural products-like chemical 53. Ruzin, A. et al. Mechanism of action of the
J. Indust. Microbiol. Biotech. 17, 505511 (1996). diversity achieved by modern combinatorial mannopeptimycins, a novel class of glycopeptide antibiotics
4. Projan S. J. Why is big pharma getting out of antibacterial chemistry. active against vancomycin-resistant Gram-positive bacteria.
drug discovery? Cur. Opin. Microbiol. 6, 42730 (2003). 28. Nicolaou, K. C. et al. Natural product-like combinatorial Antimicrob. Agent. Chemother. 48, 728738 (2004).
5. Ajay, W. P. W. & Murcko, M. Can we learn to distinguish libraries based on privileged structures. 2. Construction of a 54. He, H. et al. Mannopeptimycin esters and carbonates,
between Drug-Like and Nondrug-like molecules? J. Med. 10,000-membered benzopyran library by directed split-and- potent antibiotic agents against drug resistant bacteria.
Chem. 41, 33143324 (1998). pool chemistry using NanoKans and optical encoding. Bioorg. Med. Chem. Lett. 14, 279282 (2004).
6. Sadowski, J. & Kubinyi, H. Scoring scheme for J. Am. Chem. Soc. 122, 99549967 (2000). 55. Sum, P. E., et al. Synthesis and evaluation of ether and
discriminating between drugs and nondrugs. J. Med. 29. Kissau, L., Stahl, P., Mazitschek, R., Giannis, A. & halogenated derivatives of mannopeptimycin glycopeptide
Chem. 41, 33253329 (1998). Waldmann, H. Development of natural product-derived antibiotic. Biorg. Med. Chem. Lett. 13, 28052808 (2003).
7. Newman, D., Cragg, G., Kingston, D. in The Practice of receptor tyrosine kinase inhibitors based on conservation of 56. Dushin, R. G. et al. Hydrophobic acetal and ketal derivatives
Medicinal Chemistry (ed. Wermuth, C. G.) 91109 protein domain fold. J. Med. Chem. 46, 29172931 (2003). of mannopeptimycin- and AC98-0053: semisynthetic
(Academic, London,2003). 30. Peczuh, M. W. & Hamilton, A. D. Peptide and protein glycopeptides with potent activity against Gram-positive
8. Feher, M. & Schmidt, J. M. Property distributions: recognition by designed molecules. Chem. Rev. 100, bacteria. J. Med. Chem. 47, 34873490 (2004).
Differences between drugs, natural products, and molecules 24792494 (2000). An extensive illustration of the capabilities of NMR
from combinatorial chemistry. J. Chem. Inf. Comput. Sci. 31. Dumont, F.J. FK506, An immunosuppressant targeting spectroscopy in the structure elucidation of complex
43, 218227 (2003). calcineurin function. Curr. Med. Chem. 7, 731748 (2000). natural products.
9. Lipinski, C. A., Lombardo, F., Dominy, B. W. & Feeney, P. J. 32. Abraham, R. T. & Wiederrecht, G. J. Immunophamacology 57. Strege, M. A. High-performance liquid chromatographic-
Experimental and computational approaches to estimate of rapamycin. Ann. Rev. Immunol. 14, 483510 (1996). electrospray ionization mass spectrometric analyses for the
solubility and permeability in drug discovery and 33. Hersperger, R. & Keller, T. H. Ascomycin derivatives and integration of natural products with modern high-throughput
development settings. Adv. Drug Del. Rev. 23, 325 (1997). their use as immunosuppressive agents. Drugs Future 25, screening. J. Chrom. B 725, 6768 (1999).
Fundamental work describing important physico- 269277 (2000). 58. Nielsen, K. F.& Smedsgaard, J. Fungal metabolite
chemical properties of drug molecules. 34. Takahashi, N. in Macrolide Antibiotics 2nd Edn screening: database of 474 mycotoxins and fungal
10. Lee, M. L. & Schneider, G. Scaffold architecture and (ed. Omura, S.) 577621 (Academic, London, 2002). metabolites for dereplication by standardised liquid
pharmacophoric properties of natural products and trade 35. Pong, K. & Zaleska, M. M. Therapeutic implications for chromatography-UV-mass spectrometry methodology.
drugs: Application in the design of natural product-based immunophilin ligands in the treatment of neurodegenerative J. Chrom. A 1002, 111136 (2003).
combinatorial libraries J. Comb. Chem. 3, 284289 (2001). diseases. Curr. Drug Targets CNS Neurol. Disord. 2, 6172 59. Freeman, R. & Morris, G. A. Two-dimensional Fourier
11. Stahura, F., Godden, J. W., Ling, X. & Bajorath, J. (2003). transformation in NMR. Bull. Magn. Res. 1, 126 (1979).
Distinguishing between natural products and synthetic 36. Jordan, M. A. Mechanism of action of antitumor drugs that 60. Bax, A., Aszalos, A., Dinya, Z. & Sudo, K. Structure
molecules by descriptor Shannon entropy analysis and interact with microtubules and tubulin. Curr. Med. Chem. 2, elucidation of the antibiotic desertomycin through the use of
binary QSAR calculations. J. Chem. Inf. Comput. Sci. 40, 117 (2002). new two-dimensional NMR techniques. J. Am. Chem. Soc.
12451252 (2000). 37. Kowalski, R. J. et al. The microtubule-stabilizing agent 108, 80568063 (1986).
12. Henkel, T., Brunne, R., Muller, H. & Reichel, F.Statistical discodermolide competitively inhibits the binding of 61. Schwalbe, H. & Kessler, H. 900 MHz NMR spectrometer
investigation of structural complementarity of natural paclitaxel (Taxol) to tubulin polymers, enhances tubulin in Munich and Frankfurt. Nachrichten aus der Chemie 51,
products and synthetic compounds. Angew. Chem. Int. Ed. nucleation reactions more potently than paclitaxel, and 412417 (2003).
Engl. 38, 643647 (1999). inhibits the growth of paclitaxel-resistant cells. Mol. 62. Sandvoss, M., Preiss, A., Levsen, K., Weisemann, R. &
13. Martin, Y. C. Diverse viewpoints on computational aspects Pharmacol. 52, 613622 (1997). Spraul, M. Two new asterosaponins from the starfish
of molecular diversity. J. Comb. Chem. 3, 231250 (2001). Fundamental work describing the mechanism of Asterias rubens: application of a cryogenic NMR probe
14. Martin, Y. C. & Critchlow, R. E. Beyond mere diversity: action of a potent antitubulin agent. head. Magn. Res. Chem. 41, 949954 (2003).
tailoring combinatorial libraries for drug discovery. J. Comb. 38. Bai, R., Cichacz, Z. A., Herald, C. L., Pettit, G. R. & Hamel, E. 63. Olson, D. L. et al. Microflow NMR: concepts and
Chem. 1, 3245 (1999). Spongistatin 1, a highly cytotoxic, sponge-derived, marine capabilities. Anal. Chem. 76, 29662974 (2002).
15. Evans, B. E. et al. Methods for drug discovery: development natural product that inhibits mitosis, microtubule assembly, 64. Serber, Z. et al. New carbon-detected protein NMR
of potent, selective, orally effective cholecystokinin and the binding of vinblastine to tubulin. Mol. Pharmacol. experiments using cryoProbes. J. Am. Chem. Soc. 122,
antagonists. J. Med. Chem. 31, 22352246 (1988). 44, 75766 (1993). 35543555 (2000).
Fundamental work describing the principle of 39. Loganzo, F.et al. HTI-286, a synthetic analogue of the 65. Satake, M. et al. Structural confirmation of maitotoxin based
privileged chemical structures. tripeptide hemiasterlin, is a potent antimicrotubule agent that on complete 13C NMR assignments and the three-
16. Chothia, C. One thousand families for the molecular circumvents P-glycoprotein-mediated resistance in vitro and dimensional PFG NOESY-HMQC spectrum. J. Am. Chem.
biologist. Nature 357, 543544 (1992). in vivo. Cancer Res. 63, 18381845 (2003). Soc. 117, 70197020 (1995).
17. Zhang, C. & DeLisi, C. Estimating the number of protein 40. Sackett, D. L. Podophyllotoxin, steganacin and 66. He, H. et al. Lomaiviticins A and B, potent antitumor
folds. J. Mol. Biol. 284, 13011305 (1998). combretastatin: natural products that bind at the colchicine antibiotics from Micromonospora Lomaivitiensis. J. Am.
18. Salem, G. M., Hutchinson, E. G., Orengo, C. A. & Thornton, site of tubulin. Pharmacol. Therap. 59, 163228 (1993). Chem. Soc. 123, 53525363 (2001).
J.M. Correlation of observed fold frequency with the 41. Silverman, L., Campbell, R. & Broach, J. R. New assay 67. McDonald, L. A. et al. FTMS Structure elucidation of natural
occurrence of local structural motifs. J. Mol. Biol. 287, technologies for high throughput screening. Curr. Opin. products: application to muraymycin antibiotics using ESI
969981 (1999). Chem. Biol. 2, 397403 (1998). Multi-CHEF SORI-CIT FTMSn, the Top-Down/Bottom-Up
19. Holm, L. & Sander, C. Mapping the protein universe. 42. Cohen, P. Protein kinases- The major drug targets of the 21st approach, and HPLC ESI capillary-skimmer CID FTMS.
Science 273, 595603 (1996). century? Nature Rev. Drug Discov. 1, 309316 (2002). Anal. Chem. 75, 27302739 (2003).
20. Hou, J., Sims, G., Zhang, C. & Kim, S-H. A global 43. Zaman, G. J. R., Garritsen, A., de Boer, T.& Capabilities of modern mass spectrometry in the
representation of the protein fold space. Proc. Natl Acad. van Boeckel, C. A. Fluorescence assays for high throughput structure elucidation of complex natural products.
Sci. USA 100, 23862390 (2003). screening of protein kinases. Comb. Chem. High 68. Gunasekera, A. P., Gunaskera, M., Longley, R. E. &
21. Anantharaman, V., Aravind, L. & Koonin, E. V. Emergence of Throughput Screen. 6, 313320 (2003). Schulte, G. K. Discodermolide: a new bioactive poly-
diverse biochemical activities in evolutionarily conserved 44. Fowler, A. et al. An evaluation of fluorescence polarization hydroxylated lactone from the marine sponge Discodermia
structural scaffolds of proteins. Cur. Opin. Chem. Biol. 7, and lifetime discriminated polarization for high throughput dissoluta. J. Org. Chem. 55, 49124915 (1990).
1220 (2003). screening of serine/threonine kinases. Anal. Biochem. 308, 69. Towle, M. J. et al. In vitro and in vivo anticancer activities of
22. Breinbauer, R., Vetter, I. R. & Waldmann, H. From protein 223231 (2002). synthetic macrocyclic ketone analogues of halichondrin B.
domains to drug candidates- Natural products as guiding 45. Turek-Etienne, T. C. et al. Evaluation of fluorescent Cancer Res. 61, 10131021 (2001).
principles in the design and synthesis of compound libraries. compound interference in 4 fluoroescence polarization 70. Kuznetsov, G. et al. Induction of morphological and
Angew. Chem. Int. Ed. 41, 28782890 (2002). assays: 2 kinases, 1 protease, and 1 phosphatase. biochemical apoptosis following prolonged mitotic blockage
Analysis and description of the importance of natural J. Biomol. Screen. 8, 176184 (2003). by halichondrin B macrocyclic ketone analog E7389.
products as scaffolds for drug design. 46. Eldridge, G. R. et al. High-throughput method for the Cancer Res. 64, 57605766 (2004).
23. Atuegbu, A., MacLean, D., Nguyen, C., Gordon, E. M. & production and analysis of large natural product libraries for 71. Schenk, T. et al. A Generic assay for phosphate-consuming
Jacobs, J. W. Combinatorial modification of natural drug discovery. Anal. Chem. 74, 39633971 (2002). or -releasing enzymes coupled on-line to liquid
products: Preparation of un-encoded and encoded libraries 47. Abel, U., Koch, C., Speitling, M. & Hansske, F. G. Modern chromatography for lead finding in natural products. Anal.
of Rauwolfia alkaloids. Bioorg. Med. Chem. 4, 10971106 methods to produce natural-product libraries. Curr. Opin. Biochem. 316, 118126 (2003).
(1996). Chem. Biol. 6, 453458 (2002). 72. Cummins, L. L. et al. Multitarget affinity/specificity screening
24. Xaio, X. Y., Parandoosh, Z. & Nova, M. P. Design and 48. Cummins, L. L. et al. Multitarget affinity/specificity of natural products finding and characterizing high-affinity
synthesis of a taxoid library using radiofrequency encoded screening of natural products:Finding and characterizing ligands from complex mixtures by using high-performance
combinatorial chemistry. J. Org. Chem. 62, 60296033 high affinity ligands from complex mixtures by using high mass spectrometry. J. Nat. Prod. 66, 11861190 (2003).
(1997). performance mass spectrometry. J. Nat. Prod. 66, 73. Schriemer, D. C., Bundle, D. R., Li, L. & Hindsgaul, O. Micro-
25. Xu, R., Grieveldinger, G., Marenus, L. E., Cooper, A. & Ellman, 11861190 (2003). scale frontal affinity chromatography with mass
J.A. Combinatorial library approach for the identification of 49. Firn, R. D. & Jones, C. D. Natural products a simple spectrometric detection: a new method for the screening of
synthetic receptors targeting vancomycin-resistant bacteria. model to explain chemical diversity. Nat. Prod. Rep. 20, compound libraries. Angew. Chem. Int. Ed. 37, 33833387
J. Am. Chem. Soc. 121, 48984899 (1999). 382391 (2003). (1998).

NATURE REVIEWS | DRUG DISCOVERY VOLUME 4 | MARCH 2005 | 2 1 9


REVIEWS

74. Chan, N. W. C., Lewis, D. F., Rosner, P. J., Kelly, M. A. & 87. Lin, S. et al. Total syntheses of TMC-95A and B via a new 101. Niggemann, J., Michaelis, K., Frank, R., Zander, N. &
Schriemer, D. C. Frontal affinity chromatographymass reaction leading to Z-enamides. Some preliminary Hfle, G. Natural product-derived building blocks for
spectrometry assay technology for multiple stages of drug findings as to SAR. J. Am. Chem. Soc. 126, 63476355 combinatorial synthesis. Part 1. Fragmentation of natural
discovery: applications of a chromatographic biosensor. (2004). products from myxobacteria. J. Chem. Soc. Perkin. Trans.
Anal. Biochem. 319, 112 (2003). 88. Wender, P. A. et al. Modeling of the bryostatins to the 1, 24902503 (2002).
75. Zhu, L., Chen, L., Luo, H. & Xu, X. Frontal affinity phorbol ester pharmacophore on protein kinase C. Proc. 102. ter Haar, E. et al. Discodermolide, A cytotoxic agent that
chromatography combined on-line with mass spectrometry: Natl Acad. Sci. USA 85, 7197201 (1988). stabilizes microtubules more potently than taxol. Biochemistry
a tool for the binding study of different epidermal growth factor 89. Wender, P. A. et al. The design, computer modeling, 35,243250 (1996).
receptor inhibitors. Anal. Chem. 75, 63886393 (2003). solution structure, and biological evaluation of synthetic 103. Nerenberg, J. B., Hung, D. T., Somers, P. K., Schreiber, S. L.
76. Wolfender, J-L., Ndjoko, K. & Hostettmann, K. Liquid analogs of bryostatin 1. Proc. Natl Acad. Sci. USA 95, Total synthesis of the immunosuppressive agent (-)-disco-
chromatography with ultraviolet absorbancemass 66246629 (1998). dermolide. J. Am. Chem. Soc. 115, 1262112622 (1993).
spectrometric detection and with nuclear magnetic 90. Wender, P. A., DeBrabander, P. G, Hinkle, K. W., Lippa, B., 104. Smith, A. B. et al. Evolution of a gram-scale synthesis of (+)-
resonance spectroscopy: a powerful combination for the Pettit, G. R. Synthesis and biological evaluation of fully discodermolide. J. Am. Chem. Soc. 122, 86548664 (2000).
on-line structural investigation of plant metabolites. synthetic bryostatin analogues. Tet. Lett. 39, 86258628 105. Paterson, I., Florence, G. J., Gerlach, K. & Scott, J. P.
J. Chrom. A 1000, 437455 (2003). (1998). Total synthesis of the antimicrotubule agent (+)
77. Exarchou, V., Godejohann, M., van Beek, T. S., 91. Wender, P. A. et al. The practical synthesis of a novel and discodermolide using boronmediated aldol reactions of
Gerothanassis, I. P. & Vervoort, J. LC-UV-solid-phase highly potent analogue of bryostatin. J. Am. Chem. Soc. chiral ketones. Angew. Chem. Int. Ed. 39, 377 (2000).
extraction-NMR-MS combined with a cryogenic flow probe 124, 1364813649 (2002). 106. Marshall, J. A. & Johns, B. A. Total synthesis of (+)-
and its application to the identification of compounds present Along with preceding articles, a good example of discodermolide. J. Org. Chem. 63, 78857892 (1998).
in greek oregano. Anal. Chem. 75, 62886294 (2003). natural-products-based drug design. 107. Mickel, S. J. et al. Large scale synthesis of the anti-cancer
Capabilities of hyphenated spectroscopic techniques 92. Hommel, U., Weber, H-P., Oberer, L., Naegeli, H. U., marine natural product (+)-discodermolide. Part 1. Org.
in the structure elucidation of complex natural Oberhauser, B. & Foster, C. A. The 3D-structure of a natural Proc. Res. Dev. 8, 92100 (2004).
product mixtures. inhibitor of cell adhesion molecule expression. FEBS Lett. 108. Mickel, S. J. et al. Large scale synthesis of the anti-cancer
78. Laude, D. A. &, Wilkins, C. L. Direct-linked analytical scale 379, 6973(1996). marine natural product (+)-discodermolide. Part 2. Org.
high-performance liquid chromatography/nuclear magnetic 93. Chen, Y., Bilban, M., Foster, C. A. & Boger, D. L. Solution- Proc. Res. Dev. 8, 101106 (2004).
resonance spectrometry. Anal. Chem. 56, 24712475 phase parallel synthesis of a pharmacophore library of HUN- 109. Mickel, S. J. et al. Large scale synthesis of the anti-cancer
(1984). 7293 analogues: A general chemical mutagenesis approach marine natural product (+)-discodermolide. Part 3. Org.
79. Albert, K. Liquid chromatography-nuclear magnetic to defining structure-function properties of naturally Proc. Res. Dev. 8, 107112 (2004).
resonance spectroscopy. J. Chrom. 856, 199211 (1999). occurring cyclic (depsi)peptides. J. Am. Chem. Soc. 124, 110. Mickel, S. J. et al. Large scale synthesis of the anti-cancer
80. Wolfender, J.-L., Ndjoko, K. & Hostettmann, K. The potential 54315440 (2002). marine natural product (+)-discodermolide. Part 4. Org.
of LC-NMR in phytochemical analysis. Phytochem. Anal. 12, 94. Smith, III, A. B., Cho, Y. S., Pettit, G. R. & Hirschmann, R. Proc. Res. Dev. 8, 113121 (2004).
222 (2001). Design, synthesis, and evaluation of azepine-based 111. Mickel, S. J. et al. Large scale synthesis of the anti-cancer
81. Schaller, F., Wolfender, J.-L., Hostettmann, K. & Mavi, S. cryptophycin mimetics. Tetrahedron 59, 69917009 marine natural product (+)-discodermolide. Part 5. Org.
New antifungal quinone methide diterpenes from (2003). Proc. Res. Dev. 8, 122130 (2004).
Bobgunnia madagascariensis and study of their 95. Lee, M. D., Durr, F. E., Hinman, L. M., Hanmann, P. R. & References 107111 provide a strong example of
interconversion by LC/NMR Helv. Chim. Act. 84, 222229 Ellestad, G. A. The calicheamicins. Adv. Med. Chem. 2, successful synthetic scale-up of a natural product
(2001). 3166 (1993). for clinical supply.
82. Lommen, A., Godejohann, M., Venema, D. P., 96. Damle, N. K. & Frost, P. Antibody-targeted chemotherapy
Hollman, P. C. H., Spraul, M. Application of directly coupled with immunoconjugates of calicheamicin. Curr. Opin. Acknowledgements
HPLC-NMRMS to the identification and confirmation of Pharm. 3, 386390 (2003). We thank N. Pilote for valuable assistance in literature and patent
quercetin glycosides and phloretin glycosides in apple peel. 97. Hamann, P. R. et al. Gemtuzumab ozogamicin, A potent and analysis.
Anal. Chem. 72, 17931797 (2000). selective anti-CD33 antibody-calicheamicin conjugate for
83. Spraul, M. et al. Advancing sensitivity for LC-NMR-MS using treatment of acute myeloid leukemia. Bioconj. Chem. 13, Competing interests statement
a cryoflow probe: application to analysis of acetominophen 4758 (2002). The authors declare no competing financial interests.
metabolites in urine. Anal. Chem. 75, 15461551 (2003). 98. Burke, M. D. & Schreiber, S. L. A planning strategy for
84. Wu, N., Webb, A., Peck, T. L., Sweedler, J. V. Online NMR diversity oriented synthesis. Angew. Chem. Int. Ed. 43,
detection of amino acids and peptides in microbore LC. 4658 (2004). Online links
Anal. Chem. 67, 31013107 (1995). 99. Koehler, A. N., Shamji, A. F.& Schreiber, S. L. Discovery of
85. Khosla, C. & Keasling, J. D. Metabolic engineering for drug an inhibitor of a transcription factor using small molecule DATABASES
discovery and development. Nature Rev. Drug Discov. 2, microarrays and diversity-oriented synthesis. J. Am. Chem. The following terms in this article are linked online to:
10191025 (2003). Soc. 125, 84208421 (2003). Entrez Gene:
86. Nicolaou, K. C. et al. Total synthesis of apoptolidin: 100. Brik, A. M. et al. Rapid diversity-oriented synthesis in http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene
construction of enantiomerically pure fragments. J. Am. microtiter plates for in situ screening of HIV protease Hap3p |TIE2
Chem. Soc. 125, 1543315442 (2003). inhibitors. Chembiochem 4, 12461248 (2003). Access to this interactive links box is free online.

220 | MARCH 2005 | VOLUME 4 www.nature.com/reviews/drugdisc

S-ar putea să vă placă și