Natural products as kinase inhibitors
Natural products have been widely used to dissect the basic mechanisms of fundamental life science and as clinical therapeutics. Recently, there has been significant interest in discovering new chemical pharmacophores in natural products to fulfil the vast demand for novel kinase inhibitors and address critical unmet medical needs with respect to signal transduction pathways. In this review, we summarize the history of several different classes of natural product-derived kinase inhibitors, discuss their kinome-wide target profiles and examine their structural binding modes based on available 3D X-ray structures. In particular, their origin, target activity, selectivity, scope and potential therapeutic development are highlighted against the backdrop of medicinal chemistry.
1 Natural products as bioactive reagents
Natural products (NPs) have a long history of being used as bioactive ingredients in herbal recipes throughout the world. Since the 19th century, biologically active NPs have been extensively explored as drug candidates for clinical applications as well as basic research tools to dissect biological processes.
Because they disturb a wide range of biological processes, including transcription, translation and post-translational modification, NPs and NP-derived synthetic agents have been shown to be effective modulators of almost all basic cellular processes. Due to their significant structural diversity and wide range of biological functions, NPs have found broad clinical application, particularly in the mid-20th century, when an explosion of antibiotic and anti-cancer drug discovery occurred.1 However, there was a decrease in NP-based drug discovery in the 1980s and 1990s, widely believed to be a result of the emergence of combinatorial chemistry, which proved that human knowledge cannot compete with Mother Nature in terms of creating chemical structural diversity.2 The decade of the 2000s witnessed a recovery of NP-based drug discovery, as exemplified by several NP anti-cancer drugs being approved for clinical application, including a microtubule-stabilizing epothilone derivative (Ixabepilone) and mTOR kinase inhibitor rapaymycin (sirolimus), as well as several NP-derived kinase inhibitors entering clinical trials, such as roscovitine and the staurosporines, among others.3–5
2 Introduction to protein kinases and kinase inhibitors
.
The Human Genome Project revealed more than 2000 kinase genes in humans, among which are more than 500 protein kinases.6 Since their discovery in the early 1950s, protein kinases have been central to dissecting signal transduction pathways and have been found to be involved in almost all critical cellular functions, including growth, development and homeostasis.7 Protein kinases exert their biological roles by transferring the gamma phosphate from ATP to tyrosine, threonine or serine residues in their downstream protein targets. The N-terminus of a protein kinase consists of several beta-sheets, while the C- terminus is composed of alpha helices; these two regions are
linked through a hinge chain. The cleft between the two termini and the hinge region form the ATP binding pocket, and most ATP-competitive inhibitors were designed to target this area (Fig. 1A).8,9 In the early 1980s, it was first shown that protein kinases played a critical role in oncogenesis and tumor progres- sion, which made them extremely attractive targets for anti- cancer therapy.10,11 Later study aided by newly developed tools, such as synthetic small molecule inhibitors, genetic modulation, RNAi technology and bioinformatics, revealed that protein kinases were involved in almost all human diseases, including cancer, diabetes, cardiovascular disease, developmental disease, neurological disease and infectious disease.6,12 It has been esti- mated that more than 180 protein kinases are disease-related, with more than 170 currently serving as drug discovery targets in preclinical and clinical evaluations of a wide spectrum of diseases.13 Small molecule kinase inhibitors have been powerful tools in understanding kinase-mediated signal transduction pathways in the past, and to date more than 10 of them have been and translational studies. In this report, we provide a concise overview of NPs bearing protein kinase inhibitory activity and describe their structure-related binding information, through which we hope to promote future NP-based drug discovery.
3 Natural products as kinase inhibitors
3.1 Purine analogues
Olomoucine (1, Scheme 1) is one of the most important natu- rally-occurring kinase inhibitors in the history of drug discovery. It was first isolated from cotyledons of the radish as a cytokinin 7-glycosyltransferase inhibitor.15 Later, it was shown that olo- moucine exhibited low micromolar anti-mitotic activity through the inhibition of cyclin-dependent kinase (CDK) in ATP- competitive mode.16,17 It moderately inhibits CDK1/2 (IC50: 7 mM) and CDK5 (IC50: 5 mM), and was found to be a relatively potent inhibitor of over 35 kinases, including AMP-activated protein kinase (AMPK) and protein kinase C (PKC).18 An X-ray crystallography study revealed that olomoucine had a different binding mode to CDK2 than did ATP, which is shown in Fig. 1B. ATP uses the purine N1 and C6-NH2 to form two hydrogen bonds with L83 and E81, respectively, in the hinge region. Conversely, olomoucine uses N7 and C6-NH2 to form two hydrogen bonds with only L83, while the purine ring is in a different orientation.19 Further biological mechanistic investi- gation revealed that olomoucine can block the transit between both the G1/S and G2/M phases of the cell cycle.20 A variety of cancer cell growth can be inhibited by olomoucine, and severe apoptosis was observed when it was combined with a DNA damaging reagent.21 The seminal discoveries of olomoucine opened a new window for kinase-targeted drug discovery. The apparent inhibition of the CDKs by olomoucine refuted the widely accepted drug discovery dogma that selective ATP- competitive kinase inhibitors were not achievable because of the high cellular concentration of ATP (in the 1–5 mM range).2 A drug discovery program based on olomoucine provided a new structurally relevant CDK inhibitor, R-roscovitine (also called Seliciclib, 2, Scheme 1), which exhibits an IC50 of around 450 nM against CDK1, 2, 7 and 9. Roscovitine is currently in phase I clinical trials for advanced solid tumors.22 Further medicinal chemistry efforts have resulted in a series of new, potent inhibi- tors, such as the purvalanols, which exhibit much higher activi- ties (IC50 values: 4–40 nM) and selectivities against the CDKs. These compounds have been used as research tools and have greatly aided cellular mitotic studies in recent decades.
Fig. 1 X-ray crystal structures of kinase inhibitors with kinases. A. ATP with CDK2 (PDB: 1FIN); B. Olomucine with CDK2 (PDB:1W0X); C. Hymenialdisine with CDK2 (PDB: 1DM2); D. Quercetin with PI3Kg (PDB: 1E8W); E. Quercetin with HCK (PDB: 2HCK); F. Myricetin with PI3Kg (PDB: 1E90); G. Myricetin with PIM1 (PDB: 1O63); H. LY294002 with PI3Kg (PDB: 1E7V).
Lymphostin (3, Scheme 1), a novel tricyclic aromatic alkaloid with a pyrroloquinoline core, was discovered while searching for immunosuppressant agents from the soil bacteria Streptomyces sp. KY11783.25,26 It was originally identified as an inhibitor of lymphocyte-specific protein tyrosine kinase (LCK) (IC50: 50 nM), a member of the Src kinase family. Later, it was revealed that the efficacy of lymphostin in blocking mixed lymphocyte reaction and delayed-type hypersensitivity was actually due to the dual inhibition of LCK and phosphoinositide 3-kinase (PI3K) (IC50: 1 nM).27 Very recently, lymphostin was shown to be a potent mTOR inhibitor (IC50: 1.7 nM).28 It is likely that the binding mode of lymphostin mimics that of olomoucine in the hinge region.
Hymenialdisine (4, Scheme 1) was first isolated from the marine sponges Acanthella sp. and Axinella sp. in 1982, and its structure was determined by X-ray analysis to be a rare bromine- containing pyrroloazepine compound.29–31 It did not draw significant attention until it was characterized in 2000 as an ATP- competitive inhibitor of multiple kinases such as CDK1, 2 and 5 (IC50: 22 nM, 40–70 nM, 28 nM), glycogen synthase kinase 3b (GSK3b) (IC50: 10 nM) and casein kinase I (CK1) (IC50: 35 nM). In cells, hymenialdisine inhibits the phosphorylation of micro- tubule-binding protein tau at sites that are hyperphosphorylated by GSK3b and CDK5/p35 in Alzheimer’s disease.32 Hyme- nialdisine is also a potent mitogen-activated protein kinase 1 (MEK1) inhibitor (IC50: 6 nM).33 The X-ray structure of CDK2 and hymenialdisine revealed that three hydrogen bonds were formed in the hinge region between amino acids E81 and L83 in the backbone and the pyrrole nitrogen and azepine amide moiety, respectively (Fig. 1C). Extensive efforts have been devoted to the development of a more selective inhibitor based on the hymenialdisine core structure, which has lead to the discovery of a selective indoloazepine checkpoint kinase 2 (CHK2) inhibitor (IC50: 8 nM) by replacing the bromine atom with a benzene ring.
Scheme 1 Chemical structures of purine analogues.
3.2 Flavonoid analogues
Quercetin (5, Scheme 2) is a flavonoid that is naturally abundant in many fruits and vegetables and is one of the first compounds in the flavonoid family determined to have kinase inhibitory activity.35,36 In the early 1980s, it was demonstrated that quer- cetin inhibits pp60vSrc both in vitro and in vivo (IC50: 60–80 mM and 3 mM, respectively) in an ATP-competitive manner.37 This was an early example illustrating that large differences between the in vitro biochemical assay and cellular assay may be due to the ATP concentration differences among the assay conditions. This also discouraged the development of ATP-competitive inhibitors because of the high concentration of ATP (1–5 mM) in the cellular environment, although it was later shown that this issue could be overcome. Quercetin was also found to be active against a variety of other kinases at low micromolar levels, including cyclic-AMP-independent protein kinase,38 calcium- and phospholipid-dependent protein kinase (Ca, PL-PK)39–41 and PI3K.42 In vivo, quercetin shows anti-tumor activity induced by 12-O-tetra-decanoylphorbol-13-acetate (TPA). Qucertein has also been crystallized with tyrosine kinase hemopoietic cell kinase (HCK)43 and lipid kinase PI3Kg,44 respectively (Fig. 1D & 1E). Quercetin adopts different binding modes with PI3K and HCK. In PI3K, three hydrogen bonds are formed with E880 and V882 in the hinge region, together with the flavonoid quinone carbonyl unit and two adjacent phenol groups. In this binding mode, the dihydroxyl benzene ring is directed into the inner hydrophobic pocket. However, in HCK, two hydrogen bonds are formed with the single amino acid M341 in the hinge region and the dihydroxyl benzene ring is orientated into the solvent. This was an early example of the same compound adopting different binding modes with multiple kinases.
Scheme 2 Chemical structures of flavonoid analogues.
Myricetin (6, Scheme 2) is another naturally occurring flavo- noid found in vegetables, fruits and plants that shows broad activity against a variety of kinases, especially PI3K. The addi- tion of one more hydroxyl group to the phenyl moiety resulted in much tighter binding to PI3K than with quercetin. The X-ray crystal structure of myricetin complexed with PI3K demon- strated a completely different binding mode compared to quer- cetin (Fig. 1F). Hinge binding was realized through a single amino acid, V882, with two hydroxyl groups on the phenyl moiety. T867 and D841 in the inner hydrophobic pocket, together with the catalytically active K833 and D964 residues in the DFG domain, provided four more hydrogen bonds with the flavonoid moiety.44 However, in Pim-1 kinase, the compound adopts a different binding conformation, where the flavonoid is rotated 180◦ and three hydrogen bonds are formed in the hinge area with P123 and E121. D186 provides another hydrogen bond contact (Fig. 1G).45
Genistein (7, Scheme 2), an isoflavone and secondary metab- olite secreted during the fermentation of Pseudomonas sp., has been characterized as a specific tyrosine kinase inhibitor.46 In contrast to its relatives in the flavone family, genistein does not show significant inhibitory activity against the PI3Ks (IC50 > 30 mg mL—1).44 As a specific ATP-competitive tyrosine kinase inhibitor, genistein exerts inhibitory efficacy against the epidermal growth factor receptors (EGFR), pp60v-Src and MEK4.47,48
The wide spectrum kinase activity of quercetin, myricetin and genistein paved the way for medicinal chemistry modification to develop more potent and selective kinase inhibitors based on the flavonoid core scaffold. This led to the discovery of LY294002 (8, Scheme 2), one of the first synthetic kinase inhibitors, by Eli Lilly in the early 1990s.49 LY294002 was the first selective class IA PI3K inhibitor (IC50: 1.4 mM) that did not inhibit most of the protein kinases at a concentration of 50 mM. X-ray crystal structure analysis of LY294002 complexed with PI3Kg revealed that it used the morpholine moiety as the hinge binding contact with V882 and that the 8-phenyl group was directed into the space where the ribose of ATP resided, which mainly accounted for its selectivity (Fig. 1H). Due to its high specificity, LY294002 has been widely used as a research tool to dissect the PI3K signaling pathway in the past and has had a significant impact on PI3K signaling pathway studies in the oncology area.50 A version of LY294002 with improved pharmacokinetics (PK), SF1126, that bears a more water soluble prodrug moiety is in clinical trials for the treatment of solid tumors.51
Hematoxylin (9, Scheme 2) is a homoisoflavonoid originally isolated in the early 1990s from the heartwood of Haematoxylon campechianum (Leguminosae) and exhibits broad activity against protein kinases.52 In an ELISA-based automated high throughput screen, hematoxylin was selected as a potent c-Src inhibitor (IC50: 440 nM) from a pool of 32 200 compounds. Further study demonstrated that it could also block various major growth factor-induced signaling pathways such as Ras- Raf-MAPK, PI3K-Akt-mTOR and STAT.53
Flavonoids and their derivatives have demonstrated great efficacy in anti-cancer preclinical studies on colorectal, ovarian, lymphoid and breast cancer, among others. Their specific activity against tumor-derived cells but not normal cells, their anti- proliferative efficacy in promoting cell cycle arrest in the G0/1 and G2/M phases as well as their synergistic effects when combined with other chemotherapeutic agents at least partially account for their comprehensive ability to block oncogenic protein kinase activity.54
3.3 Polyphenol analogues
Balanol (10, Scheme 3) is a fungal secondary metabolite origi- nally isolated in 1993 from Verticillium balanoides and identified as a pan PKC isoform kinase inhibitor (IC50: 4–9 nM).55 Subsequent biological characterization demonstrated that bal- anol was potent against a variety of serine/threonine kinases, especially the AGC, CaMK and CMGC groups (IC50: protein kinase G (PKG), 1.6 nM; protein kinase C beta II (PKCbII), 1.8 nM; protein kinase A (PKA), 3.9 nM; protein kinase C alpha (PKCa), 6.4 nM; P34cdc2, 30 nM; caMKII, 74 nM; mitogen- activated protein kinases (MAPK), 742 nM).56 Recently, it was also found that balanol binds G protein-coupled receptor (GPCR) kinases (GRK) (IC50: 42–490 nM), especially GRK2 (IC50: 42 nM), which is currently being pursued as a drug target for cardiovascular disease.57 The X-ray crystal structure of bal- anol with PKA was solved in 1999 and showed that the p- hydroxybenzamide mimicked the adenine moiety of ATP and formed a hinge contact with E121 and V123 via two hydrogen bonds. Perhydroazepine served as a mimic of the ribose ring and benzophoenone occupied the space where the triphosphate was located (Fig. 2A).58 T183 and E170 formed two hydrogen bonds, which helped stabilize the conformation of the perhydroazepine ring. D184 and E91, together with S53, F54 and F55 in the p- loop, helped orient the benzophenone ring. Extensive medicinal chemistry effort has been devoted to modifying the benzophe- none subunit, and selectivity between PKA and PKC has been achieved via a structure-aided drug design approach.59–61
Scheme 3 Chemical structures of polyphenol analogues.
Scytonemin (11, Scheme 3) is a polyphenol derivative isolated from the extracellular sheath of cyanobacteria as a yellow-green pigment and has long been known as a UV-protective pigment; however, the structure was not solved until 1993.62 Later, it was shown that scytonemin exhibited moderate activity against polo- like kinase 1 (PLK1) (IC50: 2.3 mM) and PKCb (IC50: 3.4 mM). Additionally, in vivo application of scytonemin in phorbol myr- istate acetate (PMA)-induced mouse ear edema demonstrated strong anti-inflammation efficacy.63
Lavendustin A (12, Scheme 3) is a triphenol derivative isolated from Streptomyces griseolavendus and identified as a potent ATP-competitive EGFR inhibitor (IC50: 11 nM) in a high throughput screening assay.64 Later mechanistic study revealed that lavendustin A was a slow, tight-binding inhibitor of EGFR.65 It was also found that lavendustin A could potently block long-term potentiation in the hippocampus through inhibitory activity against Src kinase (EC50: 500 nM).66
Calphostin-C (13, Scheme 3), also named UCN-1028C, is a fungal secondary metabolite isolated in 1989 from Cladospo- rium Cladosporioides and an allosteric PKC inhibitor (IC50: 50 nM).67,68 In contrast to many natural products that exert their inhibitory effects by competing with ATP, calphostin-C blocks PKC activity through interaction with the regulatory domain, thereby showing significant selectivity over protein kinase A (1000-fold). This is an early example of naturally occurring allosteric kinase inhibitors. Further detailed mechanistic study revealed that calphostin-C competed with diacylglycerol (DAG) in a light-dependent manner (5-fold difference between light and dark conditions) at the unique DAG binding site in the PKC regulatory domain, which is also the target of several natural activators such as the phorbol esters, teleocidin and bryostatin. In addition, it was shown that the inactivation of PKC by cal- phostin-C is irreversible, possibly through covalent modification of the DAG binding site using the singlet oxygen produced by light excitation.69 Calphostin-C exhibits strong inactivation effi- cacy in PKC-regulated signal pathways in both benign and malignant tumor cell lines, which indicates a strong photody- namic potential of calphostin-C for anti-cancer therapy.70
Jadomycin-B (14, Scheme 3) is a polyphenol-like antibiotic produced by the bacteria Streptomyces venezuelae ISP5230 under stress conditions.71 Through virtual screening, jadomycin-B was found to have a moderate, inhibitory effect on Aurora B activity (IC50: 10.5 mM) in an ATP-competitive manner.72 In cellular assays, jadomycin-B blocked the phosphorylation of histone H3S10 and induced apoptosis in tumor cells without affecting the cell cycle.
Fig. 2 X-ray crystal structures of kinase inhibitors with kinases. A. Balanol with PKA (PDB: 1BX6); B. Staurosporines with CDK2 (PDB: 1AQ1); C. Wortmannin with PI3Kg (PDB: 1E7U); D. Hypothemycin with ERK2 (PDB: 3C9W).
3.4 Terpenoid analogues
Nakijiquinones A–D (15, 16, 17 and 18, Scheme 4) belong to a family of sesquiterpenoid quinones isolated from an Okinawan marine sponge (family Spongiidae) and exhibit strong cytotoxic effects against leukemia and epidemoid carcinoma cancer cells, as well as potent anti-fungal activity.73–76 The nakijiquione family members differ from each other in the substituents at the quinone moiety, and to date, 17 derivatives have been discovered. Further biochemical investigation showed that nakijiquinones A–D bore moderate kinase inhibitory activities against HER2 (IC50: 30 mM, 90 mM, 26 mM, 29 mM), EGFR (IC50: 400 mM, 250 mM,170 mM, 400 mM) and PKC (IC50: 270 mM, 200 mM, 23 mM, 220 mM), respectively.77 As the only family of natural products inhibiting HER2 kinase activity, nakijiquinions A–D were further developed to search for more selective and potent kinase inhibitors, which led to the discovery of Tie2 kinase-active derivatives (IC50: 70 mM).78,79
Triterpene 3,21-dioxo-olean-18-en-oic acid (19, Scheme 4) was discovered in a bioassay-guided fractionation of the plant Acacia aulacocarpa and was identified as the first naturally occurring small molecule inhibitor of protein tyrosine kinase Tie2 (IC50: 4.2 mM). It exhibits moderate cytotoxic effects on human diploid fibroblasts (EC50: 2 mM), cervical carcinoma cells (EC50: 2 mM) and mouse endothelial cells (EC50: 10 mM), respectively.80
Celastrol (20, Scheme 4), a quinone methide triterpene, was isolated from the Chinese herb Thunder God Vine root and has been used as a remedy for inflammatory and autoimmune diseases.81 Further elucidation of the molecular mechanism showed that celastrol exerted its anti-inflammatory effect on RBL2H3 cells through binding to the ATP binding site of ERK and blocking its phosphorylation activity.82 Celastrol has also demonstrated anti-proliferative activity in a variety of cancer cell lines, including leukemia, gliomas and prostate cancer. It also exhibits malignant tissue invasion prevention and angiogenesis inhibitory effects.83–85 A synergistic effect was observed in combination with both temozolomide against resistant mela- noma cancer cells and radiotherapy against prostate cancer cells.86,87 Detailed mechanistic studies revealed that celastrol might operate through suppression of the NF-kB-mediated signaling pathway by inhibition of the kinase activity of I- Kappa-B Kinase (IKK).88 IKKa and IKKb are inhibited by celastrol in a dose-dependent manner with IC50 values of 1.5 and 0.5 mg ml—1, respectively. C179 in the activation loop was shown to be critical to the inhibition, presumably through irreversible covalent bond formation between the thiol group and the elec- trophilic quinone methide moiety, which shares the same inhi- bition mechanism of celastrol as kinome co-chaperone protein Cdc37 (Scheme 5).89 However, details of the binding mode are not yet clear, although a high resolution X-ray structure of IKKb was recently released, which will facilitate further exploration of the binding mechanism.90 As a specific IKK kinase inhibitor, celastrol has shown great therapeutic efficacy both in vitro and in a W256 cell mediated breast cancer model.91 Due to its inter- esting potential medical applications, celastrol is one of the most attractive naturally occurring kinase inhibitors, although its PK/ ADME properties need to be further manipulated.92
Scheme 4 Chemical structures of terpenoid analogues.
3.5 Indolocarbazole analogues
Staurosporine (21, Scheme 6) is a naturally occurring indolo- carbazole microbial alkaloid that was first isolated from the soil bacteria Streptomyces staurosporeus AM-2282T in the late 1970s as an anti-fungal reagent.93 More than 50 indolocarbazole analogs, including UCN-01 (22, Scheme 6) and K252a/b (23, 24, Scheme 6), were subsequently isolated from a variety of organ- isms. It was not until almost a decade later that staurosporine and K252 were found to be potent protein kinase C (PKC) inhibitors (IC50: 2.7 nM), at which point a significant amount of attention was directed at this series of natural products.94,95 Later, detailed studies demonstrated that staurosporine and its derivatives were actually pan-kinome inhibitors from the nano- molar IC50 range (e.g., c-APK and CDK2) to the micromolar IC50 range (e.g., CK1/2, CSK and MAPK).96,97 Due to its kinome-wide inhibitory activity, staurosporine is used as a control in many kinase activity assays. The first crystal struc- ture of staurosporine with a kinase was solved with CDK2, which revealed its reversible ATP competitive binding mode (Fig. 2B). The binding was strengthened via two hydrogen bonds in the hinge region involving E81 and L83 contacting the lactam moiety, and one hydrogen bond with D86 contacting the methoxyl group of the fused tetrahydropyran.98 To date, there are more than 70 X-ray crystal structures of staurosporine complexes available in the public domain.99 Despite their poor selectivity profile, staurosporine and its natural derivatives have attracted wide interest in the drug discovery community due to their strong anti-tumor activity, and these investigations have led to a set of drug-like candidates for further drug development. Six of the derivatives, including midostaurin (PKC412, 25, Scheme 6), lestaurtinib (CEP-701, 26, Scheme 6), CEP-751 (27, Scheme 6), CEP-1347 (28, Scheme 6), edotecarin (29, Scheme 6) and becatecarin (30, Scheme 6), have advanced to clinical trials. The discovery and development of staurosporine triggered significant interest on behalf of pharmaceutical companies to screen natural products and synthetic derivatives as selective protein kinase inhibitors for oncology programs, which has resulted in protein kinases being one of the most important drug discovery targets of the last 20 years.
Scheme 5 A schematic illustration of the celastrol binding mechanism to IKKb.
Scheme 6 Chemical structures of staurosporine and synthetic derivatives.
3.6 Furanosteroid analogues
Wortmannin (31, Scheme 7) is a steroidal furanoid toxin first isolated in the 1950s from Penicillium wortmannii, although the structure was not elucidated until 1972.100,101 It did not attract much attention until 1992, when it was found that wortmannin could potently inhibit smooth muscle myosin light chain kinase (MLCK), a serine/threonine kinase, with an IC50 of 0.17 mM.102 A primary enzymatic study demonstrated that the inhibitory activity of wortmannin against MLCK occurred at or close to the catalytic site in a dose- and time-dependent manner, and could be prevented by a high concentration of ATP. It was later discov- ered that wortmannin could inhibit PI3K kinase more potently (IC50: 1.9 nM) and also had strong inhibitory activity against other PI3K-like kinases, such as mTOR (IC50: 250 nM) and DNA-PK (IC50: 16 nM).102–104 In addition, human type III PI3K kinase Vps34p was also sensitive to wortmannin (IC50: 3 mM).105
Further studies revealed that wortmannin inactivated PI3K through covalent bond formation between K833, which is a highly conserved catalytic residue responsible for g-phosphate transfer from ATP, and C20 in the furan ring (Fig. 2C).106 Detailed X-ray crystal structure analysis showed that the cova- lent modification induced a large conformational change compared to the reversible inhibitor quercetin.44 In an effort to search for other targets of wortmannin via an activity-based protein profiling approach, PLK1, a serine/threonine protein kinase that is involved in the cell cycle, was found to be strongly inhibited (IC50: 24 nM).107 Due to its high potency and selectivity over PI3K kinase, extensive effort has been devoted to the development of wortmannin as a clinically useful drug for anti- cancer therapy, which finally led to the identification of the more drug-like candidate PX-866 (32, Scheme 7), which has entered clinical trials for solid tumors as either a single agent or in combination with other therapy.
Halenaquinol (33, Scheme 7) and its derivatives (including halenaquinol sulfate (34, Scheme 7), halenaquinone (35, Scheme 7), xestoquinone (36, Scheme 7), tetrahydrohalenaquinone A (37, Scheme 7), etc.) are a class of polyketides bearing a furanoid moiety that were isolated from marine sponges Xestospongia exigua and sapra, and exhibit moderate inhibitory activity against tyrosine kinase v-Src.110–113 Enzymatic study of these compounds determined their IC50 values against v-Src to be 0.55 mM, 28 mM, 1.5 mM, 60 mM and over 200 mM, respec- tively.114 Compared to halenaquinol and other derivatives, the significant loss of inhibitory potency of tetrahydrohalenquinone A indicated that the Michael acceptor furan ring played a critical role in the inhibition. Presumably, a covalent bond is formed with the nucleophilic cysteine residue in the catalytic site. Hale- naquinone has also been shown to moderately inhibit PI3K (IC50: 3 mM) and induce apoptosis in the PC12 cell line.115
Scheme 7 Chemical structures of furanosteroid analogues.
Cercosporamide (38, Scheme 7) is a furanoid derivative iso- lated from plant fungal pathogen Cercosporidium henningsii and bears a wide spectrum of antifungal activity.116 It was later revealed that cercosporamide exerted its biological activity through the inhibition of fungal kinase PKC1 (IC50: <40 nM) in an ATP-competitive manner.117 More recently, it was found that cercosporamide could inhibit MAP kinase interacting kinases (MNK1/2) with IC50 values of 116 nM and 11 nM, respectively. It can effectively block eIF4E phosphorylation in vitro and in vivo and exhibits significant tumor growth suppression in B16 mela- noma pulmonary metastases and HCT116 colon carcinoma xenograft models.118 Broader profiling showed that Jak3 (IC50: 31 nM), ALK4 (IC50: 356 nM), GSK3b (IC50: 516 nM) and Pim1 (IC50: 732 nM) are also protein targets of cercosporamide. 3.7 Resorcylic acid lactones Hypothemycin (39, Scheme 8) belongs to the family of resorcylic acid lactones and was originally isolated as a moderately active anti-fungal and anti-malaria agent from a fungal fermentation culture of Hypomyces trichothecoides.119,120 The discovery that hypothemycin could inhibit cancer cell proliferation and suppress tumor growth renewed interest in the molecule, and further study demonstrated that it was an effective protein kinase inhibitor.121–123 Detailed enzymatic studies revealed that hypothemycin inhibits a variety of kinases at low nanomolar levels, including MEK1/2 (Kd: 17 nM and 38 nM), beta-type platelet-derived growth factor receptor (PDGFRb) (Kd: 900 nM), FMS-like tyrosine kinase-3 (FLT-3) (Kd: 90 nM), vascular endothelial growth factor receptor 1 (VEGFR1) (Kd: 70 nM) and VEGFR2 (10 nM), and at low micromolar levels, including extracellular-signal-regulated kinase (ERK) (Kd: 8.4 mM).124 Biochemical study with human ERK kinase revealed that it was an ATP-competitive irreversible inhibitor that exerted its inhibitory activity via covalent modification of C164 of ERK through the cis-enone moiety. It was further determined that all of the targets strongly inhibited by hypothemycin contained a reactive nucleophilic cysteine in the catalytic site. This irreversible binding mode was clearly demonstrated in an X-ray structure of hypothemycin with ERK2 (Fig. 2D). Scheme 8 Chemical structures of resorcylic acid lactone analogues. Several other classes of NPs have been isolated and exhibit promising inhibitory effects against kinase activity, including LL-783,277 (40, Scheme 8), which was obtained from the fungus. Phoma sp. and identified as an irreversible ATP-competitive MEK4 inhibitor (IC50: 4 nM).126 LL-783,227 exhibits weak reversible inhibitory activity against lymphocyte-specific protein tyrosine kinase (LCK) (IC50: 750 nM), which lacks the corre- sponding cysteine at the catalytic site. LL-783,227 was found to be effective at blocking the proliferation of a set of human epithelial tumor cell lines with EC50 values in the 100–200 nM range, and it also showed significant tumor suppression activity at 100 mg kg—1. Fig. 3 Triad X-ray crystal structure of rapamycin with FKBP-12 and mTOR (PDB: 2FAP). Scheme 9 Chemical structures of rapamycin and analogues. LL-Z1640-2 (41, Scheme 8) belongs to the same NP class as hypothemycin and was isolated in the late 1970s from an unidentified fungus, but did not attract attention because of a lack of apparent biological activity.127 Interest in this molecule was renewed in the early 1990s when its activity against trans- forming growth factor b-activated kinase 1 (TAK1) (IC50: 8 nM) was discovered.128 Sharing the same inhibitory mechanism as hypothemycin, LL-Z1640-2 blocks TAK1 activity in an ATP- competitive and irreversible manner. It also weakly inactivates MEK1 (IC50: 411 nM) with 50-fold selectivity over TAK1. Through topical application, LL-Z1640-2 has exhibited anti- inflammation efficacy in animal models. Despite the promising results from in vitro studies, poor stability in blood plasma and liver microsomes hindered its preclinical development, which resulted in an extensive medicinal chemistry effort to improve its pharmacokinetic properties. These studies lead to a more drug-like candidate, E6201 (42, Scheme 8), which inhibits lipopoly- saccharide (LPS)-induced tumor necrosis factor-alpha (TNF-a) transcription with an IC50 of 50 nM, but does not significantly affect b-actin. E6201 is in phase II clinical trials for the treatment of plaque-type psoriasis and phase I clinical trials for solid tumor therapy.129 3.8 Rapamycin and analogues Rapamycin (43, Scheme 9), also known as sirolimus, is a natural macrocyclic polyketide originally isolated from the soil bacteria Streptomyces hygroscopicus in the early 1970s and identified as an anti-fungal reagent.130 Later, it was found that this compound exhibited immunosuppressant effects, which resulted in success- ful clinical application for organ transplantation in 1999, and anti-tumor activity, which lead to approval for anti-cancer treatment in 2007. Interest in rapamycin dramatically increased in the early 1990s, when the seminal discovery of the cellular targets of rapamycin in yeast and mammalian cells as target of rapamycin (TOR) and mammalian target of rapamycin (mTOR), respectively, were reported.131,132 Detailed study revealed that rapamycin indirectly inhibited mTOR activity through the formation of a complex with FK506-binding protein (FKBP12) and subsequent perturbation of mTOR functional protein complex 1 (mTORC1) in the cellular background. X-ray crystal structure analysis clearly demonstrated this triad interaction (Fig. 3).133 This unique inhibitory mechanism endowed rapa- mycin with extreme selectivity and attracted wide interest from the drug discovery community. Due to its poor PK properties, including water solubility and bioavailability, extensive medic- inal chemistry effort was expended to modify the C-40-O posi- tion, which resulted in several drug-like molecules such as CCI-779 (Termsirolimus, 44, Scheme 9), RAD-001 (Everolimus, 45, Scheme 9) and AP23573 (Ridaforolimus, 46, Scheme 9).134 Rapamycin and its analogues have been applied in a variety of clinical evaluations for anti-cancer treatment, either as a single agent or in combination with other chemotherapies. Among all of the naturally occurring kinase inhibitors, rapamycin has likely been studied more than any other compound, and is the only one to be applied in clinical practice. 4 Perspective Protein kinases are the key regulators of signal transduction and have been widely explored in recent decades as drug targets for a variety of diseases, including cancer, type II diabetes, inflam- mation, cardiovascular disease and neurological disease, among others. It is no surprise that NPs have played a critical role in this new era of life science. The aforementioned NPs are good examples of how people can benefit from manipulating naturally occurring compounds, in terms of both basic research and clin- ical applications. Kinase inhibitors created by Mother Nature have greatly aided studies of signal transduction pathways by providing valuable research tools. As previously discussed, NPs possessing protein kinase inhibitory activity are derived from a wide spectrum of natural sources, including marine sponges, bacteria, fungus and plants (Table 1). Additionally, a variety of protein kinases are targeted by these NPs, including all subfamilies of the currently known kinome, such as the tyrosine kinase family (TK), the tyrosine kinase-like family (TKL), homologs of the yeast sterile 7, 11 and 20 kinases (STE), the casein kinase 1 family (CK1), the PKA, PKC and PKG kinase families (AGC), the calcium/calmodulin-dependent kinase family (CAMK), the CDK, MAPK, GSK2 and CLK kinase families (CMGC) and other (Fig. 4). The inhibitory effi- cacies of NPs against kinases range from single digit nanomolar to submillimolar levels. The binding modes vary from ATP- competitive to allosteric, and the enzymatic kinetics can be reversible or irreversible. Generally speaking, the ATP-compet- itive reversible inhibitors tend to have weaker binding affinities compared to ATP-competitive irreversible inhibitors, i.e., flavo- noid analogues versus hypothemycin. The ATP-competitive inhibitors usually have poorer selectivity profiles than the allo- steric inhibitors, as the similarity of the 3-D structure of the highly conserved ATP binding domain across the kinome makes it more challenging to achieve selectivity. Irreversible inhibitors tend to have better selectivity profiles than reversible inhibitors since one more filter is added to the selectivity criteria. Compared to synthetic irreversible inhibitors, which are more likely to bear an acrylamide moiety as an electrophile,135 the electrophiles in NPs tend to be enones, which is likely a reflection of their biosynthetic origin. Hinge binding is one of the most critical factors for inhibitory affinity, and various moieties have demonstrated this capability, including the 1,3-N,N system (1–3), 1,3-O,N system (4, 10, 21–29) and single O/N system (5–9, 11–12, 14–20, 31–42). Although a combination of the allosteric and irreversible binding modes is very rare, it can afford better inhibitory profiles in terms of potency and selectivity, as observed with calphostin-C. Fig. 4 NP kinase inhibitor targets throughout the kinome; the figure was generated with DiscoverRX TREEspot™ v4.0. A significant concern with NPs serving as kinase inhibitors in terms of drug development is their poor drug-like properties, including high molecular weight, low water solubility and poor stability. Moreover, their structural complexity hinders the development of practical syntheses, which makes it very chal- lenging to produce enough material for preclinical studies. Further, the emerging efforts to develop kinase inhibitor thera- pies have limited the availability of heterocyclic core structures and their corresponding intellectual property, which has adversely impacted development. NPs that contain proper core scaffolds could serve as excellent starting points for medicinal chemistry, significantly increase the structural diversity of kinase inhibitors and provide new clinical candidates for current unmet medical needs.