|Home | About | Journals | Submit | Contact Us | Français|
Through the syntheses of its C-1 desvinyl, C-7 methylene, C-7 exocyclic ethylidene, and various C-3 phenylmethyl analogues, the structure–activity relationship of antimitotic ottelione A (4) against tubulin and various cancer cells was established. The results indicated that compound 4 was a colchicine-competitive inhibitor and that the C-1 vinyl group is unnecessary for its potency, whereas the C-7 exocyclic double bond is essential, possibly because of its irreversible interaction with tubulin. Further optimization of the substituents on the phenylmethyl group at the C-3 position generated compound 10g with a 3′-fluoro-4′-methoxyphenylmethyl substituent, which was 6–38-fold more active against MCF-7, NCI-H460, and COLO205 cancer cells relative to 4. Results from in vitro tubulin polymerization assay confirmed the potency of compounds 4, 10g, and 11a.
After the success of antimitotic taxanes and vinca alkaloids as anticancer drugs,1,2 antimitotic agents that target the colchicine-binding site to inhibit tubulin polymerization have attracted much attention. Many structurally diverse compounds binding to the colchicine-binding site have been discovered from natural or synthetic sources.3,4 Despite no compound of this class having been approved for clinical use, the prototype colchicine (1, Figure Figure1)1) has been used to treat gout for many years.5 Combretastatin A-4 (2)6−9 and ABT-751 (3)10,11 in Figure Figure11 are representative examples of colchicine-competitive inhibitors currently in clinical trials to treat various cancers.
Ottelione A (4, Figure Figure1), isolated1), isolated from the fresh water plant Ottelia alismoides collected in the Nile Delta,12 is among the most potent natural products that possess in vitro antiproliferative activity. Its IC50 values lie in the pM–nM range against a panel of 60 human cancer cell lines through acting as a colchicine-competitive inhibitor.12,13 Because of its promising antitumor activity and intriguing core structure with four stereogenic centers, several synthetic groups have investigated the total synthesis of 4,14−21 but no analogue has been reported and assayed for anticancer activity, perhaps because of the many steps in its preparation.
We already achieved an enantioselective total synthesis of 4 using our α-carbonyl radical cyclization method.21 The bicyclo[4.3.0]nonane core structure was constructed via radical cyclization. The C-3 3′-hydroxy-4′-methoxyphenylmethyl group, C-1 vinyl group, and C-7 exocyclic double bond were then sequentially introduced. We thus envisaged that our scheme could be adaptable for the preparation of various analogues of 4, which would be useful to establish its structure–activity relationship.
As shown in Scheme 1, we first prepared compounds 10a–g and 11a–c according to our scheme for the total synthesis of 4. For economy, these compounds were all synthesized in their racemic forms. Compound 5, prepared via α-carbonyl radical cyclization,21 served as the starting material. Hydroboration of 5 with 9-borabicyclo(3.3.1)nonane (9-BBN) at its exocyclic double bond, followed by Suzuki–Miyaura coupling with various aryl iodides and subsequent removal of the tert-butyldimethysilyl (TBDMS) protective group, gave compounds 6a–g in 23–65% yields. Oxidation of 6a–g with pyridinium chlorochromate (PCC) in CH2Cl2 generated 7a–g in 61–93% yields. Treatment of 7a–g with Ph3P+CH3Br– and tBuOK afforded compounds 8a–g possessing a C-7 exocyclic double bond. Removal of the acetyl group from 8a–g with NaOH followed by oxidation with PCC afforded compounds 9a–g in 38–95% yields. To introduce the double bond at the C-5 and C-6 positions, we treated compounds 9a–g first with LiN(SiMe3)2 and PhSeBr then mCPBA to give compounds 10a–g in 45–74% yields. Removal of the methoxymethyl (MOM) protective group in 10a–c provided compounds 11a–c in 55–65% yields (Scheme 1).
Compound 19, lacking a C-1 vinyl and C-7 exocyclic double bond, was prepared from 2-cyclohexen-1-one (12) according to our method similar to that for 11a–c (Scheme 2).22 Reaction of 12 with 4-(trimethylsilyl)-3-butynylmagnesium chloride in the presence of CuI and TMSCl followed by iodination with NaI and mCPBA gave α-iodoketone 13 in 96% yield. Radical cyclization of 13 followed by reduction with NaBH4 and protection with Ac2O afforded 15 through the key intermediate 14. Compound 15 was converted to compound 16 by hydroboration and Suzuki–Miyaura coupling in 30% yield. Deprotection of 16 followed by oxidation gave ketone 17 in 74% yield. Compound 17 was converted to α,β-unsaturated enone 18 in 33% yield via treatment with LiN(SiMe3)2 and PhSeBr followed by oxidative elimination. Acidic hydrolysis of the MOM protecting group in 18 provided 19 in 63% yield.
To prepare C-7 exocyclic ethylidene derivative 23, we treated compound 7a with Ph3P+CH2CH3Br– and tBuOK to give compound 20 in 88% yield (Scheme 3). Removal of the acetyl group in 20 followed by oxidation gave 21 in 81% yield. Introduction of the double bond between C-5 and C-6 positions gave 22 in 63% yield. Compound 22 was subsequently hydrolyzed in acidic conditions to remove the MOM group to afford the desired exocyclic ethylidene derivative 23 in 80% yield. Compounds 20–23 were all obtained in Z and E mixtures and could not be separated by simple column chromatography; we therefore employed NOE experiments to determine the E/Z ratio of 23. Two doublet peaks at 7.34 and 6.82 ppm, corresponding to the C-6 hydrogen with an integral ratio 1/3.4, were recorded in the 1H NMR spectrum of 23. Irradiation of the peak at 7.34 ppm showed a positive NOE effect with 7.58% enhancement at the signal of CH3 at 1.83 ppm, indicating that the peak corresponded to (E)-23; irradiation of the peak at 6.82 ppm, however, showed no positive NOE effect, revealing it to correspond to (Z)-23. The E/Z ratio of 23 was hence determined to be 1/3.4.
After these compounds were prepared, assays of the percentage of colchicine binding inhibition (CBI %) and antiproliferation against human cancer cells including KB (nasopharyngeal), MCF-7 (breast), NCI-H460 (lung), HCT-116 (colon), A375 (melanoma), and COLO205 (colon) were employed to evaluate their potencies. Compound 4 and its C-1 desvinyl derivative 11a, C-7 methylene derivative 19, and C-7 exocyclic ethylidene derivative 23 were assayed first (Table 1). Enantiomerically pure 4, prepared according to our published method,21 showed strong inhibition of the growth of the cancer cells with IC50 values of 0.40–92.8 nM; the results are consistent with the reported potency of 4 from natural12 and synthetic sources.17 The cells most sensitive toward 4 were HCT-116 and A375 cells, which were inhibited at subnanomolar concentrations (IC50, 0.74 and 0.40 nM, respectively). Compound 4 also showed great CBI % with a near-plateau value of 98.3%, revealing its binding to the colchicine-binding site that contributed to its remarkable antiproliferative activities.
The C-1 desvinyl analogue 11a was also active against the cancer cells with IC50 values ranging from 1.07 to 243.9 nM (Table 1). Although slightly less potent than 4, compound 11a also exhibited a satisfactory CBI % (95.8%). As a result, the C-1 vinyl might be less important for the potency of 4. We subsequently prepared all analogues without the vinyl substituent. Compound 11a also displayed a spectrum of activities against the cancer cells similar to that of 4, being more active to inhibit HCT-116 and A375 cells but less potent against COLO205.
Unlike the C-1 vinyl group, the C-7 exocyclic double bond was crucial for the potency of 4; the evidence came from the loss of the antimitotic (CBI %, 39.2%) and antiproliferative activities (IC50 > 1.0 μM) of 19 that did not bear the moiety (Table 1). As the C-7 exocyclic double bond is conjugated with the C-4 carbonyl, it might be susceptible to nucleophilic 1,6-addition23 with amino acid residues in the binding site, for example, serine, threonine, and cysteine residues.24 As a result, an irreversible inhibition of tubulin polymerization caused by the C-7 exocyclic double bond was hinted. To provide further evidence, we introduced a methyl substituent on the double bond to generate analogue 23 that was more hindered for nucleophilic addition. It showed decreased antiproliferative activities (1.5–8.6-fold) relative to 11a, which indicated that the 1,6-addition was possible. The nucleophile would be a thiol, for T138067, a pentafluorobenzenesulfonamide that possesses an N-aryl group similar to that of 2, is reported to covalently modifies tubulin at a conserved cysteine residue in the colchicine-binding site.25
After clarifying the roles of the C-1 and C-7 moieties, we examined the effect of substituents on the phenyl group of 4 on the antimitotic and antiproliferative activities. Compounds 10b–d, 11b, and 11c were subjected to the in vitro assays; the results are summarized in Table 2. The unsubstituted 10d (R1 = R2 = R3 = H) was less potent against the cancer cells with IC50 values of 237–678 nM, which were 2.2–221-fold less potent than 11a, especially on the sensitive HCT-116 and A375 cells (85- and 221-fold less potent, respectively). Its CBI % was also decreased to 43.2%. For 11b bearing a para-hydroxyl group, the antiproliferative activities and CBI % were greater than unsubstituted 10d. Blocking the para-hydroxyl in 11b with a MOM group gave compound 10b with decreased antimitotic and antiproliferative activities.
As the structure of 4 is similar to that of 2, we synthesized compound 10e with a trimethoxyphenyl substituent, but a complete loss of activity was observed (CBI % = 8.93% and IC50 > 1.0 μM). The 3′-hydroxyl-4′-methoxyphenyl in 4 evidently did not occupy the same space as the trimethoxylphenyl group in 2 in the colchicine-binding site.
Introduction of a m-fluoro group to 10b gave compound 10c with the retention of CBI % (69.3% for 10b and 68.9% for 10c; see Table 2), but 10c showed increased antiproliferative activity against cancer cells relative to 10b, except MCF-7. Removal of the MOM group in 10c afforded compound 11c with a ~2-fold decrease of CBI % (37.3%) as well as its antiproliferative activities. Compound 10f with 3′,4′-difluoro substituents in the phenyl ring was less active than 11c, in both CBI % and antiproliferative activities. We eventually obtained compound 11g with a 3′-fluoro-4′-methoxylphenyl substituent that showed a CBI % value, 97.1%, comparable with that of 4 and 11a. It was most potent among the compounds in this work against various cancer cells, with IC50 values of 0.130–2.77 nM. Even on the least sensitive COLO205 cells, its activity was also remarkable, possessing an IC50 value of 2.45 nM, which was 38 times as active as 4.
The change of the hydroxyl group in 2 to a fluoro atom was found to retain its antiproliferative activity against K562 cells.26 In our work, the switch from OH to F in 11a also provided 10g with retained or improved potency. As a result, compound 4 might bind in tubulin in a position similar to that of compound 2. The 3′-hydroxyl-4′-methoxylphenyl substituent in 2 and 4 would occupy the same space in the colchicine-binding site.
To confirm the drug target of the compounds in Tables 1 and 2, potent compounds 4, 10g, and 11a were tested using in vitro tubulin polymerization assay at concentration of 10 μM with 1 as the positive control. As presented in Figure Figure2,2, all compounds significantly inhibited the polymerization of tubulin; 4 and 10g showed potency similar to that of 1. In combination with their CBI % values as shown in Tables 1 and 2, the results confirmed that the antiproliferative activity of the compounds arose from inhibiting tubulin polymerization through binding to the colchicine-binding site.
We summarize our work on the structure–activity relationship of 4 against various cancer cells in Figure Figure3.3. As 11a showed activity comparable with that of 4, the C-1 vinyl is unnecessary for the activity, but the C-7 exocyclic double bond is essential for the potency. The exocyclic double bond in conjugation with the C-4 carbonyl might make it susceptible to nucleophilic addition, thus irreversibly binding with tubulin. Among the hydrogen, hydroxyl, methoxy, and MOM groups at the para position of the phenyl group in 4, only the methoxy group provided potent compounds. For the meta position, a fluoro or a hydroxyl group enhanced the potency, whereas trimethoxy substituents on the phenyl group deactivated compound 10e.
In conclusion, we have established the structure–activity relationship of 4 focusing on the C-1 vinyl, C-7 exocyclic double bond, and the substituents on the phenyl ring to inhibit colchicine binding and the growth of cancer cells. Compound 10g, a less complicated analogue of 4, show potency comparable with, or superior to, that of 4. Extensive syntheses and biological evaluation are now in progress in our laboratory. Using the information provided in this work, we are seeking more potent and druggable molecules for further development.
Experimental procedures for compounds 6a–g, 7a–g, 8a–g, 9a–g, 10a–g, 11a–c, and 13–23 and procedures for bioassays. This material is available free of charge via the Internet at http://pubs.acs.org.
National Science Council of the Republic of China (Taiwan) provided financial support.
The authors declare no competing financial interest.