Novel Epothilone Lactones by an Unusual Diversion of the Grubbs’ Metathesis Reaction

Abstract

An unusual reaction with Grubbs’ catalyst during the synthesis of bridged epothilones yielded five-membered internal lactones instead of the expected metathesis products. Three of the lactones have comparable activities to epothilone D.

The discovery of the taxol-like tubulin-polymerization activity of epothilones A-D (1 – 4)¹ in 1995² led to great scientific interest in their development as potential anticancer agents. The chemistry, biology, and structure activity relationships (SAR) of the epothilones have been extensively reviewed,^1,3,4 and epothilone-based drug discovery research has delivered seven compounds in development, including the lactam analogue ixabepilone, which has been approved for clinical use to treat certain forms of breast cancer.⁵

Despite the structural differences between the epothilones and paclitaxel, the former have been found to bind to the same site on microtubules as paclitaxel and to enhance tubulin polymerization more effectively than paclitaxel. They also displace [³H]-paclitaxel from the binding site on tubulin.⁶

Knowledge of the bioactive tubulin-binding conformation(s) of epothilones would help to guide the design of structurally simpler and more potent drugs. It could also help to explain the SAR data for these compounds. Two clever approaches used to assess the bioactive conformation of the epothilones integrate NMR and molecular modeling: conformation-activity relationships⁷ and INPHARMA.⁸ In additon, several efforts have been made to design pharmacophore models common to paclitaxel and the epothilones.^9-12

Very recently, two groups independently presented epoA binding conformations based on TR-NOESY and TR-CCR rates from NMR^13,14 and analysis of electron crystallographic (EC) density maps of Zn-stabilized tubulin sheets that diffract electrons to 2.9 Å.¹⁵

While the NMR work posits a bound conformer very similar to the single crystal X-ray structure of epoA^13,16 the EC study suggests a uniquely different conformer.¹⁵ The latter accounts for the drug resistance of mutated cell lines, however, both structures appear to satisfy various aspects of the SAR.^13,15,17,18 Inspection of the epoA EC model reveals juxtaposition of the C4 methyl carbon and C12-H at a distance of 4.5 Å (Fig. 1a). In epoB the corresponding separation between C4-Me and C12-Me is 5.5 Å. To test the proposed EC binding mode, we sought the two bridged epoD and epoB analogs 5 and 6 (Fig. 1b) as synthetic targets. However, the intervention of an unusual ruthenium-catalyzed internal lactonization of the epothilone scaffold under Grubbs-metathesis reaction conditions, compels us to report the synthesis of a new class of active analogs.

Fig. 1.

(a) The epoA model based on EC density (ref. ¹⁵). (b) MMFF-energy minimized structures of the EC template and the designed bridged epoD analog, 5. The two structures were overlayed with 3 point selection by PyMol.

The syntheses of the proposed bridged epothilones 5 and 6 were envisioned to proceed from diene precursors using a ring-closing metathesis reaction¹⁹ as the penultimate step (Scheme 1) by analogy with our successful synthesis of bridged paclitaxel derivatives.²⁰ Subjection of alcohol 7²¹ to Sharpless epoxidation conditions²² furnished 12α,13α-epoxide 8 (d.e. ca. 80%) or 12β,13β-epoxide 10 (d.e. > 95%) respectively. The primary hydroxyl groups of 8, 10, and 7 were selectively acylated with acryloyl chloride to yield the diene precursors 9, 11, and 14 in good yields. These compounds were subjected to ring closing metathesis with the second-generation Grubbs catalyst. To our surprise, the three novel epothilone analogs 12, 13, and 15 with internal lactones between C3 and C4 were obtained instead of the expected bridged analogs 5 and 6.

Scheme 1.

Reagents and conditions: (a) D-DET, Ti(OiPr)₄, t-BuOOH, 63% (8) L-DET, Ti(OiPr)₄, t-BuOOH, 83% (10); (b) CH₂=CHCOCl, Et₃N, DMAP, CH₂Cl₂, 79% (8 --> 9), 79% (10 --> 11), 79% (7 --> 14); (c) 2nd generation Grubbs catalyst, CH₂C₂, rt, 21 h, 44% (14 --> 15), 24% (9 --> 12), 25% (11 -->< 13)

The structure of the novel epothilone analogue 12 was elucidated using 1D and 2D NMR experiments, as well as by high resolution mass and IR spectrometry. The molecular formula of 12 was deduced as C₃₁H₄₁NO₉S from its HRFABMS. Compared to its starting material, 12 lacks one terminal double bond based on ¹H- and ¹³C NMR spectra, and possesses an additional lactone carbonyl group (δ_C = 174.4 ppm). Its IR spectrum shows absorption at 1786 cm^-1, typical for a 5-membered lactone ring. In the HMBC spectrum, the lactone carbonyl carbon is correlated with a methylene (δ_C = 41.0 ppm, δ_H = 2.57, 3.53 ppm, d, J = 16.4 Hz), which can be assigned as the methylene of a 5-membered lactone since it correlates with C-5, C-4, 4-CH₃ and C-3. Furthermore, the H-3 signal is downshifted from 4.5 ppm to 5.18 ppm relative to the starting material, indicating that the hydroxyl group at C-3 is esterified. Compounds 13 and 15 were fully characterized on the basis of similar observations.

This unexpected outcome of the Grubbs metathesis reaction for 9, 11, and 14 must result from a combination of the decreased rate of the normal metathesis, due to the steric strain of the putative bridged product and the decreased reactivity of the acryloyl double bond, coupled with the availability of an alternate reaction pathway. It should be noted that the homologated 3-butenoate analog of 14 did undergo normal metathesis to give a bridged epothilone D analog, as did the 4S isomer of 14, consistent with the importance of strain effects.²¹

Treatment of analog 7 and the new analogs 19 and 23 (Scheme 1) with Grubbs’ catalyst occurred as before to give the internal lactones of 26-hydroxy epothilone D (16), epothilone D (17), and epothilone B (18), indicating the lack of involvement of the acryloyl double bond in the reaction.

The mechanism of this unusual reaction is probably analogous to those proposed for the ruthenium-catalyzed transformations of amino alcohols to lactams²³ or of amines and alcohols to amides.²⁴ It is proposed that the ruthenium reagent reacts initially in the normal way with the allyl double bond of an epothilone derivative such as 23 to give the intermediate 24 (Scheme 2). This compound could then lose styrene to give complex 25, which is presumably too sterically encumbered to react with the acryloyl double bond. Complex 25 is thus trapped by the internal C3 hydroxyl group to give 26, which undergoes oxidation by loss of dihydrogen²⁴ to give 27 and subsequent hydrolysis to give 17. The use of a stoichiometic quantity of Grubbs’ catalyst gave a yield of 88% of 17 from 23, indicating that the catalyst is actually a reagent in this reaction.

Scheme 2.

Proposed mechanism of lactone formation.

Interestingly, the internal epothilone lactones 17-18 and the open-chain C4-allyl analog 19 all showed antiproliferative activities against the A2780 ovarian cancer cell line similar to those of epothilones B and D (Table 1). These compounds also enhance in vitro tubulin assembly approximately twice as potently as epothilone D, although the K_a values should be considered only rough estimates of the relative affinity of the ligands for assembled tubulin. The inhibition constants for binding to GMPcPP-stabilized microtubules do however demonstrate association with the paclitaxel site on microtubules,

Table 1.

Bioactivity of epothilone D and analogues.

Cmpd.	IC50 (nM) A2780a	ED50, Tb polym. (μM)b	Ka (× 107 M-1)c
2 (Epo B)	1.9 ± 0.2	ND	ND
4 (Epo D)	6.8 ± 1.7	0.44 ± 0.02	15.2 ± 0.44
7	2050 ± 1140	1.24 ± 0.07	0.72 ± 0.49
9	1040 ± 130	1.03 ± 0.06	0.67 ± 0.04
11	1070 ± 190	1.12 ± 0.05	0.71 ± 0.04
12	260 ± 83	0.55 ± 0.05	2.02 ± 0.11
13	265 ± 66	0.62 ± 0.05	2.95 ± 0.12
14	238 ± 17.5	0.58 ± 0.06	2.38 ± 0.34
15	39 ± 10	0.47 ± 0.03	2.21 ± 0.12
16	250 ± 28	0.60 ± 0.06	1.26 ± 0.09
17	4.6 ± 2.3	0.18 ± 0.02	8.98 ± 0.17
18	4.2 ± 0.81	0.14 ± 0.01	8.50 ± 0.23
19	8.9 ± 0.94	0.20 ± 0.01	6.84 ± 0.15
23	39 ± 9.7	0.47 ± 0.03	6.99 ± 0.37

^aAntiproliferative activity to A2780 ovarian cancer cells.

^bPromotion of tubulin polymerization.

^cInhibition constants for epothilone analogs binding to GMPcPP-stabilized microtubules

Docking the lactones onto the electron crystallographic (EC) pose within the tubulin binding site¹⁵ illustrates that the fused lactone is compatible with this binding model and causes no steric conflict with the protein (Figure S1, ESI). The NMR-derived binding conformer¹³ is likewise compatible with fusion of the lactone at C3-C4. However, in the absence of a completely credible protein binding model for both structures, the new analogs do not discriminate between the epo conformers.

Cytotoxicities against the A2780 cell line show a pattern that differs from the in vitro analyses. In particular, 16 exhibits one of the least favorable IC₅₀ values against the A2780 cell line, while 17 and 18 are slightly more potent than epoD. Calculated MDCK and Caco-2 membrane permeabilities for the four compounds suggest that 17 and 18 are like epoD in their ability to penetrate the cell wall readily, while 16 would have a relatively poor penetration ability by comparison (See ESI). Structural variations within the epothilone architecture would seem to be responsible for relative cytotoxicities independent of tubulin binding.

The newly prepared internal lactones not only represent a novel fused epothilone family, but three of the analogs likewise exhibit antiproliferative activities against A2780 cells equal or superior to epoD. One important outlier, 16, would appear to exhibit diminished activity as a result of an unfavorable membrane permeability. The result suggests that target design coupled with predicted permeabilities offers a means to improve cell-kill by routine filtering of conceived structures by this molecular property estimate prior to laboratory synthesis.

We are grateful for support of this work by NIH Grant CA-69571 and by NSF Grant CHE-0619382 for purchase of the Bruker Avance 600 spectrometer. We thank Dr. Dennis Liotta (Emory University) for support and Dr. Gerhard Höfle (Helmholtz HZI, Braunschweig) for a gift of samples of epothilones B and D.