Total synthesis of (-)-CP₂-disorazole C₁

Abstract

The total synthesis of a bis-cyclopropane analog of the antimitotic natural product (-)-disorazole C₁ was accomplished in 23 steps and 1.1% overall yield. A vinyl cyclopropane cross-metathesis reaction generated a key (E)-alkene segment of the target molecule. IC₅₀ determinations of (-)-CP₂-disorazole C₁ in human colon cancer cell lines indicated low nanomolar cytotoxic properties. Accordingly, this synthetic bioisostere represents the first biologically active disorazole analog not containing a conjugated diene or polyene substructure element.

The disorazoles compromise a family of ∼30 closely related polyketide microtubule disruptors isolated since 1994 from the fermentation broth of the myxobacterium Sorangium cellulosum by Jansen, Reichenbach, Höfle and co-workers.¹,2 Members of the disorazole class have displayed anti-cancer activity in the nano- and picomolar range against a variety of transformed cell lines, including multi-drug resistant cells.³ To date, only (-)-disorazole C₁ has yielded to total synthesis,⁴ although several simplified analogs and segments have been reported.⁵ As part of our investigations of structure-activity relationships (SAR) of biologically active natural products,^2d,6 we designed a cyclopropane analog of (-)-disorazole C₁ (i.e., CP₂-disorazole C₁), with the goal to replace the labile⁷ (E,Z,Z)-triene subunit with an isosteric and biologically similarly effective moiety. The choice of absolute configuration for the cis-cyclopropanes in CP₂-disorazole C₁ reflected the stereochemistry observed for the epoxide present in disorazole A₁.^2a

Retrosynthetically, the C₂-symmetrical macrodiolide CP₂-disorazole C₁ (1) could be constructed from two key fragments, 15 and 9 (Scheme 1). Diene 15 is derived from β-hydroxy ester 12 which can be obtained from an asymmetric Mukaiyama aldol reaction of crotonaldehyde. By taking advantage of a vinyl cyclopropane cross-metathesis reaction,⁸ our plan was to access alkene 9 from the readily available cyclopropane 5 and the oxazole precursor 3. This approach was realized in excellent yields beginning with the straightforward preparation of ester 3 in two steps from the known⁹ alkyne 2 (Scheme 2).

Scheme 1

Retrosynthetic approach for CP₂-disorazole C₁1

Scheme 2

Synthesis of ester 3

The vinyl cyclopropane 5 was prepared in 74% yield in a Wittig condensation from known aldehyde 4^5d (Scheme 3). Cross-metathesis reaction of alkenes 5 and 3¹⁰ in the presence of a 2 ^{generation ruthenium catalyst11} provided alkene 6 in 70% yield as an inseparable 10:1 mixture of E/Z-stereoisomers. O-Methylation of 6 with silver oxide¹² and methyl iodide, followed by enzymatic hydrolysis¹³ with pig liver esterase (PLE) and subsequent coupling of the carboxylic acid with serine methyl ester yielded amide 8. A two-step cyclodehydration/oxidation procedure¹⁴ led to oxazole 9. Oxidative deprotection of the PMB ether followed by further oxidation of the resulting primary alcohol under Dess-Martin conditions afforded the desired aldehyde 10 in excellent yield.

Scheme 3

Cross-metathesis and synthesis of aldehyde 10

After considerable experimentation, we found that Kiyooka's conditions were best suited for the asymmetric Mukaiyama aldol reaction between the sterically hindered silyl ketene acetal 11 and crotonaldehyde (Scheme 4).^5a,15

Scheme 4

Synthesis of hydroxy ester segment 18

The α,α-dimethylated aldol product 12 was obtained in 79% yield and 93% ee as determined by chiral HPLC analysis. Protection of the secondary alcohol in 12 with 3,4-dimethoxybenzyl (DMB) bromide¹⁶ followed by ester reduction with LAH and oxidation of the primary alcohol under Swern conditions provided aldehyde 13. Among contemporary allylation protocols,¹⁷ we found that the Duthaler-Hafner allylation¹⁸ of aldehyde 13 gave alcohol 14 in the highest (87%) yield and with a satisfactory dr=4.8:1.0, favoring the desired anti-configuration of the 1,3-diol. The minor isomer was present in diminishing amounts in subsequent intermediates until it was finally completely removed in the Wittig reaction of 17 and 10.

At first, the advancement of homoallylic alcohol 14 to the desired phosphonium salt 17 proved problematic. Upon formation of the alkyl iodide at C12, an in situ S_N2-displacement of the iodide by the electron-rich DMB ether at C16 afforded the corresponding tetrahydropyran as the major product.¹⁹ Although small quantities of the desired alkyl iodide were isolated, immediate formation of the tetrahydropyran occurred upon heating. Consequently, TES protection of the alcohol at C14 followed by the exchange of the DMB with a TBS ether at C16 was used to access the orthogonally silyl-protected diol 15 (Scheme 4). Stepwise Johnson-Lemieux oxidation followed by NaBH₄ reduction afforded the primary alcohol 16. Iodination and phosphonium salt generation at elevated temperature in a sealed vessel proceeded now without incident to furnish 17 in good overall yield. The Wittig condensation of the freshly formed phosphonium salt with aldehyde 10 in a vigorously degassed THF solution yielded 86% of the desired Z-alkene as a single stereoisomer by ^{H NMR analysis. Finally, selective TES deprotection with PPTs in the presence of the allylic TBS ether proceeded in 65% yield to afford the advanced segment} 18.²⁰

Originally, our end game strategy had envisioned the direct cyclo-dimerization of the carboxylic acid corresponding to 18. However, all efforts to realize this cyclo-dimerization resulted in the exclusive formation of the 15-membered macrolactone arising from the direct lactonization at the C14 hydroxyl group, i.e. the formation of the cyclic monomer.²¹ In order to circumvent this undesired reaction, a stepwise approach was adopted (Scheme 5). Hydrolysis of ester 9 followed by DCC-mediated coupling to the secondary alcohol 18 provided the bis-cyclopropane 19 in 89% yield. Oxidative PMB deprotection with DDQ and subsequent Dess-Martin oxidation of the resulting primary alcohol led to the requisite cyclopropyl aldehyde in good yield. Condensation of this aldehyde and the ylide generated from phosphonium salt 17 after vigorous degassing afforded the seco-dimer 20 in 74% yield as a single alkene stereoisomer.

Scheme 5

Segment condensation and Shiina macrolactonization

Selective TES ether deprotection of 20 revealed the secondary alcohol at C14′ (Scheme 5). With all functional groups properly installed, the stage was now set for the final lactone formation. Selective saponification of the methyl ester in the presence of the internal ester linkage of 20 with barium hydroxide followed by Shiina macrolactonization²² with 2-methyl-6-nitrobenzoic anhydride (MNBA) and DMAP provided the desired 30-membered macrocycle in 55% yield over the two steps. Finally, double TBS ether deprotection in the presence of HF-pyridine proceeded in 64% yield to the desired (-)-CP₂-disorazole C₁.

Table 1 illustrates our preliminary analysis of the antiproliferative properties of this novel disorazole analog. In three human colon cancer cell lines, 1 displayed low IC₅₀ values ranging from 25-50 nM, and therefore proved to be only slightly less active than its parent, (-)-disorazole C₁.^3a-d,23 Furthermore, 1 was equipotent to the clinically used anticancer vinca alkaloid vincristine in these colon cancer cells. In contrast, cyclomonomer 21 was inactive at concentrations up to 50 μM in all three cell lines, thereby serving as a negative control and confirming the importance of the 3-D architecture of the disorazoles.^3d

In summary, we have developed an efficient route to (-)-CP₂-disorazole C₁1, a bis-cyclopropane analog of the antimitotic natural product disorazole C₁.²⁴ The target agent was obtained in 23 steps and 1.1% overall yield for the longest linear sequence. It contains 10 stereogenic carbons, 4 more than disorazole C₁, and 6 double bonds, 2 less than the natural product. Most importantly, by replacing the central cis-alkenes at C9-C10/C9′-C10′ of disorazole C₁ with cis-cyclopropane rings, 1 no longer has a labile conjugated (E,Z,Z)-triene; yet, it shows nearly the same potent level of low nanomolar in vitro cytotoxicity in human colon cancer cell lines as the natural myxobacterium metabolite. Further biological studies of 1 will be reported in due course.

Footnotes