|Home | About | Journals | Submit | Contact Us | Français|
A tandem ring-closing metathesis of a silaketal-based dienyne substrate proceeded efficiently to provide a bicyclic siloxane, which upon removal of the silicon tether afforded an (E,Z)-1,3-dienediol. Further manipulation of this key functional motif rendered synthesis of the entire C1-C19 linear skeleton of (—)-cochleamycin A, a late stage intermediate employed in the previous total synthesis of (+)-cochleamycin A by Roush and coworkers.
Cochleamycin A was isolated in 1992 by a team headed by Shindo and Kawai during a screening programme for antitumor antibiotics from a cultured broth of Streptomyces DT136.1 Subsequently, they reported significant antimicrobial activity against grampositive bacteria and cytotoxicity against P388 leukemiacells (IC50 = 1.6 μg/mL) for cochleamycin A.2 The relative stereochemistry and the 5,6-fused and 10,6-bridged tetracyclic core were later revealed by them from exhaustive NMR studies.3 Not surprisingly, the combination of architectural complexity and favorable biological activity led to a number of impressive initital synthetic studies,4 finally resulting in the first total synthesis and establishment of the absolute configuration of naturally occurring (+)-cochleamycin A by Tatsuta and coworkers.5 Soon after, Roush and Dineen also reported their total synthesis of (+)-cochleamycin A.6 These synthetic studies were based on the proposed biosynthetic pathway of cochleamycins involving a putative transannular Diels-Alder reaction of (E,Z,E)-1,6,8-nonatrienes to construct the cochleamycin A skeleton.7
We envisioned that the prowess of enyne metathesis8 to form 1,3-dienes can be implemented in the synthesis of cochleamycin A if we can effectively control the E/Z stereochemistry of 1,3-diene. Previously, we reported on a tandem dienyne ring-closing metathesis (RCM) of alkynyl silaketals to generate bicyclic siloxanes9 in the presence of Grubbs NHC-ruthenium catalyst.10 Removal of the silicon tether through protodesilylation allowed for the generation of stereochemically defined 1,4-substituted (E,Z)-1,3-dienes. The group-selectivity (ring-closure from right to left vs. that from left to right) in tandem RCM to obtain only the desired regio-isomer would be achieved by differential substitution on the alkene as demonstrated in the construction of the carbon framework of tatrolon B.11 Herein, we report a formal total synthesis of (—)-cochleamycin A by intercepting the Roush’s advanced synthetic intermediate.
As shown in the retrosynthetic analysis in Scheme 1, β-keto ester 2, an advanced intermediate employed in the Roush total synthesis of (+)-cochleamycin A,6 would be derived from dienediol precursor 3. The tandem RCM12 of silaketal 4 followed by desilylation would deliver dienediol 3. Silaketal 4 could be obtained from the base catalyzed stepwise alcoholysis of cyclopentyltrialkynyl silane 6 with alcohols 5 and 7. The key intermediate 5 was prepared by utilizing the anion relay chemistry developed by Smith and coworkers.13 A general route to a precursor of 5 was developed starting with the formylation of the lithium (triethylsilyl)acetylide,14 which afforded the known 3-triethylsilyl propynal15 in 78% yield (Scheme 2). This aldehyde was then subjected to double Michael addition of 1,3-propane dithiol in the presence of basic alumina to give dithiane aldehyde 8 in 85% yield.16 Asymmetric allylation utilizing the pseudoephedrine derived strained silacycle 9 developed by Leighton and coworkers, afforded homoallylic alcohol 10 in 74% yield and 77% enantiomeric excess (ee).17 Many other conventional aymmetric allylation protocols including Brown’s allylation,18 Roush allylation,19 Keck allylation,20 or Loh’s indium mediated allylation21 failed to provide acceptable yields or enatiomeric excess. We suppose, this should be caused by possible chelation of the dithiane moiety to the allylating agents disabling the complexation of the aldehyde. The lithium alkoxide generated from alcohol 10 with butyllithium induced a solvent (HMPA)-controlled 1,4-Brook rearrangement to provide intermediate dithiane anion 11, which was reacted with the diethylacetal bromoacetaldehyde to afford triethylsilyl ether 12.13 Removal of the silyl group with TBAF afforded alcohol 5 in 64% yield over two steps.
The synthesis of another key building block 7 began with the asymmetric (E)-crotylboration22 of trans-crotonaldehyde, providing known allyl alcohol 1323 in 54% yield with 15:1 anti/syn selectivity and 95% ee (Scheme 3). TBS-protection of the hydroxyl group, regioselective hydroboration-oxidation gave the silyl ether 14 in 63% over two steps. Pivalation of the resulting primary hydroxyl group proceeded in 77% yield. This was followed by cleavage of the TBS ether furnishing alcohol 7 in 78% yield.
With all the required building blocks for silaketal 4 in hand, we first silylated alcohol 5 with trialkynylsilane 6 in the presence of a catalytic amount (10 mol %) of NaH in hexanes,24 affording silyl ether 15 in 81% yield (Scheme 4). Reaction proceeded to completion in 15-20 minutes at room temperature with 1 equivalent of 6 and formation of the symmetrical silaketal was not observed. Subsequent coupling of silyl ether 15 with alcohol 7 required higher reaction temperature (60 °C, toluene) to provide desired silaketal 4 in 68% yield as a mixture of diastereomers originating from the newly created stereocenter at the silicon. Formation of a symmetrical silaketal by double addition of 7 was also observed, which could not be suppressed even with 1.5 equivalent of 15. Silaketal 4 was designed such that the ring closure can occur in a group-selective manner25 by initiaing the RCM at the most accessible terminal alkene. Treatment of the bicyclic siloxane RCM product with TBAF in THF under reflux provided dienediol 3 in 61% yield over two steps. Varying amounts of a prematurely terminated monocyclic siloxane, obtained via an RCM from the mono-substituted olefin, was also isolated from the RCM-desilylation reaction sequence. Longer reaction times or higher loading of catalyst did not reduce the formation of this monocylic siloxane.
With dienediol 3 in hand, we proceeded to the synthesis of β-keto ester 2 as shown in Scheme 5. Removal of the dithane using an excess amount of iodomethane in aqueous acetone at 60—65 °C provided ketone 16 in 79% yield.26 The diastereoselective carbonyl reduction of 16, according to Paterson’s procedure27 gave syn-1,3-diol 17 in 76% yield, which was then converted to tri-TBS ether 18 in 73% yield. Reduction of the pivaloate in 17 with DIBAL-H gave alcohol 19 in quantitative yield. Oxidation of primary alcohol 19 by using the Parikh—Doering protocol28 gave the corresponding aldehyde, which was subjected to Horner—Wadsworth—Emmons olefination29 to give ester 20 in 90% yield for the two steps. Again, reduction of ester 20 with DIBAL-H in 85% yield followed by carbonate protection of the resulting allylic alcohol 21 afforded carbonate 22 in 96% yield. Removal of the acetal groups with p-TsOH in acetone gave the corresponding aldehyde, which was converted to the β-keto ester 23 (79% over two steps) by subjecting it to ethyl diazoacetate in the presence of catalytic amount (10 mol %) of SnCl2 according to Roskamp’s procedure.30 Finally, we attempted to remove the carbonate group in the presence of 2% K2CO3 in methanol to obtain the Roush intermediate 2. However, methyl β-keto ester 24 was isolated in 82% yield instead of the corresponding ethyl ester 2, due to a concomitant transesterfication. The subtle difference between methyl and ethyl β-keto esters 2 and 24 is inconsequential for the remaining transformations toward a total synthesis of cochleamycin A. Nevertheless we decided to synthesise ethyl ester 2 to directly compare its spectroscopic data to that reported by Roush. The carbonate deprotection was attempted again using K2CO3 in ethanol. The reaction was much slower than in methanol and thus required longer reaction time or higher temperature. This led to significant decomposition as evident from the 1H NMR of the isolated crude products. Other stronger bases such as sodium ethoxide reduced the overall reaction time, but increased amounts of unidentified byproducts were generated. Trimethyltin hydroxide mediated hydrolysis of methyl ester according to Nicolaou’s procedure31 resulted in deesterification-decarboxylation of the β-keto ester moiety. However, ethyl ester 2 could be isolated in fairly high purity (~90%) by stopping the reaction at low conversion (30-40%) of 24 with K2CO3 in ethanol. The 1H, 13C NMR, and optical rotation of 2 were in good agreement with that reported by Roush, except for the opposite sign in optical rotation.6
In conclusion, we have achieved a formal total synthesis of (—)-cochleamycin A. The key feature of the synthesis is to form a silaketal from two alkenyl alcohols and trialkynyl silane followed by tandem enyne RCM to establish the (E,Z)-1,3-diene moiety required for a Diels-Alder reaction. Further development and application of this silaketal-based enyne RCM will be reported in due course.
We thank the NIH (CA106673) for financial support of this work.