Enantiospecific Total Synthesis of N-Methylwelwitindolinone D Isonitrile

Abstract

We report the enantiospecific total synthesis of N-methylwelwitindolinone D isonitrile. Our route features a double C-H functionalization event involving a keto oxindole substrate to introduce the tetrahydrofuran ring of the natural product.

The welwitindolinone family of natural products (e.g., 1–2, Scheme 1) has attracted tremendous attention from the synthetic community over the past two decades.^[1,2,3,4,5] Interest in these compounds stems from their promising biological profiles, in addition to their compact, yet daunting structures. Synthetic efforts toward the welwitindolinones have led to at least ten methods for building the bicyclo[4.3.1] core that is common to most of these natural products.^[1,4] However, the sheer difficulty associated with late-stage manipulations has plagued most synthetic routes and only a few completed syntheses have been reported in recent years.^[5]

Scheme 1.

Welwitindolinones 1 and 2.

One exceptionally challenging synthetic target is N-methylwelwitindolinone D isonitrile (2).^{[6, 7]} The compound possesses five stereocenters, two quaternary carbons, and a heavily substituted cyclohexyl ring. Compared to other related family members, 2 also possesses an ether linkage between C3 and C14. Thus, a successful synthesis of 2 would not only have to assemble the congested oxindole-fused bicyclo[4.3.1] framework, but would also have to allow for introduction of the ethereal linkage on the sterically congested face of the bicycle. Highlights of synthetic efforts toward 2 include the Wood group’s assembly of the spirocyclic oxindole^[8] and Rawal’s elegant total synthesis of (±)-2 in 2011.^[5a] Herein, we report our synthetic forays toward 2, which culminate in an enantiospecific synthesis.

Our retrosynthetic plan for the synthesis of 2 is presented in Scheme 2. The natural product would be accessed from 3 via late-stage manipulations. In a key disconnection, the tetrahydrofuran ring would be installed from keto-oxindole derivative 4. Of note, the ability to elaborate 4 to 3 would hinge on our ability to perform chemoselective and diastereoselective manipulations adjacent to the two carbonyls. The cyclic carbamate was thought to be accessible using an intramolecular nitrene insertion reaction^[9] involving oxindole substrate 5. Substrate 5 would be derived from ketone 6, which in turn can be readily prepared from known carvone derivative 7^[10] in just four steps using our previously established procedure involving an indolyne cyclization.^[5b,11]

Scheme 2.

Retrosynthetic analysis of 2.

Our approach toward implementing the retrosynthetic plan is highlighted in Scheme 3. Indole 6 was converted to oxindole 8 using a one-pot oxidation/hydrolysis sequence. As the acidic conditions led to desilylation, reprotection of the alcohol was necessary to provide 9. Deuteride reduction and carbamoylation proceeded without event to furnish 5 in quantitative yield. To our delight, exposure of 5 to Ag-promoted nitrene insertion conditions^[12,5e] furnished 10 in 70% yield. It should be noted that attempts to use the proteo analog of 5 gave only 44% yield of the corresponding insertion product, along with 19% of recovered ketone 9. Thus, consistent with our previous findings on an alternate substrate,^[5e] the strategic use of deuterium minimizes an undesirable competitive reaction, thus giving synthetically useful yields of the desired insertion product 10. From 10, a standard deprotection/oxidation sequence delivered key intermediate 4.

Scheme 3.

Elaboration of 6 to keto oxindole 4; TBS=tert-butyldimethylsilyl, NBS=N-bromosuccinimide, DMAP=4-dimethylaminopyridine, DMF=dimethylformamide, THF=tetrahydrofuran, Tf=trifluoromethanesulfonyl, OAc=acetate, bathophenanthroline=4,7-diphenyl-1,10-phenanthroline, Dess–Martin=1,1,1-triacetoxy-1,1-dihydro-1,2-benziodoxol-3(1H)-one.

Many attempts to introduce the tetrahydrofuran ring from 4 were put forth. Unfortunately, efforts toward site-selective functionalization of one carbonyl over the other via enol ethers were unsuccessful. After considerable experimentation, it was found that the keto carbonyl could be α-functionalized first upon treatment of 4 with CuBr₂ in THF at ambient temperature to yield 11 as a single diastereomer (Scheme 4). It was hoped that C3-oxidation would provide an alcohol intermediate that would cyclize to give the necessary tetrahydrofuran ring. However, upon treatment of 11 with C3 oxidation conditions,^[5b] the desired oxidation and cyclization did not occur. Instead, we unexpectedly obtained cyclobutane 13 in high yield, presumably via direct cyclization of the oxindole enolate (see transition structure 12).^[13] X-ray analysis of a single crystal of 13 validated our structural assignment.^[14,7]

Scheme 4.

Unexpected formation of cyclobutane 13; THF = tetrahydrofuran.

As a workaround, we opted to introduce a protected hydroxyl group directly onto C3 of substrate 11. Mn(OAc)₃ was deemed a potential reagent for selective C3-acetoxylation, based on its use in benzylic acetoxylation reactions.^[15] As shown in Table 1, treatment of oxindole 11 with Mn(OAc)₃ in AcOH at 80 °C provided acetoxylated product 14 (entry 1). Interestingly, when the corresponding reaction was conducted at 150 °C, we obtained a 53% yield of 3, which possesses the desired tetrahydrofuran ring. Alternatively, 3 could also be prepared in one-pot by performing the acetoxylation at 80 °C, removing the volatiles, and exposing the crude intermediate to K₂CO₃ in MeOH and H₂O at 70 °C.

Table 1.

Conversion of 11 to acetate 14 and cyclized product 3.

entry	conditions	conversion to products 14 and 3
1	Mn(OAc)3 (4.0 equiv), AcOH, 80 °C	74	0
2	Mn(OAc)3 (4.0 equiv), AcOH, 150 °C	2	53
3	Mn(OAc)3 (4.0 equiv), AcOH, 80 °C; K2CO3, MeOH, H2O, 70 °C	0	56

We also explored the feasibility of directly converting keto oxindole 4 to 3 (Scheme 5). Of note, the Wood group was able to elegantly install a tetrahydrofuran ring from a keto oxindole substrate using basic conditions and O₂.[8] Despite the modest yield, this key precedent laid the groundwork for additional experimentation. To our delight, we found that simple exposure of 4 to tetrabutylammonium fluoride in acetonitrile in the presence of air efficiently delivered 3.^[16] In previous studies, we^[17] and others^[18] have found that TBAF/air can facilitate C3 oxidation of oxindoles containing the welwitindolinone scaffold, but the use of TBAF/air to build an ethereal linkage via double C-H functionalization was unknown. It should be noted that the use of other bases in place of TBAF, such as K₂CO₃ and Cs₂CO₃, also promoted the formation of 3, albeit in lower yields. It is likely that this efficient method for introducing the tetrahydrofuran ring proceeds by initial diastereoselective C3 oxidation, followed by cyclization.^[19] Related C3-peroxy compounds have been observed in our studies^[20] and in Wood’s.^[8]

Scheme 5.

Double C-H functionalization of substrate 4 to install the tetrahydrofuran ring.

To complete the total synthesis, it remained to elaborate the cyclic carbamate to the ketone and isonitrile functional groups present in 2 (Scheme 6). Unexpectedly, attempted hydrolysis of 3 led to cyclohexyl ring fragmentation, a process that was attributed to the reactivity of the ketone. To circumvent this, ketone 3 was reduced to alcohol 15 with LiAlH₄. Fortunately, upon exposure of 15 to hydrolysis conditions, cyclohexyl ring fragmentation was not observed. Hydrolysis gave the desired diol intermediate, which was oxidized with IBX to provide diketone 16. Finally, formylation provided 17, which was directly exposed to standard dehydration conditions to deliver (+)-2.

Scheme 6.

Completion of (+)-2; THF=tetrahydrofuran, dioxane=1,4-dioxane, IBX=2-iodoxybenzoic acid, TFA=trifluoroacetic acid, DMSO=dimethylsulfoxide, Burgess reagent=methyl N-(triethylammoniumsulfonyl)carbamate.

In summary, we have completed the enantiospecific total synthesis of N-methylwelwitindolinone D isonitrile. Several unexpected hurdles, including the formation of the unusual cyclobutane-containing compound 13 were overcome en route to the natural product. Our total synthesis features a double C-H functionalization event of keto oxindole 4 to introduce the tetrahydrofuran ring of 2 and is achieved in 17 steps from readily available carvone derivative 7.