Initial Efforts to Synthesize Gephyrotoxin

Model studies helped us to develop the basic Schmidt reaction for the synthesis of the core of gephyrotoxin. An efficient assembly of the core was developed, and the issues of regioselectivity of the rearrangement and the stereoselectivity of the iminium ion reduction were explored. For the "real system", two new issues arise (Scheme 4). First, if we use the same indene alkylation approach that we developed in the model studies, methoxyindene (6) is needed as the starting material, and the regioselectivity of its alkylation becomes important. Second, we must choose a group R, a synthetic equivalent of a 2-hydroxyethyl group, that will survive the strongly acidic conditions of the Schmidt reaction. Our initial efforts to address these issues are outlined here.

While the routes described in this section were not ultimately useful for a formal synthesis of gephyrotoxin, some interesting chemistry was uncovered. But, if you want to skip ahead to the successful route, click here.

Scheme 4

C. Alkylation of Methoxyindene and Schmidt Reactions of Regioisomeric Azidoindenes

The desired methoxyindene 20 was prepared from commercially available 4-methoxy-1-indanone (19) by sodium borohydride reduction and dehydration according to Grieco's selenoxide method (Scheme 5).[Ref.21] Metalation of 20 with n-butyllithium followed by alkylation with 21, easily prepared from commercially available 4-bromotetrahydropyran (24),[Refs 22,23] followed by azide displacement produced a 1:3 ratio of the desired indene 22 and the undesired regioisomeric indene 23. Chelation of lithium to the methoxy group may explain this regioselectivity, and has been observed in other alkoxyindene metalation/alkylations.[Refs 19,24] Attempts to inhibit this internal chelation and thus alter the regioselectivity of the alkylation [e.g., KN(TMS); NaH; silyloxy rather than methoxy] were unsuccessful. Kelly's method for regioselective indene alkylation (t-butyllithium/TMSCl) could be used to make the undesired indene 23 selectively.[Ref.19]

Scheme 5
. Attempt at Regioselective Alkylation of Methoxyindene 20.

Pure 23, the "wrong" regioisomer, was subjected to the Schmidt reaction (Scheme 6), producing 26 in moderate yield as a single regioisomer (i.e., no benzylic amine was observed). A sample of pure 22 could not be obtained by chromoatography, so a mixture of 22 and 23 was subjected to the Schmidt reaction. Compound 26 (derived from 23) was the only product; no 28 was detected, perhaps indicating that 22 does not undergo the Schmidt reaction. An alternate synthesis of a structure related to 22, the "correct" regiosomer, was developed to verify this conclusion.

Scheme 6. Schmidt Reactions of Regioisomeric Azido-Alkenes 22 and 23.

D. Regiocontrolled Routes to the Cyclization Precusors: Avoid Indene Alkylations

The related azido-alkene 32 was synthesized as shown in Scheme 7. Addition of dibromolithiomethane[Ref.25] to the indanone 19 followed by treatment of the resulting adduct with zinc dust[Ref.26] gave the vinyl bromide 29. Metal halogen exchange on 29 followed by Lewis acid mediated opening[Ref.27] of the epoxide 30[Ref.28] provided the alcohol 31 in good yield. Replacement of the alcohol with azide was accomplished with a modified Mitsunobu reaction.[Ref.29] When the azido-alkene 32 was subjected to our usual Schmidt reaction/iminium ion reduction sequence, no identifiable products were obtained, consistent with the observations in Scheme 6. Considering the data in Schemes 5 and 6 and the model studies already described (i.e., no methoxy on the ring), we conclude that the electron density in aromatic ring is an important influence in the success/regioselectivty of the Schmidt migration. Regarding rate, we would expect the most electron rich aromatic ring to migrate fastest, i.e. o-methoxyaryl (23/25, Scheme 6) > no methoxy (8/9, Scheme 2; 16, Scheme 3) > m-methoxyaryl (22/27, Scheme 6 and 32/33, Scheme 7), which might explain why there is no cyclization for 22 and 32. Regarding regioselectivity, the more electron rich aromatic ring will be better able to compete effectively with alkyl migration. Thus we expect the ratio of aryl:alkyl migration to decrease in the order given above. Indeed, the o-methoxyaryl case is completely aryl selective, while the non-methoxy cases gave mostly aryl migration (although the ratio was not high). The lack of success in the m-methoxyaryl case does not allow us to discuss this data point.

Scheme 7. Schmidt Reaction of Azido-Alkene 32.

We next proposed that a side-chain lacking an oxygen atom might lead to a successful Schmidt reaction in the desired methoxyaryl case, since the azide would be slightly more nucleophilic. Thus, we chose to examine an allyl group, since this group might be transformed later into the desired 2-(hydroxy)ethyl group of gephyrotoxin. The synthesis and cyclization of such a compound, 36, is shown in Scheme 8. A cerium-mediated[Ref.30] addition of allylmagnesium chloride to methoxyindanone 19 gave an alcohol, which was to furnish a diene (not shown). Selective hydroboration of the terminal alkene with 9-BBN[Ref.31] followed by oxidation, provided the alcohol 34 in excellent overall yield. A one-pot Swern oxidantion/organometallic addition using Ireland's method[Ref.32] gave 35, which was transformed into the desired azide 36 with standard chemistry. Schmidt reaction of 36 followed by hydride reduction of the resultant iminium ion proceeded in excellent yield (86%) to produce a 1:1 mixture of regioisomeric products, 37 (desired) and 38 (undesired). Hence, the m-methoxyaryl group can migrate in good yield if a more nucleophilic azide is used (compare with Scheme 6 and Scheme 7), although the regioselectivity is poor, as predicted (vide supra). Nonetheless, oxidative cleavage of the alkene of 37 and reduction of the resultant aldehyde would complete a formal synthesis of gephyrotoxin by intercepting Ito's compound 2 (Scheme 1). However, all attempts at oxidative transformations on 37 were unsuccessful, leading to decomposition, perhaps due to the highly electron rich aromatic ring.

Scheme 8. Synthesis and Schmidt Reaction of Azido-Alkene 36.

Our search for the proper 2-hydroxyethyl equivalent had thus led us to uncover several important aspects of the Schmidt reaction required for the synthesis of gephyrotoxin, but each attempt had a flaw. However, we had now learned enough to narrow our search to a 2-haloethyl group, which would survive the acidic conditions of the Schmidt reaction, but would be easily transformed into a 2-hydroxyethyl group without relying on an oxidative reaction. A successful formal synthesis of gephyrotoxin is outlined in the next section.


Links to the Rest of the Paper:

1. Introduction
2. Model Studies: The Basic Schmidt Reaction, wherein we study the regioselectivity of the rearrangement and the stereoselectivity of the iminium ion reduction.

A. Regioselectivity of the Schmidt Reaction: Assembly of the Basic Core of Gephyrotoxin.
B. The Use of a Secondary Azide: Stereoselectivity of the Iminium Ion Reduction.

3. Initial Efforts to Synthesize Gephyrotoxin (THIS PAGE), wherein we struggle with the regioselectivity of the indene alkylation and the choice of a synthetic equivalent for the 2-hydroxyethyl side chain.

C. Alkylation of Methoxyindene and Schmidt Reactions of Regioisomeric Azidoindenes
D. Regiocontrolled Routes to the Cyclization Precusors: Avoid Indene Alkylations

4. Formal Synthesis of Gephyrotoxin, wherein we finally get it right.
5. References