An unexpected Prins desymmetrisation reaction driven by silyl migration

Prins desymmetrisation reactions of cyclohexa-1,4-diene derivatives have been investigated as a route to the core of the cladiellin diterpenes. During the course of this work, we observed the formation of a partially-reduced benzofuran 18 , which is clearly derived from oxocarbenium ion 21 . This can only be rationalised by an unexpected primary to secondary silyl group migration.


Introduction
We have recently reported model studies for the Prins desymmetrisation of cyclohexa-1,4-dienes to give systems related to the cladiellin diterpenes. 1 In our initial work, compound 1 underwent a rather dramatic transformation into compound 2 by way of formation and reaction of an oxocarbenium ion followed by rearrangement.In this way, the core functionality and stereochemistry of the cladiellin diterpenes, for example 7-deacetoxyalcyonin acetate (3) 2 was rapidly established (Scheme 1).

Scheme 1
This reaction involves the double-deprotection of compound 1 to give intermediate 4, followed by formation of oxocarbenium ion 5, cyclisation to give 6 which rearranges to give 7 and deprotonation/tautomerisation to give the product 2 (Scheme 2).

Scheme 2
In seeking to extend these studies, we elected to fuse an additional ring onto the precursor as shown in Scheme 3.While the fusion of an aromatic ring is appropriate for such a study, we envisage that eventually this will be replaced with a system that can be cleaved to give the complete cladiellin core.It was therefore envisaged that the Prins desymmetrisation of compound 8 would give rise to compound 9.

Results and Discussion
It was envisaged that the key precursor 12 would be accessible by formation of an organolithium compound from 11 and then reaction with the epoxide 10 (Scheme 4).

Scheme 4
Compound 11 was prepared according to a literature method. 3Compound 10 was prepared as shown in Scheme 5. Birch reduction of benzoic acid followed by esterification gave compound 13.Deprotonation and acylation with methyl chloroformate was followed by lithium aluminium hydride reduction to give diol 15.Mono-silylation and Swern oxidation then gave aldehyde 17.Addition of bromomethyllithium to the aldehyde was rather troublesome.With an excess of bromomethyllithium complex mixtures of products were obtained.Therefore it was better to use only a slight excess.Under these conditions the reaction did not proceed to completion, but the unreacted aldehyde was readily removed by treatment of the crude reaction mixture with sodium borohydride prior to chromatography.This gave the desired epoxide 10 in satisfactory yield.
Lithiation of compound 11 with t-BuLi for 10 minutes at -78 °C prior to addition of epoxide 10 initially appeared to have been successful.Purification gave a product in which the acetal had been retained along with a 1,2-disubstituted benzene ring, and that the epoxide had been opened.This was therefore subjected to the conditions of the Prins desymmetrisation (Scheme 6).It rapidly became clear that the product of this two-step process was the partiallyreduced benzofuran 18 rather than the desired product.The structural assignment of compound Scheme 5 18 was by no means straightforward.Extensive analysis of 1 H and 13 C NMR spectroscopic data, and 1 H-1 H and 1 H-13 C correlation data (COSY, HMBC and HSQC) enabled determination of the carbon-hydrogen framework connectivity.Mass spectrometric studies identified the presence of the two bromine atoms in the compound 18.The stereochemistry of compound 18 was assigned by analogy with that of related compounds.1b

Scheme 6
In a previous study, we reported the formation of ketones during Prins desymmetrisation reactions.1b These arise by protonation of an acetal 19 on the more hindered oxygen followed by the Prins reaction, and compounds 20 were invariably the minor products (Scheme 7).In this case compound 14 was the only product formed, which suggests that if an acetal such as 19 is formed, it is opened regioselectively but in the "wrong" direction.This seems rather unlikely.

Scheme 7
However, since the structure of compound 18 is secure, it is clear that it must be formed from oxocarbenium ion 21 (R = H or TBS).Since it is unlikely that opening of an acetal forms this intermediate, the most likely explanation is opening of epoxide 10 by bromide derived from the partial lithiation of compound 11.

Scheme 8
In order to investigate this process further, the coupling of compounds 10 and 11 was repeated.Extensive chromatography led to the isolation of an epoxide-opening product that lacked the aromatic ring (Scheme 9).This was assigned structure 22 or 23, although we could not a priori deduce the location of the silyl group.Examination of the spectra from the previous coupling reaction showed that the same epoxide-opening product was present.As a result of the subsequent formation of compound 18, it seems overwhelmingly likely that the structure is 23 and not 22, so that a silyl migration has taken place during the epoxide-opening.

Scheme 9
Explaining this apparent silyl migration is not straightforward.While there are many examples of silyl groups migrating from secondary to primary alcohols in the literature, 4 we could find no examples of primary to secondary migration, and we assume that such a process would be kinetically disfavoured.However, if we assume that the BF3 coordinates to the epoxide oxygen in order to assist the epoxide-opening, we can consider the intermediacy of a hypervalent silicon compound 26a (Scheme 10).The BF3 should then be readily transferred to the lesshindered oxygen which will give intermediate 27a which should then undergo opening with effective silyl-transfer to give 29a.In fact, this process is very reminiscent of the regioselective opening of the corresponding acetals with Lewis acids. 1

Scheme 10
Density functional theory calculations (Spartan 10, B3LYP 6-31+G*) have provided some insight into this transformation.For simplicity the calculations were carried out with a trimethylsilyl group in place of TBS.Both 26b and 27b minimise at this level of theory with cleavage of a Si-O bond, so that it was not possible to obtain minimum energy structures for these intermediates.However, a single-point DFT calculation based on a molecular mechanics minimised structure in each case indicated that 27b is considerably more stable than 26b (235 kJ mol -1 , although since these numbers are not based on optimised structures, they should be interpreted cautiously) (Figure 1).In both cases the BF3 is axial to avoid severe steric interactions with the trimethylsilyl group.However, upon minimisation, the product 29b is more stable than 28b (by 11.6 kJ mol -1 ).The lowest energy conformers of these two structures are shown in Figure 2. Therefore, we would tentatively attribute the formation of intermediate 29 as being due to rapid migration of the BF3 to the less-hindered oxygen.

Conclusions
The formation of compound 18 was unanticipated on the basis of our previous work.However, it can be rationalised as a result of an unexpected primary to secondary silyl migration as follows.Metal-halogen exchange on compound 11 generates bromide.Presumably the organolithium reagent is also formed, but the fate of this species is unclear.Boron trifluoride promotes the opening of epoxide 10 with bromide to generate intermediate 25a, which rearranges under the influence of the boron trifluoride to give the primary alcohol 23.Oxocarbenium ion 21 is then formed by reaction with 11, and cyclisation of this ion initially gives the secondary carbenium ion 30 before rearrangement to the more stable allylic carbenium ion 31.Loss of a proton will be followed by hydrolysis and tautomerisation of the silyl enol ether 32 to ultimately give the observed product 18.

Experimental Section
General.Melting points were determined on a Gallenkamp melting point apparatus.Infrared spectra were recorded on a Perkin Elmer 1600 FTIR spectrophotometer.Mass spectra were recorded on a Fisons VG Platform II spectrometer and on a Micromass Q-TOF Micro spectrometer.NMR spectra were recorded on a Bruker DPX 400 spectrometer operating at 400 MHz for 1 H and at 100 MHz for 13 C at 25 °C, or on a Bruker Avance 500 spectrometer operating at 500 MHz for 1 H and 125 MHz for 13 C at 25 °C.All chemical shifts are reported in ppm downfield from TMS. Coupling constants (J) are reported in Hz.Multiplicity in 1 H NMR spectroscopy is reported as singlet (s), doublet (d), double doublet (dd), double triplet (dt), double quartet (dq), triplet (t), and multiplet (m).Multiplicity in 13 C NMR spectroscopy was obtained using the DEPT pulse sequence.Flash chromatography was performed using Matrex silica 60 35-70 micron.Solvents for moisture-sensitive reactions were dried by distillation; THF over sodium benzophenone ketal and CH2Cl2 over CaH2.Such reactions were carried out under an atmosphere of nitrogen.