Influence of solvents and catalysts on the formation and hydrolysis of polyfunctional enoxysilanes derived from aucubin

Aucubin, a natural iridoid widely extracted from Aucuba japonica presents a strong interest as semisynthetic raw substantial, thanks to its numerous features. In this study, we describe an unprecedented reaction between trimethylsilyldiazomethane and the aldehyde-derived aucubin ( 3a ) leading to different original non-natural skeletons exemplified by the tricyclic compounds 6 and 9 (Figure 1). The influence of catalysts, solvents and temperature on the formation of the products together with the hydrolysis of enoxysilane 4 will be presented

Herein, in continuation of our work concerning the reaction between iridoids skeletons and TMSDM, we describe the influence of various process parameters as solvent, catalyst or temperature on the nature of the products.The best conditions to form original non-natural heterotricyclic skeletons 6 and 9 (Figure 1) obtained from perpivaloyltarennoside (3a) derived from aucubin are reported.

Figure 1.
List of the compounds obtained from the reaction between 3a and TMSDM.
Previously, we published that the reaction of 3a, obtained by Vilsmeier procedure from perpivaloylaucubin 2 (Scheme 1), 23,25 with TMSDM in presence of TMSOTf and molecular sieves in dichloromethane at -40°C led to a quantitative homologation of the pyran ring to oxepin forming the enoxysilane 4 (Table 1, Entry 1).The impact of process parameters has been showed and the role of molecular sieves has been described.Indeed, the lack of molecular sieves drove a significant decrease of the enoxysilane yield along with the formation of the minor product 5, resulting in the homologation on C-11 of the starting material. 16ith the goal to demonstrate that the experimental conditions control the products of the reaction, we studied at first the influence of the solvent polarity.As shown in Table 1, the use of non-polar solvents without sieves did not significantly affect the yield of the formation of 4 and 5 (Table 1, Entries 2-5).In a protic solvent as MeOH, 3a does not react (Table 1, Entry 6).Surprisingly, in a more polar solvent such as acetonitrile, the reaction does not stop to the formation of 4, because an original tricyclic glucoside 6 was isolated with an acceptable yield (60%) (Table 1, Entry 7).With this solvent, the use of sieves does not seem to have a clear effect on the formation of the tricyclic compound (Table 1, Entry 8).Regarding the 1 H NMR spectrum, the structure of 6 was deduced from the loss of the pivaloyl group in position 6.This allegation has been confirmed by the molecular ion at m/z [M + Na] + 813.4013 in the HR-ESI-MS.The important correlation on the HMBC spectrum between H-5' and C-6 justified the formation of the dihydrofuran ring 5-5'-O-6-5a.Furthermore, the broad doublet at 5.50 ppm is in good agreement with a proton H-6 up to the plan and characteristic of the configuration.On the NOESY spectrum, examination of correlations between H-6 and H-5a, and between H-6 and H-8a ensured this affirmation.The second part of the study concerned the influence of the catalyst.As described before, the reaction of 3a with TMSDM in presence of TMSOTf has led specifically to the enoxysilane 4 having the oxepin ring with high yield compared to the experience with AlCl3 (Table 2, Entry 1). 16Using a catalytic amount of a solution of Yb(OTf)3 in THF, the stereoselectivity of the formation of enoxysilanes has been lost because 4 and 8 were formed with equal ratio (Table 2, Entry 2).Enoxysilanes 4 and 8 presented a structural analogy.They differed by the stereochemistry of the double bond C-5-C-5' which was deduced thanks to NOESY spectrum.For 4, a H-5'-H-4 correlation was observed while that for 8, a H-5'-H-5a correlation was displayed.It must be noted that with those two latter catalysts, no homologation of the formyl group at position C-11 and no formation of tricycle 6 have been observed.Various catalysts which were defined previously to react with TMSDM and carbonyls, 26 have also been tested.No reaction occurred with AlMe3 and MgBr2 while the use of BF3.Et2O generated the homologated methylketone 7.
To explain the formation of 5, a proton source is required.The hypothesis that TMSOTf is able to generate triflic acid (TfOH) without molecular sieves was formulated.So, we sought to optimize the process to obtain 5 by using a Brønsted acid as catalyst.With Tf2NH, a mixture of enoxysilane 4 and aldehyde 5 has been obtained in a poor yield (Table 2, Entry 6).Trifluoroacetic acid failed to react with 3a (Table 2, Entry 7) while the acidity of TfOH has been proved efficient in a stoichiometric amount.These latter conditions lead to the unique aldehyde 5 in 55% of yield (Table 2, Entry 8).In order to have a better understanding of the mechanism of formations of 5 and 7, the deuteriated aldehyde 3b 16 has been engaged in similar processes (Scheme 2).We first demonstrated that ring-enlargement products exemplified by enoxysilane 4a proceeded by the 1,4-addition of TMSDM (way a).In accordance to the Table 2 (Entries 6 and 8), the protonic catalysis to form 5a suggests an activation of the aldehyde followed by the exclusive attack of TMSDM on the carbonyl group.Then the hydride migrates to furnish ketosilane 5b which was hydrolyzed to 5a (way b).Otherwise, we found that the use of BF3.Et2O as catalyst promotes the methylketone 7a formation.In that event, TMSDM reacts on the carbonyl giving a double homologation.Although the first steps of formation of compounds 5a and 7a are similar, for 7a, BF3.Et2O also participated to the hydrolysis of the trimethylsilyl group. 29Once the intermediate 5a formed, TMSDM reacts again with the carbonyl activated by BF3.Et2O (way c).Thereby, to have a total reaction of the starting material and optimize the results, two equivalents of TMSDM were required in this case.The formation of tricyclic product was also thoroughly studied.The preferred configuration of enoxysilane observed for 4 suggests that 6 was obtained from a rearrangement of this latter.In order to support this hypothesis, we carried out the acid catalysis of 4 with TMSOTf, Tf2NH, AlCl3, Yb(OTf)3 and BF3.Et2O.Regarding the results presented in Table 3, we observed an undeniable influence of the nature of the catalyst on the formation of tricyclic compound.Thus, using 10% of TMSOTf in dichloromethane at -78°C the original tricyclic compound 6 was obtained quantitatively in an instant way (Table 3, Entry 1).With Tf2NH, 20% of catalyst has been necessary to have a total reaction of 4 and the tricycle 6 has been formed in a lower yield (Table 3, Entry 2).AlCl3 has not allowed the rearrangement of the enoxysilane even at room temperature by using a stoichiometric quantity of catalyst (Table 3, Entry 3). 4 did not react at low temperature with Yb(OTf)3 and BF3.Et2O.Nevertheless, at room temperature, an unexpected rearrangement occurred (Table 3, Entries 4 and 5) and an original heterotricyclic heteroside aldehyde 9 produced.The nature of this skeleton has been established from several NMR experiments.In addition to the loss of one pivaloyl group, a quaternary carbon at 63.4 ppm was attributed to C-8 by a 3 J-HMBC correlation with H-1.The presence of the double bond Δ6-7 was shown firstly by the 13 C NMR chemical shifts at δ 145.1 and 130.2 ppm together with COSY correlation between H-7 at δ 5.66 ppm and H-6 at δ 6.51 ppm.Moreover HMBC correlations H-7/C-8', H-7/C-5a, H-6/C-8 and H-6/C-5a confirmed the insaturation's position.These data considered, the tricyclic skeleton was deduced on one hand from the HMBC correlations between the aldehydic signal at δ 9.41 ppm and the ethylenic carbons C-4, C-5 and also the carbon sp 3 C-5a and on other hand by the 3 J correlations H-3/C-8a and H-8a/C-3.The final part of this study concerned the hydrolysis of enoxysilane 4 by cleavage of the trimethylsilyl group following three different methods that are TBAF in THF, a hydromethanolic mixture or a combination of acetic acid, THF, water.Whatever the conditions used, an unselective oxydative addition at C-3 or C-5 has furnished, in low yield, two original compounds 10 and 11 possessing an aldehyde function in C-5.Although the mechanism of formation of these latter is still obscure, it could be also explained by the presence of dioxygen during the reaction which has not been carried out under inert atmosphere.Aldehyde 5. To a solution of 300 mg of aldehyde 3a (0.34 mmol) in dichloromethane (10 mL) at -40 °C was added 30 µL of TfOH (0.34 mmol) dissolved in dichloromethane (C = 0.1 M). 171 µL of trimethylsilyldiazomethane has then added.The reaction was stirred for 2h.After the complete conversion of the aldehyde the mixture was warmed to rt, washed with 2 x 10 mL of distillated iced water.Then the organic phase was dried on MgSO4, filtered and the solvent removed under reduced pressure.The crude product was purified by column chromatography over silica gel (20-45 μM, cyclohexane/EtOAc: 98/2) to afford corresponding aldehyde 5 as white amorphous solid (yield: 55%).Spectral data of 5 were consistent with the literature. 16ricycle 6.To a solution of enoxysilane 4 (100 mg, 0.11 mmol) in 5 mL of dichloromethane at -78°C, 2 µL of TMSOTf dissolved in dichloromethane (0.011 mmol) were added dropwise.After 1 min, the catalyst was neutralized with 1.6 µL of Et3N (0.011 mmol) dissolved in dichloromethane.After at room temperature, 3 mL of pentane was introduced to precipitate the formed salts.The 1h30 stirring solution was then filtered and the solvent was removed under reduced pressure.The residue was purified by column chromatography over silica gel (20-45 μM, cyclohexane/EtOAc: 9/1) to afford tricycle 6 as white powder, yield 98%, 85 mg, [α]D 20  The reaction was stirred for 2h.After the complete conversion of the aldehyde the mixture was warmed to rt, washed with 2 x 10 mL of distillated iced water.Then the organic phase was dried on MgSO4, filtered and the solvent removed under reduced pressure.The crude product was purified by column chromatography over silica gel (20-45 μM, cyclohexane/EtOAc: 98/2) to afford corresponding methylketone 7 as an amorphous white solid, yield 42%, 165.6 mg, [α]D 20 -60.4 (c 0.065, CH2Cl2); IR (νmax, cm -1 ): 1666 (C=O). 1 H NMR (300 MHz, CDCl3): δH 6.22 (br.d, 1H, 4 J3-5 1.5 Hz, H3), 5.69 (m, 1H, H7), 5.4 (br.ddd, 1H, 5 J6-10 2 Hz, 3 J6-7 4 Hz, 3 J6-5 6 Hz, H6), 5.35 (t, 1H, 3 J3'-4' = 3 J3'-2' 9.5 Hz, H3'), 5.14 (t, 1H, 3 J4'-5' = 3 J4'-3' 9.5 Hz, H4'), 5.06 (dd, 1H,

Scheme 2 .
Scheme 2. Proposed mechanism of ring expansion and homologation of 3b.

Table 2 .
Influence of catalysts for the reaction of 3a with TMSDM