Kevin R. Munro, Louise Male, Neil Spencer and Richard S. Grainger*
School of Chemistry, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK. E-mail: r.s.grainger@bham.ac.uk; Fax: +44 (0)121 4144403; Tel: +44 (0)121 4144465
First published on 30th August 2013
α-Hydroxyalkylidene carbenes, generated from thermolysis of α,β-epoxy-N-aziridinylimines, undergo diastereotopic group selective 1,5 C–H insertion reactions on 2,4-dimethyl-8-oxabicyclo[3.2.1]oct-6-ene ring systems. Protection of a tertiary alcohol at C-3 of the bridged oxabicycle as a trimethylsilyl ether reverses the sense of diastereoselectivity. 1,5 C–H insertion into a methine adjacent to an OBn group, 1,5 O–R insertion into a tertiary alcohol (R = H) or silylether (R = TMS) at C-3 to form spirocyclic dihydrofurans, 1,2-rearrangement to an alkyne and fragmentation to a ketone are competing major pathways for 2-benzyloxy-substituted 8-oxabicyclo[3.2.1]oct-6-ene systems. Dihydrofuran formation is shown to be a result of substitution on the oxabicyclic ring system through comparison with other methods of alkylidene carbene formation.
We are interested in the use of this reaction for the preparation of the A-ring of ingenol, a structurally complex diterpene of considerable synthetic and biological interest.3 Our previous model studies established that 1,5 C–H insertion to give cyclopentenes 3 and 6 overrides the potentially competing 1,5 O–Si insertion pathway on structurally rigid 8-oxabicyclo[3.2.1]oct-6-ene ring systems 1 and 4 (Scheme 1).4,5 However, these studies did not address the presence of an additional hydroxyl group within the cyclopentene A-ring of the natural product. Keen to avoid the need to introduce the required oxygenation at C-3 (ingenol numbering) post-annulation, we were attracted to the methodology of Kim for the generation of α-hydroxyalkylidene carbenes 8 through thermolysis of α,β-epoxy-N-aziridinylimines 7 (Scheme 2).6 Depending upon substituents R1–R3, alkylidene carbenes 8 can undergo a number of different reaction pathways,7,8 with 1,5 C–H insertion leading to hydroxy-substituted cyclopentenes.6
![]() | ||
Scheme 1 Previous chemo- and regioselective 1,5 C–H insertion of alkylidene carbenes on 8-oxabicyclo[3.2.1]oct-6-ene ring systems. |
![]() | ||
Scheme 2 Formation of α-hydroxyalkylidene carbenes 8 from α,β-epoxy-N-aziridinylimines 7. |
In this paper we report our studies on the use of the Kim α-hydroxyalkylidene carbene for hydroxycyclopentene annulation on two model 8-oxabicyclo[3.2.1]oct-6-ene ring systems.9 In addition to establishing the compatibility of this methodology on systems bearing oxygen functionality at C-3 of the oxabicycle (a potential site of reactivity),5,7 these studies were designed to probe two additional issues. The directing effect of the hydroxyl group in 8, initially masked in the form of an epoxide 7, is investigated through a formal diastereotopic group selective 1,5 C–H insertion reaction.10,11 The influence of additional oxygenation, of relevance to the 5-hydroxyl group in ingenol, is investigated on C-2 benzyloxy-substituted oxabicycles. The formation of products arising from 1,5 O–Si and O–H insertion reactions on these latter systems are further probed using alternative methods of alkylidene carbene formation.
![]() | ||
Scheme 3 Synthesis of alkylidene carbene precursors 16a and 16b. |
![]() | ||
Scheme 4 Diastereotopic group selective 1,5 C–H insertion reactions of α-hydroxyalkylidene carbenes 17a and 21a. |
Deprotection of the trimethylsilylether in 16a with TBAF gave the tertiary alcohol 20a in excellent yield (Scheme 4). The tertiary alcohol 20a also gave rise to a mixture of cyclopentenols 22 and 23 upon heating to reflux in toluene, in lower overall yield (42%) and with the reversed sense of diastereoselectivity (1:
2 ratio 22
:
23) compared with the thermolysis of silylether 16a to give cyclopentenols 18 and 19 (64%, 2
:
1 ratio 18
:
19). Cyclopentenols 22 and 23 were isolable by column chromatography, and relative stereochemistry was again determined by nOe correlation in the major diol 23, with the structure of the minor diol 22 further confirmed by X-ray analysis (Fig. 1).‡1H NMR of the crude reaction mixture also showed the presence of dihydrofuran 24, the product of 1,5 O–H insertion, and surprisingly the ketone 9a, the product of C–C bond fragmentation. The ratio of C–H insertion products 22 and 23 to both 24 and 9a was approximately 6
:
1 by integration of characteristic peaks in the 1H NMR of the crude reaction mixture.
![]() | ||
Fig. 1 Crystal structure of hydroxycyclopentene 22 with ellipsoids drawn at the 50% probability level. |
The diastereoselectivity of the 1,5 C–H insertion reaction is clearly dependent on the C-3 substituent of the oxabicycle, and while modest, is notably reversed for silylether 16a and alcohol 20a. The alcohol stereocentre in alkylidene carbenes 17a and 21a renders the two potential bonds for 1,5 C–H insertion, C–Hα and C–Hβ, diastereotopic. The diastereotopic group selectivity20,21 is controlled by the preferred orientation about the HO–C–C–OR carbon–carbon bond within the carbene (Scheme 5). For the TMS ether 17a, insertion into C–Hβ is disfavoured compared to insertion into C–Hα, presumably due to unfavourable steric and/or electronic interactions between the silyl ether and the alcohol. For the tertiary alcohol 21a, preferential insertion into C–Hβ can be rationalized by invoking a hydrogen bonding interaction, which fixes the conformation about the vicinal diol and orients the alkylidene carbene towards C–Hβ.
![]() | ||
Scheme 5 Stereochemical rationale for diastereotopic group selective C–H insertion. |
Attempts were made to increase the diastereoselectivity through functionalisation of the tertiary alcohol in 13a with either a bulkier silyl group or an ester, which might alternatively act as a H-bond acceptor to the developing alcohol. Unfortunately the hindered nature of the tertiary alcohol in 13a made it recalcitrant to reaction with TIPSOTf, TESOTf, AcCl, Cl2CHCOCl or CF3COCl.
Further attempts to exploit an attractive interaction such as that proposed for 21a (Scheme 5) instead promoted the unusual C–C bond fragmentation pathway leading to ketone 9a. Running the thermolysis of 20a in the presence of LiHMDS (to form a metal chelate) or phenylboronic acid (to form a cyclic boronate in situ22) gave 9a in 22% and 52% isolated yield respectively, with no evidence of formation of 1,5 insertion products of an alkylidene carbene (Scheme 6). Alkoxysilylcyclobutane 26, expected to be Lewis acidic at Si through relief of ring strain,23 was prepared in two steps from 13a (yields unoptimized). Again refluxing 26 in toluene resulted in fragmentation to form the ketone 9a. Although the mechanism of this fragmentation has not been elucidated, a common factor may be the possible impedance of the proton transfer step in alkylidene carbene formation (Scheme 2), allowing alternative pathways to dominate.24 This unusual pathway was further demonstrated to be a function of the aziridinylimine – the epoxy aldehyde 25 proved stable to refluxing toluene, and could be recovered in good yield.
![]() | ||
Scheme 6 Fragmentation to ketone 9a. |
![]() | ||
Scheme 7 Regio- and chemoselectivity in alkylidene carbene insertion on 2-benzyloxy-substituted 8-oxabicyclo[3.2.1]oct-6-ene ring systems. |
nOe analysis of 27 showed a correlation between the proton adjacent to the hydroxyl group and one of the benzylic hydrogens, thus establishing the stereochemistry of major epoxide 12b in the synthetic route to 16b (Scheme 3). 1H NMR analysis of the crude reaction mixture showed 27 to be the major product, obtained in an approximately 2.4:
1
:
1.2 ratio along with silyldihydrofuran 28, the product of 1,5 O–Si insertion, and the ketone 9b, the product of C–C bond fragmentation. The structure of 28 within the reaction mixture was assigned on the basis of comparison with related tetrahydrofurans (vide infra), however it could not be separated by column chromatography. The combined yield of 27, 28 and 9b was 79%.
Removal of the TMS group from 16b with TBAF gave the tertiary alcohol 20b, which could not be adequately purified by column chromatography and so was instead taken directly into the thermolysis reaction to form the α-hydroxyalkylidene carbene 21b. Heating a toluene solution of the crude tertiary alcohol 20b at reflux also gave rise to a relatively complex mixture of products, which could, however, be separated by column chromatography. Hydroxycyclopentene 29, arising from 1,5 C–H insertion next to oxygen, and dihydrofuran 30, the product of O–H insertion, were the two major compounds, formed in an approximately 1:
1 ratio. Significant amounts of ketone 9b were again formed, and it was also possible to isolate small quantities of the alkyne 31, the product of 1,2-rearrangement of the intermediate alkylidene carbene.
Alkylidene carbenes show a general preference for insertion into more substituted C–H bonds, tertiary > secondary > primary, a trend which is rationalized by the interaction of the more electron-rich C–H bond with the electron-deficient carbene.25 The preferential 1,5-insertion of alkylidene carbenes into C–H bonds adjacent to heteroatoms is also well established in the literature. A recent comprehensive study by Lee has shown there is a directionality to the activating effect of an oxygen on an adjacent C–H bond. An ether oxygen can activate an adjacent C–H bond towards insertion through lone-pair donation into the σ*C–H antibonding orbital when appropriately aligned.2 For both 17b and 21b 1,5 C–H insertion is completely selective for insertion into the methine group adjacent to the benzylic oxygen over the alternative 1,5 C–H insertion into the methylene group. The electronic and stereoelectronic activation of the tertiary C–H apparently overrides, or acts in concert with, any potential directing effect of the alcohol stereocentre in 17b and 21b.
More surprising is the increased predominance of additional reaction pathways observed for the 2-benzyloxy-substituted oxabicyclo[3.2.1]oct-6-ene ring systems 16b and 20b compared to the 2,4-dimethyl-substituted systems 16a and 20a. Analogous O–Si/O–H insertion, fragmentation to ketone 9a, and 1,2-migration are all pathways available to 16a and 20a, yet only occurred to a minor extent, or not at all. 1,2-Alkyl migration in particular is rarely observed for dialkyl-substituted alkylidene carbenes.10e,26,27 Either 1,5 C–H insertion is slower than might be expected based on the additional stereoelectronic activation by the OBn group compared to the methyl groups in 17a and 21a, or structural features in the benzyloxy-substituted system promote these competing reaction pathways.
We have previously5 ascribed the lack of O–Si insertion in the alkyl-substituted oxabicycles 1 and 4 (Scheme 1) to the preferred geometry about the C–OSiMe3 bond, which orients the lone pairs on oxygen away from the alkylidene carbene (represented as the free carbenes 2 and 5, but better envisioned as metal carbenoids29). Why then should O–Si and O–H insertion be favoured to such an extent for the benzyloxy-substituted systems? While the nature of the alkylidene carbene is different to that shown in Scheme 1, and the higher temperature at which it is being generated should increase bond rotation and may allow these alternative pathways to operate, this does not explain the lack of O–Si insertion in 16a, and the relatively small amount of O–H insertion observed for 20a. In order to separate any potential role of the developing hydroxyl group in directing the product outcome, e.g. through hydrogen bonding with the benzylic oxygen which may orientate the carbene away from the C–H insertion site, we have extended our prior studies in systems lacking this feature. In this way we hoped to determine the relative importance of the structure of the oxabicyclic vs. the nature of the carbene generation method on the product outcome.30 Two classical methods of carbene generation were chosen, the base-mediated α-elimination of a chloroalkene, and the reaction of a ketone with lithiated trimethylsilyldiazomethane, which generates the carbene through sequential Peterson olefination and loss of nitrogen.2,31
β-Silyloxyketone 34 was prepared in three steps from 2,4-dimethyl-8-oxabicyclo[3.2.1]oct-6-ene (9a) as previously described (Scheme 8).5 Treatment of ketone 34 with lithium (trimethylsilyl)diazomethane gave cyclopentene 3, the same product of C–H insertion we observed when generating the carbene from deprotonation of chloroalkene 1 (Scheme 1). Hence all three carbene generation methods on the 2,4-dimethyl-3-trimethylsilyloxy-substituted oxabicyclic systems give products of 1,5 C–H insertion, with no evidence of significant dihydrofuran formation through 1,5 O–Si insertion.
![]() | ||
Scheme 8 Alternative methods of alkylidene carbene generation. |
For the 2-benzyloxy-substituted system we employed a route previously developed for the synthesis of chloroalkene 4 (Scheme 1).5 Addition of the dianion of methylacetoacetate to ketone 9b gave β-ketoester 35 in good yield. Protection of the tertiary alcohol as a trimethylsilylether followed by Krapcho decarboxylation gave the β-silyloxyketone 36. Subsequent Wittig reaction gave chloroalkene 37 as a 1:
1 mixture of double bond isomers (Scheme 8).
The alkylidene carbenes generated from both 36 and 37 gave a mixture of compounds arising from both 1,5 C–H insertion and 1,5 O–Si insertion, which were separable after treatment of the crude reaction mixture with TBAF to liberate the tertiary alcohol 39. The ratio of 38:
39 changes with the method of carbene generation. Using lithium (trimethylsilyl)diazomethane O–Si insertion to give 38 is the dominant reaction pathway, whereas cyclopentene 39 is slightly favoured through deprotonation of chloroalkene 37. Although there is a difference in product ratio with the method of generation of the carbene,30 these results, and those using the Kim alkylidene carbene precursor, show a clear overall trend. The presence of a benzyloxy-substituent at C-2 of the oxabicyclic ring system controls the regioselectivity of 1,5 C–H insertion, but also leads to significant quantities of 1,5 O–Si insertion, irrespective of the method of carbene generation. In neither case was fragmentation to ketone 9b nor 1,2-rearrangement to an alkyne observed, suggesting these pathways can be ascribed to the nature of the α-hydroxyalkylidene carbene, or the conditions under which it is generated.
Footnotes |
† Electronic supplementary information (ESI) available: Experimental procedures and copies of 1H and 13C NMR for all new compounds. For ESI see DOI: 10.1039/c3ob41390j |
‡ 22: C13H18O3, M = 222.27, orthorhombic, a = 13.4833(5), b = 11.3687(5), c = 14.6074(8) Å, U = 2239.13(18) Å3, T = 120(2) K, space group Pbca, Z = 8, 24![]() |
This journal is © The Royal Society of Chemistry 2013 |