Ring expansion of substituted norbornadienes for the synthesis of mono- and disubstituted 2-azabicyclo[3.2.1]octadienes

Abstract

We have studied the conversion of substituted norbornadienes into a substituted 2-azabicyclo[3.2.1]octadiene system via reaction with toluenesulfonyl azide. We have found that both and mono- and disubstituted norbornadienes will undergo the cycloaddition/rearrangement sequence to provide the bicyclooctadiene ring system as a single regioisomer. The 2-azabicyclo[3.2.1]octane ring system can be prepared from the unsaturated 2-azabicyclo[3.2.1]octadiene ring system.

The 2-azabicyclo[3.2.1]octane ring system 1 (Figure 1) is found in a number of natural products including nordipertenoid alkaloids such as methyllycaconitine^1–3 as well as himandrine.⁴ In addition a number of nonnatural biologically active ring systems contain the 2-azabicyclo[3.2.1]octane ring system.^5,6 The development of new methods for the synthesis of this ring system is important both for natural product synthesis as well as the preparation of simpler analogues of these natural products. An attractive option for the synthesis of substituted derivatives of 1 would be the functionalization of the readily available N-phenylsulfonyl 2-azabicyclo[3.2.1]octadiene 2a. N-phenylsulfonyl 2-azabicyclo[3.2.1]octadiene was first reported in 1965.⁷ This unique compound was prepared by the reaction of phenylsulfonyl azide with norbornadiene (3a). While the yields are only moderate the reaction is readily scalable and easy to carry out. It has been reported that the reaction proceeds through an initial dipolar cycloaddition of the azide followed by loss of nitrogen to form the fused-ring aziridine. This then undergoes a ring opening reaction to a bicyclo[3.1.0]hexene imine. This highly strained bicyclic ring system then undegoes a Cope rearrangement to form the observed 2-azabicyclo[3.2.1]octadiene 2a.⁸

Figure 1.

2-Azabicyclo[3.2.1]octane ring system.

Two options exist for the conversion of 2 into the saturated and substituted ring system 1. The first is the chemical modification of the parent bicyclo[3.2.1]octadiene ring system. Few examples of such modifications have been reported. These modifications are largely limited to the reduction of one or both double bonds,^9–11 dihyroxylation of the 6,7-olefin¹² and cycloaddition reactions with the 6,7-olefin.^13,14

Given the limited methods for the functionalization of 2a, an alternate option for the synthesis of substituted systems such as 1 is the use of substituted norbornadienes 3b in the addition/rearrangement reaction. The product of this reaction, a substituted 2-azabicyclo[3.2.3]octadiene 2b could then be further modified via reduction, dihydroxylation, or cycloadditions to provide highly substituted derivatives of 1. This strategy has several advanatages, including the relative ease with which some substituted norbornadienes can be prepared and the ability to further substitute the products of such an addition/rearrangment process. This would then produce derivatives of 1 with a much wider range of substitution and functionality. A key unknown is the regioselectivity of the cycloaddition/rearrangement process with a substituted norbornadiene.

Given a unique substituted norbornadiene 4 (Figure 2), the reaction with a sulfonyl azide could produce four possible products when R¹ ≠ R² (or two when R¹ = R²) Reaction on the opposite side of the substitution would provide regioisomeric products 5 and 6, while reaction on the same side would provide regioisomers 7 and 8. One report has briefly examined this reaction.¹⁵ Treatment of a compound where R¹ = R² = CN or R¹ = R² = CO₂Me provide largely a single product 5, where the addition of the phenylsulfonyl azide occurred distal to the substitution. We wished to determine if other functional groups, where R¹ = R², would provide the same level of regiocontrol. More interestingly, would be an examination of the reaction of 4 where R¹ ≠ R². Such compounds would likely provide only 4 and 5 but a priori it is not obvious which of these two regioisomers will be formed.

Figure 2.

Possible products derived from the reaction of a substituted norbornadiene with a sulfonyl azide.

In order to begin to address these questions several monosubstituted norbornadienes were prepared as outlined in Scheme 1. The Diels Alder cycloaddition of ethyl propiolate with cyclopentadiene provided monosubstituted norbornadiene 4a in 83% yield.¹⁶ While the Diels Alder method provides an excellent route for the synthesis of norbornadienes with electron withdrawing substituents, an alternate approach was needed for the synthesis of non-electron withdrawing substituted norbornadienes. Alcohol 4b can be readily prepared via metallation and reaction of norbornadiene with formaldehyde.¹⁷ The resulting alcohol was acylated or silylated to provide monosubstituted norbornadienes 4c,¹⁸ 4d, and 4e.

Scheme 1.

(a) CH₂Cl₂, reflux, 18 h, 83%. (b) CH₂Cl₂, AcCl, pyridine, DMAP, 98%. (c) CH₂Cl₂, Me₂tBuSiCl, Et₃N, DMAP, 98%. (d) CH₂Cl₂, pivaloyl chloride, Et₃N, DMAP, 98%.

With a group of monosubstituted norbornadienes in hand, their reaction with tosyl azide was investigated. Based on previous studies,¹⁵ it was expected that addition would occur distal to the substitution. In concert with the original reaction conditions and in contrast to the report by Umano and co-workers,¹⁵ all reactions were carried out at room temperature.¹⁹

The reaction of 4a with tosyl azide resulted in only a 6% yield of an 85:15 mixture of 5a and 6a were obtained. This reaction provided multiple products. The reaction of the acetate provided a significantly better yield of the product in a 94:6 mixture of 5c:6c. Changing the ester to a silyl ether (4d) lowered the yield somewhat (50% to 30%) but dramatically improved the product ratio yielding only the 7-substituted product 5d. Suspecting that this moderate change in regioisomer ratio might be due to the size of the group on the oxygen, the pivalate ester 4e was prepared. The yield returned to the moderate yield observed with the acetate and retained the improved ratio of regioisomers. Clearly the substitution of these monosubstituted norbornadienes is of paramount importance. Electron withdrawing substitution provides little to no product while the substituted hydroxymethyl derivatives provided a moderate yield of a single regioisomer.

In order to investigate regioselectivity in the reaction of disubstituted norbornadienes, a series of disubstituted norbornadienes were prepared via the Diels Alder reaction of cyclopentadiene with substituted alkynes (Scheme 2). The reaction of diethyl acetylene dicarboxylate with cyclopentadiene provided norbornadiene 4f in 97% yield. Reaction of the diacetate of butyne diol with cyclopentadiene required microwave heating of the reaction to 220 °C in DMF and provided 4g²⁰ in 50% yield.

Scheme 2.

4f, R¹ = COOEt, R² = COOEt, toluene, reflux, 18 h, 97%. 4g, R¹ = CH₂OAc, R² = CH₂OAc, DMF, microwave, 220 °C, 2 h, 50%. 4h, R¹ = Ph, R² = COOMe, toluene, sealed tube, 160 °C, 42 h, 99%. 4i, R¹ = nC₅H₁₁, R² = COOEt, toluene, sealed tube, 160 °C, 60 h, 56%*. 4j, R¹ = CH₂OPh, R² = COOMe, toluene, sealed tube, 160 °C, 60 h, 56%*.

In order to prepare norbornadienes where R¹ ≠ R², cyclopentadiene was heated with several substituted propiolate esters in a sealed tube at 160 °C. Norbornadiene 4h²¹ was obtained in an almost quantitative yield. Compounds 4i and 4j were prepared in moderate yield, however they could not be completely purified due to contamination with ~40% of an unidentified non-polar impurity that could not be readily removed.

The ring expansion of symmetrically disubstituted norbornadiene 4f was examined first. Following the reaction conditions that worked well for norbornadiene itself as well as the monosubstituted norbornadienes provided only a 4:1 mixture of aziridine 10 and rearrangement product 5f. Heating the reaction to 80 °C for 18 hours provides the same mixture of aziridine and rearrangement product. Increasing the heat to 110 °C (refluxing toluene) shows a steadily increasing amount of the rearrangement product. After 3 days in refluxing toluene only the rearrangement product was observed.

With a viable procedure in hand a range of disubstituted norbornadienes were examined. The diester 4f provided 5f in 94% isolated yield. The yield obtained in refluxing toluene is similar to that obtained by Umano et al in refluxing 1,2-dichlorobenzene.¹⁵ However the use of toluene provides a somewhat more convenient procedure than using 1,2-dichlorobenzene as the solvent. The diacetate 4g provided 5g in 50% yield.

Several nonsymmetrical derivatives were next examined. Our prediction was that the larger of the two groups R¹ or R² would end up on the same side of the bicyclic ring as the N-Ts group. A single product, 5h, was obtained upon treatment of 4h with tosyl azide. The regiochemistry was determined through NOESY experiments that showed a crosspeak between H1 and H1’ on the phenyl ring. This is consistent with the reactions of the monosubstituted norbornadienes 4a – 4e in which the larger group was on the same side of the ring as the N-Ts group. The reaction of unsymmetrically disubstituted norbornadienes 4i and 4j provided identical regiochemical results. While the yields of 5i and 5j were not as good as 5h, these were clean reactions providing only a single regioisomer. The regiochemistry of 5i and 5j was also assigned by the presence of a similar crosspeak in a NOESY spectrum, between the bridgehead H1 and a methylene proton of the alkyl group at C7.

We also wished to examine the reaction of the benzofused norbornadiene 11²³ with toluenesulfonyl azide. Reaction of 11 with tosyl azide provided only the aziridine 12²⁴ in 77% yield. All attempts to convert 12 to the desired 13 gave either no reaction or complete decomposition. These reaction conditions include heating 12 to over 250 °C, or the addition of a number of Lewis acids with or without heating. In trying to determine the reason for the lact of reaction of 12, Umano and coworkers determined that endo aziridines (e.g. endo-10) do not undergo the rearrangement reaction to provide the [3.2.1]bicyclic ring system.¹⁵ We carried out a NOESY experiment with 12 in order to determine the relative stereochemistry of the aziridine. We observed no crosspeaks between the aziridine protons and the bridgehead protons, indicating a likely exo-aziridine. Conseuqently, it may be that the rigidity of the system compled with the loss of aromaticity in the rearrangment reaction preclude the rearrangement of this system.

In conclusion, we have found that mono- and disubstituted norbornadienes undergo an addition/rearrangement reaction with tosyl azide. The reaction proceeds with excellent levels of regiocontrol in modest to excellent yield. This reaction provides an excellent route for the synthesis of substituted 2-azabicyclo[3.2.1]octadienes. These substituted 2-azabicyclo[3.2.1]octadienes can be readily converted to the reduced bicyclooctane ring system found in a number of natural products and pharmacologically active molecules.

Figure 3.

Observed NOESY crosspeak for product 5h.

Scheme 3.

Time and temperatures studies on 4f

Scheme 4.

Attempted reaction of benzonorbornadiene with tosyl azide

Table 1.

Reaction of monosubstituted norbornadienes^a, ^b

R1	Yield (%)	Ratio 5:6
4a, CO2Et	6	85:15 5a:6a
4c, CH2OAc	50	94:6, 5c:6c
4d, CH2OTBS	30	Only 5d
4e, CH2OPiv	50	Only 5e

^aWhen R² = H, R² is not shown for clarity.

^bThe regiochemistry of products 5a – 5e was determined by the observation of coupling between H5 (δ 2.62 – 3.01) and H6 (δ 5.93 – 6.09) as determined via COSY experiments.

Table 2.

Reactions of disubstituted norbornadienes.²²

Starting material	R1	R2	% Yield,a Product
4f	CO2Et	CO2Et	94, 5f
4g	CH2OAc	CH2OAc	50, 5g
4h	CO2Me	Ph	98, 5h
4i	CO2Et	nC5H11	40, 5i
4j	CO2Me	CH2OPh	33, 5j

^aYields are relative to tosyl azide.