Bicyclic peroxides

Chemistry of Saturated Bicyclic Peroxides (The Prostaglandin Connection)1... 

This review is concerned with the synthesis, spectroscopic characterisation, and reactions of saturated bicyclic peroxides. Four years ago such compounds were virtually unknown, yet the first example 2a had been prepared as early as 1938 by catalytic hydrogenation of the naturally occurring peroxide ascaridole 1 (Eq. 1) l). 

The stimulus for the recent surge of activity in this previously dormant area of organic chemistry can be traced to the prostaglandin connection . That is to the discovery that saturated bicyclic peroxides are key intermediates in the biosynthesis of prostaglandins and other physiologically active substances by the enzymatic oxygenation of polyunsaturated fatty acids. 

As a reasonable biogenetie pathway for the enzymatic conversion of the polyunsaturated fatty acid 3 into the bicyclic peroxide 4, the free radical mechanism in Equation 3 was postulated 9). That such a free radical process is a viable mechanism has been indicated by model studies in which prostaglandin-like products were obtained from the autoxidation of methyl linolenate 10> and from the treatment of unsaturated lipid hydroperoxides with free radical initiators U). 

Dioxabicyclo[2.2.1]heptane naturally assumed the role of the principal target molecule. It represented a considerable synthetic challenge, for not only is it a strained bicyclic molecule containing the weak and labile 0—0 bond, but it is also a di(secondary-alkyl) peroxide which is the most difficult type to make by classical procedures 12). New synthetic methods of exceptional mildness were clearly needed to solve this problem. In the course of the development of such techniques and from a desire to establish their scope, a variety of saturated bicyclic peroxides have been obtained in addition to 2,3-dioxabicyclo[2.2.1]heptane. The question of how substitution patterns and ring sizes affect the reactivity of bicyclic peroxides has further served to broaden interest in the subject. 

With synthetic principles now well established, a progress report seems timely. It is hoped that this will encourage further work in the area, particularly on the reactions of bicyclic peroxides where studies are still in their infancy. 

Subsequently we demonstrated the generality and effectiveness of this synthetic approach to bicyclic peroxides. Among the basic skeletons that have been prepared, figure the [2.2.1]-, [2.2.2]-, [3.2.2]-, [4.2.1]- and [4.2.2]-bicycloperoxides, whose structures are shown below. 

This important modification enabled us to prepare a number of bicyclic peroxides possessing the [2.2.1]-skeleton, of which the important ones are shown below. 

Of the substrates that have worked well, let us first illustrate the 7-alkylidene-2,3-dioxabicyclo[2.2.1]heptane system 10. It was known that fulvenes react with singlet oxygen at low temperatures to afford the corresponding endoperoxides however, attempts to isolate these labile compounds led to decomposition, although NMR identification was possible at —70 °C 19>. When reduction of the singlet oxygenates with diimide was performed at —50 °C, the bicyclic peroxides 10 were obtained in high yield (Eq. 7) 20). 

Furthermore, ozonolysis in the presence of tetracyanoethylene (TCNE) afforded the novel bicyclic peroxide 15 which, as stated already, could not be prepared via the singlet oxygen-diimide route starting from cyclopentadienone. Peroxide 15 was too unstable for isolation, but the characteristic proton resonances at 8 2.0 (m, 4 H) and 4.38 (m, 2 H) ppm are consistent with the assignment. 

The bicyclic peroxide 11 was prepared via diimide reduction of the endoperoxide derived from spirocyclopentadiene (Eq. 8)21>. As before, at elevated temperature the labile endoperoxide rearranges into diepoxide and ketoepoxide,22) but diimide reduction at —78 °C allows trapping leading to the highly strained bicyclic peroxide 11. 

Similarly, the cyclobutane-fused bicyclic peroxide 19 was prepared by diimide reduction of the corresponding bicyclic endoperoxide derived from 1,3,5-cyclooctatriene (Eq. 14)31a). 

Again, the bicyclic valence isomer coexists in sufficient concentration, that the bicyclic peroxide 19 was readily accessible in ca. 20% yield. Alternatively, the thermally labile bicyclic valence isomer of cyclooctatetraene, namely bicyclo[4.2.0]-octa-2,4,7-triene, was converted into the corresponding endoperoxide on low temperature singlet oxygenation and reduced with diimide to yield 19. 

Alternatively, the (2 + 4)-tropilidene endoperoxide, which is the major product in the singlet oxygenation of cycloheptatriene 30 a) affords on diimide reduction the desired bicyclic peroxide 20. The double bond in the two-carbon bridge is reduced selectively, but on exhaustive treatment with excess diimide, the fully saturated substance is obtained. A number of substituted derivatives have been prepared in this way30). 

The strained dienic endoperoxide is readily reduced by diimide, leading to the relatively stable bicyclic peroxide in high yield. Again, aprotic solvents such as CH2C12 or CFClj are essential for the diimide reduction, because in MeOH complex rearrangements take place 30d e>. 

An intramolecular variation of this method, employing 3-bromocycloalkyl hydroperoxides 24, has proved a versatile procedure for the synthesis of bicyclic peroxides with an [n.2,1] skeletal arrangement, 25 (Eq. 20). 

Peroxide ring closures were effected by stirring the 2,3-dibromocycloalkyl hydroperoxides with silver trifluoroacetate, and the bromo-substituted bicyclic peroxides were isolated by silica chromatography at —25 °C. Yields (based on 2-cycloalkenyl hydroperoxide) of 56 and 38% were achieved respectively for the [3.2.1]- and [4.2.11-compounds, but only 16% of the [2.2.1]- and 13% of the [5.2.1]-peroxide was obtained. The main reason for the low yield of the [2.2.1]-peroxide was that substitution by trifluoroacetate, which competes with the desired dioxabicyclization, is particularly prevalent with the 5-membered ring. 

Only one bicyclic peroxide was obtained for each ring system and for 26 and 27 we confirmed that the bromide is cis to the peroxide linkage. In the [2,2.1]-compound this was established from its H n.m.r. spectrum by the absence of long range W-plan coupling for the CHBt proton. The cw-configuration of the [3.2.1] compound was indicated by the lH n.m.r. spectrum of the 2-bromo-3-butoxy-cyclohexanol obtained quantitatively upon reaction with butyllithium (Eq. 24), and has now been confirmed by an X-ray crystal structure determination39). 

Thus, as expected from earlier work 36,37), the dioxabicyclization proceeds with inversion of configuration at the 3-position of the m-2-frans-3-dibromide 30, the stereochemistry at the 2-position being unaffected. However, experiments with individual diastereoisomers unexpectedly showed that the franj-2-m-3-dibromide 31, also reacts with silver trifluoroacetate, albeit less efficiently, to give the same bicyclic peroxide. We feel that this probably proceeds via an isomerisation (Eq. 25). 

The behaviour of the tram-3-bromide 38 closely resembled that of its cyclopentyl analogue 32. Thus with silver oxide only the cis-2-bromo-[3.2.1]peroxide 40 expected for a SN2 ring closure was obtained, and although some 40 was also formed in the reaction of 38 with silver trifluoroacetate, the predominant (90 %) bicyclic peroxide obtained was 41, i.e. the [3.2.1]peroxide available via a bromonium ion mechanism. The behaviour of the tran.v-4-bromide 39 was very revealing, for it did not react with silver oxide and 41 was the only bicyclic peroxide formed with silver trifluoroacetate. 

As the first isolable intermediate in the bioconversion of arachidonic acid into prostaglandins and thromboxanes (Eq. 3), PGG2 is a bicyclic peroxide of immense biological importance. It is difficult to obtain pure from natural sources and the presence of the 15-hydroperoxide group adds a further dimension of chemical lability to that associated with the 9,11-peroxide bridge. The chemical synthesis of PGG2 is thus a landmark in prostaglandin chemistry. It also represents a pinnacle of success for the silver-salt route to bicyclic peroxides. 

The reductive demercuration was marred by the loss of about half of the peroxide due to competing deoxymercuration which afforded 4-cycloocten-l-ol. An additional complication was the formation of a small amount of trans-1,2-epoxy-cw-cyclooct-5-ene. The bicyclic peroxide 50 was readily separated from the unsaturated alcohol by silica chromatography, but complete removal of the epoxide was more difficult. Preservation of the peroxide linkage was markedly higher in the bromodemercuration. The diastereoisomeric dibromoperoxides 51 were separated by HPLC, although only one isomer was fully characterised. 

A bicyclic peroxide was isolated in 1.8% yield by HPLC of the bromodemercuration product, and was identified as a single diastereoisomer of 2,4-dibromo-6,7-dioxa-bicyclo[3,2.1]octane 57 with the iransjrans- or cis,cis-configuration. By analogy with the cyclooctadiene reactions, formation of the other two diastereoisomers of 57 can be expected, but although additional peroxides with similar HPLC characteristics were detected, they were not identified. Thus the presence of [2.2.2]-compounds cannot be ruled out, and no comment can be made on the regioselectivity of the dioxabicyclization. 

Thus the range of bicyclic peroxides available via peroxymercuration may be quite limited. Nevertheless where the method works best, namely with 1,5- and 1,4-cyclo-octadiene, it makes a valuable contribution in that each peroxymercuration is regiospecific and leads to a dioxabicyclodecane that is isomeric with the [4.2.2] compound 23 available via photooxygenation (Eq. 18). Furthermore, the [3.3.2] compounds derived from 1,5-cyclooctadiene are, to the best of our knowledge, the only bicyclic peroxides obtained to date that do not contain either a 5- or a 6-membered dioxacycloalkane ring. 

Although no other examples have been reported, oxygen trapping of azo-derived triplet diradicals provides a potentially versatile strategy for the synthesis of bicyclic peroxides under neutral conditions. 

The structures of the new bicyclic peroxides have been established by the usual combination of physical techniques and chemical transformations. Here we highlight features of the H and 13C n.m.r. spectroscopic data that provide the best characterization of these compounds their reactions are discussed later. Information about the C-O-O-C dihedral angle in organic peroxides is potentially available from photoelectron (PE) spectroscopy. Measurements on comparatively rigid systems play an important part in establishing a soundly based experimental correlation, and the results obtained on several of these bicyclic peroxides are presented in this section also. 

The most distinctive features of the H and 13C n.m.r. spectra of bicyclic peroxides are provided by their bridgehead nuclei. An analysis of the data on over thirty compounds indicates that the characteristic chemical shift ranges are 5 3.7-4.8 for bridgehead protons and 8 72-89 for bridgehead carbons. 

Diagram 1. Correlation between the split (AI) of the first two ionisation potentials associated with the peroxide moiety and the C—O—O—C dihedral angle (cos 0) for selected bicyclic peroxides. 

---