|Home | About | Journals | Submit | Contact Us | Français|
Triphenylphosphine reduction of saturated endoperoxides derived from 6,6-dimethylfulvene and spiro[2.4]hepta-4,6-diene in the presence of nucleophiles results in the formation of products that mainly stem from deoxygenation followed by carbocation formation. Nucleophilic attack by solvent proceeds by an SN1 like mechanism; allyl shifts and cyclopropylcarbinyl-cyclobutyl rearrangements also occur. With the systems lacking carbocation-stabilizing groups, the deoxygenation step is preceded by attack of H2O at the phosphorus.
The reduction of cyclic peroxides1 and ozonides2 with triphenylphosphine has been known for a long time. In the case of unsaturated endoperoxides, the reaction results in deoxygenation, leading to ene epoxides,3 although a few exceptions have been observed.4 In the case of 1,2-dioxetanes, the reaction with PPh3 leads to a cyclic phosphorane that undergoes O-P cleavage followed by an intramolecular SN 2 reaction to give epoxides in a stereospecific manner.5 With bicyclic 1,2-dioxetanes, on the other hand, where backside attack is impeded, eliminations to allylic alcohols are observed.6 Samuelsson and co-workers reported that prostaglandin endoperoxides PGG2 (1a) or PGH2 (1b) upon reduction with PPh3 give the corresponding cis-1,3-diol (2), PGF2α.7 Clennan and Heah found that saturated bicyclic endoperoxides give with PPh3 in the presence of H2O trans-1,n-diols;8 moreover, they were able to detect the cyclic phosphorane intermediates by NMR and postulated for their formation a biphilic insertion of PPh3 into the O-O linkage. They also offered a mechanism for phosphorane decomposition involving heterolytic cleavage of the phosphorane to give a zwitterion that reacts with water via backside displacement (SN2) to give the trans-diol 5 (Scheme 1). On the other hand, the studies by Westheimer and co-workers,9 as well as McClelland and co-workers10 have shown, with careful mechanistic work, that the hydrolysis of pentaary-loxyphosphoranes occurs by an “inner sphere” mechanism (attack at phosphorus).
The base hydrolysis appears to be an associative process proceeding through a hexacoordinated phosphoroxanide ion, whereas the acid-catalyzed reaction is a dissociative pathway resembling the acetal or orthoester hydrolysis. Also contrary to the “outer sphere” mechanism, Taylor and Greatrex reported that a cis-1,4-diphenyl substituted 4,5-epoxy-1,2-dioxine resulted in the hydrolysis of the phosphorane intermediate at phosphorus rather than at the carbon center to give a meso-1,4-diol, in addition to deoxygenation products.11
Herein we disclose our results from PPh3 reductions of saturated endoperoxides derived from 6,6-dimethylfulvene and spiro[2.4]hepta-4,6-diene that have unraveled new pathways involving carbocation intermediates and rearrangements. We also report that the origin of the above-mentioned discrepancy regarding the stereochemistry of the diols 2 and 5, and hence the mechanism of phosphorane decomposition during PPh3 reductions of saturated endoperoxides in the [2.2.1] and [2.2.2] series in the presence of H2O is likely just due to a misassignment of the 1,n-diol stereochemistry.
In the course of our studies on peroxides derived from fulvenes and fulvene analogs,12 we investigated the triphenylphosphine reductions of 6,6-dimethylfulvene and [2.4]spirohepta-4,6-diene since our systems promised some unusual behavior due to the presence of the vinyl and cyclopropyl groups adjacent to the peroxo carbons, respectively.
After singlet oxygenation of 6,6-dimethylfulvene at -78 °C in CH2Cl2, and subsequent diazene reduction at low temperature, as described previously,13 the saturated endoperoxide 6 was treated with 1.2 equivalents PPh3 at 0 °C, and the mixture was stirred at room temperature for 10 h. The product mixture was chromatographed on silica gel to give one major product, 7, and a minor product, 8, a known aldehyde.14 For characterization purposes the latter was converted to its oxime 8a (syn and anti isomers in ca. 1:1 ratio) (Scheme 2).
We then decided to carry out the PPh3 reduction in the presence of a nucleophile, and for reasons of easier purification/ isolation of products we chose acetic acid instead of water. The reaction was conducted in CH2Cl2 at 0 °C in the presence of a slight excess of acetic acid. Under these conditions, a mixture of six products was formed, four carrying the acetoxy group. They were separated by silica gel chromatography (after treating the mixture with cold ether/petroleum ether and removing most of the Ph3 P=O by filtration) and identified as the products shown in Scheme 3.
The results presented above deserve some comment: Oxetane 13 is most likely formed from 12 by an SN2′ reaction. Under the conditions, 13 undergoes oxetane cleavage to give the allenic aldehyde 8 (Scheme 4). Compound 7 is formed from 14 by proton loss. Azine 9 in Scheme 3 is formed from 8 with hydrazine, a disproportionation product of diazene,15 in a serendipitous manner: the latter was present in the mixture in large excess since after reduction andfiltration of KOAc, the solution was treated with PPh3 without aqueous workup.
The hydroxyacetates 10a and b (cis and trans) were isolated as mixtures, whereas 11a was obtained in pure form. Through repeated chromatography the major isomer 10a could be isolated free of 10b. An attractive mechanism for the formation of 7, 8 as well as 10 and 11 would feature oxetane 13 as the key intermediate (Scheme 4). The assignment of the cis and trans isomers of 10 is based on independent synthesis of an authentic sample of 10a from 6 by NaBH4 reduction in MeOH at 0 °C followed by monoacetylation of the resulting cis-1,3-diol with acetyl chloride in CH2Cl2 in the presence of NEt3.
The intermediacy of oxetane 13 and its role as a precursor of 8 as well as 11a/b has precedent. The diphenyl analog of 13 has been prepared by Abe and Adam et al.16 and also by Abe et al.17 (Scheme 5). The latter group observed that 16 undergoes rearrangement to form 17, the corresponding allene aldehyde of type 8. On the other hand, with CH3CO2H the same oxetane was reported to give mostly the cis-acetoxy alcohol 18 with traces of the hydroxyacetates 19 and 20.In our case, products 10a, b and 11a, b do not have to stem from 13 and can directly be formed from 12 by loss of Ph3 P=O, followed by capture of the allyl cation 14 by the acetate ion. We believe that 11b is a secondary product stemming from 11a by an intramolecular acyl transfer.18
Next, we studied the triphenylphosphine reduction of the saturated endoperoxide 21 derived from [2.4]-spiro-4,6-heptadiene 21.19 Considering the significant stabilizing effect of a cyclopropyl group on adjacent carbocations when the three-membered ring and the carbocation are fixed in a bisected conformation,20 we expected a similar mechanism as with 6 with possible carbocation rearrangement involving the cyclopropyl group. These expectations were borne out by the following experiments. The saturated endoperoxide 21 derived from spiro[2.4]hepta-4,6-diene, prepared as previously described,19 reacted with PPh3 in the presence of acetic acid at room temperature to give a mixture of three products all of which contained an acetoxy group and which were separated from one another by silica gel chromatography. They were identified as 22, 23a and 23b (in order of their Rfvalues with 1:1 pet.ether/EtOAc, respectively) by means of their spectral and accurate HRMS data. The formation of the hydroxy acetates 23a and 23b is analogous to that of 10a and 10b from 6, and is indicative of a carbocation intermediate derived from the initial bicyclic phosphorane by way of Ph3 P=O extrusion. In particular the formation of the 1-acetoxybicyclo[3.2.01,5]heptan-2-ol (22, a single stereoisomer) constitutes clear-cut evidence for the intervention of a carbocation intermediate of the type 26, undergoing a cyclopropyl-carbinyl-cyclobutyl rearrangement,21 followed by capture of the cyclobutyl cation by the acetate ion.
Both 23a and 23b were stable toward AcOH under the reaction conditions applied to 21 and did not undergo rearrangement to 22. The mechanism outlined in Scheme 7 satisfactorily accounts for all three products.
The question remained whether the parent 2,3-dioxabicyclo[2.2.1]heptyl system 3 and its [2.2.2]octyl homologue 27 lacking carbocation stabilizing groups indeed undergo PPh3 reduction by an SN2-like attack of the nucleophile (e.g., H2O) at the bridgehead carbon. Our results, as presented above,do not support the SN2 mechanism, though it is clear that carbocation-stabilizing groups at the α-position would dramatically influence the mechanistic pathway in the deoxygenation step. We therefore subjected the saturated endoperoxide derived from cyclopentadiene as well as 1,3-cyclohexadiene to PPh3 reduction in the presence of water.
In our hands, the sole products formed from both reactions were exclusively the cis-diols 28a and 28b (in 92 and 79% yields, respectively, from the corresponding dienes) with no traces of the trans isomers. These results confirm Samuelsson’s earlier report on the isolation of the cis-diol 2 from prostaglandin endoperoxide 1 with PPh3 and render the SN2 pathway less likely. We propose the mechanism shown in Scheme 9 for PPh3 reduction of endoperoxides of the type 3 and 27 lacking carbocation-stabilizing groups adjacent to the peroxo carbons in the presence of H2O.
Authentic samples of 28a and 28b were independently synthesized by thiourea reduction of the respective unsaturated endoperoxides 3 and 27, followed by diazene reduction of the resulting unsaturated cis-diols in methanol (diazene generated in situ from potassium azodicarboxylate with AcOH in MeOH at 0 °C).
In conclusion, we have shown that PPh3 reductions of saturated endoperoxides such as 6 and 21 containing vinyl or cyclopropyl groups α to the peroxo carbon facilitate direct deoxygenation and carbocation formation. The initial biphilic insertion of PPh3 into the peroxo bridge had elegantly been demonstrated by Clennan and Heah.8 However, it is quite likely that the deoxygenation step in the case of 28a or 28b proceeds by heterolytic O-P cleavage of the cyclic phosphorane intermediate of the type 4 followed by attack of water at the phosphonium ion before 31 collapses to Ph3 P=O and the cis-diol (Scheme 9). With a vinyl or cyclopropyl group present at the carbon adjacent to the bridgehead position (as in 12 and 24), facile Ph3 P=O loss from intermediate 30 occurs leading to an allyl or cyclopropylcarbinyl cation that undergoes SN1 substitution either directly or after allyl shift (both cis and trans hydroxyacetates are formed in each case), or cyclopropylcarbinyl-cyclobutyl rearrangement, respectively.
This paper is dedicated to Professor Lawrence T. Scott on the occasion of his 65th birthday. This work was supported by funds from the National Institutes of Health, MBRS-SCORE Program-NIGMS (Grant No. GM52588). We thank Professor Edward L. Clennan, University of Wyoming, for valuable discussions and Mr. Wee Tam, SFSU, for recording the NMR spectra.