The chemistry of some West African plants

• Main Author: Toft, Peter
• Published: University of Oxford 1966
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.732553

The main subjects of this thesis are the chemistry and structure of two triterpenes, methyl angolensate (1) and turraeanthin (2), which were isolated from species of the family Meliaceae. A review is included of other triterpenes isolated from this family and from the related plant families Rutaceae and Simaroubaceee. Methyl angolensate has been obtained from several species of the family Meliaceae found in West Africa,¹ of the principal source being Entandrophragme angolense² where it occurs with gedunin (3).⁹ Initial studies had been carried out by Taylor and Powell³ and continued by Welford.⁴ Methyl angolensate was also isolated by Chan,⁵ from Cedrela odorate obtained from the West Indies, with whom there has been an interchange of information. Methyl angolensate was assigned the structure (1) on the basis of the following studies. Spectral evidence and chemical studies indicated the presence of a ketone, a methyl ester, a &delta;-lactone and a &beta;-substituted furan, and showed the absence of a hydroxyl group. Reduction of methyl angolensate with potassium borohydride gave a diol (4) in which the 3-oxo group had been reduced to an alcohol and the &delta;-lactone had been converted into a hemiacetal. The diol (4) underwent a retropinacolic rearrangement in the presence of phosphorus pentachloride, and a product (5) was obtained which on ozonolysis afforded acetone and a cyclopentanone derivative. This indicated that methyl angolensate had the partial structure (6).⁶ Reduction of methyl angolensate with lithium aluminium hydride gave a mixture of two triols (7) and a diol (8). On oxidation with chromic acid all of these compounds gave angolensic acid (9).^{4, 5} Oppenauer oxidation of the diol (8) gave a hydroxy-aldehyde.⁵ The n.m,r. spectrum of the latter compound showed the signal due to the aldehydic proton as a triplet, indicating the presence of -CH₂.CHO in this compound, and hence suggesting the grouping -CH₂.CO₂Me in methyl angolensate. Hydrogenation of methyl angolensate in acetic acid over palladised charcoal produced an octahydro acid (10) which was still a methyl ester but no longer a furan. ^of.7 This reaction indicates the presence of the lactone - furan system as found in gedunin (3).⁹ The pK^*<sub style='position: relative; left: -.5em;'>MCS</sub> (7.2) of this acid showed that it was not an &alpha;-oxy-acid while the n.m.r. spectra of the methyl angolensate derivatives in which the lactone group had been reduced to a hemiacetel group showed that there was a -CH₂- group next to the carbonyl group of the lactone. The functional groups so far described account for six of the seven oxygen atoms. The n.m.r. spectrum of methyl angolensate had a band at 6.53 &Gamma; (quartet) due to one proton which is indicative of the grouping. This suggested that the seventh oxygen atom was in a cyclic ether. After allowance for the functional groups the formula C₂₇H₃₄O₇ requires that methyl angolensate has three carbocyclic rings and/or double bonds. Spectral evidence indicated the presence of an exocyclic methylene group which could be hydrogenated to give a secondary methyl group. The presence of the -CH₂.CO₂Me group and the CH₂ group in methyl angolensate can be explained if ring B or ring C of the gedunin skeleton has been opened oxidatively. Evidence from the n.m.r. spectre indicated that ring C was intact, and therefore ring B must be open. The nature of the ether bridge was demonstrated by treating with sodium isobutoxide which caused a base-cataysed elimination of the ether linkage at C₍₁₎ to yield the &alpha;,&beta;-unsaturated ketone (11).⁵ Biogenetic and steric considerations point to the other end of the ether bridge being at C₍₁₄₎. Thus the structure (1) was deduced for methyl angolensate. Extraction of the heartwood of Turraeanthus africanus afforded a new triterpene, turraeanthin, which was assigned the structure (2) on the basis of the following evidence. Mass spectral and combustion analysed indicated the formula C₃₂H₅₀O5. Spectral evidence indicated the presence of a hydroxyl group, an acetoxy group, a trisubstituted double bond and an epoxy group. The n.m.r. spectrum showed that there were seven tertiary methyl groups. Mild oxidation with chromium trioxide - pyridine complex gave a &gamma;-lactone (12). This indicated that a five-membered cyclic hemiacetal must be present in turraeanthin. The trisubstituted double bond could not be hydrogenated under mild conditions and ozonolysis did not afford any volatile fragment. The double bond is therefore in a ring. Alkaline hydrolysis of turraeanthin gave deacetylturraeanthin (13), the oxidation of which gave a ketone (14) and a keto-lactone (15). Acetylation of turraeanthin afforded an acetate (16), also obtained by acetylation of deacetylturraeanthin (13). Turraeanthin is therefore a monoacetate. The n.m.r. spectrum had a band at 5.45 &Gamma; due to the proton on the acetoxy carbon atom. Its band width and shape were typical of that of the 3&alpha;-proton of triterpene 3&beta;-acetate. Treatment of turraeanthin with methanolic hydrochloric acid converted the hemiacetal grouping to an acetal and opened the epoxide ring to give the chlorohydrin (17) and the methoxyhydrin (18). A similar reaction with methanolic sulphuric acid also caused hydrolysis of the acetoxy group affording compound (19) which with chromium trioxide - pyridine gave the ketons (20). The negative Cotton curve of the latter compound (20) was almost superimposable upon that of flindissone lactone (24).⁸ The relationship of the epoxide group to the hemiacetal group was demonstrated by a spin decoupling experiment on the lactone (12). Irradiation of the proton at C₍₂₃₎ absorbing at 5.85 &Gamma; caused collapse of the doublet at 7.15 &Gamma; due to the epoxy proton at C₍₂₄₎. The Presence in Meliaceae and the related Rutacese of a number of degraded triterpenes such as limonin (21)⁷ and gedunin (3)³, all of which can be formally derived by oxidative degradation of tirucalla-7,24-dien-3&beta;-cl (&minuseq; 20&alpha;H-butyrospermol) together with the evidence described suggested that turraeanthin was (2) and closely related to flindissol (22).⁸ The relationship was established in the following way. The Keto-lactone (15) on being treated with sulphuric acid in ether gave the glycol (23). This was converted ^{of. 10} into the cyclic thionocarbonate and hence to a product which proved to be identical with flindissone lactone (24).⁸ No melting point depression was observed on admixture with an authentic sample, and the infrared, mass, and n.m.r. spectra were identical. The possible biogenetic origin of turraeanthin (2), methyl angolensate (1) and other oxidised triterpenes found in the related Mellaceae, Rutaceae, and Cimaroubaceae plant families is discussed.