Iridoids from Vitex cymosa

a Núcleo de Pesquisas de Produtos Naturais, Universidade Federal do Rio de Janeiro, Ilha do Fundão, Centro de Ciências da Saude, Bloco H, 21941-590, Rio de Janeiro RJ, Brazil b Setor de Química de Produtos Naturais, Universidade Estadual do Norte Fluminense, LCQUI/CCT, Av. Alberto Lamego, 2000, 28015-620, Campos RJ, Brazil c Centro Chimica dei Recettori, C. N. R., Università Cattolica S. Cuore, Largo Francesco Vito, 1, 00168, Rome Italy d Departamento de Produtos Naturais e Alimentos, Faculdade de Farmácia, Universidade Federal do Rio de Janeiro, Ilha do Fundão, CCS, Bloco A , 21941-590, Rio de Janeiro RJ, Brazil


Introduction
Vitex species (Verbenaceae) are small trees or shrubs occurring in tropical and subtropical regions.Plants of this genus have been used for a great variety of medicinal purposes as well as an incense with insecticidal properties 1 .Vitex cymosa Bertero is a small tree widely distributed in the Central and Amazon regions of Brazil where it is popularly known as "Tarumã-do-Igapó" and "Tarumã-do-alagado" 2 .
In 1995, Watanabe 3 et al. isolated the monoterpene dialdehyde rotundial (3) as a new natural mosquito repellent from V. rotundifolia.In other studies, derivatives of aucubin and mussaenosidic acid were found in various Vitex species 4,5,6 .
We now report the isolation and identification, from the leaves of V. cymosa, of a new iridoid, closely related to 3 which was named tarumal (1), as well as the known compounds viteoid II (2) and agnuside (4).

Results and Discussion
From the dichloromethane extract from the leaves of V. cymosa a new iridoid aglycone was isolated as a colourless oil which was named tarumal, (1).The 1 H NMR data of 1 showed some similarities with those of 3 (Table 1).In particular, the signals arising from the aldehyde conjugated with the endocyclic double bond at δ 9.98 and the methyl group located on the same double bond at δ 2.17.However, instead of the other aldehyde signal of 3, a signal was observed at δ 5.00, indicating the presence of an acetal.Further, a doublet of doublets at δ 4.65 suggested the additional presence of an oxygen function and a three proton singlet at δ 3.33 showed the presence of a methoxyl group.The 13 C NMR spectrum of 1 displayed 10 signals (Table 2).In comparison with the spectrum of 3 the replacement of the aldehyde by an acetal was confirmed by the absence of the signal at δ 201.9 and the presence of an additional signal at δ 105.8.Further, one methylene group was replaced by an oxygen bearing carbon (δ 78.4).On the basis of these data the structure of 1 was obtained.The configuration at C-5 was considered to be R, (with a 5β-oriented H) on biosynthetic grounds (its relation with 3, in which the same stereochemistry has been determined, see below).Furthermore, nearly all iridoids which have been found in nature have this configuration.Only a few exceptions are kown, none of which are within the family Verbenaceae.The ring closure with the substituent at C-6 requires the S configuration for the latter.By analysis of the coupling constants, the configuration of C-3 was determined to be R (Figure 1).This was confirmed by molecular modelling using the program Serena Software for PC, v. 4. 0. This program uses a molecular mechanics approach to calculate the best conformation.In contrast to the programs using molecular orbital interactions the molecular mechanics approach can be used satisfactorily for the determination of configurations of 5-membered ring systems.For each of the C-3 configurations, two least energy conformations were found.One with the C-3 methoxyl-substituent in an axial position and another with the substituent in a equatorial position.Only 3R-tarumal with the methoxyl group in an axial position (Figure1) showed complete agreement of the calculated coupling constants with the experimental values.In this comparison all coupling constants of the complete molecule were considered.Especially the coupling constants of H-4α and H-4β with both H-3 and H-5 showed large differences between the experimental and calculated coupling constants (see Table 3).According to the heats of formation calculated by PC-model for the conformers of 3R-tarumal, the conformer with the methoxyl group in an axial position should also be the most stable and thus be the predominant conformer.
From the butanol extract of the leaves of the same plant, 1 was reisolated together with another iridoid aglycone which was identified as viteoid II (2).This extract also yielded the known iridoid agnuside (4), which has been isolated before from V. rotundifolia 5 . 1 H and 13 C NMR data reported in the literature for viteoid H 5 are almost identical to those obtained for compound 2 (Tables 1 and 2).The relative stereo-chemistry proposed by the authors 5 is also in good agreement with our nOe difference experiments (Figure 2).Irradiation of H-6 yielded strong enhancements of the signals of H-7α (6. 5 %) and H-4β (3. 2 %), while no enhancement of the H-5 signal was observed.Irradiation of H-5 yielded enhancements of H-4β (3.0 %), H-3β (4.0 %) and H-7β (6.1 %).Based on the same arguments as mentioned above for 1, the configuration at C-5 should be R. Molecular modeling of 2 showed that the configuration at C-6 should be R.This was based on the comparison of the predicted values (data not shown) of all coupling constants of the entire molecule of both stereoisomers with the experimental values.The presence of a bicyclic structure was confirmed in the HMBC spectrum of 2 by a long range correlation of C-1 (δ 165.9) with H-3α and H-3β at δ 4.41 and δ 4.31, respectively.
Upon standing in methanol, we observed that 2 decomposed to another product.This was also described for viteoid II 5 and the decomposition product was identified as 1-oxo-eucommiol (5).At the time of the submission of this paper, the same structure 5 was reported for another product, crescentin II, isolated from Crescentia cujete 7 .
Our present study confirms the structure published for 1oxo-eucommiol ( 5) and consequently the structure of crescentin II needs revision.However, we cannot suggest an alternative structure for crescentin II based on the published spectral data.
All the proton and carbon assignments of 1, 2 and 5 (Tables1 and 2) were confirmed by 2D NMR measurements (HSQC and HMBC).
In order to verify the possibility of 1 being an artifact, reisolation of this substance was performed without purification on Sephadex LH-20 using MeOH.After fractionation of the dichloromethane extract by CC using a CH 2 Cl 2 -EtOAc gradient, the fractions eluted with CH 2 Cl 2 -EtOAC (9:1) which presumably contained 1 were analyzed by HPLC.The presence of 1 could be confirmed by comparison of retention times, UV spectra and coinjection with pure 1.
Compounds 1 and 2 are closely related to 3. Compound 1 might well be biosynthetically derived from 3 by a hydroxylation on C-6, subsequent hemiacetal formation with the C-3 aldehyde, and a reaction with a molecule of MeOH.
The occurrence of ciclopentene iridoids has been reported from various families.Recently the presence of rotundial was reported in Alberta magna 8 (Rubiaceae).The crescentins reported from Crescentia cujete 7 (Bignoniaceae), borreriagenin from Borreria verticillata 9 (Rubiaceae), the eucommiosides from Eucommia ulmoides 4 (Eucommiaceae)  and Aucuba japonica 4 (Cornaceae) and the dolichodial derivative reported from Centranthus ruber 10 (Labiatae) also show similar structures.The occurrence of these compounds does not seem to have much chemosystematic value.