Further Triterpenes, Steroids and Furocoumarins from Brazilian Medicinal Plants of Dorstenia genus (Moraceae)

As frações solúveis em hexano dos rizomas ou das folhas de cinco espécies do gênero Dorstenia (Moraceae) ( D. bahiensis Kl., D. bryoniifolia Mart ex. Miq., D. carautae C.C.Berg., D. cayapiaa Vell. e D. heringerii Car. & Val.) foram analisadas por HRGC-MS (cromatografia gasosa de alta resolução acoplada a espectrometria de massas). Foram identificados triterpenos pentacíclicos, esteróides e furocumarinas. HRGC-MS mostrou ser uma ferramenta valiosa para a análise dos terpenóides de Dorstenia spp. Estas substâncias podem estar ligadas à utilização tradicional de Dorstenia spp. como plantas anti-ofídicas.


Introduction
The genus Dorstenia L. (Moraceae) includes several Brazilian species which are utilized in folk medicine due to their alleged medicinal properties, mainly in skin diseases and as antiophidics 1 . In these plants, furocoumarins are the most abundant compounds 2 and these compounds support topical utilization of Dorstenia rhizome preparations in the treatment of skin diseases such as psoriasis and vitiligo 3 .
Despite the strong ethnobothanical indication concerning the antiophidical properties of some Dorstenia species, their pharmacological evaluation is still inconclusive. However, it should be pointed out that the antiophidical utilization of some other Brazilian medicinal plants may be associated to the presence of widespread terpenoids (triterpenoids and phytosterols) with venom-inactivating action 4,5 . The folk indication of triterpenoid-containing plants as "antiophidics" may also be explained by pharmacological action related to the relief of snake bite symptoms, such as analgesic and antiinflamatory effects. These effects may be related to the presence of triterpenoids, as these effects were found even in some triterpene-containing plants with no folk indication as antiophidics 6 .
ln face of the above mentioned data, we have focused the present work on the analysis of further terpenoidal compounds from five Dorstenia species, using HRGC-MS. This technique may be an excellent alternative to the clas-sical phytochemical analysis procedures based on preparative-scale chromatographic isolation combined with spectroscopic identification, for the analysis of low-polarity plant extracts. HRGC-MS both avoids time-consuming and relatively expensive purification steps and gives relevant structural information about compounds not easily identified by NMR, IR or UV 7 . However, in the study of medicinal plants, HRGC-MS is still usually restricted to the analysis of low-polarity compounds of low molecular weight (MW 150-250 a.m.u.) such as monoterpenes from essential oils.

Extraction of plant material and fractionation of extracts
General Plant material was dried at 40 °C, with forced ventilation, during 3 days, before being powdered and sieved. Only material between 0.210-0.350 mm size was utilized. Plant material (5 g) was macerated with the solvent indicated for each species, at the ratio of 1:10 (w/v), room temperature, during 7 days. When necessary, the extracts obtained were fractionated, as described as follows.
The hexane (Hex) fraction was obtained from the crude extracts: the extract was filtered over active charcoal (Reagen, Rio de Janeiro, Brazil); the solvent was removed under vacuum; the residue was shaken with hexane (Merck), and the hexane-soluble fraction was filtrated, dried over anhydrous Na2SO4 (Merck) and analyzed by HRGC.
The terpenoidal (Terp) fractions which contained mainly triterpenes and/or steroids were obtained by silica gel open-column preparative liquid chromatography (PLC) of the crude plant extracts and selected by NMR ( 1 H-or 13 C-) analysis of the mixture. Since that the avaiable amount of each hexanic extract obtained as above was very reduced (between 10-100 mg), all the PLC procedure was performed in reduced scale, utilizing glass Pasteur pipettes containing around 500 mg silicagel, and around 5 mL of each solvent for elution.
D. bryoniifolia -the rhizomes were macerated with hexane. The crude extract was directly analyzed by HRGC.
D. carautae -the whole plant was macerated with EtOAc. The Hex fraction was obtained as described above (General) D. cayapiaa -the leaves were macerated with CHCl3-EtOH 1:1 (v/v). The Hex fraction was obtained as described above.

Chromatographic analysis
HRGC-MS analysis were performed by using a HP 5970 MSD, with electron impact ionization (70 eV) coupled to a HP 5890 GC. The column used was a cross-linked 5% phenyl, 95% methylsilicone HP-5 (25 m x 0.2 mm x 0.33 µm), furnished by Hewlett-Packard, U.S.A.. Samples were injected using the split mode (split ratio 1:30), with injector temperature and GC-MS interface temperature both at 280 °C. Column temperature was programmed from 150 °C (2 min), at 8 °C/min to 290 °C (held during 20 min). Hydrogen was used as carrier gas. MS scan range was 50 to 500 a.m.u.. Data were processed on a HP 7946/HP 9000-300 CPU.
HRGC-FID analysis were performed on a HP 5890 GC, using the same column with the same temperature programming utilized for HRGC-MS analysis. Detector temperature was 290 °C, and data were obtained on a HP 3396A integrator. Hydrogen was utilized as carrier gas, at the average linear velocity of 40 cm/s.

Analysis of fatty acids
Fatty acids were identified by HRGC-MS as their methyl esters, after esterification with diazomethane of the fractions containing these compounds.

ldentification of the chemical constituents
Identifications were made by comparison of retention times, obtained at identical chromatographic conditions of the analyzed samples and authentic standards or by co-injection. In both cases, chromatographic data were considered together with the comparison of MS data with those from authentic standards, obtained at similar ionization conditions (EI, 70 eV) in our laboratory. Tentative identi-fications were made by analysis of the fragmentation data, comparison with literature MS data 8,9,10 and computer search in the NBS library.

Results and Discussion
HRGC-MS was applied both to the analysis of crude extracts and of fractions obtained from PLC. For the analysis of crude extracts, due to the presence of more polar compounds, only the hexane-soluble compounds were analyzed, since this solvent can selectively extract compounds appropriate to GC analysis, including low-functionalized steroids and triterpenoids. Despite a relatively longer analysis time, preliminary work with standards showed that derivatization of the triterpene and steroid alcohols is not essential, with the advantage of allowing direct comparison with literature data of underivatized compounds 9 .  The HRGC-MS -TIC profile of the Dorstenia crude extracts showed some similarity between the chemical pattern of leaves and rhizomes, even from different species: the furocoumarins are eluted first, followed by the steroids, while the triterpenes are eluted later. FID integration of the peaks indicated a predominance of the furocoumarins over the aliphatic compounds.
HRGC-MS-TIC allowed identifying several compounds from Dorstenia samples (Table 1 and Fig. 1). The chemical composition of all Dorstenia samples herein examined showed to be very similar (Table 1), and some of the identified compounds have already been reported from other Dorstenia species 2 . All the triterpenes were found in small amounts (yield from vegetal material: 0,1% or less).
The procedure herein adopted limited the scope of the compounds analyzed to the less polar constituents. More highly funtionalized or polar molecules such as syriogenin, which was previously found in D. contrajerva 11 might not be detected, due to the selective extraction of low-polarity compounds and also due to the lower volatility of the polar compounds; the knowledge of this limitation is important for the final option for HRGC-MS analysis. Therefore, we are studying the more polar compounds from Dorstenia hydroalcoholic extracts by using conventional procedures.

Identification of the compounds
The prenylated derivative of bergapten, 1d, was found only in D. bahiensis. Its identity was confirmed by comparison of retention time and MS data with those of an authentic sample isolated from D. cayapiaa 12 .
Using only MS data, it was not possible to identify the furocoumarins 1a-c and 2a-b, due to the similarity of the fragmentation patterns of the linear and angular isomers. Their identification also required comparison with tR data of authentic samples of some furocoumarins using the same chromatographic conditions as in this work.
Most of the triterpenes found belong to the oleanane/ursane series, characterized by a base peak at m/z = 218 9 . Unequivocal differentiation between α and β-amyrin re-quired co-injection with authentic samples, since these triterpenes have very similar MS spectra (in the chromatographic conditions utilized, β-amyrin elutes first). In the operational conditions of the MSD system utilized in this work, αand β-amyrin could be differentiated by examination of the relative intensities of the peaks at m/z 189 and 203: β-amyrin has a m/z 203 peak around twice the intensity of the m/z 189 peak, while α-amyrin spectra shows both peaks with similar intensity.
ln the hexane extract of D. bryoniifolia, two triterpenes were found with base peaks at m/z 218, but both with a peak at m/z 424, supposed to be the molecular peak, instead of m/z = 426 as in the case of αand β-amyrin. Exchange of an hydroxy group for a keto group leads to a decrease of molecular weight and polarity, which leads to a lower tR in the column utilized for the HRGC analysis of these compounds. Besides, MS spectra of these compounds shows a peak at m/z 205, due to the fragmentation of a 3-keto-system involving rings A to C, and at m/z= 203, corresponding to the fragment originating from methyl loss of the diene portion originated from retro Diels-Alder fragmentation of a ∆ 12 -unsaturated oleanane or ursane 9 . With these data, they were tentatively identified as being αor β-amyrone (3e-f). Final identification was made by using αand β-amyrin fragmentation as models: in the keto compounds, a difference was also observed in the relative intensity of the peaks at m/z = 189 and 203, which originate from further fragmentation of the m/z = 218 ion, the latter indicating the C-ring opening of a D,C-unsubstituted ∆ 12 -oleanene or ursene 9 . ln an analogous way to αand β-amyrin, we have assigned the compound with the more intense m/z = 203 peak as being β-amyrone (3f) and the α-amyrone structure (3e) was assigned to the compound with m/z = 189 and 203 peaks with similar intensity. Chromatographic data also supports these assignments, since the peak corresponding to 3e in the TIC has a lower tR than 3f in the same elution sequence as for the 3-OH terpenes. Glutinol acetate was identified by analysis of its fragmentation data and comparison with an authentic sample of simiarenol, isolated from Dorstenia asaroides 13 . Despite the lupane-type ring E of the latter, which leads to some characteristic peaks related to the isopropyl group, both glutinol and simiarenol have great similarities in their MS spectra due to the similar structure of rings A to D 14 , the most important feature being the base peak at m/z 274, followed by a (274-Me) peak at m/z 259, which characterizes ∆ 5 -unsaturated skeleta 10 .
Taraxerol acetate was identified only by MS, mainly due to the base peak at m/z 204, which originates from rings D and E of an ∆ 14 -taraxerene. Another important peak is m/z 344, which originates from a retro Diels-Alder decomposition with ring-D opening and confirms both the unsaturation and the presence of an acetoxy group at C-3 9 .
The other triterpenes which are not discussed herein were identified by co-injection with authentic samples; other compounds were identified by direct comparison of MS and tR data with authentic standards (sitosterol, sitosterone, α-tocopherol) or by chromatographic data, confirmed by interpretation of MS fragmentation (fatty acids, analyzed as their methyl esters). Some significant MS data of the identified compounds from Dorstenia are listed in Table 2.
A compound detected in D. heringerii, that was identified tentatively as a steroid, requires further MS studies, possibly by using soft ionization techniques (not available Vol. 8, No. 5, 1997 Triterpenes, Steroids and Furocoumarins from Brazilian Medicinal Plants 533  in our laboratory during the present work), since this compound is present at a very low level, which prevents its isolation through usual phytochemical procedures. Figure 2 shows a typical chromatographic profile from dechlorophyllated crude Dorstenia extracts. The extracts obtained from the leaves require a clean-up procedure for removal of the chlorophylls (filtration over active charcoal, described in Experimental), while the crude extracts obtained from the rhizomes could be directly analyzed by HRGC-MS. ln a previous work we have shown that SFE-CO2 (supercritical fluid extraction using CO2) also furnishes rhizome extracts suitable for direct analysis 15 .
The TIC-MS data obtained from the Hex-fraction and from the Terp-fractions were almost the same for the terpenoidal compounds (Fig. 3), and the identification of most of the triterpenes and steroids was possible even in Hex-fraction. As the Terp-fractions required more isolation steps (with greater consumption of solvents and silica gel adsorvent) than the Hex-fractions, the HRGC-MS analysis of a crude dechlorophyllated hexane extract from Dorstenia samples is sufficient to obtain relevant information concerning their chemical composition. The utilization of small amounts of plant material is very important in the case of Dorstenia genus, since the analysis of the rhizomes (the most valuable part of these plants, according to folk medicine indications) leads to the destruction of the plant, and the native Dorstenia population is diminishing in Brazil.

Conclusions
Bergapten, as well as the triterpenes β-amyrin and sitosterol, have already been associated with the snakevenom inactivating action of some Brazilian medicinal plants 5 . Thus, the utilization of Dorstenia plants as antiophidicals may be inferred to be both due to a venom-inactivating action and to analgesic and antiinflamatory properties of the several triterpenes. These results also pose the question if biologically-guided fractionation is an appropriate procedure in the search for the active compounds of medicinal plants such as those of Dorstenia genus, since these active molecules may be minor compounds, with small and/or non-specific pharmacological actions when studied as isolated compounds.
HRGC-MS has proved to be a very powerful tool affording both the separation and the individual characterization of isomers which could not to be separated by conventional PLC procedures. Further to the preliminar report of some of the data herein presented 16 , a paper concerning the analysis of triterpenes from Brazilian Polypodiaceae was published, but with the aid of preparative purification steps and chemical transformations, for obtention of reference samples of some triterpenes not found commercially 17 .
In the screening of medicinal plants from the same genus, which often requires the analysis of many species with similar chemical composition (as in the case of Dorstenia genus), and specially in the particular case of the pentacyclic triterpenes, MS data furnish a fast differentiation among important skeleta, some of them with potential biological interest.  Table 1).