Leaves of T. stans were collected from the Lowveld National Botanical Garden, Nelspruit, South Africa, in summer. The leaves were stored hanging in open mesh bags, used for selling oranges, in the shade with good ventilation to dry. Voucher specimens in the garden herbarium verified the identity of the species.

Powdered dried leaves of T. stans were exhaustively extracted with acetone at a ratio of 10:1, dried under a stream of air in a fume cupboard at room temperature. The extract was stored in preweighed labelled containers to quantify extraction. Acetone was used as an extractant, based on its efficacy with reference to quantity and diversity of compounds extracted, ease of removal, safety and toxicity to assay organisms in the case of bioassay work (Eloff 1998a).

Solvent-solvent extraction is one of the popular techniques used in the preparation of samples for qualitative and quantitative analysis. It is a process of separating one constituent from a mixture by dissolving it into a solvent, in which it is soluble, while the other constituents of the mixture are not, or are at least less soluble (Holden 1999). The acetone crude extract of T. stans was separately suspended in 20% methanol in distilled water in a separatory funnel (2000 mL) and successively partitioned with hexane (Hex), dichloromethane (DCM), ethyl acetate (EtOAc) and n-butanol (BuOH) respectively, in this order. These solvents have varying polarities to extract different constituents of the plant with different polarities. The chemical composition (by thin layer chromatography), the number of antifungal compounds (by bioautography), antifungal activity (by serial microplate dilution) and cytotoxicity against Vero cells of the four partition fractions were determined.

A bioautographic method, developed in our laboratory (Masoko & Eloff 2005), was used to determine activity. Thin-layer chromatography (TLC) plates were prepared and loaded with extract and oleanolic acid dissolved to a final concentration of 10 mg/mL and 1.0 mg/mL respectively. Ten microlitres was applied and the plates were developed in mobile phase, dried and sprayed with conidia of the fungal organisms until they were just wet, incubated overnight and then sprayed with a 2 mg/mL solution of p-iodonitrotetrazolium violet (Sigma^®) (INT), and further incubated overnight or longer at 35 °C in a clean chamber, at 100% relative humidity in the dark. White areas indicated where reduction of INT to the coloured formazan did not take place, due to the presence of compounds that inhibited the growth of tested fungi. To minimise fungal spreading and infection in our laboratory, the bioautograms were sealed in clear plastic envelopes before scanning in for permanent record. R_f values of active compounds were calculated using the following formula:

R_f = distance moved by compound/distance moved by solvent front

The microdilution method developed by Eloff (1998b), and modified by Masoko, Picard and Eloff (2005), was used to determine the minimum inhibitory concentration (MIC) values of T. stans extracts, fractions and isolated compound against 10 plant pathogenic fungi viz. Rhizoctonia solani, Fusarium oxysporium, Penicillum janthinellum, Penicillum expansum, Aspergillus parasiticus, Aspergillus niger, Collectotrichum gloeosporoides, Trichoderma harzianum, Pythium ultimum and Phytophthora nicotiana. This method was applied as follows.

Dried residues of the Tecoma stans DCM extract, fractions and oleanolic acid were dissolved in acetone to a concentration of 10 mg/mL and 1 mg/mL respectively. The Tecoma stans DCM extract and oleanolic acid (100 μL) were serially diluted 50% with water in a 96-well microtitre plate (Eloff 1998b). Fungal cultures were transferred into a fresh potato dextrose broth, and 100 μL of this was added to each well. Amphotericin B was used as the reference antibiotic and positive control, and appropriate solvent blanks were included. As indicator of growth, 40 μL of 0.2 mg/mL of INT dissolved in water was added to each of the microplate wells. The covered microplates were incubated for 1 day at 35 °C and 100% relative humidity. The experiment was performed in triplicate and repeated three times. The MIC was recorded as the lowest concentration of the extract that inhibited antifungal growth after 24 h of incubation. Acetone was used as a solvent and negative control, because fungi are not affected by acetone concentrations lower than 50% (Eloff, Picard & Masoko 2007). The highest acetone concentration the fungi was subjected to in this procedure was 25% in the first well.

The samples were tested for cytotoxicity against the Vero monkey kidney cell line (obtained from the Department of Veterinary Tropical Disease, University of Pretoria). The cells were maintained in minimal essential medium (MEM, Highveld Biological, Johannesburg, South Africa) supplemented with 0.1% gentamicin (Virbac) and 5% foetal calf serum (Adcock-Ingram). To prepare the cells for the assay, cell suspensions were prepared from confluent monolayer cultures and plated at a density of 0.5 × 10³ cells into each well of a 96-well microtitre plate. After overnight incubation at 37° C in a 5% CO₂ incubator, the subconfluent cells in the microtitre plate were used in the cytotoxicity assay. Stock solutions of the plant extract were prepared in growth medium (1 μg/mL–1000 μg/mL). The viable cell growth after 120 h incubation with plant extracts/samples was determined, using the tetrazolium based colorimetric assay (MTT assay) described by Mosmann (1983). The absorbance was measured on a Titertek Multiscan MCC/340 microplate reader at 540 nm test wavelength and reference wavelength of 690 nm. Berberine chloride (Sigma Chemical Company) was used as a positive control. Tests were carried out in quadruplicate and each experiment was repeated three times

The DCM extract obtained from solvent-solvent extraction was subjected to column chromatography on silica gel 60. The column was eluted with 100% DCM and, subsequently, the polarity of the eluting solvent was increased with methanol (MeOH). A volume of 1000 mL of 100% DCM was initially used, followed by solvent mixture of 10% MeOH, 20% MeOH, 30% MeOH, 40% MeOH, 50% MeOH, 60% MeOH, 80% MeOH, all in DCM, and finally the column was eluted with 100% MeOH. A total of nine fractions, each 1000 ml, were collected. Fractions were pooled according to their similarity in behaviour on TLC. The fractions were concentrated and chromatographed to remove impurities to finally obtain the compound. To confirm that the isolation of the antifungal compound present in the extract is not an artefact of the isolation, bioautography was carried out on the isolated compound and the DCM crude extract.

Nuclear Magnetic Resonance Spectroscopy (1H NMR and 13C NMR) was carried out to determine the structure of the compound, using a Varian Inva 500 MHz spectrometer. The isolated compound from T. stans was dried, weighed (15 mg–20 mg) and dissolved in 2 mL of deuterated chloroform into NMR tubes, using a clean Pasteur pipette and analysed in the Department of Chemistry, University of Botswana.

A quantity of 46.67 g of serially extracted acetone extracts of T. stans leaves powder was subjected to solvent-solvent extraction leading to four fractions of increasing polarity viz. hexane, DCM, EtOAc and BuOH. Hexane was the best extractant in terms of quantity extracted with 46.23% (21.58 g) extracted, followed by DCM with 16.75% (7.82 g), EtOAc with 1.79% (0.84 g) and BuOH with 0.64% (0.3 g) respectively. This indicated that these leaves are very rich in non-polar compounds.

The DCM fraction was selected to isolate active compounds by bioassay guided fractionation, based on its activity. Fractions eluted from the DCM extract in open column silica gel chromatography were pooled according to their TLC composition, concentrated and chromatographed to remove impurities which yielded a white precipitate. To examine the activity of the antifungal compounds, bioautography was undertaken of the combined fractions by TLC and the DCM extract that was fractionated and sprayed with P. janthelium suspension (Figure 1).

R_f values of active compounds may help in dereplicating isolation of antifungal compounds from other plant extracts. The R_f value of the extract and isolated compound was 0.08 with BEA (benzene/ethanol/ammonium hydroxide 90/9/1) (Kotze & Eloff 2002) as eluant (Figure 1). This confirms that the compound isolated was probably the antifungal compound present in the crude extract and not an artefact of the isolation process.

The compound was isolated as a white precipitate. NMR spectroscopy identified the compound as oleanolic acid (Figures 2 and 1-A1). Oleanolic acid is a triterpenoid compound present in the leaves and bark of many plant species. This isomer of ursolic acid has been recognised to have hepatoprotective, anti-inflammatory, and antihyperlipidemic properties and has recently been noted for antitumor promotion effects. It has been marketed as an oral drug in China for human liver disorders (www.alibaba.com/product-gs/203129384/Oleanolicacid.html). The compound is well known but literature reveals little about its antifungal activity. It has not been isolated from Tecoma stans before.

Many antifungal compounds are toxic, because several more similar metabolic pathways exist between fungi and mammals than between bacteria and mammals. Therefore there are fewer specific targets that might be addressed. The important question is how the toxicity to the target organism relates to the cellular toxicity. The selectivity index can be defined here as the LC₅₀ in (μg/ml)/MIC in (μg/ml). The higher this value is, the safer the product is to use under controlled conditions (Table 3).

The crude extract had a high selectivity index value of 21 against microorganisms T. harzianum, R. solani, F. oxysporium and P. expansum which means crude extracts can be used against infection of these fungi and pose little toxicity if used under controlled conditions. It is interesting to note that P. expansum, which is a major problem in fruit spoilage, can be controlled by the crude extract at a dosage level with low toxicity. Fusarium spp. which can be a big problem in maize contamination and damping off of seedlings, can also be controlled with little toxicity.

The high selectivity index of 16 and 64 of oleanolic acid against T. harzianum and R. solani, respectively, stood out as the most promising for the use of oleanolic acid against infections, caused by these two fungi, at the correct dose. With many other phytopathogens the extract or oleanolic acid were more toxic to Vero cells than to the fungus.