Antimicrobial flavonoids and diterpenoids from Dodonaea angustifolia

a Department of Chemistry, School of Physical Sciences, University of Nairobi, P. O. Box 30197-00100, Nairobi, Kenya b Department of Pharmaceutical Chemistry, School of Pharmacy, University of Nairobi, P. O. Box 19676-00202, Nairobi, Kenya c Center for Biotechnology and Bioinformatics, University of Nairobi, P. O. Box 30197-00100, Nairobi, Kenya d Institut für Chemie, Universitat Potsdam, P. O. Box 60 15 53, D-14415 Potsdam, Germany


Introduction
Dodonaea angustifolia L.f. (Sapindaceae) is a medium-sized shrub or small tree 0.5 to 7.5 m high with characteristic glossy green leaves covered by sticky surface exudates. The new leaves are stickier than the old ones which have characteristic rough, and sand papery texture (Beentje, 1994). D. angustifolia is an extremely variable species throughout its natural range; in Australia, Africa, Asia and South America, but many distinctive populations have been described as separate species (Beentje, 1994); in Kenya it is reported to exist along with Dodonaea viscosa (Beentje, 1994). D. angustifolia is used in traditional medicine to treat a number of ailments including tuberculosis and pneumonia (Watt and Breyer-Brandwijk, 1962;Cano et al., 1980). The leaf surface exudates (up to 13% dry leaf weight) of D. angustifolia constitute of mainly methylated flavonoids in a clerodane and labdane diterpenoid milieu (Ghisalberti, 1998). There is great geographical variability in the composition for this substance both in quality and yields of each component in Dodonaea populations as observed from their thin layer chromatography (TLC) profiles. This created the interest to study the phytochemistry of D. angustifolia from Ngong forest to compare with the Voi population investigated earlier.
Previous phytochemical investigations have shown that lipophilic flavonoids, with structural features akin to those isolated from D. angustifolia surface exudates in this study, display antimicrobial activity due to their ability to penetrate biological membranes (Harborne, 1983). It was also suggested that the hydroxyl groups on flavonoids may interact with biological structures through hydrogen bonding, and that the relative positions of the hydroxyl group on the flavone skeleton is important in determining antimicrobial activity (McClure, 1975). Here the antimicrobial activities of the exudates and constituents from D. angustifolia collected from Ngong forest, Kenya is reported. South African Journal of Botany 91 (2014) 58-62 2. Materials and methods

General experimental procedures
Column chromatography was carried out using Merck silica gel 40 (70-230 mesh) and Sephadex LH-20. Analytical TLC and preparative TLC were done using Merck pre-coated 60 F 254 and Merck 60 PF 254 respectively. 1 H NMR (500 MHz) and 13 C NMR (125 MHz) were run on AVANCE-500 (Bruker) machine. Heteronuclear multiple quantum coherence (HMQC) and heteronuclear multiple bond correlation (HMBC) spectra were acquired using standard Bruker software. Electron Ionization Mass Spectroscopy (EIMS) spectra were recorded on 70 eV, on SSQ 710 MAT mass spectrometer. UV values were obtained using SP8 150 ultraviolet visible (UV/VIS) spectrophotometer. Melting points were recorded using a Gallenkamp melting point apparatus with capillary tubes. Inhibition zone diameters were read using a Wezu electronic digital caliper (Messzenge GmbH, Germany).

Plant collection and identification
The fresh leaves of D. angustifolia were collected from Ngong forest (6 km from Nairobi city center) in December, 2010. The plant material was identified by Mr. S.G. Mathenge of the University of Nairobi Herbarium, School of Biological Sciences (SBS), where voucher specimen (Mathenge-012/December, 2010) is deposited.

Extraction and isolation
Extraction of the surface exudates of the leaves of D. angustifolia from Ngong forest (450 g), was done by successive dipping fresh aerial parts into fresh portions of acetone for short periods (less than 15 s) thus avoiding the extraction of the internal tissue components. The extracts obtained were filtered under pressure and concentrated in vacuo using a rotary evaporator to yield 52 g of crude extract. A portion of the crude extract (45 g) was dissolved in 2% dichloromethane (CH 2 Cl 2 ) in methanol (MeOH) and adsorbed on silica gel (45 g). The adsorbed silica gel was loaded onto a column packed with silica gel (450 g) under 50% CH 2 Cl 2 in normal hexane (n-C 6 H 12 ). Separation was carried out by stepwise gradient elution with mixtures of CH 2 Cl 2 in n-C 6 H 12 and then with CH 2 Cl 2 containing increasing amounts of MeOH. The total number of fractions collected in the main column was 20 of 200 ml each which was subsequently combined based on the similarities of their TLC profiles (50% CH 2 Cl 2 in n-C 6 H 12 and 5% MeOH in CH 2 Cl 2 ) into only 7 fractions. Yellow armophous solids of 3,5-dihydroxy-4′,7dimethoxyflavone (2, 30 mg) precipitated out of the fraction eluted with 50% CH 2 Cl 2 in n-C 6 H 12 . The solids were filtered, dried and weighed. The fraction eluted with 60% CH 2 Cl 2 in n-C 6 H 12 was purified further by column chromatography using silica gel, eluting with increasing gradient of CH 2 Cl 2 in n-C 6 H 12 up to 100% and then MeOH in CH 2 Cl 2 up to 5%. The eluants were collected in 100 ml Erlenmeyer flasks leading to 20 fractions. TLC (2% MeOH in CH 2 Cl 2 ) analysis of fraction 6-13 of this minor column showed similar profiles and therefore were combined. Removal of the solvent of the combined fractions and recrystallization (80% CH 2 Cl 2 in n-C 6 H 12 ) afforded yellow needles of 5-hydroxy-3,4′7trimethoxyflavone (1, 204 mg). Santin (3, 330 mg) precipitated out of the fraction eluted with 90% CH 2 Cl 2 in n-C 6 H 12 . The solids were filtered, dried and weighed. Purification of the mother liquor of the above fractions using PTLC (silica gel, 100% CH 2 Cl 2 multiple development) afforded (ent-3β,8α)-15,16-epoxy-13(16),14-labdadiene-3,8-diol (11, 10 mg). The fraction eluting with 1% MeOH in CH 2 Cl 2 after column chromatography on Sephadex LH 20 (MeOH in CH 2 Cl 2 ; 1:1) resulted to 2 minor fractions of 100 ml of each. Fraction 1 after filtration yielded white crystals of dodonic acid (9, 500 mg), while that eluting with 2% MeOH in CH 2 Cl 2 after column chromatography on Sephadex LH 20 (MeOH in CH 2 Cl 2 ; 1:1) gave pinocembrin (8, 120 mg) in the third eluant of 100 ml each. The fraction eluted with 3% MeOH in CH 2 Cl 2 afforded kumatakenin (4, 200 mg) from fractions 5-9 and rhamnocitrin (13, 184 mg) from fractions 12-15 (100 ml each) of a silica gel column eluted with increasing gradient of CH 2 Cl 2 in n-C 6 H 12 and then MeOH in CH 2 Cl 2 . The fraction eluted with 4% MeOH in CH 2 Cl 2 after purification using column chromatography on Sephadex LH 20 (MeOH in CH 2 Cl 2 ; 1:1) afforded 2β-hydroxyhardwickiic acid (10, 778 mg), rhamnocitrin (5, 60 mg) and isokaempferide (6, 40 mg) in the first, second and third fractions (100 ml each) respectively. The TLC analysis of the fraction eluted with 5% MeOH in CH 2 Cl 2 indicated only two spots, one of which was minor and therefore was subjected to PTLC (silica gel, 100 ml of 2% MeOH in CH 2 Cl 2 multiple development) to yield 3,4′,5,7-tetrahydroxy-6-methoxyflavone (7, 60 mg).

In vitro antimicrobial assay
Evaluation of antimicrobial activity of extracts and pure compounds was accomplished using the agar well-diffusion method (Bauer et al., 1966). The extracts and pure compounds were tested for activity against three strains of bacteria; Escherichia coli (American Type Culture Collection, ATCC25922), Staphylococcus aureus (ATCC29737) and Bacillus pumilus (local strain) and a local strain of fungus, Sacchromyces cerevisiae. The bacterial test organisms were cultured on tryptone soya agar and the fungi on Saboraud's dextrose agar. The nutrient agar was inoculated uniformly with standardized test organisms. Reservoir wells were formed by cutting out cylindrical plugs from the solidified nutrient agar at equidistant points, using a sterile cork borer, to produce wells (diameter 5 mm, depth 2 mm). The wells were each filled with 50 μl of the stock solutions in dimethylsulfoxide (DMSO): 50 mg/ml (2500 μg/well) for plant extracts and 10 mg/ml (500 μg/well) for pure compounds. The standard drugs gentamicin 0.3 mg/ml (15 μg/well), nystatin 0.25 mg/ml (12.5 μg/well) used as the antibacterial and antifungal positive controls respectively, while the solvent, DMSO, used as the negative control were similarly introduced into their respective wells. For determination of minimum inhibition concentration (MIC) of the extract and pure compounds, serial dilution of the stock solution was carried out resulting in concentration range from 625-2500 μg/well for the extract and 3.9 to 500 μg/well for each compound. All determinations were carried out in triplicate. The inoculated petri-dishes with test solutions in wells were allowed to diffuse for 30 min before overnight (18 h) incubation at 37°C and 25°C for bacteria and fungi, respectively. The antimicrobial activity was recorded as the diameter (mm) of the clear circular zone of inhibition surrounding the agar well after incubation. The MICs of the test microorganisms was similarly determined by the agar welldiffusion method and is defined as the lowest concentrations of the compounds that visually showed no growth compared with growth in control wells.
The 1 H and 13 C NMR spectra of compound 3 were similar to those of compound 12 except that ring B was mono-substituted as shown by the 1 H NMR spectrum which displayed an AA′XX′ system resonating at δ Η 8.11 and 7.12 (each 2H, d, J = 8.8 Hz). A singlet at δ 6.60 (1H) was assigned to H-8 on the tri-substituted ring A. The 1 H NMR further displayed signals for three methoxyls at δ 3.91 3.87 and 3.86, two of which were di-ortho substituted (δ 60.0 and 59.6 as shown in the 13 C NMR spectrum). The location of one of the methoxyl group was assigned to C-4′ due to NOE interaction between the aromatic protons at C-3′/C-5′ (δ Η 7.12) with the methoxyl group at δ 3.91. The spectral data of this compound is in close agreement with what has been reported in literature (Barbera et al., 1986) for santin (3), especially the 13 C NMR for C-6, C-8 and C-9 showing identical ring A. Hence, the compound was identified as 5,7-dihydroxy-3,6,4′-trimethoxyflavone (3, santin) that was previously isolated from the leaf extract of three Dodonaea species  (Sachdev and Kulshreshtha.,1984).
The 13 C NMR attached proton test (APT) spectrum of compound 14 (m/z 332, C 20 H 28 O 4 ) corroborated the presence of two methyls, seven methylenes and six methines and five quaternary carbon atoms. The fragmentation peaks at m/z 95 and 81 suggested the presence of furan ring with an alkyl chain (Spanevello and Vila, 1994). These results indicated that compound 14 is a diterpene with a furan ring. The 13 C NMR spectrum exhibited signals at δ 19.1 and 16.7 due to tertiary and secondary methyl groups at C-9 and C-8, respectively, in agreement with the data of compounds having both of these substituents as alpha (α) on a trans-clerodane skeleton (San-Martin et al., 1986;Manabe and Nishino, 1986). The 1 H NMR spectrum of compound 14 displayed broad singlets at δ 6.28, 7.26 and 7.37 attributed to the H-14, H-16 and H-15 protons of the β substituted furan ring. The presence of an α,β-unsaturated γ-lactone moiety is evident in this compound from the 1 H NMR signals at δ 6.63 (dd, J = 7.4, 2.0 Hz) the olefinic β-protons, δ 4.30 (d, J = 8.1 Hz) and 3.92 (dd, J = 8.0, 2.0 Hz) for oxymethylenes at C-19. The corresponding carbons in the 13 C NMR for the lactone moiety appeared at δ 169.3 for CfO; δ 71.7 for the oxymethylene and the olefinic carbons resonated at δ 135.8, 138.4. The methylene protons at C-19 had an AB spin system. The pro-19S diastereotopic proton of this group (δ 3.92) was also ω-coupled ( 4 J = 2.0 Hz) with the H-6β proton, indicating an α-axial orientation for C-19 (Bruno et al., 1981;Esquivel et al., 1986;Stapel., 1980). In the 1 H NMR the pro-19R proton resonated  at δ 4.30 which is in agreement with the lack of a substituent at C-7 position in this compound (Herz, 1977;Zdero et al., 1989;Esquivel et al., 1988). In addition, a three proton doublet at δ 0.87 (J = 6.6 Hz) was attributed to the secondary methyl and a three proton singlet at δ 0.79 attributed to the tertiary methyl group typical of clerodane-type diterpenes. The correlation spectroscopy (COSY) experiment showed coupling between the methyl at δ 0.87 and the H-8 proton at δ 1.63. Furthermore, the COSY experiment showed coupling between the protons at δ 6.28 and δ 2.42 assigned to H-12 and between the proton at δ 7.26 and the H-12 methylene protons at δ 2.42 and δ 2.20. There were also cross peaks from the protons at δ 6.28, 7.26 and 7.34. The structure of 14 was confirmed from the HMBC experiment, with the olefinic proton at δ 6.63 showing correlations to C-4 (δ 138.9), C-5 (43.3), C-2 (δ 27.7). Similarly, the proton at δ 6.28 assigned to H-14 showed cross peak correlations to the C-13 (δ 126.7), C-15 (δ 144.0), C-16 (δ 139.7) and C-12 (δ 18.1*/δ 19.1*). The relative configuration was established on the basis of nuclear overhauser and exchange spectroscopy (NOESY) cross peaks observed between H-20/H-17 and H-20/H-19 (the two protons). However, there were no cross peaks between H-20/H17/H-19 (the two protons) and H-10. These results can be rationalized only if C-20, C-17, C-19 are on the same face of the molecule and H-10 on another face of the molecule. All the data are in agreement with compound 14 being hautriwaic acid lactone previously isolated from D. viscosa (Hsu et al., 1971).

Bioactivity analysis
The exudates of D. angustifolia exhibited antimicrobial activity against Gram-negative (E. coli), Gram-positive (S. aureus and B. pumilus) bacteria and the fungus S. cerevisiae (Table 1). The isolated compounds were also tested and showed varied antimicrobial activities against the Gram-positive bacteria S. aureus, B. pumilus and the fungus S. cerevisiae but were inactive against the Gram negative bacteria E. coli. Interestingly, the flavonoid 3,4′,5-trihydroxy-3′,7-dimethoxyflavone (12) isolated from the surface exudates of Senecio roseiflorus (Omosa et al., 2013) showed good activity against E. coli with MIC b 31.25 μg/well. Rhamnocitrin (5) with a similar oxygenation pattern; except for the absence of a methoxy group at 3′ position; was inactive, indicating the importance of a 3′-methoxy (OMe) group for activity against E. coli in this type of flavonoids. All the 3-methoxyflavones tested; 1, 3, 4 and 15 (Omosa et al., 2010) were inactive against S. aureus even at the highest concentration tested, 500 μg/well. For good activity against S. aureus, hydroxyl group at C-3 position in the flavone skeleton, as encountered in compounds 5 and 12, which exhibited activity with MIC b 62.5 μg/well, seems to be important. Santin (3) was the most active flavone against S. cerevisiae with an MIC b 3.9 μg/well while penduletin (16) was less active with an MIC b 125 μg/well. The presence of 5,7-dihydroxy substitution seems to substantially improve activity of flavones against the fungus S. cerevisiae. This is evident from the observation that compound 7, having 5,7-dihydroxy substituents, was more active than compound 16 with 5,4′-dihydroxy substituents. The presence of two hydroxyl groups and a minimum of two methoxy groups, appears to be necessary for good activity while complete methylation, even if C-5 hydroxy is free, negates activity as encountered in 5-hydroxy-3,7,4′-trimethoxyflavone (1).
The three diterpenoids 9, 10 and 15 have the same carbon skeleton and one hydroxyl substituent each, but differ in the position of this group around the ring; the hydroxyl group being at C-6, C-2, and C-19 in compounds 9, 10 and 15, respectively. The activities of these diterpenoids are dependent on the position of the hydroxyl group, vis a vis OH-6 (9) N OH-2 (10) N OH-19 (15). Compound 1 and 4 exhibited no antimicrobial activity against the three strains of bacterial and one fungal strain. Santin (3) and hautriwaic acid lactone (9) which showed good antifungal activity with MIC values b 3.9 and 7.8 μg/well respectively exhibited poor antibacterial activity with MIC value N 250 μg/well. The results on antibacterial activities of compounds 1 and 3 are consistent with those reported by Teffo et al. (2010). In general no marked changes in inhibition zones were observed relative to dilution. The compounds that were isolated in small quantities were not tested for activity. "-" not active. a Inhibition zone in mm.