Antibacterial and antioxidant activities of four kaempferol methyl ethers isolated from Dodonaea viscosa Jacq. var. angustifolia leaf extracts

Fractionation of dichloromethane and acetone fractions obtained by serial extraction from the leaf powder of Dodonaea viscosa Jacq. var. angustifolia (Sapindaceae) resulted in the isolation of four kaempferol methyl ethers. The compounds were identified by spectral data (H NMR, C NMR and MS) as: 3, 5, 7-trihydroxy-4'-methoxyflavone (1); 5, 7, 4'-trihydroxy-3, 6-dimethoxyflavone (2); 5, 7-dihydroxy-3, 6, 4'trimethoxyflavone (santin) (3); and 5-hydroxy -3, 7, 4'-trimethoxyflavone (4) together with 3,4',5,7-tetrahydroxy flavone (kaempferol) (5). Antioxidant potential of the compounds was evaluated using a DPPH spectrophotometric assay, while antibacterial activity was determined using a serial dilution microplate technique. The isolates demonstrated varying degrees of antioxidant and antibacterial activities. Of all the compounds investigated, compounds 1 and 5 demonstrated some antioxidant activity (EC50=75.49±1.76 μM and 35.06±0.85 respectively) but lower than Lascorbic acid (EC50=13.55±0.28 μM) used as a standard antioxidant agent. The minimum inhibitory concentration (MIC) of isolated compounds against Staphylococcus aureus, Enterococcus faecalis, Escherichia coli and Pseudomonas aeruginosa varied from 16 μg/ml to more than 250 μg/ ml. Some structure activity relationships could be established for these compounds. © 2009 SAAB. Published by Elsevier B.V. All rights reserved.


Introduction
Dodonaea viscosa Jacq. var. angustifolia (Sapindaceae) is a medicinal plant used in folk medicine to treat various ailments; the leaves have traditionally been administered to treat the following: sore throat, wounds, fever (Jansen, 1981); piles (Tadese, 1994); fever, malaria, angina, cold, arthritis, sinusitis flu (Lemordant, 1971); and boils, dressing for skin diseases of the head and face (Sandhya et al., 2006;Abate, 1989). Getie et al. (2003) reported that a methanol leaf extract of D. viscosa from Ethiopia demonstrated antibacterial activity against Staphylococcus aureus, Streptococcus pyogenes and Corynebacterium diphtheriae and exhibited antiviral activity against Coxsackie virus B3 and Influenza A virus. However, their quantitative data were obtained by using the agar diffusion technique. This method is not useful to quantify antimicrobial activity of plant extracts for comparative purposes. We therefore decided to reinvestigate the antibacterial activity of D. viscosa leaf extracted with solvents of various polarities by using a microplate serial dilution assay (Eloff, 1998).that would provide MIC data.
Antioxidants are important in maintaining good health and there is a growing interest in the investigation of antioxidant activity of secondary metabolites from medicinal plants for compounds with higher potency and lower toxicities than the synthetic ones currently available (Thabrew et al., 1998). Plants offer a wide range of natural antioxidants due to the structural diversities of their secondary metabolites. Many medicinal plants have now been recognized as sources of natural antioxidant compounds which are mainly phenolic compounds (Wangensteen et al., 2004;Langley-Evans, 2000;Aruoma, 2003).
There has been much phytochemical work done on D. viscosa. We were interested in the interaction between edible insects feeding mainly on this plant on the possibility that the antibacterial compounds could be sequestered by the insect (Teffo, 2006;Teffo et al., 2007), There has been some reports on the antibacterial activity of D. viscosa extracts but we found no publication on the compounds responsible for the antibacterial and antioxidant activities of this widely used medicinal plant. To evaluate the medicinal potential of compounds present in D. viscosa and validate the traditional use of extracts to treat infections, we used bioassay guided fractionation method to isolate and characterize antibacterial compounds. Because plant extracts may also play an indirect role in protecting humans against infections by stimulating the immune system through antioxidant constituents (Craig, 1999); we also isolated antioxidant compounds present in the leaf extracts.

Chemicals
All chemicals used were of analytical grade purchased from Sigma, Merck and Fluka.

Test micro-organisms
Gram-positive micro-organisms used were Staphylococcus aureus (ATCC 29213) and Enterococcus faecalis (ATCC 29212); Gram-negative micro-organisms used were Pseudomonas aeruginosa (ATCC 25922) and Escherichia coli (ATCC 27853). These strains are responsible for nosocomial infections and are recommended by the National Committee for Clinical Laboratory Standards (NCCLS, 1990), Villanova, Pennsylvania, USA for comparing antibacterial activity of different compounds.

Plant collection and identification
Leaves of D. viscosa were collected from Modjadji in Limpopo Province in August 2004 and allowed to dry under the shade. The identification and voucher specimen number of the plant was sorted out and deposited at the National Botanical Institution, Pretoria with voucher specimen no. 1 Gen Spec no. 48310004. The plant was identified as Dodonaea viscosa Jacq. var. angustifolia.

Isolation of the antioxidant and antibacterial compounds
D. viscosa leaf powder (5 kg) was serially extracted with 3 × 5 L of hexane, dichloromethane (DCM), acetone and methanol respectively. This afforded four crude extracts on evaporation of the solvents using a rotary evaporator at 40°C. Qualitative antioxidant assay on TLC plate with 0.2% DPPH in methanol (Glavind and Holmer, 1967) and bio-autography (Begue and Kline, 1972) showed that the DCM and acetone fractions (results not shown) mainly contained the antioxidant and antibacterial compounds.

Fractionation of the DCM extract
The DCM extract (30 g) was fractionated on a silica gel column eluted with hexane (100%); followed by a mixture of hexane/ethylacetate (EtOAc) with an increasing gradient of EtOAc up to 100% and an increasing gradient of methanol (MeOH) up to 100% into EtOAc. The eluent was collected in 50 ml test tubes and analysed by TLC sprayed with vanillinsulphuric acid (Stahl, 1969). The fractions were pooled into 15 fractions based on the TLC profiles and concentrated. The active fractions were detected by MIC, antioxidant analysis and by bio-autography.
2.4.1.1. Purification of pooled fractions 4 + 5. A mixture of fractions 4 and 5 (2.1 g) was fractionated on a silica gel column using an increasing gradient of EtOAc in hexane up to 100%. The eluent was collected in 50 ml test tubes and a total of 80 test tubes were collected. Yellow crystals were formed in test tubes 48-60. The crystals were collected and allowed to dry. This afforded compound (3) (8.0 mg) with R f 0.79 when developed in chloroform/ethylacetate/formic acid (CEF) (10:8:2) solvent system.

Fractionation and purification of the acetone serial extract
The serial acetone crude extract (3.2 g) was fractionated on silica gel as in above into 80 and 50 ml fractions. Fractions 24-26 were combined and purified on Sephadex LH-20 using DCM:EtOAc (2:1). A pure (yellow powder) compound 5 (9 mg) was obtained; with an R f 0.41 in CEF solvent.

DPPH spectrophotometric (quantitative) assay
Antioxidant activity was determined as described by Mensor et al. (2001) and modified by Aderogba et al. (2007). Sample stock solution of each of the isolated compounds (200 µM) from D. viscosa leaf extracts was diluted to a final concentration of 125, 100, 50, 25 and 12.5 µM in MeOH. Ten microlitres of 0.2 mM DPPH MeOH solution was added to 50 µl of sample stock solution of different concentration and allowed to react at room temperature for 30 min in a dark chamber. The blank solutions were prepared with sample solution 50 µl and 20 µl of MeOH only while the negative control was a DPPH solution, 20 µl plus 50 µl MeOH. The absorbance values of the compounds changing from violet to yellow were measured at 515 nm on a microplate reader and converted to percentage antioxidant activity (AA%) using the formula: AA% = 100−f½ðAbs sample −Abs blank Þ × 100=Abs control g (Abs sample is the absorbance of the sample, Abs blank is the absorbance of the blank and Abs control is the absorbance of the control). L-ascorbic acid (vitamin C) diluted to final concentrations of 50, 25, 12.5, 6.25 and 3.125 µM in MeOH was used as a positive control (antioxidant agent).
EC 50 is the concentration of the test sample leading to a 50% inhibition of the DPPH free radicals. Its value was calculated from the separate linear regression of plots of the mean percentage of the antioxidant activity against concentration of the test compounds (µM) obtained from three replicate assays.

Statistical analysis
The results were expressed as mean ± SEM (standard error of mean). The software for regression plots (SigmaPlot® 2001, SPSS Science) was used.

Antibacterial activity assay
To determine the minimum inhibitory concentration (MIC) values, the microdilution assay as described by Eloff (1998) was followed. Gentamicin (SIGMA Aldrich, Germany) was used as a positive control; acetone and water were used as negative controls. Isolated compounds were tested at an initial concentration of 1000 µg/ml (1 mg/ml) and the experiment was done in triplicate. The following test micro-organisms were used: Gram-positive-S. aureus and E. faecalis; Gram-negative-P. aeruginosa and E. coli. Briefly, 100 µl of sterile distilled water was added to 96-well microtitre plates followed by the addition of 100 µl of 1000 µg/ml isolated compounds and serially diluted two fold after which 100 µl of the above mentioned 4 actively growing test micro-organisms were added to each microtitre plates in all the wells to give the final volume of 200 µl. The prepared microtitre plates were sealed so that they do not dry and incubated overnight at 37°C in 100% relative humidity. After overnight incubation an indicator of bacterial growth, 40 µl of 0.2 mg/ml p-iodonitrotetrazolium violet (INT) (SIGMA Aldrich, Germany) dissolved in water was then added to all the microtitre plate wells and incubated for a further 30 min-2 h. Bacterial growth was indicated by the red colour and in the wells that were colourless or the colour intensity decreased the indicated inhibition of growth. The MIC values were then recorded as the lowest concentration at which a decrease in red colour is apparent compared to the next dilution after approximately 24 h of incubation.

DPPH qualitative assay
All the isolated compounds except compound 4 bleached the DPPH solution (0.2% in MeOH) purple colour in the TLC autography qualitative assay. Quantitative assay revealed varying degrees of antioxidant activity of the compounds. Kaempferol with free phenolic -OH groups (without any methoxyl substitution) was the most active. Compounds 1-4 are kaempferol derivatives with different methoxylation patterns. Of the four derivatives, compound 1 4'-OMe derivative was the most active. Compound 1 has the same oxygenation pattern as the remaining derivatives but with only C-4'-methoxylation. The presence of these free-OH groups could be linked to higher antioxidant activity of the compound compared to other derivatives with more methoxylated groups. Methylation of free-OH groups in flavonoids substantially reduces the antioxidant activity of the compounds (Op de Beck et al., 2003). In line with this observation, compound 2 with 3-OMe and 6-OMe and also compound 3 with 3-OMe, 4'-OMe and 6-OMe did not inhibit 50% of DPPH free radical hence their EC 50 was higher than the highest concentration tested. The weak activity exhibited by these compounds could be linked to the methylation of some of their phenolic-OH groups (Martini et al., 2004;Op de Beck et al., 2003). At the highest concentration (200 µM) tested, compounds 2 and 3 had 19.06 and 8.23% antioxidant activities respectively. The LD 50 values were therefore substantially higher than 200 µM. It appears that methylation of 3-OH in both compounds reduced activity significantly when compared with compound 1 with free 3-OH. Compound 5 has the same oxygenation pattern as its four derivatives (1-4). All its -OH groups are free and this significantly enhances its antioxidant activity. This observation is in agreement with the earlier structural activity study (Op de Beck et al., 2003;Rice-Evans et al., 1996;Saskia et al., 1996). The effect of 6-OMe on activity is not clear. Compound 4 did not demonstrate activity in the qualitative assay apparently due to extensive methoxylation (3-OMe, 4'-OMe and 7-OMe). This observation suggests that 5-OH does not play a strong role in antioxidant activity of the investigated compounds. This is also in line with the earlier report on flavonoids (Rice-Evans et al., 1996). Our results indicate that only compounds 1 and 5 had strong activity (EC 50 = 75.49 ± 1.76 µM and 35.06 ± 0.85 respectively) but it was still less than that of the positive control L-ascorbic acid (EC 50 = 13.55 ± 0.28 µM) ( Table 1). The EC 50 values obtained from the regression plots (SigmaPlot ® 2001, SPSS Science) showed good correlation coefficient (r 2 ≥ 0.947).

Antibacterial assay
The MIC values for the isolated compounds on the four bacteria varied from 23 to more than 250 µg/ml (Table 2). Compound 5 was in general the most active against all the test organisms with MIC values between 16 and 63 µg/ml. Compound 2 was the second most active and was the only isolated compound with reasonable activity against P. aeruginosa. Compound 1 had the best activity (23 µg/ml) against E. faecalis, but had very low activity against the other organisms. Compound 3 had poor activity against all the test organisms (MIC between 63 and 125 µg/ml) and compound 4 had no activity at the highest concentration tested (250 µg/ml).
The antimicrobial activity of the isolated compounds against some of the tested organisms could be linked to varying degrees of methylation of the -OH groups of these compounds. This is consistent with earlier structure activity related study on the antimicrobial activity of flavonoids (Cushnie and Lamb, 2005).
The activity of the different pathogens was not associated with the Gram-positive-Gram-negative classification indicating that the effect was not cell-wall related. In all cases a hydroxyl was required at C-7 for antibacterial activity. A hydroxyl was required at C-4' for activity with P. aeruginosa but this was apparently not the case with S. aureus. It does not seem as if substitution at C-6 plays an important role.
No clear correlation between antioxidant and antibacterial activities of these compounds was apparent. Flavonoids are known for substantial antimicrobial activity (Alcaraz et al., 2000). Low antibacterial activity demonstrated by some of these compounds could be ascribed to methylation of the phenolic -OH groups in our compounds.

Conclusion
The compounds responsible for the antibacterial and antioxidant activity in D. viscosa leaf extracts are all flavonoids related to kaempferol. Some conclusions on the relationship between structural and biological activity could be drawn that supported earlier reports. The data also confirm the rationale for the traditional use of D. viscosa extracts to treat infections. Compound 5 was in general the most active against all the test organisms with MIC values between 16 and 63 µg/ml. Compound 2 was the second most active and was the only isolated compound with reasonable activity against P. aeruginosa. Compound 1 had the best activity (23 µg/ml) against E. faecalis, but had very low activity against the other organisms. Compound 3 had poor activity against all the test organisms (MIC between 63 and 125 µg/ml) and compound 4 had no activity at the highest concentration tested (250 µg/ml).