Saponins are a major family of secondary metabolites that occur in a wide range of plant species. Bioassay - guided fractionation of extract of the leaves of Polyscias fulva led to the isolation of three known saponins named, 3-O-α-L-arabinopyranosyl-hederagenin (1), 3-O-[α-L-rhamnopyranosyl(1-2)-α-L-arabinopyranosyl]-hederagenin (2) and 3-O-[rhamnopyranosyl-(1→2)-xylopyranosyl]-Olean-12-en-28-O-[rhamnopyranosyl-(1→4)-glucopyranosyl-(1→6-glucopyranosyl] ester) (3). Leaves of the plant were collected from Kakamega rain forest in Kenya, dried under shade and ground into fine powder and extraction was done using methanol. The methanol extract was subjected to column chromatography and the fractions purified using preparative high performance liquid chromatography (HPLC). The bioactivity of the pure compounds was done using disc diffusion method. The three compounds exhibited moderate activities against Gram positive bacterium (Staphylococcus aureus ATCC25922) and Gram negative bacterium (Klebsiella pneumoniae ATCC13883). Compound 1 was found to be the most active against K. pneumoniae (8.00±1.00 mm) and S. aureus (10.00±1.73 mm) followed by compound 2 with inhibition zones of 7.66±0.57 and 7.33±0.57 mm against K. pneumoniae and S. aureus, respectively. Compound 3 was the least active against both K. pneumoniae (7.33±0.57 mm) and S. aureus (7.00±1.00 mm). The results obtained indicate that compounds 1, 2 and 3 exhibit potential as possible sources of antibacterial agents.
Saponins are a major family of secondary metabolites that occur in a wide range of plant species (Hostettman et al., 1995). They are naturally occurring glycosides characterized by their strong foam forming property in aqueous solution (Gl-stndag and Mazza, 2007; Man et al., 2010; Negi et al., 2013. Various members of this important family of plant secondary metabolites are exploited commercially for a variety of purposes including drugs and medicines, precursors for hormone synthesis, adjuvants, foaming agents, sweeteners, taste modifiers and cosmetics. Since many saponins have potent antimicrobial activity, the natural role of these molecules in plants is likely to be in conferring protection against attack by potential pathogens (Morrissey and Osbourn,1999; Negi et al., 2013).
Bacterial infections constitute a major public health problem in developing countries where the high cost of antibiotics makes them unaffordable to the majority of the population (Adwan et al., 2010). Population increase, traditional inadequate supply of drugs, prohibitive cost of treatments, side effects of several synthetic drugs and development of resistance to currently used drugs for various diseases have led to increased emphasis on the use of plant materials as a source of medicines for a wide variety of human ailments (Sharma and Manu, 2013).
Among these plants is P. fulva which belongs to the family Araliaceae, a medium size and fast growing deciduous tree of the tropical forests of sub-Saharan Africa (Bedir et al., 2001). In Kenya, the plant grows around Elburgon, North of Mt. Elgon, West of Mt. Kenya, North of the Nandi forests and wetter highlands areas like Kakamega forest (Orwa et al., 2009). The plant has been used traditionally to cure many diseases, for instance, in west and central Africa it is traditionally used to treat malaria, fever and mental illness (Ndaya et al., 2002). It has been reported that the Araliaceae family and precisely the genus Polyscias are considered as a rich source of triterpene glycosides (Gopalsamy et al., 1990).
This study reports the isolation and characterization of three compounds of triterpene saponins from the leaves of P. fulva collected from Kakamega forest in the western part of Kenya.
Plant material
Leaves of P. fulva were collected from Kakamega rain forest which is at an altitude of 0°10’ to 0°21’N 34°58’ E. The plant materials were then taken to the Centre for Herbal Research at Egerton University, Kenya where they were dried under shade for three weeks.
Extraction of phytochemicals
Leaves of P. fulva were dried under shade and at room temperature to prevent the loss of labile compounds and to retain their natural active compounds. The plant materials were turned over periodically during drying to avoid growth of moulds. The materials were ground separately to fine powder using a Thomas-wiley mill model 4. Six kilograms of the P. fulva ground materials were soaked in methanol at room temperature for 24 h with periodical shaking. The contents were then filtered through Whatman no. 1 filter paper and the filtrate was concentrated in vacuum at 50°C using Buchi Rotavapor R-205 rotary evaporator. The methanol crude extracts were placed in the fume hood to total dryness.
Preparative high performance liquid chromatography
The methanol extracts of P. fulva were purified using preparative high performance liquid chromatography equipped with uv-vis detector. The stationary phase used was C-18 column (250 mm by20 mm, 10 um). Anisocratic mobile phase consist of acetonitrile: Water (5:95 v/v) was delivered at a flow rate of 15.000 ul/min and the elution profiles were read at different wavelengths. The methanol extract of P. fulva yielded three major fractions namely 1, 2 and 3. The compounds 1, 2 and 3 were divided into two portions each; one portion of each compound was used for 1 and 2D high field NMR spectroscopy and mass spectroscopy while the other portions were subjected to assays against selected bacteria.
Disc diffusion assay of pure compounds
The disc diffusion method for antibacterial susceptibility testing was carried out according to the standard method by Zaidan et al. (2003). The pure extracts were screened for antibacterial activity against S. aureus and K. pneumoniae. Nutrient agar mixed with bacteria at a concentration of 1×106 cfu/ml were poured in Petri dishes and allowed to cool. The plant extracts equivalent to 1 mg/ml, dissolved in methanol were applied to sterile paper discs (6 mm diameter). The solvent were then allowed to evaporate and the discs deposited on the surface of the inoculated agar plates and incubated for 24 h at 37°C. Zones of inhibition were measured in millimeter after 24 h of growth. The negative control used in this experiment was 1% dimethyl sulfoxide (DMSO) whereas 30 µg/disc chloramphenicol discs were used as the positive control. All tests were performed in triplicates.
Data analysis
Mean inhibition zones were calculated and equality of means was analyzed using one-way analysis of variance (ANOVA). Tukey’s Honestly Significant Difference (HSD), a Post-Hoc Analysis, was used to determine if there was any significant difference between the means of the isolates. Data analysis was performed using R statistical software version 3.3.1.
Nuclear magnetic resonance (NMR) spectroscopy
The 1H, 13C, DEPT, HSQC, COSY and HMBC NMR spectra were recorded on the Bruker Advance 500 MHz NMR spectrometer at the Technical University of Berlin, Germany. The readings were done in DMSO and chemical shifts assigned by comparison with the residue proton and carbon resonance of the solvent. Tetramethylsilane (TMS) were used as an internal standard and chemical shifts were given as δ (ppm). The structures were then simulated using ACD NMR manager program to obtain the chemical shifts of proton. The off- diagonal elements were used to identify the spin - spin coupling interactions in the 1H-1H COSY (Correlation spectroscopy). The proton-carbon connectivity, up to three bonds away, was identified using 1H-13C Heteronuclear Multiple Bond Correlation (HMBC) spectrum. The 1H-13C Heteronuclear Single Quantum Coherence (HSQC) spectrum were used to determine the connectivity of hydrogen to their respective carbon atoms.
Mass spectrometry
Mass spectra of the compounds were recorded on FinniganTripple Stage Quadrupol Spectrometer (TSQ-70) with electron spray ionization (ESI) method in negative and positive ion mode. Thermo XcaliburQual computer software was used in analysis of the mass chromatograms.
The methanol extract was subjected to extensive spectroscopic studies such as 1H NMR, DEPT NMR and mass spectrometry as described in the methodology resulting in the elucidation of three triterpene glycosides (saponins; 1-3).
Compound 1 (10.27 mg) was obtained as a dark brown oily substance (Figure 1). The 1D and 2D NMR spectral data of compound 1 are summarized in Table 1. The mass spectral data of compound 1 gave a molecular ion peak at m/z 605.40 corresponding to its (M+H)+ ion suggesting the molecular formula as C35H56O11. The 1H-NMR spectra of compound 1 showed signals of six methyl groups at δH 0.58 (H-24), 0.88 (H-25), 0.71 (H-26), 1.10 (H-27), 0.86 (H-29), 0.87 (H-30), an olefinic group at δH 5.16 (H-12), which were characteristics signals for the oleanane skeleton with a hydroxyl group at C-23. These signals indicate a pentacyclic structure hence this compound was identified as an olean-12-ene type pentacyclic triterpene and this was confirmed by comparison of its NMR data with those of known olean-12-ene type derivatives (Maillard et al., 1992; Mahato and Kundu, 1994; Beaudelaire et al., 2016). Additionally, 1H NMR spectra of compound 1 also showed the presence of anomeric protons at δ 4.19 (H-1’) indicating the presence of a sugar in its structure.
DEPT NMR spectra displayed a total of 27 carbons which consisted of two carbons of a trisubstituted double bond, one anomeric carbon, six methyl groups, 12 methylene groups, nine methine groups and the other missing seven signals were found to be quaternary carbons. The presence of a signal of a carbonyl ester group at δ 173.0 (C-28) suggested the compound as oleanolic acid (Onoja and Ndukwe, 2013).
The HSQC spectrum was used to assign protons directly attached to carbon atoms. This spectrum showed correlation between proton δH - 0.85/1.49 (C-1), 0.99/1.65 (C-2), 3.49 (C-3), 1.51(C-5),1.17 (C-6), 1.19/1.43 (C-7), 1.18 (C-9), 1.48 (C-11), 5.16 (C-12), 1.81 (C-15), 1.72 (C-16), 2.74 (C-18), 1.05/1.61 (C-19), 2.28 (C-21), 1.61 (C-22), 3.08/3.41 (C-23), 0.58 (C-24), 0.88 (C-25), 0.71 (C-26), 1.10(C-27), 0.86(C-29) and 0.87 (C-30). Additionally, one anomeric signal at δ 4.19 (H-1’) was observed giving HSQC correlations with one anomeric carbon at δ 105.2.
The proton - proton COSY correlations for compound 1 were also determined. COSY spectrum gave information on the correlation between protons attached to adjacent carbon atoms. The protons H-11 (δH - 1.48) correlated with proton H-12 (δH - 5.16) and proton H-18 (δH - 2.74) correlated with proton H-19 (δH - 1.05/1.61).
The NMR values for all the protons and carbons were assigned on the basis of 1H-1H COSY, HSQC and HMBC experiments. The attachment of the sugar molecule at position C-3 of the aglycone was established by the HMBC correlation between the δH 4.19 (H-1’) and δc-80.25 (C-3). The 1H NMR spectrum of the glycone portion showed the presence of three oxymethine protons at δH- 3.30, 3.60, 3.31 together with a methylene proton at 3.32/3.65 for H-5’ and anomeric proton at δH - 4.19. The sugar molecule was therefore identified as Arabinose (Joshi et al., 1992; Njateng et al., 2015). Thus, with all the correlations considered, the compound was identified as a monosaccharide triterpenoid saponin of olean-12-en-28-oic acid aglycone with the molecular formula C35H56O8. This compound has been isolated before from the stem bark of the same plant (P. fulva) and was established to be 3-O-α-L-arabinopyranosyl- hederagenin (Joshi et al., 1992; Njateng et al., 2015), however, it is the first time to be isolated from the leaves of the same plant.
Compound 2 was also obtained as a dark brown oily substance with a mass of 30.81 mg (Figure 1). Its molecular formula was established as C41H66O12. The aglycone region in the DEPT NMR spectra showed great similarity to that of compound 1. The six sp3 hybrid carbon signals at δc 13.4, 16.0, 17.3, 23.8, 26.0 and 33.2, and the two sp2 hybrid carbon signals at δc 122.0 and 144.0 (Onoja and Ndukwe, 2013) together with the information from 1H NMR analysis (six methyl proton singlets at δH 0.57, 0.70, 0.87, 0.87, 0.88 and 1.10 and a vinyl proton at δH 5.16) indicated that the aglycone possesses an olean- 12-ene skeleton.
The HSQC correlations showed correlation between protons absorption at δH-1.49 (C-1), 0.98/ 1.66 (C-2), 3.50 (C-3), 1.52 (C-5), 1.38 (C-6), 1.15(C-7), 1.19 (C-9), 1.91 (C-11), 5.16 (C-12), 1.81 (C-15), 1.72 (C-16), 2.74 (C-18), 1.03 (C-19), 1.11/2.28 (C-21), 1.01 (C-22), 3.09/3.31 (C-23), 0.57 (C-24), 0.87 (C-25), 0.70 (C-26), 1.10 (C-27), 0.88 (C-29) and 0.87 (C-30). The presence of six methyl signals were characteristics signals for the oleanane skeleton with a hydroxyl group at C-23. Additionally, anomeric protons signals in NMR spectrum at δH 4.34 and δH 5.06 together with carbon signals at δC 100.38 and 103.40 in the DEPT NMR data suggested that compound 2 was a glycoside with two sugar units.
The DEPT NMR spectrum showed a total of 14 methine (CH) signals at δc- 41.2, 46.6, 47.5, 66.2, 68.5, 70.8, 70.9, 72.5, 73.3, 74.6, 79.7, 100.3, 103.4 and 122.0 and a total of 12 methylene carbons (CH2) signals absorption at δc- 17.5, 23.0, 23.3, 25.7, 27.6, 32.3, 32.5, 33.7, 38.6, 46.1, 62.9 and 64.8. It also showed a total of seven methyl (CH3) signals at δc- 13.4, 16.0, 17.3, 18.2, 23.8, 26.0 and 33.2. The other missing seven signals were found to be quaternary carbons. The HMBC spectrum of Compound 2 was almost similar to the one for compound 1 with the only difference occurring with two sugar moieties substituent at position C-3. Similar to compound 1 the aglycone of compound 2 was confirmed to have a pentacyclic olean-12-ene type triterpene skeleton.
The signals of the aglycone’s C-12 and C-13 at δ 122.0 and δ 144.0, respectively show the presence of two olefinic carbons (Mehta et al., 2004; Xu et al., 2010) and a signal at δ 178.6 (C-28) which shows the presence of a carboxylic acid at that position, thus, confirming that the aglycone is of olean-12-ene skeleton and therefore olean-12-en-28-oic acid aglycone.
The presence of two sugar moieties is evidenced in the signals at δc 103.4 and 100.3 confirming the presence of two anomeric carbons. The 1H NMR spectrum showed anomeric proton signals at δH 4.34 and δH 5.06 and one methyl group signal at δH 1.07 suggesting the occurrence of rhamnopyranosyl unit.
The sugar moieties were assigned mainly from 1H-1H COSY, HSQC and HMBC experiments which allowed the identification of one rhamnopyranose unit with anomeric protons resonating at δH5.06 and one arabinopyranose unit with the anomeric protons at δ 4.34. The position of the sugar chain at C-3 was determined by the HMBC correlations. A correlation between proton H-1’ signal (δH 4.34) and C-3 signal (δC 79.7) of the aglycone indicated that this pentose is directly attached to the aglycone. Further correlations were observed between signals of ara C-2’ (74.6) and Rha H-1” (δ 5.06). 1H-1H COSY correlations was also observed between proton H-1’ (δH4.34) and H-2’ (δH3.52), proton H-1” (δH 5.06) and H-2” (δH 3.67) and between proton H-5” (δH 3.71) and H-6” (δH 1.07) (Table 2).
According to electrospray ionization - mass spectrometry (ESI-MS), the positive electron mass spectrometry showed peaks at m/z 751 [M+H]+, 773 [M+Na]+, 789[M+K]+ and 1523 [2M+Na]+ upon positive ionization mode and at 749[M-H]-, 785[M+Cl]- and 1499[2M-H]- upon negative ionization mode. Thus, on the basis of the above evidence and analysis, compound 2 has been identified as a disaccharide triterpenoidsaponin of olean-12-en-28-oic acid aglycone with the molecular formula of C41H66O12. As compound 1, compound 2 was isolated before from the stem bark of P. fulva and given the name 3-O-[α-L-rhamnopyranosyl (1-2)-α-L-arabinopyranosyl]- hederagenin (Joshi et al., 1992; Njateng et al., 2015) however, it is isolated for the first time from the leaves from the this plant.
Compound 3 (Figure 3) was also isolated as a brownish amorphous powder and its molecular formula was determined as C59H96O25 based on molecular ion peak at 1227[M+Na-H]+ and 1242[M+K+H]+ upon positive ion mode and at 1203[M-2H]- upon negative ionization mode.
The 1H NMR spectrum of compound 3 displayed seven methyl signals at δ 0.68, 0.74, 0.85, 0.86, 0.87, 0.93 and 1.07 which is a characteristic of a typical oleanane type triterpene acid, an oxymethine at δ 3.01 and olefinic protons at δ 5.17 indicating that the aglycone is of olean -12- ene skeleton and therefore olean-12-en-28-oic acid aglycone (Mehta et al., 2004; Xu et al., 2010). It also displayed five anomeric proton signals at δ 4.31, 4.27, 5.21/5.33, 4.69 and 5.04 signifying the presence of five sugar units.
The DEPT NMR spectrum showed a total of 51 signals and the other missing eight signals were quaternary signals. Out of the total of 59 signals, 30 signals were assigned to the oleanolic acid moiety and the remaining 29 signals to the saccharide portion. The DEPT spectra of the aglycone displayed seven methyl signals at δ 15.7, 16.7, 17.1, 23.8, 25.9, 27.8 and 33.2; ten methylene (CH2) signals at δ 18.1, 22.9, 23.3, 26.1, 27.7, 32.1, 32.6, 33.6, 38.7 and 46.0. It also showed a total of five methine (CH) signals at δ 41.1, 47.5, 55.5, 88.2 and 122.1. The other missing eight signals were quaternary signals. The first carbon was found to resonate at δ 38.7 (C-1), 27.7 (C-2) and 88.2 (C-3). Quaternary carbons at C-4, C-8, C-10, C-13, C-14, C-17, C-20 and C-28 were found to be absorbed at δ 40.9, 39.3, 36.2, 143.8, 41.0, 45.6, 30.6 and 175.6, respectively. Carbon 5 were found to be absorbed at δ 55.5 while C-6, C-7, C-9, C-11, C-12, C-15, C-16, C-18, C-19, C-21 and C-22 were found to resonate at δ 18.1, 32.6, 47.5, 22.9, 122.1, 26.1, 23.3, 41.1, 46.0, 33.6 and 32.1 respectively. Carbons C-23, C-24, C-25, C-26 and C-27 resonated at δ 27.8, 16.7, 15.7, 17.1 and 25.9 while carbon C-29 and C-30 were found to be absorbed at δ 33.2 and 23.8, respectively.
The HMBC spectrum of compound 3 was almost similar to the one of compounds 1 and 2 with difference occurring with two sugar moieties substituent at position C-3 and in addition to an ester linkage between a trisaccharide chain and the aglycone at position C-28. Further confirmation with 1H - 1H COSY correlations showed that there was a strong correlation between protons H-9 (δ 1.47) and H-11 (δ 1.59) and proton H-11(δ 1.59) and H-12 (δ 5.17) and proton H-18 (δ 2.74) and H-19 (δ 1.07/1.61).
The sugar portion of compound 3 contained in the 1H NMR spectrum five anomeric proton signals at δ 5.04, 4.27, 5.21, 4.69, 5.04 and two methyl signals at δ 1.08 and 1.09 suggesting the occurrence of two rhamnose units. The sugar moieties were assigned mainly from 1H-1H COSY, HSQC and HMBC experiments which allowed the identification of one Arabinopyranose unit with the anomeric proton signal at δ 5.04, two glucopyranose units with anomeric protons resonating at δ 4.69 and 5.21/5.33 and two rhamnopyranose units with anomeric protons resonating at δ 4.27 and 5.04. Considering the δ values of the signals due to C-3 (δ 88.2) and C-28 (δ-175.6), saponin compound 3 was a 3, 28 bisdesmoside, thus, it has been identified as a bisdesmosidic pentasaccharide triterpenoid saponin of olean-12-en-28-
oic acid aglycone. The structure of the sugar chain at C-3 was unambiquously defined by the HMBC correlations and compared with those from previous study (Mehta et al., 2004). A correlation between C-3 signal (δ 88.2) of the aglycone and Ara H-1 signal (δ 4.30) indicated that this pentose was directly attached to the aglycone. Further correlations were observed between signals of Xyl C-2 (δ 76.9) and Rha I H-1 (δ 4.27). 1H-1H COSY correlations was also observed between proton H-1 (δ 4.27) and H-2 (δ 3.57). The structure of the oligosaccharide chain at C-28 was identified from the HMBC correlations between signals of Glc I C-4 (δ 69.0), Glc II H-1 (δ 4.69), Glc II C-4 (70.8) and Rha II H-1(δ 5.04).Correlating signals due to Glc I H-1 (δ 5.21/5.33) and aglycone C-28 (δ 175.6) provided a definitive evidence of an ester linkage between a trisaccharide chain and the aglycone. Thus, on the basis of the aforementioned evidence and analysis, compound 3 was found to be 3-O-[rhamnopyranosyl-(1→2)-xylopyranosyl]-olean- 12-en-28-O-[rhamnopyranosyl-(1→4)-glucopyranosyl-(1→6) glucopyranosyl] ester (Figure 2). This compound is isolated for the first time from this plant, however, it has the same skeleton with Hederasaponin B isolated from the leaves of Hedera helix L. (Nanyoung et al., 2017), the only difference occurring at position C-23.
Biological activities of the compounds
Secondary metabolites isolated from P. fulva leaf extracts were subjected to antibacterial activity against Gram negative bacteria (K. pneumoniae) and Gram positive bacteria (S. aureus) using disc diffusion assay test and inhibition zones were measured and recorded (Table 3). The crude extract and the isolated compounds generally demonstrated antibacterial activities. The most sensitive bacterium was S. aureus. Compound 1 was the most active against both K. pneumoniae and S. aureus with inhibition zones of 8.00±1.00 and 10.00±1.73 mm, respectively. Compound 3 on the other hand, was the least active against both K. pneumoniae and S. aureus with inhibition zones of 7.33±0.57 and 7.00±1.00 mm, respectively.
The antibacterial properties for the crude extract and compounds can be explained by the presence of potentially active secondary metabolites detected in them. Among the compounds isolated from the crude methanol extract, compound 1 was more active on the two tested microorganisms. This difference in activity may be attributed by the presence of hydroxyl group in position 23 and presence of 3-O-[xylopyranosyl] group in the compound (Njateng et al., 2017). Compound 2 that result from the addition of 3-O-[rhamnopyranosyl] group to compound 1 was less active. This modification may have slightly reduced the antibacterial activity of compound 2. The substitution of position C-23 with a methyl group and the addition of O-[rhamnopyranosyl-(1→4)-glucopyranosyl-(1→6)-glucopyranosyl] group in position 28 to compound 3 could have attributed to being less active against both bacteria compared to compound 1 and 2 (Njateng et al., 2017). According to Pavithra et al. (2010) saponins possess antimicrobial activities. The mean inhibition zones of methanol extract and pure compounds in comparison with chloramphenicol showed that they were significantly less active than chloramphenicol. The means of compounds 1, 2 and 3 against K. pneumoniae are not significantly different while compounds 1 and 3 and the methanol extract are not significantly different against S. aureus.
Within a column, extracts sharing the same letter(s) are not significantly different while those with different letter
(s) are significantly different (α =0.05, Turkey’s test).
