Inhibitory Effect of Triterpenoids from Dillenia serrata (Dilleniaceae) on Prostaglandin E2 Production and Quantitative HPLC Analysis of Its Koetjapic Acid and Betulinic Acid Contents

The crude methanol extracts and fractions of the root and stem barks of Dillenia serrata Thunb. showed 64% to 73% inhibition on the production of prostaglandin E2 (PGE2) in lipopolysaccharide-induced human whole blood using a radioimmunoassay technique. Three triterpenoids isolated from the root bark of the plant, koetjapic (1), 3-oxoolean-12-en-30-oic (2), and betulinic (3) acids, exhibited significant concentration-dependent inhibitory effects on PGE2 production with IC50 values of 1.05, 1.54, and 2.59 μM, respectively, as compared with the positive control, indomethacin (IC50 = 0.45 μM). Quantification of compounds 1 and 3 in the methanol extracts and fractions were carried out by using a validated reversed-phase high performance liquid chromatography (RP-HPLC) method. The ethyl acetate fraction of the stem bark showed the highest content of both compound 1 (15.1%) and compound 3 (52.8%). The strong inhibition of the extracts and fractions on cyclooxygenase-2 (COX-2) enzymatic activity was due to the presence of their major constituents, especially koetjapic and betulinic acids.


Isolation and Characterization of Compounds
After successive partition of the crude methanol extract of the root bark with petroleum ether and ethyl acetate, serial chromatography yielded two known oleanene-type triterpenoids, koetjapic acid (1) and 3-oxoolean-12-en-30-oic acid (2) and a lupene derivative, betulinic acid (3). Compound 3 was identified as a major compound of this plant. Koetjapic acid (1, also known as a seco-triterpenoid) was unexpectedly identified in the extracts of D. serrata. This compound was first reported from Sandoricum koetjape (Meliaceae) by Kaneda et al. [44] and, to the best of our knowledge, the occurrence of koetjapic acid (1) in Dillenia species has not been previously reported. The presence of seco-triterpenoids seems to be a common feature of Dillenia species, as reported by previous works [18,19]. Figure 1 displays the structures of these isolated triterpenoids. Identification of all triterpenoids was accomplished using a combination of physicochemical and spectroscopic experiments, viz. IR, 1D NMR as well as 2D NMR and HRESI-MS. All obtained values were similar to those reported in the literature [18,44,45].  (1) and Betulinic Acid (3) Previous HPLC studies on quantification of betulinic acid (3) in plants have been reported in some papers [46][47][48]. Bae et al. [46] quantified betulinic acid (3) in Ziziphus fructus using a Nova-Pak C18 column, eluted with phosphate buffer (Na2HPO4 0.05 M, pH 2.5)-methanol at a ratio of 19:81 and showed retention time at 28 min. Considering that the background noise resulted from methanol, Oliveira et al. [47] improved the quantification of this compound in Doliocarpus schottianus by using acetonitrile instead of methanol. The condition was isocratic with acetonitrile-water pH 3.0 (9:1) and the retention time was at 11.5 min. Kumar et al. [30] applied this method for quantification of betulinic acid (3) from D. indica in a study on anticancer activity of this plant. A modification of the method was also performed using a Diamonsil C18 column, eluted with acetonitrile-water (86:14) for quantification of betulinic acid (3) in Betula platyphylla and showed that the retention time was at 16.5 min [48]. To the best of our knowledge, quantitative analysis of koetjapic acid (1) has not previously been reported.

Quantitative Analysis of Koetjapic Acid
In our study, a pH modification of a method [47] was conducted in order to shorten the retention time of betulinic acid (3), resulting in a rapid analysis. The method was also able to give a good separation for koetjapic acid (1) at a retention time of 10.801 min (see Figure 2). Validation of the reversed phase HPLC method for quantification of these compounds was determined by regression equation, coefficient correlation (r 2 ), limit of detection (LOD) and limit of quantification (LOQ) (see Experimental Section). The calibration curves plotted were linear over the concentration range of 62.5 to 1000 μg/mL. LOD and LOQ values were found to be reliable for the method according to Bretnall et al. [49]. In addition, the precision of HPLC method regarding reproducibility and repeatability was satistacfory as indicated by the relative standard deviation (RSD) not greater than 5.0% [50] of peak area and retention time by intraday and interday analyses. Thus, the linearity, precision and accuracy of this modification were acceptable for quantitative analysis of koetjapic acid (1) and betulinic acid (3) in D. serrata. Distribution of detectable amounts of koetjapic acid (1) and betulinic acid (3) in the crude methanol extracts and fractions of D. serrata was determined by using the validated HPLC method. As shown in Table 1, koetjapic acid (1) and betulinic acid (3) were mainly concentrated in the ethyl acetate fraction while they were found in least amounts in the methanol fractions. From the quantitative analysis, we also found that betulinic acid (3) preferentially accumulated in the stem bark rather than the root bark. Koetjapic acid (1) together with betulinic acid (3) are suggested as the chemical markers of D. serrata. Betulinic acid has been reported from several genera of the Dilleniaceae family, including Dillenia, Wormia, and Acrotrema [16]. The presence of a considerable amount of betulinic acid in these genera suggests that this compound may be a chemotaxonomic marker of the family. Koetjapic acid (1) can also be used as an additional marker since its presence in this plant seems unique among the Dillenia species.

Inhibition of Prostaglandin E2 (PGE2)
Production of PGE2 induced by lipopolysaccharide (LPS) in human whole blood has been measured as a reflection of cyclooxygenase-2 (COX-2) activity of blood cells such as monocytes [51]. The inhibition of PGE2 production in human whole blood can be expressed as an inhibition of the enzymatic activity of COX-2 and/or inhibition of the expression of COX-2 protein. Crude methanol extracts and fractions of D. serrata root and stem bark were investigated for their ability to inhibit PGE2 in LPS-induced human whole blood and showed various percentage inhibitions at a concentration of 10 μg/mL ( Table 2). This inhibitory activity indicated the presence of bioactive compounds in the crude methanol extracts and fractions of D. serrata.
The three triterpenoids 1-3 isolated from the root bark of D. serrata were also able to inhibit PGE2 production induced by LPS. This concentration-dependent inhibition was observed at five serial concentrations ranging from 10 to 0.625 μg/mL ( Figure 3). All compounds showed significant inhibitory activity on PGE2 production as indicated by their IC50 values (Table 3). Of the three compounds, koetjapic acid (1) was identified as a promising inhibitor, with an IC50 value of 1.05 μM, comparable to that of a potent cyclooxygenase inhibitor, indomethacin (0.45 μM). In contrast to our study, koetjapic acid (1) was considered inactive in mouse ear inflammation model induced by tetradecanoylphorbol acetate [52]. Betulinic acid (3) had a higher IC50 value (2.59 μM) among the three compounds. The inhibition of LPS-induced PGE2 production by betulinic acid has been reported in human peripheral blood mononuclear cells (hPBMCs), and this inhibition was due to the suppression of COX-2 protein expression induced by LPS [53]. On the other hand, Wenzig et al. [54] investigated inhibition of COX-2 enzyme activity by betulinic acid and the results showed that the inhibition against this enzyme was not significant (IC50 ˃ 125 μM). Based on both studies, we can suggest that betulinic acid (3) plays role as an inhibitor of LPS induced expression of COX-2 protein, hence it can inhibits the production of PGE2 induced by LPS. Koetjapic acid (1) and 3-oxoolean-12en-30-oic acid (2) may also probably inhibiting the COX-2 enzyme by similar mechanisms. However, phospholipase A2 (PLA2) may also be a potential target for these three compounds.  Since betulinic acid (3) was found as a major compound in all the crude extracts and fractions, the LPS-induced PGE2 production inhibitory activity might be contributed mainly by this compound. However, as shown in Table 2 24 μg/mL, respectively), but they contained very different amounts of compounds 1 and 3. Interestingly, fraction RB-MF with IC50 of 3.33 μg/mL contains about 0.06 μg (1.98%) of compound 1 and 0.30 μg (9.09%) of compound 3, and these amount were considerably lower than the IC50 values of pure compound 1 and 3. Based on these results, we suggest that substantial amounts of koetjapic (1) and betulinic (3) acids in the crude methanol extracts and fractions of D. serrata may affect LPS-induced PGE2 production. However, the presence of other compounds in the extracts and fractions may also contribute to the activity, either in inhibiting or assisting LPS-induced PGE2 production. Koetjapic acid (1) has various pharmacological activities, including growth inhibition of multi-drug resistant bacteria Staphylococcus aureus and Pseudomonas aeruginosa (minimum inhibitory concentration, MIC of 12.5 and 6.25 μg/mL, respectively) [34], inhibition on Epstein-Barr virus antigen activation [36] and inhibition of DNA polymerase beta (IC50 20 μM) [37]. This compound also preserved high viability of the Raji cells [36]. Meanwhile, Nick et al. [18] investigated the antibacterial activity of 3-oxoolean-12-en-30-oic acid (2) from D. papuana that inhibited the growth of Bacillus subtilis, Escherichia coli, and Micrococcus luteus. Some papers have reported the anti-inflammatory, antileukemic and anti-HIV activities of betulinic acid (3) [38][39][40][41][42][43]. Treatment of hPBMCs with this compound alone did not elicit any PGE2 production and was non-cytotoxic, as indicated by trypan blue assay [53]. Chromatography techniques (Thin Layer Chromatography; TLC, Vacuum Liquid Chromatography; VLC and Column Chromatography; CC) were performed using Merck Si-gel. Melting point was observed using a Stuart Melting Point SMP10. UV-Vis and IR spectra were recorded on a Shimadzu UV1800 and a PerkinElmer GX IR (ATR), respectively. High-resolution electrospray ionization mass spectrometry (HRESI-MS) was performed on a Bruker MicroTOF-Q mass spectrometer. Spectra of 1 H-NMR (600 MHz) and 13 C-NMR (150 MHz) were recorded on a Bruker Advance NMR. Radioactivity was measured using a Tri-Carb 3110 TR PerkinElmer Liquid Scintillation Analyzer. HPLC technique was performed on a Waters 2535 Quaternary Gradient Module HPLC using an XBridge TM RP C-18 column (4.6 × 250 mm, 5 μm).

Plant Material
The barks of root and stem of D. serrata were collected from secondary forest in Onewila village, a region of Southeast Sulawesi, Indonesia and authenticated by Herbarium Bogoriense, Bogor, Indonesia (voucher number: BO-1902181).

Extraction and Isolation of Compounds
The dried root barks (1.15 kg) and stem barks (2.5 kg) were macerated with MeOH (4 and 8 L, respectively) for 24 h. The extracts were filtered and the solvent was evaporated under vacuum. The steps were executed three times to yield 500 g crude (43.5%) extract of root barks (CRB) and 600 g crude (24%) extract of stem barks (CSB). 250 g of CRB was dissolved in methanol (2.5 L) and left to stand overnight to re-crystallize compound 3. Compound 3 (993.4 mg) was collected using a vacuum filter and the residue was evaporated. The remaining CRB was dissolved in a small volume of MeOH (~100 mL) and partitioned three times with petroleum ether (PE) and ethyl acetate (EtOAc) (~500 mL), to give PE (1.5 g) and EtOAc (29.7 g) soluble fractions, respectively. The PE fraction was subjected onto VLC with hexane-EtOAc to give 28 sub-fractions. Sub-fractions were combined into 7 fractions (F1-F7) based on TLC analysis. Compound 2 (9.0 mg) was precipitated from F3. The EtOAc fraction was then subjected onto VLC with hexane-EtOAc to yield 32 sub-fractions. Compound 1 (102.2 mg) was precipitated from sub-F29 and sub-F31. For HPLC and bioactivity samples, 10 g of each crude extract was suspended in methanol and partitioned successively with petroleum ether, ethyl acetate and methanol. Six fractions were then evaporated under vacuum to yield residues of petroleum ether (RB-PEF) (0.6 g, 6%), ethyl acetate (RB-EAF) 1.9 g, 19%), and methanol (RB-MF) (7.2 g, 72%) fractions of root barks followed by petroleum ether (SB-PEF) (0.3 g, 3%), ethyl acetate (SB-EAF) (1.7 g, 17%), and methanol (SB-MF) (7.9 g, 79%) fractions of stem barks. These residues were stored in a refrigerator at 4 °C until analyses.  -30). NMR spectral data were identical to those given in [44]. (2)
The HPLC method for koetjapic acid (1) and betulinic acid (3) was validated by determination of linearity, precision and accuracy in accordance with ICH guidelines [55]. Linearity was evaluated from the linear regression equation and correlation coefficient (r 2 ) of calibration curves constructed for both compounds within the concentration range of 62.5 to 1000 μg/mL (Table 4). Precision was determined by the LOD and LOQ by injecting a series of known concentrations of the compounds. The values of LOD and LOQ were calculated from the relative standard deviation (RSD) and slope (S) of the calibration curves. The accuracy of the method regarding reproducibility and repeatability was evaluated by intra-and inter-day variation on three consecutive days with three repetitions each. The reproducibility and repeatability were demonstrated by the RSD of peak area and retention time.
The content of koetjapic acid (1) and betulinic acid (3) in the D. serrata extracts and fractions was quantified using the validated HPLC method. Precise amount of samples (10 mg for each of CRB, RB-PEF, RB-EAF, RB-MF, CSB, SB-PEF, SB-EAF, and SB-MF) were sonicated in methanol (1 mL) and filtered through a 0.45 μm filter. An aliquot of 20 μL of each sample was injected onto the HPLC.

Radioimmunoassay for Prostaglandin E2 (PGE2)
The inhibition of PGE2 production indicated by the concentration of PGE2 in human whole blood was measured according to the validated radioimmunoassay (RIA) method [51]. The application of human blood was permitted by the Ethics Committee of Universiti Kebangsaan Malaysia (UKM) with approval number NF-016-2013.
Human whole blood was drawn using aseptic vein puncture from the same donors of healthy volunteers when they had not taken any medicine or supplements during the last two weeks and fasted for 8 h prior to blood being withdrawn. The blood sample was prevented from coagulation by adding 10% (v/v) of 2% EDTA in a polypropylene tube. Duplicate 1 mL aliquots of EDTA-whole blood samples were transferred into test tubes and incubated with 10 μL of sample or indomethacin (1 mg/mL in 1:1 of DMSO-ethanol) for 15 min (37 °C) before LPS addition. The effects of samples or indomethacin on PGE2 production were studied by incubating each sample with whole blood-EDTA in the presence of LPS (10 μg/mL in 0.9% normal saline) for 24 h. For IC50, the concentration of samples were adjusted in five serial dilutions over a concentration range of 0.625 to 10 μg/mL. After incubation at 37 °C for 24 h, the plasma was separated by centrifugation at 2600 × g for 15 min at 4 °C. RIA buffer (phosphate buffered saline [0.01 M, pH 7.4] containing 0.1% BSA and 0.1% sodium azide) was used as the standard diluent of the assay. The plasma (100 μL) was added to anti-PGE2 (100 μL; diluted with ratio of 1:50,000) and [ 3 H]-PGE2 (100 μL; 5000 cpm) and incubated for 18-24 h at 4 °C. After incubation, dextran-charcoal (200 μL) was added to the mixture and incubated for 10 min at 0 °C. The supernatant was then separated by centrifugation at 3000× g for 15 min at 4 °C and pipetted (300 μL) into liquid scintillation cocktail (3 mL). The radioactivity was measured using a liquid scintillation analyzer.
Concentration of PGE2 (pg/0.1 mL) in the blood was calculated using a semi-logarithmic graph of standard PGE2. Previously, standard PGE2 (1 mg/mL) had been serially diluted to concentrations ranging from 2.45 to 400 pg/0.1 mL. The interference of compounds in the crude methanol extracts and fractions towards RIA method was checked by adding crude extracts and fractions to the standards and found not to interfere with the measurements. The average count per min (cpm) values of standards and samples (B) resulting from antibody-antigen (labeled PGE2) binding in the plasma were subtracted from the non-specific binding (Nc) together with the total binding between antibody and antigen (Bo). The normalized percent bound (% B/Bo) was then determined using Equation (1) The calculated % B/Bo values were plotted against their respective concentrations of standard PGE2 in picograms (pg) semi-logarithmically. Thus, the interpolation of % B/Bo values for samples and indomethacin using PGE2 standard curve resulted in the determination of PGE2 concentration in the blood. The percentage inhibition (% I) was then calculated using Equation (2).

Statistical Analysis
The HPLC analysis and bioassay were performed in triplicate and the data were expressed as means ± SD. Empower software (Waters) was used to construct a calibration curve of koetjapic and betulinic acids as well as their quantification. The bioactivity data were analyzed using Statistical Package for Social Sciences (SPSS) software Version 17. Data were analyzed using one way ANOVA analysis with a probability p < 0.05 representing a significant difference as compared to control. GraphPad Prism 5 was used to determine the IC50 value of active extracts and compounds.

Conclusions
Our study revealed that D. serrata possesses a promising inhibitory effect on LPS-induced PGE2 production. Three triterpenoids from this plant, koetjapic acid (1), 3-oxoolean-12-en-30-oic acid (2), and betulinic acid (3), were found to inhibit LPS-induced PGE2 production concentration-dependently. Although substantial amounts of koetjapic (1) and betulinic (3) acids in the crude methanol extracts and fractions of D. serrata may affect LPS-induced PGE2 production, the presence of other compounds in the extracts and fractions may also contribute to the observed activity. Further studies need to be carried out to investigate the effect of these compounds on COX-2 especially in vitro prostaglandin biosynthesis catalysed by COX-2 in cell free assays and expression of COX-2 protein.