Bioassay-Guided Fractionation of Melastoma malabathricum Linn. Leaf Solid Phase Extraction Fraction and Its Anticoagulant Activity

The aims of this study were to examine the bioactive component(s) responsible for the anticoagulant activity of M. malabathricum Linn. leaf hot water crude extract via bioassay-guided fractionation and to evaluate the effect of bioactive component(s) on the intrinsic blood coagulation pathway. The active anticoagulant fraction of F3 was subjected to a series of chromatographic separation and spectroscopic analyses. Furthermore, the effect of the bioactive component(s) on the intrinsic blood coagulation pathway was studied through immediate and time incubation mixing studies. Through Activated Partial Thromboplastin Time (APTT) assay-guided fractionation, Subfraction B was considered the most potent anticoagulant fraction. Characterisation of Subfraction B indicated that anticoagulant activity could partly be due to the presence of cinnamic acid and a cinnamic acid derivative. APTT assays for both the immediate and time incubation mixing were corrected back into normal clotting time range (35.4–56.3 s). In conclusion, cinnamic acid and cinnamic acid derivative from Subfraction B were the first such compounds to be discovered from M. malabathricum Linn. leaf hot water crude extract that possess anticoagulant activity. This active anticoagulant Subfraction B prolonged blood clotting time by causing factor(s) deficiency in the intrinsic blood coagulation pathway.

magnetic resonance (NMR). The nature of the single active fraction together with the activity of identified chemical components on the intrinsic blood coagulation pathway was also evaluated.

Results and Discussion
Commonly, commercial anticoagulants are often associated with bleeding and other life threatening side effects. These reasons have encouraged development of alternative anticoagulants, especially from medicinal plants [24][25][26][27]. These medicinal plants were postulated to exert similar or better anticoagulant effects than commercial anticoagulants coupled with minimal side effects [16,27]. Previously, M. malabathricum Linn. leaf hot water crude extract was reported to show prolongation in blood clotting time through an activated partial thromboplastin time (APTT) assay, but did not prolong the blood clotting times of prothrombin time (PT) and thrombin time (TT) assays [19]. In addition, polyphenolic glycoconjugates were identified as contributors to the anticoagulant activity of M. malabathricum Linn. leaf hot water crude extract [20]. Other than the aforementioned information, there was no further research regarding the biological active compound(s) present in M. malabantricum Linn. leaf hot water crude extract that might be involved in its anticoagulant properties. In this study, in continuation of the Khoo et al. [20] study, the active anticoagulant solid phase extracted fraction from M. malabathricum Linn. leaf hot water crude extract was subjected to chromatographic separation. Among the active anticoagulant fractions F1, F2 and F3, fractions F1 and F3 were selected as candidates for further fractionation. Of the successive fractions obtained from F1 and F3, one (F3.3) possessed the most active anticoagulant activity when compared to the fractions of F1. Hence, only fractionated F3 was reported in this study. Meanwhile, F2 displayed a complex HPLC chromatogram and not well resolved peaks (data not shown). Chromatographic separations of F3 were guided by activated partial thromboplastin time (APTT) assays. Besides, the effect of the active anticoagulant fraction together with identification of potential component(s) in the intrinsic blood coagulation pathway was examined. F3, the active anticoagulant fraction with its high phenolic acids content and moderate amounts of carbohydrate and uronic acid moieties [20], was fractionated based on the separation methods for isolating polyphenolics through preparative high performance liquid chromatography (HPLC) into six fractions (F3.1-F3.6). The polyanionic characteristics of polyphenolics portray a favourable binding affinity towards blood coagulant factors [16,27]. According to Figure 1A, seemingly, all separated fractions of F3 exhibited significant anticoagulant activity except for F3.6. Subfraction F3.1 displayed the highest anticoagulant activity with a blood clotting time of 268.2 ± 7.8 s (p < 0.001). Meanwhile, prolongation of blood clotting times for F3.3 and F3.4 were similar. F3.3 prolonged the clotting process for about 238.1 ± 6.8 s (p < 0.001); while F3.4 prolonged it for about 228.8 ± 11.3 s (p < 0.001). The active anticoagulant fraction of F3.3 with the highest yielded was fractionated over a size exclusion Sephadex LH-20 chromatography column, and produced 21 fractions (F3.3.S1-F3.3.S21). Among these 21 fractions (F3.3.S1-F3.3.S21), ten fractions (F3.3.S1-F3.3.S10) underwent chromatographic separation and together yielded a pooled fraction designated as Subfraction A. In reference to Figure 1B, no significant difference of anticoagulant activity was observed for Subfraction A when compared to the vehicle as a negative control (p > 0.05). The intrinsic blood clotting times for Subfraction A and vehicle were 76.3 ± 6.6 s and 58.3 ± 2.4 s, respectively. The anticoagulant activity of Subfraction A was thus lost after the chromatographic separation.    3.S11 to F3.3.S21; (D) APTT of fractions F3.3.S11.1 to F3.3.S11.10; (E) APTT of fractions F3.3.S11.6(a) to F3.3.S11.6(c); (F) APTT of Subfraction B and cinnamic acid standard. Heparin sodium salt from porcine intestinal mucosa (140 USP units mg −1 ) was used as a positive control. Data represent means ± S.E.M. of three independent experiments. Significant differences compared to control group (standard human normal pooled plasma with distilled water, vehicle) were designated as ** p < 0.05 and *** p < 0.001. NC: No Coagulation. Details of the bioassay-guided fractionation are described in Experimental Section 3.3.
It is presumed that the active anticoagulant compound(s) in Subfraction A, which contributed to its anticoagulant activity were eliminated. Distributions of low concentrations of the active anticoagulant compounds over two adjacent fractions might also be a cause for the observed loss of anticoagulant activities [28]. Our results paralleled those reported by Liu et al., [29] and Senejoux et al., [30] that the anti-inflammatory of Rehmania glutinose's root and vasodilating properties of whole plant of Ziziphora clinopodioides Lam. were lost or lower than that of the crude extracts after being subjected to bioassay-guided fractionation, respectively. Both research groups suggested that the occurrence of interactions between chemical compounds, known as synergistic effects, might play a crucial role in contributing to the observed biological activities.
Hence, Subfraction B was suggested to contain cinnamic acid and a cinnamic acid derivative as major components in M. malabathricum Linn. leaf hot water crude extract. However, we postulate that the anticoagulant activity of Subfraction B might not be solely due to cinnamic acid and the cinnamic acid derivative, but might also be enhanced by other active and as yet unidentified components. Similar findings were observed by Han et al., who reported that Xanthium strumarium L. n-butanol fraction showed more potent anti-inflammatory and analgesic effects than isolated caffeoylquinic acid. Synergism between identified phenolic acids, which included caffeoylquinic acid, heterocyclic compounds and other unidentified secondary metabolites, enhanced the anti-inflammatory and analgesic activities for the n-butanol fraction [33]. Meanwhile, synergistic or additive effects for triterpene taraxasteryl myristate, taraxasteryl acetate and fern-7-en-3-β-one within n-hexane fraction from methanolic extract of Scorzonera latifolia were also observed [34]. Therefore, together with these literatures, the better blood prolongation time of Subfraction B was proposed to be due to cinnamic acid and cinnamic acid derivative, and also other different compounds.
The subsequent aim of this study was to understand the effect of the active anticoagulant fraction on the intrinsic blood coagulation pathway. Subfraction B, which either acted as an inhibitor or caused coagulant factor deficiency, was determined through mixing studies. Interestingly, blood clotting times of Subfraction B mixing studies were corrected back to normal. According to Figure 5, the blood clotting time of Subfraction B was 286.0 ± 5.3 s (p < 0.001) at 1 mg/mL. After adding 50% of standard human normal pooled plasmas into spiked plasmas, the blood clotting times for immediate mixing and time of incubation mixing were corrected to 65.3 ± 1.8 s (p < 0.001) and 66.8 ± 0.9 s (p < 0.001), respectively. Both mixing studies also revealed that additional standard Subfraction B μg/mL C lotting Tim e (s) human normal pooled plasmas to spiked plasmas in a ratio 1:1 for concentrations at 500 µg/mL, 250 µg/mL and 125 µg/mL successfully corrected the initial blood clotting time back into the reference range (35.4-56.3 s). A plausible explanation for these results could be that Subfraction B-treated plasma exerted its anticoagulant activity through reduction of coagulant factor levels. By means of additional equal volumes of standard human normal but platelet-poor plasma in mixing studies, an insufficient amount of coagulant factors were replenished [35,36]. Hence, the blood clotting times resumed to normal, and the coagulation activity was restored.
The immediate mixing and time incubation mixing clotting times were further evaluated with the respective set of formulae, known as the Rosner index and percent correction. These interpretation methods are simple and robust to distinguish between the categories of presence of inhibitor or factor deficiency [37,38]. By using these two methods in our data interpretations, misclassification of plasma can be avoided. Based on our results, as anticipated, the Rosner indexes for immediate mixing and time incubation mixing were less than 15 (<15) ( Table 1), while, their percent corrections were more than 70% (>70%) ( Table 2). Both results highlighted that Subfraction B-treated plasma prolonged APTT assay abnormally due to factor deficiency.  Taken together, the comparison of mixing studies blood clotting times with reference range, Rosner index and percent corrections, the anticoagulant activity of Subfraction B was attributable to coagulant factor(s) insufficiency in the treated plasma. These results were further supported by a previous report, whereby M. malabathricum Linn. leaf hot water crude extract prolonged initial blood clotting time through reduction of intrinsic coagulant factor level(s) [19].

Plant Material, Extraction, and Solid Phase Extraction
Leaves of Melastoma malabathricum Linn. were obtained continuously from Lebuh Silikon, Universiti Putra Malaysia between September 2009 to December 2012. The plant material was authenticated by a botanist, Dr. Shamsul Khamis, and deposited as a voucher specimen at the Herbarium Biodiversity Unit, Institute of Biosciences, Universiti Putra Malaysia with reference number SK1717/09. The leaves were first extracted with hot water in accordance to the methodology suggested by Manicam et al., [19]. Briefly, cleaned and air dried fine pieces of M. malabathricum Linn. leaves (1 kg) were refluxed with 1 L of deionised water at 100 °C. After 5 h of extraction, a hot water extract was obtained, which was lyophilised, and subjected to solid phase extraction.

Bioassay-Guided Fractionation
Solid phase extraction of crude extract was carried out according to the method described by Khoo et al., [20]. The hot water crude extract (100 g) was separated through a 10 g C18, Sep-Pak Cartridge (Water, Milford, MA, USA) into four fractions (F1-F4). Active anticoagulant fractions of F1, F2 and F3 were then lyophilised and stored at −20 °C until further analyses.
The active anticoagulant fraction, F3 (8.5 g), was fractionated using an Agilent 1200 series LC system (Agilent Technologies, Santa Clara, CA, USA) preparative high performance liquid chromatography (HPLC) equipped with multi-wavelengths detector with a Zorbax Stable Bond Preparative High Throughput SB-C 18 column (21.2 × 250 mm, 7 µm). Briefly, the fraction F3 was dissolved in water: methanol (20:80, v/v) for the preparation of the final concentration of 50 mg/mL. The injection volume was 900 µL, and the mobile phase contained (A) water and (B) acetonitrile. The gradient elution profile was as follows: 0% B from 0 to 10 min, 15% B from 10 to 15 min, 20% from 15 to 30 min, 20% B from 30 to 35 min, 23% B from 35 to 40 min, 25% B from 40 to 45 min, and lastly 50% B from 45 to 50 min. The flow rate was 15 mL/min. The detection wavelength was 254 nm, and fractionated into six fractions (F3.1-F3.6). All fractions were air dried and stored in desiccators at room temperature (25 °C) prior to anticoagulant assays and consecutive fractionation to be carried out.
F3.3 (4 g) was later purified over on a size-exclusion Sephadex LH-20 column chromatography. F3.3 (400 mg) was dissolved in water: methanol (20:80, v/v) before loading onto the Sephadex LH-20 column, which was initially eluted with water: methanol (20:80, v/v) and later with water only to afford 21 fractions (F3.3.S1-F3.3.S21). These fractions were analysed and grouped accordingly to different retention factors showed on reverse phase C 18 thin layer chromatography (TLC, Merck, Darmstadt, Germany), which were eluted with water: methanol (70:30, v/v). The TLC plates were observed under UV at 254 nm and 366 nm wavelengths followed by spraying with 2 M of sulphuric acid, and heated on a hotplate at 120 °C for 5 min. Fractions (F3.3.S1-F3.3.S21) were air dried and stored in dessicator at room temperature (25 °C). Fractionation of F3.3 was repeated for a total amount of 4 g in order to achieve sufficient yield for the following fractionation steps and anticoagulant assay. Chromatographic separation of samples (F3.3.S1-F3.3.S10) and F3.3.S11 were performed on an Agilent 1100 series LC system (Agilent Technologies) equipped with multi-wavelengths detector with a Hypersil Gold TM C 18 column (150 × 4.6 mm, 5 µm) (Thermo Scientific, Tewksbury, MA, USA). Each sample was prepared individually by dissolving in water: methanol (20:80, v/v) to a final concentration of 2 mg/mL. The injection volume was 20 µL and the flow rate was 1 mL/min. The mobile phase contained (A) water and (B) acetonitrile. For separation of fractions (F3.3.S1 to F3.3.S10), the gradient elution for sample separation was as follows: 0% B from 0 to 5 min, 15% B from 5 to 10 min, 20% B from 10 to 20 min, and lastly 50% from 20 to 30 min. The detection wavelengths were 210 nm and 254 nm. Chromatographic elution from 1 to 3 min of each sample (F3.3.S1-F3.3.S10) was collected and pooled together. This pooled fraction was designated as Subfraction A. Subfraction A (5 mg) was air dried and stored in a desiccator at room temperature (25 °C) for subsequent experiments. Anticoagulant activity of Subfraction A was evaluated.

Analytical Identification of Active Principle
The mass spectrum of active anticoagulant Subfraction B was recorded on a Shimadzu gas chromatograph-mass spectrometer (GCMS-QP5050A, Shimadzu, Nakagyo-ku, Kyoto, Japan) coupled with a direct insertion probe (DI). Analysis using this mass spectrometer was performed using 70 eV electrospray ionisation (EI) in the positive mode.   The nuclear magnetic resonance (NMR) spectra ( 1 H and 13 C) were recorded on a Varian INOVA NMR spectrometer (Agilent Technologies, Santa Clara, CA, USA). This NMR was operated at 500 MHz for 1 H and 125 MHz for 13 C, and controlled by the Vnmrj software. Subfraction B was analysed by dissolving in deuterium oxide (D2O). The chemical shifts (δ) were expressed in ppm (parts per million), while coupling constants (J) are reported as Hz. Multiplicities of proton signals were described as singlet (s) and doublet (d). All NMR spectra were analysed using offline ACDLabs 12.0 software (Advanced Chemistry Development, Toronto, ON, Canada).
Subfraction B was analysed using an Agilent 1100 series LC system (Agilent Technologies) with a Hypersil Gold TM C 18 column (150 × 4.6 mm, 5 µm) (Thermo Scientific). Subfraction B and cinnamic acid standard were dissolved in water: methanol (20:80, v/v) for preparation of the final concentration at 1 mg/mL. The injection volume was 20 µL. The mobile phase contained (A) water: formic acid (99.9:0.1, v/v) and (B) acetonitrile. The gradient elution profile was as follows: 0% B from 0 to 5 min, 5% B from 10 to 15 min, 10% B from 15 to 20 min, 20% B from 20 to 30 min, and lastly 50% B from 30 to 45 min. The detection wavelength was 254 nm.

Anticoagulant Assay
Anticoagulant activities of all fractions, subfractions, cinnamic acid, and mixed plasmas were evaluated using activated partial thromboplastin time (APTT) assays. Fractions, subfractions and cinnamic acid were dissolved in distilled water for preparation of the final concentration of 1 mg/mL. Samples were spiked with reconstituted standard human pooled plasma in a ratio 1:1. Spiked samples were later added with reconstituted STA-PTT automated reagent and 0.025 M of calcium chloride. Blood clotting time was recorded until the formation of blood clot was observed. These APTT assays were recorded using a STA Compact coagulation analyser (Diagnostica Stago, Asnieres sur Seine, France). Spiked heparin sodium salt (10 mg/mL) and spiked distilled water (vehicle) were used as positive and negative controls, respectively.

Mixing Studies
Mixing study is a testing guideline used to distinguish presence of inhibitor or factor deficiency for causing abnormal prolongation in activated partial thromboplastin time (APTT) assay [35,37]. In this study, two types of mixing studies, immediate mixing study and time incubation mixing study, were performed.
3.6.1. Immediate Mixing Study Different concentrations of Subfraction B (1 mg/mL, 500 µg/mL, 250 µg/mL and 125 µg/mL) were spiked with equal volume of standard human normal pooled plasmas. Next, spiked samples were mixed again with standard human normal pooled plasma in a 1:1 ratio, and designated as mixed plasmas. Immediately, the mixed plasmas were subjected to APTT analyses.

Time Incubation Mixing Study
Subfraction B with different concentrations (1 mg/mL, 500 µg/mL, 250 µg/mL and 125 µg/mL) were spiked with standard human normal pooled plasmas. Spiked samples and standard human normal pooled plasmas were incubated separately at 37 °C water bath for 2 h in air tight tubes. After that, both incubated spiked samples and standard human normal pooled plasma were mixed in a 1:1 ratio, and identified as mixed plasmas. APTT assays for time incubation mixed plasmas were determined by a STA Compact coagulation analyser (Diagnostica Stago).

Rosner Index
Rosner index was calculated for all 1:1 mixed plasmas results from the immediate and time incubation mixing studies with modified formula. This is an index of sample correction with a prolonged APTT profile. The formula of this index is as follows: Index = APTT 1: 1 mix − APTT of vehicle APTT of SS × 100 where SS = Spiked sample An index of correction of less than 15 (<15) suggests a factor deficiency present in the mixed plasma, whereas an index of correction of more than 15 (>15) suggests the presence of an inhibitor in mixed plasma [38].

Percent Correction
Modified percent correction from Chang et al., was used to interpret the mixing study results. Blood clotting effects of spiked samples after mixed with standard human normal pooled plasma in a ratio (1:1) were evaluated. The formula of percent correction was displayed as follows: Percent Correction = APTT of SS − APTT 1: 1 mix APTT SS − APTT of Vehicle × 100% where SS = Spiked sample Percent correction of more than 70% indicates the presence of a factor deficiency, and causes prolongation of the APTT assay. A low percent correction of less than 58% indicates the presence of an inhibitor. A percent correction between 70%-58% is classified as an undefined result [39].

Statistical Analysis
Data shown were expressed as mean ± S.E.M. in triplicates. Significance differences for all tests were referred as * p < 0.05, ** p < 0.01 and *** p < 0.001. All data except Rosner index and percent correction formula were analysed using one way analysis of variance (ANOVA) followed by post-hoc Dunett's multiple range tests, while the Rosner index and percent correction formula were analysed by student's t-test (GraphPad Prism v.5.0, La Jolla, CA, USA).

Conclusions
In conclusion, this is the first report to show that cinnamic acid and a cinnamic acid derivative identified in M. malabathricum Linn. leaf hot water crude extract exhibited anticoagulant activities. Bioassay-guided fractionation of M. malabathricum Linn. leaf hot water crude extract suggested that cinnamic acid and the cinnamic acid derivative appeared to be the major compounds in Subfraction B, although the anticoagulant activity was not totally dependent on cinnamic acid and the cinnamic acid derivative. The presence of other unidentified compounds in Subfraction B could also be responsible for the prolonged blood clotting time. Moreover, Subfraction B normalised blood clotting times in both the immediate mixing and time incubation mixing studies. Subfraction B prolonged the blood clotting time by causing a factor deficiency in the intrinsic pathway. Therefore, the underlying intrinsic blood coagulation pathway affected by Subfraction B remains to be further examined in future studies. This study may provide new insights as Subfraction B may be a potential candidate in developing a safe and natural anticoagulant for preventing and treating thromboembolism.