We have carried out Infrared spectra on a Perkin branded Spectrum One device (Norwalk, CT, USA) using KBr pellets at a concentration of 4% by mass of the product. The mass spectrometry was conducted on Micromass Autospec 3F Mass Spectrometer using 70 eV by electronic impact for molecular ionization and fragmentation. The NMR spectra were recorded at 19˚C on a Bruker Avance 400 device with a frequency of 400.13 MHz for the ¹H nucleus and 100.61 MHz for the ¹³C nucleus in solutions of chloroform and methanol deuterated containing tetramethylsilane trace. The chemical shifts were reported in ppm using the tetramethylsilane signal as a reference and the coupling constants are expressed in Hz. The complete elucidation of the molecular structure was based on ¹H-NMR, ¹³C-NMR, DEPT-135 experiments and the 2-D experiments: COSY, NOESY, HSQC and HMBC. 1D- and 2D-NMR spectra were recorded at 400 MHz for ¹H and 100.6 MHz for ¹³C on a Bruker AM-400 NMR spectrometer (Billerica, MA, USA). Silica gel (200 - 300 mesh, Merck) was used for column chromatography. The 96-well microtiter plates (Becton Dickinson and Company, Lincoln Park, NJ, USA), and a microplate reader (Molecular Devices, Sunnyvale, CA, USA) were used for cell proliferation assays.

Bridelia duvigneaudii was collected from December 2010 to January 2011 by Kisimba Kibuye, a botanist technician in the Department of Geography of the Faculty of Science at University of Lubumbashi. The voucher specimens were stored at the Herbarium of the Faculty of Science (Voucher number KM4223). After drying in the sun the samples, at the Faculty of Sciences laboratory, were crushed in a metal mortar, sieved and stored in plastic bags at room temperature.

We have dissolved 0.85 g of bismuth nitrate (Merck) in 10 mL of glacial acetic acid (Merck) and 40 mL of distilled water to yield solution A. We have also dissolved 8 g of potassium iodide (Merck) in 20 mL of distilled water to generate solution B. 5 mL of solution A and 5 mL of solution B were transferred into a 100 mL flask. We added 20 mL of acetic acid to the mixture and diluted it with distilled water to the mark [13] .

Dry leaves of B. duvigneaudii (2.528 kg) were macerated with petroleum ether at room temperature for 24 h and this operation was performed twice. The dry marc (solid residue) mixed with Na₂CO₃ (1 kg) is moistened with water (1 L). The basified residue was kept for 48 h and then extracted with CHCl₃ for 24 h. The filtration was evaporated under vacuum to dryness with a rotavapor (BiBBy RE) and the dry residue was acidified with 0.5 M citric acid (Merck). The aqueous phase was basified with 20% Na₂CO₃ to reach the pH 9 and extracted three times with CHCl₃. After decantation, phase separation and evaporation of the organic phase, 6512 mg of total alkaloids (0.26% from dry leaves) were obtained. The total alkaloidal extract (6.512 g), chromatographed on open silica gel column 60 F₂₅₄, was successively eluted with EtOAc:MeOH system (100:0, 90:10, 80:20, 70:30, 60:40, 0:100). The different fractions were subjected to thin-layer chromatography (TLC) using the EtOAc:MeOH:Isopropylamine (IPA) system (9:1:0.05) and the plates were revealed with Dragendorff’s reagent. Fractions (2040 mg, 0.081% of dry material) with similar chromatograms (Positive to Dragendorff’s reagent) were chromatographed again on silica gel column eluting with EtOAc: MeOH (100:0, 90:10, 0:100) to afford 11 fractions, of which three fractions were positive to Dragendorff’s reagent. After evaporation under vacuum on a rotary evaporator at 30˚C - 40˚C, this process yielded 3 positive Dragendorff fractions: BRD4 (269 mg, 0.011%), BRD5 (165 mg, 0.006%) and BRD10 (584 mg, 0.023%). The three obtained fractions (BRD4, BRD5 and BRD10) were then analyzed by NMR, but only BRD4 was found pure enough for a full characterization by FTIR, MS and NMR (¹H, ¹³C, DEPT, COSY, NOESY, HSQC and HMBC).

BRD4 was obtained as an amorphous reddish solid. IR (KBr) νmax (cm⁻¹): 3209, 3078, 2962, 1690, 1589, 1551 - 1412, 1358, 879, 779. ESI-MS: 203 (27%, M^+.), 188 (73%, M^+.-˚CH₃), 135 (100%), 108 (49%), 81 (7%), 69 (23%) and 53 (9%). CI-MS: 204 (100%, (M + H)⁺), 188 (22%), 135 (20%). HREIMS: calculated mass for C₁₀H₁₄N₅ (M + H) was 204.1124 Da; found 204.1243 Da. ¹H-NMR: δ (ppm, CDCl₃, 400 MHz): 1.75 (s, 3H, CH₃-13), 1.90 (s, 3H, CH₃-14), 4.98 (d, 6.6 Hz; 2H, CH₂-10), 5.45 (t, 6.5 Hz, 1H, CH-11), 8.08 (s, 1H, CH-8), 8.13 (s, 1H, CH-2). ¹H-NMR: δ (ppm, Acetone-d₆, 400 MHz): 1.76 (s, 3H, CH₃-13), 1.90 (s, 3H, CH₃-14), 5.05 (d, 6.6 Hz; 2H, CH₂-10), 5.55 (t, 6.5 Hz, 1H, CH-11), 7.94 (s, 1H, CH-8), 8.34 (s, 1H, CH-2). ¹H-NMR: δ (ppm, CD₃OD, 400 MHz): 1.80 (s, 3H, CH₃-13), 1.88 (s, 3H, CH₃-14), 5.01 (d, 6.6 Hz; 2H, CH₂-10), 5.50 (t, 6.5 Hz, 1H, CH-11), 8.04 (s, 1H, CH-8), 8.33 (s, 1H, CH-2). ¹³C-NMR: δ (ppm, CDCl₃/CD₃OD, 400 MHz): 18.2 (C13), 25.7 (C14), 47.5 (C10), 116.2 (C11), 120.4 (C5), 141.5 (C12), 142.0 (C2), 150.1 (C4), 153.1 (C8), 154.6 (C6). ¹³C-NMR: δ (ppm, Acetone-d₆, 400 MHz): 18.2 (C13), 21.2 (C14), 48.1 (C10), 118.6 (C11), 118.7 (C5), 140.4 (C12), 145.0 (C2), 150.6 (C4), 151.4 (C8), 156.0 (C6). ¹³C-NMR: δ (ppm, CD₃OD, 400 MHz): 18.2 (C13), 21.1 (C14), 48.7 (C10), 118.2 (C11), 119.0 (C5), 141.6 (C12), 145.8 (C2), 150.4 (C4), 151.4 (C8), 156.4 (C6).

The antiproliferative assay was performed using 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)2-(4-sulfophenyl)-2H-tetrazolium (MTS) (Cell Titer 96 Aqueous, Promega, Nepean, ON, Canada), which allowed us to measure the number of viable cells. Prostate cancer LAPC-4 cells were kindly provided by Robert E. Reiter from the University of California (Los Angeles, CA, USA) while leukemia HL-60 cells were obtained from the American Type Culture Collection (Manassas, VA, USA). In brief, triplicate cultures of 1 × 10⁴ cells in a total of 100 µL of medium (RPMI-1640, Sigma, Saint Louis, MO, USA) in 96-well microtiter plates (Becton Dickinson and Company, Lincoln Park, NJ, USA) were incubated at 37˚C under 5% CO₂ humidified atmosphere. Compounds were tested at different concentrations (0.1, 1, 10 and 100 µg/mL) and the aminosteroid RM-581 [12] - [16] was used as a positive control. They were first solubilized in DMSO and diluted at multiple concentrations with culture medium, added to each well and incubated for 3 days. Control cells were treated with the medium and dimethyl sulfoxide (DMSO) only (final DMSO concentration < 0.5%). Following each treatment, MTS (20 µL) was added to each well and incubated for 4 h. MTS is converted to water-soluble colored formazan by dehydrogenase enzymes present in metabolically active cells. The plates were subsequently analyzed at 490 nm using a microplate reader (Molecular Devices, Sunnyvale, CA, USA). The results are expressed as cell proliferation and the control (basal proliferation of culture medium and DMSO) is fixed at 100%.

The statistical analysis of the data was done using the GraphPad Prism version 6 software. In fact, for the univariate analysis, the results are expressed in the form of numbers, percentages, and mean ± standard deviation. For multivariate analysis, ANOVA One Way was used to compare several variables. The Tukey test was applied with a 95% significance level (p < 0.05). The concentration inhibiting 50% of cell proliferation (IC₅₀) was calculated using the graphical method.

The alkaloid extraction from 2528 g of B. duvigneaudii leaves yielded 6512 mg of total alkaloids, which accounted for 0.257% of the plant material used. The alkaloid extract was purified using column chromatography on silica gel (60F₂₅₄ silica from Merck). The purification was carried out utilizing a gradient of chloroform (CHCl₃) and methanol (MeOH) from 100:0 to 0:100 as eluent, and we obtained fractions F8 to F12 that were positive to Dragendorff’s reagent. The mixture (F8 to F12) was subjected to an additional silica gel column chromatography with EtOAc:MeOH (100:0, 90:10, 0:100) as eluent. After silica gel column chromatography, the obtained fractions were subjected to TLC in the solvent system EtOAc:MeOH:IPA (8:2:0.05), and three fractions were positive to Dragendorff’s reagent (BRD4: 269 mg, 13.21%, Rf 0.3), (BRD5: 165 mg, 8.10%, Rf 0.2), and (BRD10: 584 mg, 28.68%, Rf 0.22) ( Table 1 ).

From the total alkaloidal extract purification works, three positive fractions to Dragendorff’s reagent were obtained. These three fractions gave 1.018 g (0.04%) of alkaloids from 2528 g of plant material (dry leaves). In fact, the purification of B. duvigneaudii allowed obtaining three alkaloidal fractions, which could be considered as three purified alkaloids: BRD4 (269 mg, 0.011%, Rf 0.3 and positive to Dragendorff’s reagent), BRD5 (165 mg, 0.006%, Rf 0.2 and positive to Dragendorff’s reagent) and BRD10 (584 mg, 0.023%, Rf 0.22 and positive to Dragendorff’s reagent). The three purified products were first evaluated as anticancer agents and secondly characterized by spectral methods such as Infrared (IR), MS and NMR) to determine their structures.

We report herein the results of our ongoing research into natural products. In fact, the three purified alkaloids (BRD4, BRD5 and BRD10) were tested for cytotoxicity on LAPC-4 prostate cancer cells and HL-60 leukemia cells. For comparison purposes, a blank test was performed upon RM-581-102 (named RM-581), an aminosteroid known for its efficiency against prostate cancer cells spreading as well as LAPC-4 cells [12] - [16] , was used as a positive control ( Table 2 , Table 3 ).

Values in the same column with different exponents (a, b, c and d) show significant differences (p < 0.05, Tukey test results).

The cytotoxicity study showed that the three extracts were cytotoxic on LAPC-4 cells ( Table 2 ). In fact, at 100 μg/mL the extracts fully inhibited the cell proliferation, namely 1.5% cell growth for BRD10, 1.7% for BRD4 and 3.6% for BRD5, thus representing 98.4%, 98.3% and 96.4% of cell growth inhibition, respectively. However, the positive control RM-581 had the best anticancer activity among the fractions tested. At 6.5 μg/mL it inhibited the cell growth to 11.4% (88.6% of inhibition). The statistical analysis (p < 0.05) carried out at 100 μg/mL showed that there are three groups of results for which the averages had any significant difference. This is the control group, the positive control group and the BRD10, BRD4 and BRD5 groups. At 10 μg/mL, statistical tests also showed that there are three groups of results without any significant difference: the control and BRD4 groups, the BRD5 and BRD10 groups and the positive control group. Besides, at 1 and 0.1 μg/mL, the statistical tests (p < 0.05) showed that there is no significant difference between the results obtained with the extracts and those of the control. However, at 10 μg/mL, the extracts BRD5 (9.8% inhibition) and BRD10 (10.3% inhibition) showed a very low activity while the BRD4 extract was inactive at this concentration. By further reducing the concentration of the extracts from 10 to 0.1 μg/mL, the cytotoxicity of all the extracts was canceled. In contrast, the positive control (RM-581) remained active from 0.06 to 6.5 μg/mL with the cell growth inhibition ranging from 29.7% to 88.6%, respectively.

The three purified extracts were also tested on HL-60 cells and statistical analysis was performed as well ( Table 3 ). They were only active on HL-60 cells at 100 μg/mL. Indeed, at this concentration, the purified alkaloids showed a strong inhibitory effect on cell growth ranging between 97.0% and 98.4%. As for the tests carried out on LAPC-4 cells, it was observed that the product used as the positive control showed the best activity than the purified alkaloids studied on HL-60 cells. Indeed, cell growth of 12.0% ± 1.0% was observed for RM-581 at 6.5 μg/mL, which represents 88.0% of cell growth inhibition. By decreasing the concentration from 10 to 0.1 μg/mL, the purified alkaloids become inactive on HL-60 cells. However, the inhibitory activity of positive control decreased. The statistical analysis (p < 0.05) showed that there was no significant difference in the activity of all extracts at 100 μg/mL. In addition, at 10 μg/mL, BRD4 and BRD5 showed significant differences compared to BRD10, which showed higher cell growth compared to the other extracts. Finally, from 1 to 0.1 μg/mL, the BRD4 and BRD5 extracts did not show any significant differences in activity compared to the control.

Values in the same column with different exponents (a, b, c and d) show significant differences (p < 0.05, Tukey test results).

The IC₅₀ values determined by the graphical method showed that the purified alkaloids were toxic to the cells tested [17] . In fact, on LAPC-4 cells, the purified fractions are active with an IC₅₀ of 17.5 μg/mL for BRD10, 19.2 μg/mL for BRD5 and 22.3 μg/mL for BRD4. Besides, on HL-60 cells, they are active with IC₅₀ values ranging between 22.5 and 24.3 μg/mL. However, the purified fractions were less toxic than the reference product RM-581, which had an IC₅₀ of 0.33 μg/mL on LAPC-4 cells and 1.15 μg/mL on HL-60 cells.

The evaluation of the anticancer activity of B. duvigneaudii alkaloids on LAPC-4 and HL-60 cancer cells confirmed the results obtained during the antimitotic activity evaluation of total alkaloids obtained from Sorghum sp seeds [12] - [16] . Indeed, the first study showed that the total alkaloids extracted from B. duvigneaudii were extremely toxic on Sorghum sp seeds with an inhibiting concentration of 50% of seed germination (IG50) ranging between 3.18 - 3.66 mg/mL. Similarly, the second study conducted by the interaction of purified alkaloids fraction of B. duvigneaudii on two types of cancer cell lines (LAPC-4 and HL-60) at 100 μg/mL showed that these alkaloidal fractions were toxic on these cancer cells with the IC₅₀ ranging respectively between 17.5 - 22.3 μg/mL on LAPC-4 cells and 22.5 - 24.3 μg/mL on HL-60 cells. These results show that seeds in germination can be used in the antiproliferative preliminary test [16] . The antiproliferative activity observed for the three alkaloid fractions encouraged us to elucidate the structure of BRD4, BRD5 and BRD10. However, since the preliminary results of NMR analysis showed that only BRD4 was completely pure, we only fully characterized this fraction.

The BRD4 infrared spectrum showed an antisymmetric elongation vibration band at 3209 cm⁻¹ assigned to the primary amine function υ NH₂. The intense band at 1690 cm⁻¹, which is associated with the in-plane deformation vibration, confirmed this assignment and the 779 cm⁻¹ bands associated with the out-of-plane deformation vibration of δ NH₂. For the different types of CH bonds, spectral data revealed the υ CH band at 2962 cm⁻¹, attributed to the aliphatic CH absorption, the 3078 cm⁻¹ band of aromatic υ CH and the 1358 cm⁻¹ band assigned to the υ CH of the methyl group. There are also bands around 1589 cm⁻¹ assigned to the imine function υ (C=N). Besides, four bands appearing between 1412 - 1551 cm⁻¹ are those of elongation vibrations of the bond υ (C=C). A similarity between the FTIR spectrum of BRD4 and the FTIR spectrum of adenine was noted ( Table 4 ).

*Masoud et al. [17] [18] .

Thus, on the FTIR spectrum of BRD4, the antisymmetric elongation vibration υ (NH₂) appears at 3209 cm⁻¹ while it appears around 3286 cm⁻¹ in the adenine spectrum. Moreover, the deformation bands δ (NH₂) appear at 1690 cm⁻¹ for BRD4 and 1668 cm⁻¹ for adenine [17] [18] . The υ (CH₂) vibration is located at 2972 cm⁻¹ for adenine but is displaced to 2962 cm⁻¹ for BRD4. Two absorption bands are observed around 1502 and 1601 cm⁻¹ on the spectrum of adenine, they respectively correspond to the vibration of elongation bands of the C=C and C=N bonds [17] - [20] . These two bands appear respectively at 1551 and 1589 cm⁻¹ on the BRD4 spectrum. The overlap of the IR spectral data of BRD4 and those of adenine highlights the existence of structural similarity for these two substances.

The electronic impact (EI) mass spectrum showed a peak of 203 Da at the highest molecular mass with an intensity of 27%; it was the molecular ion. To confirm that the ion observed at the highest molecular mass is not a fragment, another spectrum has been obtained by chemical ionization (CI). Then, the CI-MS spectrum showed that the molecular ion (M⁺) was also the base peak (100%) at m/z 204 Da. This mass corresponds to that of the molecular ion (M + H)⁺, which confirms that the molecule studied has the molecular mass of 203 Da. Its HREIMS revealed a (M + H)⁺ peak at m/z 204.1243 (calc. 204.1244), consistent with a molecular formula of C₁₀H₁₃N₅ with seven degrees of unsaturation.

The analysis of MS fragments revealed 7 characteristic peaks (EI mass spectrum). The peak of 203 Da is the molecular ion ${\text{C}}_{\text{10}}{\text{H}}_{\text{13}}{\text{N}}_5^{+}$ (M⁺) and the peak of 188 Da reflects the loss of a methyl group, which gave the ion ${\text{C}}_{\text{9}}{\text{H}}_{\text{10}}{\text{N}}_5^{+}$ (M-CH₃)⁺. Moreover, the fragment represented by the peak of 135 Da ( ${\text{C}}_{\text{5}}{\text{H}}_{\text{5}}{\text{N}}_5^{+}$) demonstrates the existence of an aromatic system following the loss of the 69 Da fragment on the molecular ion [19] . The peaks of 108 and 81 Da correspond respectively to the fragment ions ${\text{C}}_{\text{4}}{\text{H}}_{\text{4}}{\text{N}}_4^{+}$ and ${\text{C}}_{\text{3}}{\text{H}}_{\text{3}}{\text{N}}_3^{+}$ obtained as the result of a loss of neutral fragments HCN and H₂C₂N₂ on the 135 Da fragment. The 69 Da (23%) fragment represents the isopent-2-enyl cation. Finally, the 53 Da fragment represents the ${\text{C}}_{\text{2}}{\text{HN}}_2^{+}$ ion and/or the C₃H₃N⁺ ion. The ${\text{C}}_{\text{2}}{\text{HN}}_2^{+}$ fragment is more likely because it results from a loss of a neutral fragment whereas C₃H₃N⁺ requires a molecular rearrangement during its metastable state [21] - [27] .

The ¹H-NMR spectrum acquired in CDCl₃ indicates 7 characteristic signals grouped in 3 zones. The first zone of aliphatic protons comprises two singlets between 1.75 and 1.90 depending on the deuterated solvent used, which indicate the methyl protons of a dimethylallyl group. In the second zone, it was observed a doublet at 4.98 ppm (J = 6.5 Hz) and a triplet at 5.45 ppm (J = 6.6 Hz), which are the signals of the two methylene protons coupled with a methine proton. The broad singlet at 6.3 ppm indicates the protons attached to the nitrogen atom, which, in turn, is also attached to the aromatic ring. Finally, the third zone at about 8 ppm contains two characteristic singlet signals at 8.08 and 8.13 ppm for the protons of an aromatic nucleus.

The ¹³C-NMR spectrum indicated the presence of 10 peaks that means the structure studied has 10 carbon atoms, which corroborates the MS data. Besides, the analysis of the DEPT-135 spectrum indicated only 6 peaks on this spectrum. Indeed, this spectrum showed two methyl signals at 18.2 and 21.1 ppm. The signals appearing at 145.8 and 151.4 ppm were those of the aromatic methines and a signal at 118.2 ppm is a non-aromatic ethylenic methine attached to the dimethylallyl group. Moreover, the peak appearing at 48.7 ppm is attributed to methylene. Finally, the signals appearing at 156.4, 150.4, 141.6 and 119.0 ppm on the ¹³C-NMR spectrum, and which are absent on the DEPT-135 spectrum, are those of quaternary carbons, two of them being located at the 2-ring junction of the aromatic system [21] . To support the different NMR correlations and couplings, two-dimensional NMR was used, including COSY, NOESY, HSQC and HMBC experiments. Different correlations obtained by 2D-NMR and mentioned in Table 5 suggest the structure of triacanthine.

The spin systems were determined by the COSY and NOESY spectra. On the NOESY spectrum, the H-14 protons were correlated to the H-10 protons, which in turn were correlated to the H-11 proton on the COSY spectrum. The correlations obtained by NOESY between H-11 and H-2 and H-14 to H-10 prove that the H-11 proton was in the trans-position concerning the C-14 carbon. This implies that there is a double bond between two carbons on the side chain attached to the aromatic system. Moreover, the HSQC spectrum allowed the identification of protons correlating with carbon (H-C correlation). Protons H-11 (5.50 ppm), H-10 (5.01 ppm) and H-2 (8.33 ppm) were carried by carbon C-11 (118.2 ppm), carbon C-10 (48.7) and carbon C-2 (145.8 ppm), respectively. HMBC correlations allow the positioning of different groups. Thus, the correlations between H-13 and C-11, C-12 and C-14 as well as between H-14 and C-11, C-12 and C-13 were detected. Other key correlations were also detected with H-2 (C-4, C-6 and C-10), H-8 (C-4 and C-5), H-10 (C-2, C-4, C-11 and C-12) and H-11 (C-10, C-12, C-13 and C-14). All these correlations allow us to know the junction point of the side chain with the N-3 of the aromatic system.

Correlations shown on the HSQC spectrum identified two methyl groups (18.2 and 21.1 ppm), one methine (116.2 ppm) and the DEPT spectrum showed methylene (48.7 ppm) in the allylic position assigned to C-10. Moreover, the correlations obtained by COSY and NOESY showed that there is a double bond between two carbons on the side chain attached to the aromatic system. So, the C₅H₇ fragment is the isopentyl, which allows the linking of all the information obtained by all the techniques used [21] - [27] .

For comparison, the ¹H-NMR and ¹³C-NMR spectral characteristics of BRD4 and triacanthine are summarized in Table 6 .

(*) Data from Boonyaratavej and Petsom [22] . (**) The C-4 and C-6 assignation seems inversed. (***). The C-5 and C-12 assignation seems inversed. CDP: ChemDraw Prediction.

The chemical shift of H-2 was 8.13 ppm in CDCl₃, 8.34 ppm in acetone-d6 and 8.33 ppm in CD₃OD. Besides, H-8 is shifted to 8.08, 7.79 and 8.04 ppm in CDCl₃, acetone-d₆ and CD₃OD, respectively. These protons are significantly deshielded due to the aromatic nature of the purine ring. Aromatic proton signals are most often located in the relatively narrow region of 7.5 to 9.5 ppm [21] . This was found analogous to the chemical shifts of protons at 8.24 ppm (H-2) and at 7.80 ppm (H-8) acquired in the CD₃OD for triacanthine [22] , but these chemical shifts were lower than the ChemDraw theoretical prediction of 8.64 ppm for H-2 and 8.70 ppm for H-8 in triacanthine. Except for H-10 protons (5.01, 5.02 and 4.08 ppm), a good similarity was also observed for experimental and predicted chemical shifts of other protons of BRD4 and triacanthine.

On the ¹³C-NMR spectrum, chemical shifts of carbon acquired in CDCl₃-CD₃OD, acetone-d₆ and CD₃OD also showed similarities. However, it was observed a difference in chemical shifts of C-4 (150.4 ppm), C-5 (119.0 ppm), C-6 (156.4 ppm) and C-12 (141.6 ppm) for BRD4 comparatively of the chemical shift of C-4 (156.7 ppm), C-5 (141.4 ppm), C-6 (150.7 ppm) and C-12 (121.2 ppm) for the same carbons in triacanthine acquired in the CD₃OD for the two compounds. Indeed, it was observed that the C-4 and C-6 as well as C-5 and C-12 assignations seemed inverted. Thus, according to the similarities observed for chemical shifts of protons and carbons of BRD4 and triacanthine, it means that BRD4 from B. duvigneaudii spectral data are consistent with triacanthine spectral data, a purine-derived alkaloid. This alkaloid has already been isolated from Holarrhena floribunda, H. mitis, H. congolensis, H. wulfsbergii and Gleditsia triacanthos (Apocynaceae) and from Childowia sanguinea (Leguminosaceae) [21] - [27] .

Triacanthine (6-amino-3-(α,α-dimethylallyl)adenine) is an adenine derivative with a dimethylallyl group in N3 position. It is the only natural substance known as an N-alkylation derivative of adenine at position 3 [21] - [27] . This study also showed that triacanthine has anticancer activity. Furthermore, various biological studies carried out on triacanthine showed that it had diverse biological activities [28] - [34] .