Introduction

Plants of the genus Tabernaemontana are known as a storehouse of indole and bisindole alkaloids. Bisindole alkaloids are consisting of two different or identical monomeric indole alkaloid units with a C–N, C–O–C, or C–C bond [1]. Interestingly, some of the bisindole alkaloids showed cytotoxicity rather than those with monomeric units [2,3,4]. Various indole alkaloids that exhibited diverse molecular skeleton and significant diverse bioactivity have been reported from our laboratory [5,6,7,8,9,10,11,12,13,14].

Tabernaemontana macrocarpa Jack is distributed in Indonesia, Malaysia, Brunei, and the Philippines. Traditionally, the exudate of this plant has been used in Borneo, Indonesia, for the treatment of dental disease, herpes, and eczema, while the wood is used for simple construction work [15, 16]. Our previous chemical research of this plant revealed two sarpagine type indole alkaloids together with a known bisindole alkaloid, 16-demethoxycarbonylvoacamine, and their antimalarial activity [17]. The results showed only the dimeric alkaloid, 16-demethoxycarbonylvoacamine, showed moderate in vitro antimalarial activity with IC50 value of 28.8 μM [17]. In order to searching for more bioactive bisindole alkaloids from this plant, the chemical investigation on the bark of T. macrocarpa Jack was carried out and two new bisindole alkaloids, bisnaecarpamines A and B (1 and 2) (Fig. 1), were isolated. Herein are reported the isolation, structure determination, and antimalarial activity of bisnaecarpamines A and B (1 and 2).

Fig. 1
figure 1

Structures of 1-2

Full size image

Results and discussions

Structure elucidation of 1 and 2

Compound 1 was obtained as an optically active yellowish amorphous solid, [α]24D − 10 (c 1.0, MeOH). The IR absorptions implied the presence of hydroxy (3395 cm−1) and ester carbonyl (1726 cm−1) functionalities, while the UV absorption bands at λmax 225 and 271 nm indicated an indole chromophore [18]. The ESIMS (pos.) of 1 showed a molecular ion peak at m/z 733 [M]+ and the molecular formula was established as C44H53N4O6 from HRESIMS. Analysis of the 1H and 13C NMR data (Table 1) and the heteronuclear single-quantum correlation (HSQC) spectrum of 1 revealed the presence of seven sp3 methines, seven sp3 methylenes, seven methyls, eight sp2 methines, one sp3 quaternary carbon, and 14 sp2 quaternary carbons.

Table 1 1H and 13C NMR data of 1 and 2 in CD3ODa
Full size table

The gross structure of 1 was deduced from analyses of the 2D NMR data, including the 1H-1H correlation spectroscopy (COSY), HSQC, and heteronuclear multiple-bond correlation (HMBC) spectra in methanol-d4 (Fig. 2). The 1H-1H COSY and HSQC spectra revealed connectivity of seven partial structures a (C-9–C-12), b (C-3, C-14–C-16, C-5–C-6), c (C-18–C-19), d (C-10′–C-11′), e (C-5′–C-6′), f (C-3′, C-14′–C-15′), and g (C-18′–C-19′). These partial structures were classified into two units, A and B, as shown in Fig. 2.

Fig. 2
figure 2

2D NMR correlations of bisnaecarpamine A (1)

Full size image

In unit A, the HMBC correlations of H-6 (δH 3.60) and H-12 (δH 7.11) to C-8 (δC 130.9), and H-9 (δH 7.54) to C-7 (δC 109.9) and C-13 (δC 137.9) revealed the attachment of partial structure a as part of indole ring, while the HMBC correlations of H-6b to C-8, C-2 (δC 138.3) and C-7, H-3 (δH 5.20) to C-2 and C-7, H-14b (δH 2.68) to C-2 revealed the connectivity of partial structures a and b. In addition, the HMBC correlations of the N-4-methyl protons to C-5 (δC 54.6) and C-21 (δC 53.1) established the connections between C-5 and C-21 through a nitrogen atom (N-4). Another partial structure c, and the presence of a carboxylate moiety at C-16 were analyzed by the HMBC correlations of H-19 (δH 5.44) to C-15 (δC 34.2) and C-21, and H-16 (δH 2.71) to C-22 (δC 172.3) as shown in the Fig. 2. These data suggested that unit A possessed a vobasine-type skeleton [19].

In unit B, the 1H and 13C NMR data are highly similar to 12-methoxy-Nb-methylvoachalotine [20], and this similarity was verified by the 2D NMR spectra analysis. The HMBC cross-peaks of H-10′ (δH 6.53) to C-8′ (δC 125.6) and C-9′ (δC 133.0), H-11′ (δH 6.59) to C-12′ (δC 147.8) and C-13′ (δC 129.5), H-5′ (δH 5.14) and H-6′a (δH 3.68) to C-7′ (δC 103.0) revealed the attachment of partial structures d and e to the indole moiety, and those of H-19′ (δH 5.54) to C-15′ (δC 31.4) and C-21′ (δC 65.5), H3-18′ (δH 1.73) to C-20′ (δC 128.5) revealed the connectivity of partial structures f and g through C-20′. In addition, the HMBC correlations of the N-4′-methyl protons to C-5′ (δC 65.7) and C-21′ established the connections between C-5′ and C-21′ through a nitrogen atom (N-4′). The presence of a methoxy group at C-12′, a methyl group at N-1, a methyl carboxylate, and a hydroxymethyl at C-16′ were analyzed by the HMBC correlations as shown in the Fig. 2, revealing a sarpagine-type skeleton. Finally, the linkage between C-3 at unit A and C-9′ at unit B was provided by the HMBC correlation of H-10′ to C-3 (δC 41.1). Thus, the planar structure of bisnaecarpamine A was assigned as shown to be in Fig. 2.

Stereochemistry of bisnaecarpamine A (1)

The stereochemistry of each monoterpene indole unit in 1 was assigned by rotating-frame Overhauser effect spectroscopy (ROESY) correlations as shown in the computer-generated 3D drawing (Fig. 3). In unit A, the ROESY correlations of H-3/H-15 (δH 3.86) and H-14a (δH 1.91) suggested that those three protons were in the same plane and the ROESY correlations of H-15/H3-18 (δH 1.72) and H-19/H-21a (δH 3.10) established the E-configuration of the ethylidene side chain. The relative configuration of C-16 could not assign by ROESY because there was no ROESY correlation of H-3/H-16. Treatment of 1 using trimethylsilyldiazomethane (TMSCHN2) gave the methyl ester derivative of 1. The configuration of C-16 was proven to be S* by the highly shielded methyl chemical shift (2.62 ppm) of this derivative, which can be explained by the anisotropic effect of the indole ring [21]. While in unit B, the ROESY correlations of H-3/H-21′a (δH 4.39) suggested that H-3′ (δH 5.13) and H-21′a were in the same plane and the correlation of H-6′a to N-4′ suggested that H-6′a was α-oriented. Furthermore, the E-configuration of the ethylidene side chain was confirmed by the ROESY correlations of H-15′ (δH 3.45)/H3-18′.

Fig. 3
figure 3

Selected ROESY correlations and conformation of bisnaecarpamine A (1)

Full size image

The relation between units A and B of 1 was deduced from the ROESY correlations of H-6a and H-14b/H-10′ (δH 6.53) and H-3/H-6′b (δH 4.59) that suggested the relative configuration of 1 was as shown in Fig. 3.

Finally, the conformation of 1 through the C-3–C-9′ bond was investigated by using molecular mechanics calculations [22]. The global minimum obtained through Monte Carlo conformational search [22] indicated the dihedral angle of C-14–C-3–C-9′–C-10′ is ca. 120° (Fig. 3) and consistent with the conformer suggested by the observed ROESY correlations of H-6a and H-14b/H-10′ and H-3/H-6′b.

The absolute configuration of 1 was then assigned by comparing the experimental CD spectrum and the calculated CD spectrum as shown in Fig. 4. CD calculation was performed by Turbomole 7.1 [23] using RI-TD-DFT-B3LYP/SVPD level of theory on RI-DFT-B3LYP/SVP optimized geometries. The experimental CD spectrum show similar CD pattern compared to calculated CD spectrum. Therefore, the absolute configuration of 1 was proposed as shown in Fig. 1.

Fig. 4
figure 4

CD and UV spectra of bisnaecarpamine A (1)

Full size image

Compound 2 was isolated as an optically active brownish amorphous solid \([\upalpha]^{25}_{\text{D}}\): − 53 (c 1.0, MeOH). The ESIMS (pos) showed a molecular ion peak at m/z 719 [M]+. The IR absorptions implied the presence of hydroxy (3383 cm−1) and ester carbonyl (1727 cm−1) functionalities, while the UV absorption bands at λmax 225 and 287 nm indicated an indole chromophore [18]. The molecular formula C43H51N4O6 was established by HRESIMS [m/z 719.3821 [M]+, Δ + 2.1 mmu]. The 13C NMR data revealed the presence of seven sp3 methines, seven sp3 methylenes, six methyls, eight sp2 methines, one sp3 quaternary carbon, and 14 sp2 quaternary carbons (Table 1).

The NMR spectra of 1 and 2 exhibited similar resonances differ in the linkage position of the two monomer units and the presence of a hydroxy group in 2 was suggested instead of a singlet methyl signal (δH 3.86) of methoxy group at C-12′ in 1. Further analysis of the two-dimensional NMR data (1H-1H COSY, HSQC, and HMBC spectra in CD3OD) of 2 revealed that it was also composed of a vobasine and sarpagine indole alkaloid as well as 1. The ROESY correlation of H-10′/H-3 and H-10′/H-14b in Fig. 5, and the carbon chemical shift of C-10′ (δC 121.1), C-11′ (δC 105.5), and C-12′ (δC 147.8) suggested that the linkage between units A and B in 2 was C-3–C-11′ instead of C-3–C-9′ in 1.

Fig. 5
figure 5

Selected ROESY correlations and conformation of bisnaecarpamine B (2)

Full size image

The stereochemistry of each monoterpene indole unit of 2 was assigned by ROESY correlations as shown in the Fig. 5. The ROESY correlations of H-15/H3-18 and H-15′/H3-18′ established the E-configuration of the both ethylidene side chains. The β-orientation of H-3 and the α-orientation of H-3′ was elucidated by the ROESY correlations of H-3/H-15 and H-3′/H-21′a, respectively. The configuration of C-16 was proven to be S* by the ROESY correlation of H-16/H-21b. Analysis of the ROESY spectral data also suggested that 2 had similar relative configuration as in 1 for both unit A and B.

The conformation of 2 through the C-3–C-11′ bond was assigned by the ROESY correlations of H-3/H-10′ and H-9′/H-6′b as shown in the Fig. 5. The difference in the linkage of units A and B in 1 and 2 resulted in the difference of their total conformation, i.e., compound 1 possessed a twist conformation (Fig. 3), while compound 2 adopted an extended conformation with dihedral angle of C-14–C-3–C-11′–C-10′ is ca. 110° (Fig. 5).

Furthermore, to define the absolute configuration of 2, the TDDFT-ECD method was applied on the global minimum of 2. The result showed that calculated CD spectrum displayed a good similarity to the experimental CD spectrum of 2 as shown in Fig. 6.

Fig. 6
figure 6

CD and UV spectra of bisnaecarpamine B (2) 1: R=OCH3 2: R=OH

Full size image

A plausible biogenetic pathway for bisnaecarpamines A and B (1 and 2) is proposed as shown in Fig. 7. Their structures consisting of a vobasine-sarpagine type skeleton might be generated through C–N bond cleavage of a quaternary sarpagine-type alkaloid [20] followed by the introduction of another sarpagine-type alkaloid as a nucleophile. Nucleophilic attack to C-3 from C-9′ in the sarpagine-type alkaloid formed bisnaecarpamine A (1) as shown in route a, while bisnaecarpamine B (2) might be generated by an attack from C-11′ as shown in route b.

Fig. 7
figure 7

Plausible biogenetic pathway of 1 and 2

Full size image

Antimalarial activity

We have reported that some bisindole alkaloids showed antimalarial activity [5, 7, 17, 24]. Bisnaecarpamines A and B (1 and 2) were tested for the antimalarial activity against P. falciparum 3D7 strain. The result showed that 1 showed potent in vitro antimalarial activity [the half-maximal (50%) inhibitory concentration (IC50) = 4.60 μM] while 2 did not show activity even at 50 µM.

Compounds 1 and 2 have a different connection between units A and B causing them to take distinct conformations. The difference in conformations is one possible reason for the contrast in their antimalarial activity.

Experimental section

General experimental procedures

Optical rotations were measured on a JASCO DIP-1000 polarimeter. UV spectra were recorded on a Shimadzu UVmini-1240 spectrophotometer and IR spectra on a JASCO FT/IR-4100 spectrophotometer. High-resolution ESI MS were obtained on a LTQ Orbitrap XL (Thermo Scientific). 1H and 2D NMR spectra were measured on 600 MHz spectrometer at 300 K, while 13C NMR spectra were measured on a 150-MHz spectrometer. The residual solvent peaks were used as internal standards (δH 3.31 and δC 49.0 for CD3OD). Standard pulse sequences were used for the 2D NMR experiments. Merck silica gel 60 (40–63 µm), amino silica, and HP-20 were used for the column chromatography, and the separations were monitored by Merck silica gel 60 F254 or Merck amino silica gel F254 TLC plates.

Plant materials

The barks of Tabernaemontana macrocarpa Jack were collected in June 2018 from Centre for Plant Conservation Botanic Gardens, Bogor, Indonesia. The GPS coordinates of the field site are 6° 35′ 59.65″S, 106° 47′ 54.00″E. Authentication and identification of plant was carried out by Mr. Ikar Supriatna at Centre for Plant Conservation Botanic Gardens, Bogor, Indonesia.

Extraction and isolation

The barks of T. macrocarpa Jack (1520 g) were extracted with MeOH, and part of the extract (8 g) was treated with 3% tartaric acid (pH 2), and then partitioned with EtOAc. The aqueous layer was treated with saturated Na2CO3(aq) to pH 10 and extracted with CHCl3 to give an alkaloidal fraction (0.8 g) and continued by partition with butanol to give n-butanol fraction (1.6 g). The CHCl3 fraction was subjected to a column chromatography over Sephadex LH-20 and eluted with CHCl3/MeOH (1:1, v/v) to give 24 fractions. Separation of fraction 17 (63 mg) by a silica gel column eluted with n-hexane/ethyl acetate (49:1–1:1, v/v) continued by CHCl3/methanol (49:1–100%), gave six fractions. The fourth eluted fraction was subjected to ODS HPLC (Cosmosil C18 MS-II, 5 µm, 10 × 250 mm; eluent 24% CH3CN/ H2O; 0.1% TFA aq.; flow rate, 2.5 mL/min; UV detection at 254 nm) to afford bisnaecarpamine A (1, 3.2 mg, 0.00021%, tR 13.1 min) and bisnaecarpamine B (2, 4.1 mg, 0.00027%, tR 11.8 min).

Methyl esterification reaction

The preparation of methyl esters from carboxylic acids can be achieved with trimethylsilyldiazomethane (TMSCHN2) in methanolic benzene. Compound 1 (0.1 mg) was dissolved in MeOH (100 µl) and TMSCHN2 (50 µl) added. The reaction was easily monitored by the disappearance of the yellow color of TMSCHN2. The 1H NMR of reaction product showed a methyl signal at 2.62 ppm.

Parasite strain culture

P. falciparum laboratory strain 3D7 was obtained from Prof. Masatsugu Kimura (Osaka City University, Osaka, Japan). For the assessment of antimalarial activity of the compounds in vitro, the parasites were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 0.5 g/L l-glutamine, 5.96 g/L HEPES, 2 g/L sodium bicarbonate (NaHCO3), 50 mg/L hypoxanthine, 10 mg/L gentamicin, 10% heat-inactivated human serum, and red blood cells (RBCs) at a 3% hematocrit in an atmosphere of 5% CO2, 5% O2, and 90% N2 at 37 °C as previously described [25]. Ring-form parasites were collected using the sorbitol synchronization technique [26]. Briefly, the cultured parasites were collected by centrifugation at 840 g for 5 min at room temperature, suspended in a fivefold volume of 5% d-sorbitol (Nacalai Tesque, Kyoto, Japan) for 10 min at room temperature, and then they were washed twice with RPMI 1640 medium to remove the d-sorbitol. The utilization of blood samples of healthy Japanese volunteers for the parasite culture was approved by the institutional review committee of the Research Institute for Microbial Diseases (RIMD), Osaka University (approval number: 22-3).

Antimalarial activity

Ring-form-synchronized parasites were cultured with compounds 1–2 at sequentially decreasing concentrations (50, 15, 5, 1.5, 0.5, 0.15, 0.05, and 0.015 µM) for 48 h for the flow cytometric analysis using an automated hematology analyzer, XN-30. The XN-30 analyzer was equipped with a prototype algorithm for cultured falciparum parasites (prototype; software version: 01–03, (build 16)) and used specific reagents (CELLPACK DCL, SULFOLYSER, Lysercell M, and Fluorocell M) (Sysmex, Kobe, Japan) [27, 28]. Approximately 100 µL of the culture suspension diluted with 100 µL phosphate-buffered saline was added to a BD Microtainer MAP Microtube for Automated Process K2 EDTA 1.0 mg tube (Becton Dickinson and Co., Franklin Lakes, NJ, USA) and loaded onto the XN-30 analyzer with an auto-sampler as described in the instrument manual (Sysmex). The parasitemia (MI-RBC%) was automatically reported [27]. Then 0.5% DMSO alone or containing 5 µM artemisinin was used as the negative and positive controls, respectively. The growth inhibition (GI) rate was calculated from the MI-RBC% according to the following equation:

GI (%) = 100—(test sample—positive control)/(negative control—positive control) × 100.

The IC50 was calculated from GI (%) using GraphPad Prism version 5.0 (GraphPad Prism Software, San Diego, CA, USA) [29].

Bisnaecarpamine A (1): yellowish amorphous solid. \([\upalpha]^{24}_{\text{D}}\): − 10 (c 1.0, MeOH). IR Vmax (KBr): 3395 and 1727 cm−1. UV/Vis λmax (MeOH) (log ε) 225 (4.30), 271 (3.34) nm. CD (MeOH) λmax (Δε) 229 (− 10.4) and 289 (− 0.57) nm. 1H and 13C NMR (CD3OD): Table 1. MS (ESI): m/z: 733 [M]+. HRESIMS m/z: 733.4026 [M]+ (Calcd. for C44H53N4O6, 733.3960).

Bisnaecarpamine B (2): brownish amorphous solid. \([\upalpha]^{25}_{\text{D}}\): − 53 (c 1.0, MeOH). IR Vmax (KBr): 3386 and 1725 cm−1. UV/Vis λmax (MeOH) (log ε) 225 (4.66), 2.87 (4.12) nm. CD (MeOH) λmax (Δε) 225 (− 20.8) and 287 (− 3.38) nm. 1H and 13C NMR (CD3OD): Table 1. MS (ESI): m/z: 719 [M]+. HRESIMS m/z: 719.3821 [M]+ (Calcd. for C43H51N4O6, 719.9025).

CD calculation

The conformations were obtained using Monte Carlo analysis with MMFF94 force field and charges on Macromodel 9.1. CD calculations were performed in Turbomole 7.1 using RI-TD-DFT-B3LYP/SVPD level of theory on RI-DFT-B3LYP/TZVP optimized geometries.