Introduction

The plants belonging to Meliaceae have been reported to produce limonoids [1]. In our search for new bioactive compounds from medicinal plants, we have reported the isolation of new limonoids from plants of this genus [2,3,4,5,6,7,8,9,10,11,12,13,14,15], and alkaloids [16,17,18,19,20,21,22,23] and coumarins [24] showing antimalarial activity or inhibiting acetylcholinesterase. Ceramicine B, in particular, has been reported to show a strong lipid droplets accumulation (LDA) inhibitory activity on mouse pre-adipocyte cell line (MC3T3-G2/PA6) [6,7,8] and also anti-melanin deposition activity [9]. With the purpose of discovering limonoids with biological activity, we further investigated the constituents of C. ceramicus which led to the isolation of four new limonoids, ceramicines Q–T (1–4), and ceramicine R (2) showed potent antimalarial activity (Fig. 1). Their structures were determined on the basis of NMR and MS spectroscopic analyses in combination with NMR chemical shifts calculations.

Fig. 1
figure 1

Structures of 1–4

Full size image

Results and discussions

Compounds 1–4 were obtained as optically active white amorphous solids. The 1H and 13C NMR data (Tables 1 and 2) suggested the identity of 1–4 as ceramicine derivatives.

Table 1 1H and 13C NMR data of 1 and 2 in CDCl3
Full size table
Table 2 1H and 13C NMR data of 3 and 4 in CDCl3
Full size table

Ceramicine Q (1) was obtained as an optically active, \([{\upalpha }]_{{\text{D}}}^{27}\) + 119 (c 1.0, CHCl3), white amorphous solid and was revealed to have the molecular formula C25H30O5, by HRESITOFMS [m/z 433.1991 (M + Na)+, ∆ − 1.4 mmu]. IR absorptions implied the presence of α,β-unsaturated ketone (1683 cm−1) and hydroxy (3461 cm−1) groups. 1H and 13C NMR data (Table 1) revealed 25 carbon resonances due to one carbonyl, two sp2 quaternary carbons, four sp3 quaternary carbons, six sp2 methines, five sp3 methines, four sp3 methylenes, and three methyls. Among them, four sp3 carbons (δC60.4, 60.4, 66.7, and 74.3) and two sp2 methines (δC 139.7 and 142.5) were ascribed to those bearing an oxygen atom.

Analyses of the HSQC and 1H-1H COSY spectra (Fig. 2) revealed the presence of four partial structures; a (C-2 and C-3), b (C-5 ~ C-7), c (C-9, C-11, and C-12), d (C-15 ~ C-17), and e (C-22 ~ C-23). HMBC correlations of H3-18 to C-12, C-13, C-14 and C-17 suggested the connectivity of c, d, and C-14 through C-13. HMBC correlations of H-17 to C-21, H-16 to C-20, and H-23 to C-20 and C-21 suggested the presence of β-furyl at C-17, and the correlation of H2-16 to C-13 and C-14 completed the structure of ring D. The presence of ring C was deduced from the HMBC cross-peaks of H3-30 to C-7, C-8, C-9 and C-14, and the connectivity of b, C-1, C-5, and C-19 through C-10 was suggested by the HMBC correlations of H3-19 to C-1, C-5, C-9 and C-10. HMBC correlations of H-2 to C-10 and C-4, and H-3 to C-1 and C-5 suggested the presence of ring A. Finally, HMBC correlations of H-3 to C-29 (δC 60.4) suggested the presence of 1-oxaspiro[2.5]oct-4-ene of 1 as shown in Fig. 2.

Fig. 2
figure 2

Selected 2D NMR correlations of 1

Full size image

The relative configuration of 1 was assigned by analyses of the 1H-1H coupling constant data and the NOESY correlations (Fig. 2). First, H-6, H-17, CH3-19, and CH3-30 were assigned to be β-axially oriented from the NOESY correlations of H-6/H3-19 and H3-30/H-17 and H3-19, while H-5, H-9 and CH3-18 were deduced to possess α-orientation from the NOESY correlations of H-9/H-5 and H3-18. Both H-6 and H-7 should possess β-orientation because of the multiplicity pattern of H-6 (dd, 10.8, 2.4) and H-7 (d, 2.4). The stereochemistry of epoxy ring was elucidated to be as shown in Fig. 2 by the presence of NOESY correlations of H-29a/H-3 and H3-19, and H-29b/H3-19.

Ceramicine R (2), \([\upalpha ]_{{\text{D}}}^{28}\) 126 (c 0.5, CHCl3), was revealed to have the molecular formula C25H30O5 by HRESITOFMS. Its NMR data are highly similar to 1. However, the signals for oxymethylene of H2-29 in 1 are not observed in 2, and a methyl signal (δH 1.86) and two sp3 quaternary carbons bearing an oxygen atom (δC 55.6 and 68.0) are observed instead. Analysis of the 2D NMR data including HMBC and NOESY (Fig. 3) correlations supported the structure of 2 to be 4,5-epoxy derivative as shown in Fig. 1. Specifically, the HMBC correlations of H3-29 to C-3, C-4 (δC 55.6) and C-5 (δC 68.0), and H3-19 to C-5 supported its functionality in the structure of 2.

Fig. 3
figure 3

Selected 2D NMR correlations of 2

Full size image

NOESY correlations of H-6/H3-19 and H3-30, H-7/H3-30 and H-15, and H-9/H3-18 supported the stereochemistry of 2. Configurations of H-6 and H-7 taking β-orientation like as in 1 also supported by proton 3J coupling constant (J = 2.7 Hz).

The remaining problem regarding stereochemistry is that of epoxy in positions 4 and 5. Only the presence of two NOESY correlations of H-3/H3-29 and H-6/H3-19 is difficult to determine unequivocally. Therefore, we compared the predicted NMR chemical shifts of the two possible diastereomers, 2a (4,5-α-epoxy) and 2b (4,5-β-epoxy) as shown below (Fig. 4), calculated for the ωB97X-V/6-311+G* optimized conformations using ωB97X-V/6-311+G*(2df,2p).

Fig. 4
figure 4

Structures of the lowest-energy conformation of the two possible isomers (2a and 2b) of ceramicine R (2)

Full size image

Conformational searches and chemical shift calculations for compounds 2a and 2b were performed with the Spartan’20 software [25]. Stable conformers up to 40 kJ/mol for 2a and 2b were initially searched using the Merck molecular force field (MMFF) method. Stable conformers suggested were optimized using Hartree–Fock (HF)/3-21G (40 kJ/mol) and ωB97X-D/6-31G* (15 kJ/mol). Final energy optimization for the Boltzmann distribution was conducted at the ωB97X-V/6-311+G*(2df,2p) level of theory. NMR properties were calculated at the ωB97X-V/6-311+G*(2df,2p) level of theory and scaled. The obtained chemical shifts were corrected using the Boltzmann distribution to give calculated 13C chemical shifts.

The 13C chemical shift differences of predicted and experimental data are shown as graph in Fig. 5. Two atoms of C-6 and C-9 in 2b shows a deviation of greater than 5 ppm, but the average deviation is less than 2 ppm in 2a. Overall, the 13C chemical shifts predicted for structure 2a (Fig. 4) correlate very well with those observed for ceramicine R (2).

Fig. 5
figure 5

Parts per million difference between the calculated 13C NMR shifts for the two possible isomers (2a and 2b) and the experimental data of ceramicine R (2)

Full size image

Thus, we propose that the correct structure of ceramicine R (2) is the structure 2a. The prediction of 13C chemical shifts by DFT calculation is a very powerful tool for screening proposed structures and should be used more widely.

Ceramicine S (3), \([\upalpha ]_{{\text{D}}}^{27}\) + 34 (c 1.0, CHCl3), was revealed to have the molecular formula C28H36O6 by HRESITOFMS. Its 1H and 13C NMR data are highly similar to ceramicine R (2) (Table 2). However, the NMR data of 3 suggested that the oxymethylene moiety [δH 4.03 (d, 10.8), 4.59 (d, 10.8); δC 72.0] with acetyl group [δH 2.01 (3H, s); δC 21.0 and 171.3] was observed instead of the epoxide function in 2. IR absorptions implied the presence of α,β-unsaturated ketone (1684 cm−1), hydroxy (3502 cm−1), and acetyl (1733 cm−1) groups. The HMBC correlations of H2-28 and H3-2’ to C-1’ (δC 171.3) supported its functionality in the structure of 3. Finally, the NOESY correlations confirmed the position of the oxymethylene moiety with the acetyl group at C-28 and the α orientation of the hydroxy at C-6 and C-7 (See Supplementary Information).

By HRESITOFMS, ceramicine T (4), \([\upalpha ]_{{\text{D}}}^{28}\) + 67 (c 0.5, CHCl3), was revealed to have the molecular formula C29H36O7. Their NMR data are also highly similar to 3, differing only on the signals assigned to the presence of one acetyl and one methoxy carbonyl groups in 4 and the disappearance of oxy-methylene protons of 3. IR absorptions implied the presence of α,β-unsaturated ketone (1683 cm−1), hydroxy (3534 cm−1), and acetyl and/or methoxycarbonyl (1732 cm−1) groups. Analysis of the NMR data revealed the presence of a cyclopentanone[α]phenanthren ring system with a β-furyl ring at C-17.

The planar structure of 4 was deduced from the 1H-1H COSY correlations and the especially HMBC correlations of H3-29 and H3-3’ to C-28 and H3-30 to C-7. The relative configuration of 4 was then deduced from the NOESY correlations of H-6/H3-19, H3-29 and H3-30, H-9/H-5 and H3-18, H-7/H3-30, and H-12a/H3-30 and H-17 (See Supplementary Information). Thus, the structure of 4 was proposed to be as shown in Fig. 1.

Considering that 1–4 were isolated from the same extract as ceramicine B [3], their absolute configurations were assumed to be similar to ceramicine B based on the biogenetic relationships. Ceramicines Q and R might be generated from a cyclopentanone[α]phenanthren ring system followed by oxidative decarboxylation at C-28.

Antimalarial activity

Ceramicines Q–T (1–4) were tested for antimalarial activity against Plasmodium falciparum 3D7 strain. The assay showed that 2 had potent in vitro antimalarial activity [the half-maximal (50%) inhibitory concentration (IC50) = 2.8 µM, whereas 1, 3, and 4 did not (> 5.0 µM).].

Previously, we also reported some limonoids, ceramicines A–D with a cyclopentanone[α]phenanthren ring system with a tetrahydrofuran ring and a β-furyl ring from the barks of C. ceramicus, exhibited antimalarial activity against P. falciparum 3D7 in vitro [2, 3].

The activity might be depending on their unique cyclopentanone[α]phenanthren ring system and also be influenced by the presence of a tetrahydrofuran ring and its substituent patterns around rings A and B.

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 JMS-T100LP (JEOL). 1H and 2D NMR spectra were measured on a 400 MHz or 600 MHz spectrometer at 300 K, while 13C NMR spectra were on a 100 MHz or 150 MHz spectrometer. The residual solvent peaks were used as internal standards (δH 7.26 and δC 77.0 for CDCl3, δH 3.31 and δC 49.0 for CD3OD).

Material

The barks of C. ceramicus were collected in Terengganu, Malaysia in July 2013. The botanical identification was made by Prof. A. Hamid A. Hadi, University of Malaya. Voucher specimens (No. HOSHI13CCB) are deposited in the department of Pharmacognosy Hoshi University.

Extraction and isolation

The barks of C. ceramicus (8 kg) were extracted with MeOH to obtain 1.43 kg of extract. The MeOH extract was successively partitioned with n-hexane, EtOAc, n-BuOH and water. The n-hexane-soluble materials were separated further by silica gel column chromatography (n-hexane/EtOAc 1:0→1:1, CHCl3/MeOH 1:0→0:1) to obtain 10 fractions (A–J). Fraction I was separated further with a ODS silica gel column (MeOH/H2O 7:3→1:0, acetone) to obtain 6 fractions (I-1–I-6). Fraction I-3 was also separated by HPLC (Shiseido ODS MGII 30 × 250 mm, 85% aqueous MeOH at 8.0 mL/min, UV detection at 210 nm) into 7 fractions (I-3-a–I-3-g). Fraction I-3-c, I-3-d, and I-3-f are ceramicine F, G and B, respectively. Separation of fraction I-3-a by HPLC (Nacalai tesque Cholester 10 × 250 mm, 65% aqueous MeCN at 2.0 mL/min, UV detection at 210 nm) yielded ceramicine Q (1) (6.0 mg, 0.000075%, tR 18.6 min) and ceramicine R (2) (0.8 mg, 0.00001%, tR 21.0 min).

Separation of fraction I-3-e by HPLC (Nacalai tesque Cholester 10 × 250 mm, 60% aqueous MeCN at 2.0 mL/min, UV detection at 210 nm) yielded ceramicine S (3) (10.6 mg, 0.00013%, tR 44.0 min) and ceramicine T (4) (1.0 mg, 0.000012%, tR 50.0 min).

Ceramicine Q (1): white amorphous solid; \([\upalpha ]_{{\text{D}}}^{27}\) + 119 (c 1.0, CHCl3); UV (MeOH) λmax (ε) 204 (10,000) nm; CD (MeOH) λmax (Δε) 337 (− 1.59) and 216 (10.3) nm; IR (Zn-Se) νmax 3461 and 1683 cm–1; 1H and 13C NMR data (Table 1); ESIMS m/z 433 (M + Na)+; HRESIMS m/z 433.1977 (M + Na)+ [calcd for C25H30O5Na (M + Na)+: 433.1991].

Ceramicine R (2): white amorphous solid; \([\upalpha ]_{{\text{D}}}^{28}\) + 126 (c 0.5, CHCl3); UV (MeOH) λmax (ε) 204 (10,700) nm; CD (MeOH) λmax (Δε) 336 (− 2.36), 250 (2.18) and 230 (1.52) nm; IR (Zn-Se) νmax 3450 and 1697 cm–1; 1H and 13C NMR data (Table 1); ESIMS m/z 433 (M + Na)+; HRESIMS m/z 433.1982 (M + Na)+ [calcd for C25H30O5Na (M + Na)+: 433.1991].

Ceramicine S (3): white amorphous solid; \([\upalpha ]_{{\text{D}}}^{27}\) + 34 (c 1.0, CHCl3); UV (MeOH) λmax (ε) 204 (11,900) nm; CD (MeOH) λmax (Δε) 333 (− 2.96), 249 (Δ 0.36), 230 (Δ − 0.35) and 212 (Δ 3.79) nm; IR (Zn-Se) νmax 3502, 1733 and 1684 cm–1; 1H and 13C NMR data (Table 1); ESIMS m/z 491 (M + Na)+; HRESIMS m/z 491.2417 (M + Na)+ [calcd for C28H36O6Na (M + Na)+: 491.2410].

Ceramicine T (4): white amorphous solid; \([\upalpha ]_{{\text{D}}}^{28}\) + 67 (c 0.5, CHCl3); UV (MeOH) λmax (ε) 204 (12,000) nm; CD (MeOH) λmax (Δε) 336 (− 2.21), 234 (− 12.6), 206 (17.0) nm; IR (Zn-Se) νmax 3534, 1732 and 1683 cm–1; 1H and 13C NMR data (Table 2); ESIMS m/z 519 (M + Na)+; HRESIMS m/z 519.2361 (M + Na)+ [calcd for C29H36O7Na (M + Na)+: 519.2359].

13C chemical shift calculations

Conformational searches and chemical shift calculations for compounds 2a and 2b were performed with the Spartan’20 software [25]. Stable conformers up to 40 kJ/mol for 2a and 2b were initially searched using the Merck molecular force field (MMFF) method. Stable conformers suggested were optimized using Hartree–Fock (HF)/3-21G (40 kJ/mol) and ωB97X-D/6-31G* (15 kJ/mol). Final energy optimization for the Boltzmann distribution was conducted at the ωB97X-V/6-311+G*(2df,2p) level of theory. NMR properties were calculated at the ωB97X-V/6-311+G*(2df,2p) level of theory and scaled. The obtained chemical shifts were corrected using the Boltzmann distribution to give calculated 13C chemical shifts.

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 [26]. Ring-form parasites were collected using the sorbitol synchronization technique [27]. 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 1–4 at sequentially decreasing concentrations (50, 15, 5, 1.5, 0.5, and 0.15 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) [28, 29]. 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 [25]. 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:

$$ {\text{GI}}\left( \% \right) = {1}00{-}\left( {{\text{test sample}}{-}{\text{positive control}}} \right)/\left( {{\text{negative control}}{-}{\text{positive control}}} \right) \times {1}00 $$

The IC50 was calculated from GI (%) using GraphPad Prism version 5.0 (GraphPad Prism Software, San Diego, CA, USA) [30].