Introduction

Numerous monoterpenoid indole alkaloids having rich structural diversity have been isolated from natural resources. Many of them possess remarkable bioactivities and have been utilized as key lead compounds in drug development.

For many years, we have been conducting the chemical investigation of biologically active alkaloids produced by botanical medicinal resources. Some of our recent studies on the monoterpenoid indole alkaloids from genera Gelsemium and Ophiorrhiza are described herein.

Studies on alkaloids in Gelsemium plants

The genus Gelsemium, which belongs to the Loganiaceae, comprises three species: G. elegans Benth., G. sempervirens Ait., and G. rankinii Small, from which more than 50 indole alkaloids have been isolated [1, 2]. Gelsemium elegans, which is distributed over Southeast Asia, is known as a toxic plant and has been used in traditional Chinese medicine as a remedy for certain kinds of skin ulcers. Recently, some pharmacological effects, including analgesic [3], anti-inflammatory [4], and anti-tumor [5] effects, of G. elegans alkaloids have been reported. Gelsemium sempervirens and G. rankinii grow in the southeastern part of the United States of America, the former having been used in the treatment of neuralgia, migraine and spasmodic disorders such as asthma and whooping cough. The Gelsemium alkaloids have markedly diverse and complex architectures and are classified into six types (Fig. 1), based on their chemical structures, i.e., sarpagine, koumine, humantenine, gelsedine, gelsemine, and yohimbane.

Fig. 1
figure 1

Representative Gelsemium alkaloids and alkaloids 14 isolated from Yakatsu stored in Shosoin Repository

Chemical study on “Yakatsu” stored in Shosoin Repository

“Yakatsu” is one of the 60 kinds of medicines stored in Shosoin Repository, with accompanying dedicatory record, “Shuju-yaku-cho” (Memorandum of Medicines), in A.D. 756. Prior to the first scientific investigation in 1948, Yakatsu, listed last in the record, had been thought to have been scattered or lost. During the first scientific investigation, however, the possibility that a bundle of woody roots with the code name “N-127, Uyaku-no-zoku” might be Yakatsu was pointed out. On the occasion of the second scientific investigation, we isolated and identified the alkaloid constituents from this 1,250-year-old medicine. As a result, we succeeded in isolating four indole alkaloids, koumine (1), gelsevirine (2), gelsemine (3), and sempervirine (4), in their pure states from the crude base obtained from the MeOH extract of sample N-127, Uyaku-no-zoku (Fig. 1) [6]. The identified alkaloids were proof that the plant of origin of this medicine was a species of Gelsemium. The isolation of koumine (1), an alkaloid characteristic of the Asian species, G. elegans, indicated that the plant species was G. elegans. This result demonstrated that the medicine is, indeed, the Yakatsu listed in the dedicatory record and that the plant of origin was G. elegans, endemic to the southern part of China.

Alkaloids in Gelsemium elegans

We next carried out a chemical investigation of the alkaloid constituents of G. elegans, the plant of origin of Yakatsu, as well as its related plant, G. sempervirens, and isolated more than 50 compounds, including new gelsedine-, sarpagine-, and yohimbane-type alkaloids, as follows.

From the leaves of G. elegans were isolated new types of gelsenicine-related oxindole alkaloids (6–9). Gelsedilam (6) [7] possessed two carbon atoms fewer than those of common Gelsemium alkaloids (Fig. 2). Its 1H and 13C NMR data were very similar to those of the gelsedine-type alkaloid, gelsenicine (5), except for the lack of proton and carbon signals ascribable to the C18 and C19 positions. Furthermore, in addition to the carbonyl carbon due to the oxindole nucleus, an sp2 quaternary carbon at δ 179.8 was observed in the 13C NMR spectra. From the above data, compound 6 was deduced to be an 18,19-nor-20-lactam compound biogenetically derived from gelsenicine (5). The structure of 6 with a lactam ring and its absolute configuration were established by the partial synthesis of 6 from gelsenicine (5) (Fig. 3). Compound 5 was treated with trichloroethoxycarbonyl (Troc) chloride to give enamine carbamate 10. The double bond that migrated to the C19–C20 position was oxidatively cleaved in two steps [(a) OsO4, (b) Pb(OAc)4] to yield lactam 12. It was found that treatment of 10 with m-CPBA also afforded lactam 12. Finally, removal of the N b-Troc group gave gelsedilam (6). The 14-acetoxy derivative of 6, 14-acetoxygelsedilam (7), was also isolated (Fig. 2). The new alkaloids, gelsedilam (6) and 14-acetoxygelsedilam (7) are unprecedented 18,19-nor-type monoterpenoid indole alkaloids.

Fig. 2
figure 2

Structures of gelsenicine (5) and new alkaloids 6 and 7 isolated from Gelsemium elegans

Fig. 3
figure 3

Chemical transformation of gelsenicine (5) into gelsedilam (6)

Gelsefuranidine (8) [7] is a derivative of 14-hydroxygelsenicine (13), having a 2-substituted furan residue and a propenyl group (Fig. 4). The structure, including the absolute configuration, was confirmed by condensation of 14-hydroxygelsenicine (13) and furfural under acidic conditions. Gelsefuranidine (8) is the first example of a monoterpenoid indole alkaloid having a furan residue on the side chain.

Fig. 4
figure 4

Structures of new alkaloids 8 and 9 and known compounds 13 and 14 isolated from Gelsemium elegans

Gelseiridone (9) [7], whose structure was deduced by spectroscopic analysis, is a new type of alkaloid having a nitrogen–carbon linkage between a gelsenicine-type monoterpenoid indole alkaloid that possesses an α, β-unsaturated ketone residue as well as the N b-C20 seco-form and a monoterpene unit having an iridoid skeleton (Fig. 4). From a biogenetic point of view, 9 would have originated from 14-hydroxygelsenicine (13) and 7-deoxygelsemide (14), a coexisting iridoid in this plant.

Gelsemoxonine was first isolated by Lin et al. from G. elegans in 1991 [8]. On the basis of spectroscopic analysis, it was proposed to be an unusual N b-C20 seco-oxindole gelsedine-type alkaloid (15). However, during the reinvestigation of the chemical constituents of G. elegans, we found that gelsemoxonine (16) had an unusual azetidine unit (Fig. 5) [9]. Acetylation of gelsemoxonine under ordinary conditions gave an unexpected diacetylated derivative. This finding raised doubts regarding the reported structure 15 of gelsemoxonine. The HMBC spectrum was measured at −30°C in pyridine-d 5 by utilizing the phenomenon that the signal of N b-H in 16 becomes sharp and well defined at low temperature. As a result, clear correlations between N b-H and the two carbon atoms, that is, the oxymethine carbon at C14 and the ketone carbon at C20, were observed. Formula 15 cannot explain these long-range couplings that enabled us to construct an azetidine ring consisting of N b, C15, C16 and C5 positions. Finally, structure 16, with an azetidine unit, was confirmed by X-ray crystallographic analysis. This is the first example of azetidine-containing monoterpenoid indole alkaloids.

Fig. 5
figure 5

Structures of gelsemoxonine and selected HMBC correlations and X-ray structure of gelsemoxonine

Some gelsedine-type alkaloids (1721), possessing an oxygen function at the C14 position or C14 and C15 positions, were isolated from the leaves of G. elegans (Fig. 6) [9, 10]. 14,15-Dihydroxygelsenicine (18) [9] was presumed to be the biosynthetic precursor of gelsemoxonine (16).

Fig. 6
figure 6

Structures of new alkaloids 1721 isolated from Gelsemium elegans

Alkaloids in Gelsemium sempervirens

Five new sarpagine-type alkaloids, gelsempervine-A (22), -B (23), -C (24), -D (25), and 19Z-16-epi-voacarpine (26), were isolated from the dried radix of G. sempervirens Ait. (Fig. 7) [11]. The structure of gelsempervine-A (22), with a 3-oxo, i.e., 2-acyl indole nucleus, was determined by spectroscopic analysis and chemical conversion from the known sarpagine-type alkaloid, 16-epi-voacarpine (27). However, perusal of the UV spectrum of 22 revealed typical absorptions of the indole nucleus in MeOH. The 2-acyl indole alkaloids are known to exhibit a characteristic absorption at around 310 nm. In order to investigate this unusual observation in the UV spectrum, we next measured UV as well as NMR spectra in protic or aprotic solvents. As a result, we found that 22 exhibited the typical absorption patterns of indole and 2-acyl indole alkaloids in MeOH and CH3CN, respectively. Quite interestingly, the signals of the protons and carbons at C5 and C21 in CD3OD were observed at low fields compared with those in CD3CN. On the other hand, the signals of H2-14 were observed at lower fields in CD3CN than in CD3OD. These spectroscopic data suggest that gelsempervine-A (22) existed exclusively as a C/D ring-opening structure with the keto-amine form (A), in such aprotic solvents as CD3CN, and as a trans-annular structure with the zwitterionic form (B), in such protic solvents as CD3OD. The above-described phenomena in the UV and NMR spectra of compound 22 were also observed in three new compounds, gelsempervine-B (23), -C (24), and -D (25). Therefore, it was found that 2-acyl sarpagine-type alkaloids possessing an N b-methyl group take a keto-amine form or a trans-annular form in solution, depending on the solvent.

Fig. 7
figure 7

Structures of new sarpagine-type alkaloids 2226 isolated from Gelsemium sempervirens and 16-epi-voacarpine (27), and two structural forms and UV spectra of gelsempervine-A (22)

Sempervilam (28), a yohimbane-type alkaloid, was isolated from the dried radix of G. sempervirens (Fig. 8) [11]. The total synthesis of 28 was achieved using 3,4-dihydroharman (29). Compound 29 was condensed with cyclohexene-1-carboxylic acid using 1-hydroxy-7-azabenzotriazole (HOAT) and 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide·HCl (EDCI) to give amide 30. Photocyclization of 30 gave cycloadduct 31. Oxidation of the D-ring with DDQ, and aromatization of the C-ring by treatment with tBuOCl and then with DBU afforded 28.

Fig. 8
figure 8

Structure and synthesis of sempervilam (28) isolated from Gelsemium sempervirens

Three gelsedine-type alkaloids, GS-1 (32), GS-2 (33), and GS-3 (34), were isolated from the dried stems of G. sempervirens (Fig. 9) [12]. GS-2 (33) was also isolated from the fresh leaves of this plant [12].

Fig. 9
figure 9

Structures of new alkaloids 3234 isolated from Gelsemium sempervirens

Evaluation of cytotoxicity of Gelsemium alkaloids

The cytotoxic effects of 14 Gelsemium alkaloids [sarpagine type, 19Z-akuammidine; koumine type, koumine (2); humantenine type, humantenine, 11-methoxyhumantenine; gelsedine type, gelsenicine (5), 14-hydroxygelsenicine (13), 14-acetoxygelsenicine (17), 14,15-dihydroxygelsenicine (18), 14-acetoxy-15-hydroxygelsenicine (19), GS-2 (33), gelsedine (35), gelsemicine (36), gelsemoxonine (16); gelsemine type, gelsemine (3)] on some tumor cells were evaluated. Among them, the gelsedine-type alkaloids, 14-acetoxygelsenicine (17), 14,15-dihydroxygelsenicine (18), gelsedine (35), and gelsemicine (36) showed relatively strong cytotoxic effects on the A431 human epidermoid carcinoma cell line (Fig. 10) [10]. Their cytotoxicities were compared with that of a positive control, cisplatin (EC50=3.5 μM), a drug used for the treatment of skin cancer. In particular, 14-acetoxygelsenicine (17) showed high cytotoxicity (EC50=250 nM). Further chemical and pharmacological studies on Gelsemium alkaloids are in progress in our laboratories.

Fig. 10
figure 10

Gelsedine-type alkaloids possessing cytotoxic activity against A431 human epidermoid carcinoma cells

Chemical studies on alkaloids in Ophiorrhiza plants distributed in Japan and on callus, regenerated plants and hairy roots of Ophiorrhiza pumila

Camptothecin (37) [13] is a well-known monoterpenoid indole alkaloid possessing remarkable anti-tumor activity and was isolated for the first time from Camptotheca acuminata (Nyssaceae) [14]. To date, several camptothecin-producing plants, including Ophiorrhiza mungos [15] and O. filistipula [16] (Rubiaceae), have been reported. Medicinal chemistry studies of camptothecin have been actively performed [17], and, at present, two semi-synthetic camptothecins, topotecan and irinotecan, are used clinically as anti-tumor agents.

Alkaloids of Ophiorrhiza plants distributed in Japan

In order to discover new camptothecin-related alkaloids, investigations of the constituents of Ophiorrhiza plants that are distributed in Japan, that is, O. pumila Champ., O. liukiuensis Hayata, and O. japonica Bl., were carried out. As a result, it was found that O. pumila produces camptothecin (37) and its related alkaloids (3843) [1821], whereas O. japonica produces β-carboline-type alkaloids (4552) and no camptothecin-related alkaloids (Fig. 11) [22, 23]. Another Ophiorrhiza plant, O. liukiuensis, was found to produce both camptothecin-related alkaloids (37, 38, 4042, 44) and β-carboline-type alkaloids (45, 48, 49, 52) [24].

Fig. 11
figure 11

Alkaloids isolated from Ophiorrhiza plants distributed in Japan

From O. pumila, new alkaloids, pumiloside (38) [18], (3S)- and (3R)-deoxypumilosides (39 and 40) [18, 21] and the first natural glyco-camptothecin, chaboside (43) [19, 20], were isolated. Pumiloside (38) and deoxypumilosides (39 and 40) showed UV absorptions indicative of a quinolone or a quinoline chromophore, respectively. In their 1H NMR spectra, a set of three protons on a vinyl group, an acetal proton, and one sugar unit containing an anomeric proton were observed, respectively, which are very similar to those of strictosamide (41). The above data indicated that pumiloside (38) and deoxypumilosides (39 and 40) are hybrid-type compounds possessing the partial structures of both camptothecin (37) and strictosamide (41), which is established as a biosynthetic intermediate of camptothecin (37). Therefore, they seemed to be plausible biogenetic precursors of camptothecin (37). To establish these structures, including the absolute configurations, we synthesized three new camptothecinoids (3840) starting from tryptamine (53) and secologanin (54) (Fig. 12). Condensation of 53 and 54 via the Pictet–Spengler reaction under acidic conditions, followed by D-ring closure by treatment with alkali, gave strictosamide (41) having 3S and vincoside lactam (55) having 3R stereochemistry, respectively. Pumiloside tetra-acetate (56) was prepared from 41 by a three-step procedure that included acetylation of the hydroxyl groups in the glucose unit, oxidative bond cleavage at C2–C7, and ring closure between C2 and C6 to give a quinolone ring. Removal of the protecting group of 56 gave pumiloside (38) having 3S stereochemistry. Next, the oxygen function at the C7 position in 56 was removed in two steps as follows. Compound 56 was treated with lithium diisopropylamide (LDA) and then with N-phenyltrifluoromethanesulfonimide. The enol triflate 59 thus obtained was treated with Pd(OAc)2, 1,1′-bis(diphenylphosphino)-ferrocene (DPPF), Et3N and HCOOH in dioxane to afford deoxygenated compound 61. Deacetylation of 61 gave (3S)-deoxypumiloside (39). The same procedure was applied to vincoside lactam (55) to give (3R)-deoxypumiloside (40). The spectroscopic data of synthetic pumiloside (38) and deoxypumilosides (39 and 40), including the CD spectra, were identical with those of the natural compounds, respectively, thereby establishing the structures of these new alkaloids.

Fig. 12
figure 12

Synthesis of pumiloside (38) and (3S)- and (3R)-deoxypumilosides (39 and 40)

Secondary metabolites in callus cultures, regenerated plants and hairy roots of Ophiorrhiza pumila

To establish an efficient method for the production of camptothecin (37), as well as finding novel secondary metabolites and a clue to clarifying the camptothecin biosynthetic pathway, we started related biotechnology research of O. pumila that was found to produce camptothecin-related alkaloids exclusively. As a result, we succeeded in obtaining callus cultures [25] from the leaves and shoots, regenerated plants [26, 27], and hairy roots [28] of O. pumila. From the investigation of their constituents, it was found that well-growing callus cultures produced anthraquinones that could not be detected in the wild-type plants, and the regenerated plants and the hairy roots produced almost the same types of alkaloids as did the wild-type plants. From the regenerated plants that were obtained from the callus cultures originally derived from the leaf segment, camptothecin (37) and a new alkaloid, 9-β-d-glucosyloxycamptothecin (63), together with five camptothecin-related alkaloids (38, 4043), were isolated. The structure, including the absolute configuration, of the new glucosyloxycamptothecin (63) was determined by chiral total synthesis (Fig. 13) [27]. The Friedländer condensation of the A-ring moiety 65 and CDE-ring counterpart 67 was adopted for the construction of the entire molecule. First, 6-glucosyloxy-2-aminobenzaldehyde (65), corresponding to the A-ring part of 63, was prepared from known phenol 64 in two steps. The CDE-ring moiety 67 was prepared from known (S)-66. The condensation of 65 and 67 in AcOH-MeOH under reflux gave camptothecin skeleton 68. Deacetylation of 68 afforded 9-β-d-glucosyloxycamptothecin (63).

Fig. 13
figure 13

Structure and synthesis of 9-β-d-glucosyloxycamptothecin (63)

Hairy roots of O. pumila were obtained from the regenerated plant stems by the direct inoculation of Agrobacterium rhizogenes (pRi15834, pGSGluc1) [28]. From the hairy roots cultivated on solid medium or in liquid medium, camptothecin (37) and its related alkaloids (3841) were isolated, together with two new alkaloids, OPHR-23 (69) and OPHR-17 (70) (Fig. 14) [29]. This result revealed that the hairy roots of O. pumila retained their alkaloid-producing ability. HPLC analysis of the liquid medium indicated that camptothecin (37), pumiloside (38) and (3R)-deoxypumiloside (40) were excreted into the culture medium [30, 31]. From spectroscopic analysis, new alkaloids (69 and 70) were determined to be a pair of diastereomers at the C17 position, having characteristic structures due to the ABC-ring of camptothecin (37) and the CD-ring of strictosamide (41), pumiloside (38), and deoxypumilosides (39 and 40), together with acetal and methoxy groups. To determine their structures, we carried out the chemical conversion from (3R)-deoxypumiloside (40) into the new alkaloids (Fig. 15). After protection of the hydroxyl groups in the glucose unit with Troc group, 40 was treated with DDQ in toluene–MeOH, followed by deprotection to give new alkaloids 69 and 70. The configuration at the C17 position in OPHR-23 (69) and OPHR-17 (70), a pair of diastereomers at C17, was determined to be S and R, respectively, by NMR techniques, including differential NOE experiments and PFG J-HMBC 2D spectroscopy.

Fig. 14
figure 14

Structure of OPHR-23 (69) and OPHR-17 (70)

Fig. 15
figure 15

Chemical transformation of (3R)-deoxypumiloside (40) into OPHR-23 (69) and OPHR-17 (70) and stereochemical analysis at C-17 in 69 and 70

Camptothecin (37), which has a quinoline nucleus, is biogenetically derived from strictosamide (41), as was proven by Hutchinson et al. [32]. However, the biosynthetic pathway from strictosamide (41) to 37, which comprised (i) transformation of an indole 6-5-6 ring system of the ABC-ring into a quinoline 6-6-5 ring system, (ii) oxidation of the D-ring to a pyridone, and (iii) removal of a glucose unit and structural conversions in the E-ring, is not yet clear. The existence of pumiloside (38) and (3S)-deoxypumiloside (39) suggests the possibility that the conversion of the ABC-ring of strictosamide (41) would be the next step after the formation of strictosamide (41), followed by D-ring oxidation and transformation of the E-ring (Fig. 16).

Fig. 16
figure 16

Speculated biogenetic pathway of camptothecin (37) from strictosamide (41)

The biosynthetic pathway of camptothecin (37) from [1-13C]glucose was investigated by in vivo tracer experiments using the hairy roots of O. pumila [33]. From the hairy roots that were grown in the presence of [1-13C]glucose, 13C-enriched camptothecin was isolated, and the 13C incorporated positions were determined by 13C NMR measurements. The 13C labeling pattern of camptothecin (37) isolated from the hairy roots clearly showed that the secologanin moiety was synthesized via the 2C-methyl-d-erythritol 4-phosphate (MEP) pathway, and not via the mevalonate pathway. Further studies are needed to clarify the complete features of camptothecin biosynthesis.