Abstract

Graphical abstract

Open in a separate window

RESULTS AND DISCUSSION

A sample of the stem bark of S. asper collected in Vietnam was extracted with MeOH, and the extract was partitioned with n-hexane and CHCl₃. When the cytotoxic CHCl₃ partition was subjected to chromatographic separation guided by inhibitory activity against the HT-29 cell line, three new cardiac glycosides, (+)-19-hydroxykamaloside (1), (+)-5-hydroxyasperoside (2), and (+)-3'-de-O-methylkamaloside (3), and two known analogues, (+)-3-O-β-d-fucopyranosylperiplogenin (4)¹⁸ and (+)-strebloside (5)⁵ were isolated.

Compound 1 was purified as an amorphous colorless powder with a molecular formula of C₃₁H₄₈O₁₀, as shown by a sodiated molecular ion peak at m/z 603.3153 (calcd 603.3140) observed in the HRESIMS, in conjunction with the ¹H and ¹³C NMR spectroscopic data (Tables 1 and ​and2).2). The UV (λ_max 216 nm) and IR (ν_max 3459, 1743 and 1621 cm⁻¹) spectra showed the absorptions of hydroxy groups and an unsaturated lactone unit.^19,20 The ¹H and ¹³C NMR data of 1 (Tables 1 and ​and22)²¹ exhibited the characteristic resonances for a cardiac glycoside, including two methyl groups, a sugar moiety, and a lactone unit.^19,20 Comparison of the ¹³C NMR data of 1 with those of (+)-strebloside (5)⁵ showed that the resonance for C-19 of 5 at δ_C 208.3 was replaced by that at δ_C 65.0 in 1 (Table 2). The resonances for C-5 and C-10 of 5 were different from those for these carbons of 1, but they were closely comparable for other carbons of both compounds. Thus, compound 1 could be defined as 19-nor-10-hydroxymethylstrebloside, an analogue of digoxin.

The conformation of the cardenolide ring system of digoxin has been proposed previously by investigation of its NMR spectroscopic parameters,²² and this was confirmed by that generated by a Mercury program,²³ based on the reported crystal structure of digoxin (Figure 1).¹⁰ Using Snatzke’s method, the ECD spectrum induced by [Mo₂(OAc)₄] of digoxin showed a positive Cotton effect (CE) at 310 nm and a positive O-C-C-O dihedral angle (Figure 2), which indicated a (3'cS, 4'cS) absolute configuration.²⁴ Thus, a (3S, 5R, 8R, 9S, 10S, 12R, 13S, 14S, 17R, 1'aR, 3'aS, 4'aS, 5'aR, 1'bS, 3'bS, 4'bS, 5'bR, 1'cS, 3'cS, 4'cS, 5'cR) absolute configuration was confirmed for digoxin from its induced ECD data and previously published crystal structure. A positive CE at 239 nm observed in the ECD spectrum of digoxin (Figure 3) correlated to a 17R configuration. This same configuration could be assigned for 1 from its ECD spectrum, which was consistent with that of digoxin (Figure 3). This determination was also supported by the resonances for H-17 and C-18 of 1, which were consistent with those reported for a cardiac glycoside comprising a 17β-lactone unit but different from those published for a 17-epi-cardiac glycoside containing a 17α-lactone unit.²⁵ These changes would result from the stereoelectronic effects of the C-17 lactone unit, the same as those reported for bicyclic alcohols.²⁶

Open in a separate window

Open in a separate window

Figure 1

A. COSY (, ¹H¹H) and key HMBC (, ¹H¹³C) correlations of 1. B. Selected NOESY (, ¹H¹H) correlations of 1. C. Partial structure showing the conformation of the central ring system as generated by Mercury CSD Version 3.1.1,²³ and based on the reported crystal structure of digoxin.¹⁰ D. Selected NOESY (, ¹H¹H) correlations of digoxin.

Open in a separate window

Figure 2

A. ECD spectrum of digoxin induced by [Mo₂(OAc)₄] (IECD, 1.0 mM digoxin in a 1.2 mM DMSO solution of dimolybdenum tetraacetate [(MoAc₂)₂], 1.2 mM, in DMSO). B. The relationship between the absolute configuration of C-3'c and 4'c of digoxin and the sign of the O-C-C-O dihedral angle.

Open in a separate window

Figure 3

ECD (A and B) and UV (C and D) spectra of compounds 1–5 and digoxin.

Conformational effects have been also observed for other protons and carbons of cardenolides. For example, the chemical shift of C-4 at δ_C 41.3 of a 3α-hydroxycardenolide appeared at δ_C 36.9 in its 3-epimer.²⁷ Also, the NMR resonances for H-7/C-7 at δ_H 2.0/δ_C 28.0 and H-9/C-9 at δ_H 0.9/δ_C 50.0 in a 5αH-cardenolide shifted to those at δ_H 1.5/δ_C 22.0 for H-7/C-7 and at δ_H 1.7/δ_C 35.0 for H-9/C-9 in its 5-epimer.²⁸ Following these observations, the resonances for C-4 at δ_C 34.9, C-7/H-7 at δ_C 19.0/δ_H 1.39 and 2.59, and C-9/H-9 at δ_C 39.1/δ_H 1.57 (Table 2) of 1 supported the presence of 3-O-β-glycosyl and 5β-hydroxy groups.^27–29 The occurrence of the same conformation for the cardenolide core of 1 as that for digoxin was also confirmed by the NOESY correlations between H-3 and H-7α, H-7α and H-9, OH-5 and H-8, H-8 and H-18, and H-18 and H-19 of 1 (Figure 1).

Owing to the overlapped signals observed for H-1'–H-3' and H-5' of 1, determination of the relative configuration of the sugar unit proved to be difficult. Fortunately, this problem was resolved when acetone-d₆ rather than CDCl₃ was employed as the NMR solvent (Table 1). The NOESY correlations observed between H-1' and H-9, H-3', H-5', and OCH₃-2', and between OH-4' and OH-5, OH-19, and H-6' (Figure 1), indicated that H-1', H-3', H-4', and H-5' are α-oriented, and H-2' and H-6' are β-oriented. Therefore, a (3S, 5S, 8R, 9S, 10R, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'S, 5'R) absolute configuration could be assigned for 1 [(+)-19-hydroxykamaloside], with the structure determined as (3β,5β)-3-[(2,3-di-O-methyl-β-d-fucopyranosyl)oxy]-5,14,19-trihydroxycard-20(22)-enolide.

Compound 2 was isolated as an amorphous colorless powder. The similar UV and IR spectra to those of 1 indicated that 2 is also a cardiac glycoside. The same molecular formula of 2 [positive-ion HRESIMS m/z 603.3152 (calcd for C₃₁H₄₈O₁₀Na, 603.3140), as observed in the HRESIMS] to that of 1 showed these compounds to be regioisomers. Comparison of their ¹H and ¹³C NMR spectroscopic data (Tables 1 and ​and2)2) suggested that both compounds contain a 5,14-dihydroxycardenolide core substituted with a 2',3'-di-O-methylglycose unit at C-3. The 19-hydroxymethyl and 6'-methyl groups of 1 were substituted by a 19-methyl group and a 6'-hydroxymethyl group in 2, as supported by the HMBC cross-peaks observed between H-19 and C-1 and C-9, H-6' and C-4', and H-3' and C-6' for 2 (Figure S14, Supporting Information).

A 17R absolute configuration could be defined for 2 from the positive CE at 236 nm observed in its ECD spectrum, consistent with that of 1, and this was supported by the resonances for H-17 at δ_H 2.81 and C-18 at δ_C 15.7, which were indicative of a 17β-lactone unit.²⁵ A 3-O-β-glycosyl moiety and a 5β-hydroxy group were evident in 2 due to the resonances for C-4 at δ_C 34.2, C-7/H-7 at δ_C 23.8/δ_H 1.17 and 1.89, and C-9/H-9 at δ_C 39.2/δ_H 1.51 (Table 2).^27–29 These were confirmed by the NOESY correlations observed between H-3 and H-7α, H-7α and H-9, H-8 and H-19, and H-18 and H-19 (Figure S15, Supporting Information). Thus, the same conformation and relative configuration of the cardenolide moiety of 1 occurred in 2. The NOESY correlations observed between H-1' and H-3, H-3', and H-5', and between H-4' and H-2' and H-6' (Figure S15, Supporting Information), indicated an α-orientation for H-1', H-3', H-5' and a β-orientation for H-2', H-4', and H-6'. These assignments were supported by the resonances for C-1'–C-6' of 2, which were closely comparable with those of asperoside.⁶ Thus, a (3S, 5S, 8R, 9S, 10R, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'R, 5'R) absolute configuration could be defined for 2 [(+)-5-hydroxyasperoside], with its structure characterized therefore as (3β,5β)-3-[(2,3-di-O-methyl-β-d-glucopyranosyl)oxy]-5,14-dihydroxycard-20(22)-enolide.

Compound 3 was isolated as an amorphous colorless powder. The similar UV, IR, and NMR spectra with those of 1 and 2 indicated that 3 is a further cardiac glycoside. The molecular formula determined from the positive-ion HRESIMS m/z 573.3054 (calcd for C₃₀H₄₆O₉Na, 573.3034) indicated an OCH₂ unit to be missing from 3, when compared with either 1 or 2, as confirmed by the NMR resonances for a single methoxy group at δ_H 3.38 s and δ_C 57.1 observed for 3 (Tables 1 and ​and2).2). Comparison of the ¹H and ¹³C NMR spectroscopic data of 3 with those of the known compound 4 (Tables 1 and ​and22 and Table S1, Supporting Information) indicated that both compounds possess the same cardenolide aglycone but a different sugar unit. A methoxy group was proposed at C-2' of 3, as indicated by the HMBC cross-peak observed between the methoxy group and C-2'. The 17R absolute configuration could be defined for 3 from a positive CE at 241 nm in the ECD spectrum, which was consistent with 1 and 2. The same conformation as that of 2 for its cardenolide unit was evident for 3, from the closely comparable NMR resonances for this moiety of 2 and 3. A (3S, 5S, 8R, 9S, 10R, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'R, 5'R) absolute configuration was characterized for 3 from the NOESY correlations observed between H-3 and H-7α and H-17, H-7α and H-9, H-8 and H-18 and H-19, H-18 and H-22, H-1' and H-3, H-3', H-4', and H-5', and between OH-4' and H-6' (Figure S15, Supporting Information). Therefore, compound 3 was assigned structurally as (3β,5β)-3-[(2-O-methyl-β-d-fucopyranosyl)oxy]-5,14-dihydroxycard-20(22)-enolide, and has been accorded the trivial name, (+)-3'-de-O-methylkamaloside.

The structures of the two known cardiac glycosides 4 and 5 isolated from the stem bark of S. asper in the present study were determined by comparison of their spectroscopic data (Tables 1 and ​and22 and Table S1, Supporting Information) with literature data for (+)-3-O-β-d-fucopyranosylperiplogenin (4)¹⁸ and (+)-strebloside (5),⁵ respectively. The (3S, 5S, 8R, 9S, 10R, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'R, 5'R) and (3S, 5S, 8R, 9S, 10S, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'S, 5'R) absolute configurations were evident for 4 and 5, respectively, from their ECD and 2D NOESY NMR spectra, which were consistent with those of 1 (Figures 1 and ​and33 and Figures S15 and S16, Supporting Information).

Compounds 1–5, along with a commercially available sample of digoxin, were evaluated for their cytotoxicity against the HT-29 human colon cancer cell line, using paclitaxel as the positive control (Table 3).³⁰ The compounds obtained in sufficient quantity were also tested for their activity toward the human MV4–11 and Kasumi-1 leukemia cell lines³¹ and the H1299 human non-small cell lung cancer cell line,³² using silvestrol and paclitaxel as the positive controls, respectively (Table 3). All compounds tested showed potent cytotoxicity that was comparable with digoxin, with IC₅₀ values in the range 93–690 nM (Table 3), indicating that they are the cytotoxic principles of the stem bark of S. asper. Interestingly, both compounds 1 and 4 showed more potent activity than paclitaxel toward H1299 cells (Table 3).

The major active compound, (+)-strebloside (5), and digoxin were also tested for their selectivity, using normal human CCD-112CoN colon, NL20 lung, and peripheral blood mononuclear cells.^31–33 The potency of both agents toward normal human cells was found to be much lower than that against human cancer cells, indicating that both 5 and digoxin showed selective cytotoxicity toward human colon and lung cancer and leukemia cells. The selectivity of 5 observed toward HT-29 cells and CCD-112CoN cells was greater than digoxin (Tables 3 and ​and44).

To characterize the functional groups that were responsible for the cytotoxicity of (+)-strebloside (5), several derivatives (Figure 4) were synthesized from 5 (Scheme 1 and Schemes S1–S4, Supporting Information), which was isolated in a relatively large quantity from S. asper with a yield of 0.15‰ w/w. The synthesized compounds were defined structurally through comparison of their spectroscopic data with those of 5 (Tables 1 and ​and22 and Tables S1–S4, Supporting Information) and evaluated by their cytotoxicity against the HT-29 cell line (Table 5).³⁰

Open in a separate window

Figure 4

Structure of synthesized derivatives of (+)-strebloside (5).

Open in a separate window

Scheme 1

Plausible mechanism of formation of 5g from the reaction of 5 and TFA.

Previously, the cytotoxic potency of a lignan lactone was found to be increased by acetylation of its saccharide unit,³³ and the presence of different glycosyl units of cardiac glycosides contributed diversely to the cytotoxicity of these compounds, as exemplified by 2 (IC₅₀ 690 nM) and 4 (IC₅₀ 90 nM) (Table 3). Thus, acetylation was included as the first step in the investigation of the structure-cytotoxicity relationship of 5.

When 5 was treated with Ac₂O and pyridine, (+)-4'-O-acetylstrebloside (5a) was produced. This product showed a sodiated molecular ion peak at m/z 643.3109 for a molecular formula of C₃₃H₄₈O₁₁Na and the NMR resonances for a C-4' acetyl group (Table S1, Supporting Information), as indicated by the HMBC cross-peak between H-4' and the acetyl carbonyl group (Figure S14, Supporting Information). A (3S, 5S, 8R, 9S, 10S, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'S, 5'R) absolute configuration was defined for 5a from its ECD and 2D NOESY NMR spectra, which were consistent with those of 5 (Figure 3 and Figures S15 and S16, Supporting Information).When tested against the HT-29 cell line, 5a was slightly less active than 5 (Table 5).

To further test the importance of the C-4' substituent, several new benzoylated derivatives were prepared from 5. When 5 was reacted with benzoyl chloride and pyridine, (+)-4'-O-benzoylstrebloside (5b) and (+)-4'-O-benzoyl-19-nor-kamaloside-10-carboxylic acid (5c) were obtained. The product 5b showed a sodiated molecular ion peak at m/z 705.3260 for a molecular formula of C₃₈H₅₀O₁₁. It also exhibited the NMR resonances for an additional benzoyl group, when compared with those of 5 (Tables 1 and ​and22 and Tables S2 and S3, Supporting Information). The C-4' benzoyl group was implied by the HMBC between H-4' and the benzoyl carbonyl group (Figure S14, Supporting Information). A molecular formula of C₃₈H₅₀O₁₂, deduced for 5c from its sodiated molecular ion peak at m/z 721.3214 (calcd 721.3194), indicated that this compound contains an additional oxygen atom, when compared with 5b. A hydroxycarbonyl rather than a formyl group was defined at C-10 of 5c, on account of the resonance for its C-19 at δ_C 176.1 rather than at δ_C 208.3 for the same carbon of 5b (Table S3, Supporting Information). The product 5c could be formed by the aerial oxidation of the C-10 formyl group of 5b.

When 5 was treated with 4-chlorobenzoyl chloride, pyridine, and dichloromethane, (+)-4'-O-(4-chlorobenzoyl)strebloside (5d), (+)-4'-O-(4-chlorobenzoyl)-19-nor-kamaloside-10-carboxylic acid (5e), and (+)-4'-O-(4-chlorobenzoyl)-14(15)-anhydrostrebloside (5f) were obtained. Compound 5d was defined structurally from its sodiated molecular ion peak at m/z 739.2862 for a molecular formula of C₃₈H₄₉O₁₁Cl and the NMR resonances for an additional 4-chlorobenzoyl group, when compared with 5 (Tables 1 and ​and22 and Tables S2 and S3, Supporting Information). The 4-chlorobenzoyl group was shown to be at the C-4' position from the HMBC between H-4' and the benzoyl carbonyl group (Figure S14, Supporting Information). The product 5e gave a sodiated molecular ion peak at m/z 755.2805 for a molecular formula of C₃₈H₄₉O₁₂Cl, 16 Da more than 5d, indicating that 5e contains an additional oxygen. A hydroxycarbonyl rather than a formyl group was linked at C-10 of 5e, as indicated by the resonance for C-19 at δ_C 176.1 (Table S3, Supporting Information). Similar to the formation of 5c, 5e might be generated from the aerial oxidation of 5d. The product 5f showed a sodiated molecular ion peak at m/z 721.2753 for a molecular formula of C₃₈H₄₇O₁₀Cl, 18 Da (H₂O) less than 5d, indicating that 5f is a dehydrated analogue of 5d. This was verified by comparison of ¹³C NMR spectroscopic data of 5f with those of 5d, in which the resonances for C-14 at δ_C 85.2 and C-15 at δ_C 32.0 of 5d appeared at δ_C 152.7 for C-14 and δ_C 117.6 for C-15 of 5f (Tables S3 and S4, Supporting Information).

The same (3S, 5S, 8R, 9S, 10S, 13R, 14S, 17R, 1'R, 2'R, 3'S, 4'S, 5'R) absolute configuration as determined for 5 was assigned for 5b–5e from their consistent ECD and 2D NOESY NMR spectra. In turn, a (3S, 5S, 8R, 9S, 10S, 13R, 17S, 1'R, 2'R, 3'S, 4'S, 5'R) absolute configuration was defined for 5f, from comparison of its ECD and 2D NOESY NMR spectra with those of 5 (Figure 3 and Figure S16, Supporting Information). Different from the ECD bands observed for 5 (Figure 3), a negative CE at 239 nm and a positive CE at 216 nm, indicative of a 17S configuration, occurred in the ECD spectrum of 5f (Figure S16, Supporting Information). This probably resulted from the impact of the c/d-ring conformational change upon formation of the C-14/C-15 double bond, as evidenced previously.³⁴ When evaluated toward the HT-29 cell line,³⁰ 5b and 5d were cytotoxic, but 5c, 5e, and 5f were all inactive in this regard (Table 5).

To test the role of other substituents on mediating its cytotoxicity, (+)-strebloside (5) was subjected to acid hydrolysis. Treatment with trifluoroacetic acid (TFA) at 70 °C for 1 h afforded a new analogue, 5g, with the C-19 formyl group transferred to the C-3 hydroxy group, which was purified from the reaction mixture. Compound 5g showed a sodiated molecular ion peak at m/z 391.1899 for a molecular formula of C₂₃H₂₈O₄, with no resonances for a saccharide residue observed from its ¹H and ¹³C NMR spectra, when compared with those of 5 (Tables 1, ​,2,2, and Table S4, Supporting Information). Three double bonds could be deduced from its ¹³C NMR resonances, for C-5 at δ_C 125.0 and C-10 at δ_C 129.6, C-14 at δ_C 153.3 and C-15 at δ_C 116.0, and C-20 at δ_C 170.8 and C-22 at δ_C 116.5 (Table S4, Supporting Information). These assignments were supported by the HMBC observed between H-1 and C-5, H-2 and C-10, H-7 and C-14, H-15 and C-14 and C-17, and H-17 and C-22 (Figure S14, Supporting Information). A formate group indicated by a resonance at δ_C 161.2 (Table S4, Supporting Information) was assigned at the C-3 position, as implied by the HMBC between H-3 and the formate carbonyl group (Figure S14, Supporting Information). Thus, 5g was characterized as (3β)-3-O-formyl-19-nor-card-5(10),14(15),20(22)-trienolide or (+)-19-nor-5(10),14-dianhydrostrophanthidin-3-yl formate.

A 17S configuration was indicated for 5g from its sequential negative and positive CEs at 242 and 213 nm in the ECD spectrum, which was consistent with that of 5f (Figure S16, Supporting Information), and a (3S, 8R, 9S, 13R, 17S) absolute configuration was assigned from its 1D and 2D NOESY NMR spectra (Figure S15, Supporting Information). When evaluated against the HT-29 cell line,³⁰ 5g did not show any activity (Table 5).

The mechanism of the formation of 5g from 5 could be proposed based on the retro-Prins reaction.³⁵ As presumed in Scheme 1, isomerization of the cis a/b-ring system of 5 to a trans a/b-ring system was followed by hemiacetalization involving the C-3 hydroxy and C-10 formyl groups. Formate 5g then could be formed via a retro-Prins reaction³⁵ of the intermediate hemiacetal. (Other related mechanisms for this reaction are also possible). Such a retro-Prins-type reaction based mechanism could also be extended to the biosynthesis of other 19-nor cardiac glycosides, as reported from Antiaris toxicaria recently.³⁶

Investigation of the cytotoxicity of 1–5 and 5a–5g showed that the C-3 saccharide residue, along with the C-5 and C-14 hydroxy and C-10 formyl groups, was important in the mediation of the cytotoxicity of 5 toward HT-29 cells, as indicated by the IC₅₀ values (µM) of 5 (0.17) and 5g (>20) (Table 5). Comparison of the data for 4 (IC₅₀ 0.09 µM) with 2 (IC₅₀ 0.69 µM) (Table 3) showed that different C-3 sugar residues contributed to the resultant activity differentially. Methylation of the 2′-hydroxy group decreased compound potency moderately, as shown by 3 (IC₅₀ 0.36 µM) and 4 (IC₅₀ 0.09 µM) (Table 3). The activity was reduced in a sequence following the bulk of the ester group when an acetyl, a benzoyl, or a 4-chlorobenzoyl group was introduced to 4'-OH of 5, as implied by the IC₅₀ values (µM) of 5 (0.17), 5a (0.47), 5b (1.2), and 5d (2.8) (Table 5). Furthermore, the C-14 hydroxy group alone plays an important role in potentiating the cytotoxic effect against HT-29 cells, as indicated by comparison of 5d (IC₅₀ 2.8 µM) and 5f (IC₅₀ >20 µM) (Table 5). The activity was retained when the C-10 formyl group of 5 was reduced to a C-10 hydroxymethyl group, as indicated by the closely similar cytotoxicity of 1 (IC₅₀ 0.16 µM) and 5 (IC₅₀ 0.17 µM) (Table 3). However, oxidation of this formyl group to a hydroxycarbonyl group led to abolishment of the activity, as shown by data for 5b (IC₅₀ 1.2 µM) and 5c (IC₅₀ >20 µM), as well as 5d (IC₅₀ 2.8 µM) and 5e (IC₅₀ >20 µM) (Table 5).

Consistent with these observations, several cardiac glycosides were found previously to be more active than the corresponding aglycones,³⁶ and the activity against a small panel of human cancer cell lines was increased when the carbohydrate moiety was attached at both C-2 and C-3 through a 1,4-dioxane moiety.²⁰ No cytotoxicity was evident for several cardiac glycosides that contain a 17-hydroxy (IC₅₀ > 40 µM)¹⁹ or a 19-hydroxycarbonyl (IC₅₀ > 10 µM) group.²⁸

(+)-Strebloside (5), the main cytotoxic compound identified from the stem bark of S. asper, was tested further for its cytotoxicity toward the MDA-MB-435 human melanoma, MDA-MB-231 human breast cancer, and OVCAR3 human ovarian cancer cell lines.³⁷ It was found to be potently cytotoxic in all cases in the present study (Tables 3 and ​and6).6). Hence, (+)-strebloside (5) exhibits a broad spectrum of action toward both human solid tumor and leukemic cells.

(+)-Strebloside (5) was then deemed worthy of evaluation for its antitumor efficacy in an in vivo hollow fiber assay.³³ Immunodeficient NCr nu/nu mice were implanted with hollow fibers containing MDA-MB-231, MDA-MB-435, or OVCAR3 cells and treated with 5 at doses of 1.0, 5.0, 10.0, 15.0, or 30.0 mg/kg, or with the vehicle control or paclitaxel (positive control, 5 mg/kg). In all cases, intraperitoneal (ip) administration for four days was used. The relative cell survival values from the treatment of 5 at 5.0, 10.0, 15.0, or 30.0 mg/kg (ip) toward MDA-MB-231 cells and at 10.0, 15.0, or 30.0 mg/kg (ip) toward OVCAR3 cells were all statistically significantly different from those of the vehicle control (Figure 5). However, no efficacy was observed for 5 toward MDA-MB-435 cells using the same dose range (Figure S18, Supporting Information). The values from the treatment of 5 at all cases were significantly different from those of paclitaxel (Figure 5), and no gross toxicities were observed for any of the animals in this in vivo study. These results indicated that compound 5 showed activity toward MDA-MB-231 and OVCAR3 cells at doses down to 5 mg/kg and 10 mg/kg, respectively, although this efficacy was less potent than paclitaxel.

Open in a separate window

Figure 5

Effect of (+)-strebloside (5) on the growth of human breast MDA-MB-231 and ovarian OVCAR3 cancer cells implanted in NCr nu/nu mice tested in an in vivo hollow fiber assay. The results are shown as the average percentage cell growth relative to control [columns, mean in each group (n = 12 for the control group and n = 6 for the treatment group); bars, SE; **p ≤ 0.05, ***p ≤ 0.01, and ****p ≤ 0.001 for significant differences from the control treatment.

The cardiac glycoside drug digoxin is known to show a narrow therapeutic index¹¹ and has been administered (i.p.) to mice at a non-toxic dose of 0.7 mg/kg/day,¹³ while its close analogue digitoxin exhibited an oral LD₅₀ value of 10–80 mg/kg in mice.³⁸ (+)-Strebloside (5), however, did not exhibit obvious toxicity to the mice at doses (i.p.) up to 30 mg/kg, which is much higher than its effective dose of 5 mg/kg. Previously, several cardiac glycosides have been evaluated in clinical trials for the treatment of cancer,^7,15 and a mixture of two cardiac glycosides that contain the same cardenolide aglycone as 5 showed an LD₅₀ value of more than 500 mg/kg in an acute oral toxicity test, using six-week old ddY mice.³⁸ Mechanistically, most cardiac glycosides mediate their cytotoxicity through inhibition of Na⁺,K⁺-ATPase followed by a Ca⁺² flux pathway.^7,8,39 Therefore, (+)-strebloside (5) seems to be a promising lead compound worthy of further investigation for development as a new anticancer agent.

EXPERIMENTAL SECTION

Supplementary Material

Acknowledgments

Footnotes

REFERENCES