New Structures, Spectrometric Quantification, and Inhibitory Properties of Cardenolides from Asclepias curassavica Seeds

Cardiac glycosides are a large class of secondary metabolites found in plants. In the genus Asclepias, cardenolides in milkweed plants have an established role in plant–herbivore and predator–prey interactions, based on their ability to inhibit the membrane-bound Na+/K+-ATPase enzyme. Milkweed seeds are eaten by specialist lygaeid bugs, which are the most cardenolide-tolerant insects known. These insects likely impose natural selection for the repeated derivatisation of cardenolides. A first step in investigating this hypothesis is to conduct a phytochemical profiling of the cardenolides in the seeds. Here, we report the concentrations of 10 purified cardenolides from the seeds of Asclepias curassavica. We report the structures of new compounds: 3-O-β-allopyranosyl coroglaucigenin (1), 3-[4′-O-β-glucopyranosyl-β-allopyranosyl] coroglaucigenin (2), 3′-O-β-glucopyranosyl-15-β-hydroxycalotropin (3), and 3-O-β-glucopyranosyl-12-β-hydroxyl coroglaucigenin (4), as well as six previously reported cardenolides (5–10). We test the in vitro inhibition of these compounds on the sensitive porcine Na+/K+-ATPase. The least inhibitory compound was also the most abundant in the seeds—4′-O-β-glucopyranosyl frugoside (5). Gofruside (9) was the most inhibitory. We found no direct correlation between the number of glycosides/sugar moieties in a cardenolide and its inhibitory effect. Our results enhance the literature on cardenolide diversity and concentration among tissues eaten by insects and provide an opportunity to uncover potential evolutionary relationships between tissue-specific defense expression and insect adaptations in plant–herbivore interactions.


Introduction
Plants produce a range of low molecular weight organic compounds, some of which are not involved in the 'primary' functions of plants but which mediate plant-environment interactions-known as secondary metabolites (or natural products) [1]. Most plant secondary metabolites have evolved to defend plants against insects and other natural enemies [2,3]. That does not mean, however, that all secondary compounds have a defensive function, and criteria for determining this are still not fully developed [1,4]. Even in some of the most well-studied systems, the structures and functions of these metabolites are still undescribed [5]. Therefore, testing the biological activity of secondary metabolites in a plant and, if they are active, whether they are of evolutionary and ecological significance is important for understanding the mechanisms and function of chemodiversity [5].
Compound 1 has the molecular formula of C29H44O10, determined by the ion peak at m/z 553.3015 [M + H] + (calculated for C29H45O10, m/z 553.3013). It is a coroglaucigenin derivative with one glycosylation, analogous to 4′-O-β-glucopyranosyl frugoside (5) [45]. However, the sugar moiety bound to the sterol at position C-3 shows, unlike for many cardenolides reported in Asclepias, an oxidized methylene at C-6′. (Supporting Information, Figures S3-S11) The molecular structures of Asclepias cardenolides have been intensively studied in the past [52][53][54]. On the basis of X-ray analysis, their principle structures have been determined: the triterpene scaffold of Asclepias cardenolides has the common feature of an α-orientation of the methine proton at C-5; therefore, the rings A and B of the scaffold are trans-fused [55]. Analysis of the NMR data led us to the conclusion that 1, 2, and 4 are coroglaucigenin-type molecules with a hydroxylation at C-19, while compound 3 is a calotropin derivative ( Figure 2). We found compounds with one or two glycosylations, whereas in the previous study of the seeds, four cardenolides containing cellobiosyl units were reported [45]. The molecular structures of Asclepias cardenolides have been intensively studied in the past [52][53][54]. On the basis of x-ray analysis, their principle structures have been determined: the triterpene scaffold of Asclepias cardenolides has the common feature of an orientation of the methine proton at C-5; therefore, the rings A and B of the scaffold are trans-fused [55]. Analysis of the NMR data led us to the conclusion that 1, 2, and 4 are coroglaucigenin-type molecules with a hydroxylation at C-19, while compound 3 is a calotropin derivative ( Figure 2). We found compounds with one or two glycosylations, whereas in the previous study of the seeds, four cardenolides containing cellobiosyl units were reported [45].
Compound 1 has the molecular formula of C29H44O10, determined by the ion peak at m/z 553.3015 [M + H] + (calculated for C29H45O10, m/z 553.3013). It is a coroglaucigenin derivative with one glycosylation, analogous to 4′-O-β-glucopyranosyl frugoside (5) [45]. However, the sugar moiety bound to the sterol at position C-3 shows, unlike for many cardenolides reported in Asclepias, an oxidized methylene at C-6′. (Supporting Information, Figures S3-S11)   [45]. However, the sugar moiety bound to the sterol at position C-3 shows, unlike for many cardenolides reported in Asclepias, an oxidized methylene at C-6 . (Supporting Information, Figures S3-S11) Further analysis of the glycosyl relative configuration revealed that the carbinolic protons H-1 and H-3 are both in equatorial position. Proton H-3 is furthermore in synperiplanar position to H-2 and H-4 . We therefore assumed an α-oriented hydroxyl function in position C-3 ( Figure 3). This stereochemistry is characteristic for allose, the C-3 epimer of glucose [56]. Compound 1 is, accordingly, 3-O-β-allopyranosyl coroglaucigenin (see Supporting Information, Figure S5 for the IUPAC name). Further analysis of the glycosyl relative configuration revealed that the carbinolic protons H-1′and H-3′ are both in equatorial position. Proton H-3′ is furthermore in synperiplanar position to H-2′ and H-4′. We therefore assumed an α-oriented hydroxyl function in position C-3′ ( Figure 3). This stereochemistry is characteristic for allose, the C-3′ epimer of glucose [56]. Compound 1 is, accordingly, 3-O-β-allopyranosyl coroglaucigenin (see Supporting Information, Figure S5 for the IUPAC name).   Figure S12-S24). The structure is similar to compound 1 but shows two signals in the 1 H NMR spectrum at δH 4.87 (δC 97.8, H-1′) and δH 4.56 (δC 103.6, H-1′') that we assigned to anomeric glycosyl positions. It suggested a glycosyl chain, where the first sugar was again an allose. For the second glycosyl moiety, however, the multiplicity of H-2′ and H-3′, both dd multiplicities with a large coupling constant ( 3 JHH > 9 Hz), revealed an anti-periplanar arrangement, consistent with the β-OH orientation at C-3′. This is characteristic for a glucosyl rest and, accordingly, compound 2 is 4′-O-β-glucopyranosyl-3-O-β-allopyranosyl coroglaucigenin (see Supporting Information, Figure S17 for the IUPAC name).
Compound 3 is a calotropin-type cardenolide with a molecular formula C35H48O14, determined by the ion peak at m/z 693.3113 [M -H2O + H] + (calcd for C35H49O14, 693.3117) (Supporting Information, Figure S25-S33). This compound is similar to compound 7, with the only difference being a hydroxylation at C-15. Analysis of the 1 H-1 H ROESY data showed that H-17 and H-15 are in syn-periplanar orientation. Given the absolute configuration of H-17 according to biosynthetic considerations, we assign H-15 as 15R [54]. We determine compound 3 as 3′-O-β-glucopyranosyl-15β-hydroxycalotropin (see Supporting Information, Figure S29 for the IUPAC name).
Compound 4 has a molecular formula of C29H44O11, determined by the ion peak at m/z 569.2970 [M + H] + (calcd for C29H45O11, 569.2956). It is of the coroglaucigenin type but, in this case, with a carbinolic proton resonating at δH 3.32 (δC 74,7, H-12). We defined the stereocenter of this oxydized methine as (R) from 1 H-1 H ROESY correlations between H- The structure is similar to compound 1 but shows two signals in the 1 H NMR spectrum at δ H 4.87 (δ C 97.8, H-1 ) and δ H 4.56 (δ C 103.6, H-1") that we assigned to anomeric glycosyl positions. It suggested a glycosyl chain, where the first sugar was again an allose. For the second glycosyl moiety, however, the multiplicity of H-2 and H-3 , both dd multiplicities with a large coupling constant ( 3 J HH > 9 Hz), revealed an anti-periplanar arrangement, consistent with the β-OH orientation at C-3 . This is characteristic for a glucosyl rest and, accordingly, compound 2 is 4 -O-β-glucopyranosyl-3-O-β-allopyranosyl coroglaucigenin (see Supporting Information, Figure S17 for the IUPAC name).
Compound 3 is a calotropin-type cardenolide with a molecular formula C 35 Figures S25-S33). This compound is similar to compound 7, with the only difference being a hydroxylation at C-15. Analysis of the 1 H-1 H ROESY data showed that H-17 and H-15 are in syn-periplanar orientation. Given the absolute configuration of H-17 according to biosynthetic considerations, we assign H-15 as 15R [54]. We determine compound 3 as 3 -O-β-glucopyranosyl-15β-hydroxycalotropin (see Supporting Information, Figure S29 for the IUPAC name).
Compound 4 has a molecular formula of C 29 H 44 O 11 , determined by the ion peak at m/z 569.2970 [M + H] + (calcd for C 29 H 45 O 11 , 569.2956). It is of the coroglaucigenin type but, in this case, with a carbinolic proton resonating at δ H 3.32 (δ C 74,7, H-12). We defined the stereocenter of this oxydized methine as (R) from 1 H-1 H ROESY correlations between H-1α↔H-9↔H-12↔H-15α↔H-17. Furthermore, the large coupling between H-12 H 2 -11β ( 3 J HH = 12.2 Hz) was consistent with their trans-periplanar orientation. The sugar moiety was identified as glucose by the large coupling constants and the 1 H-1 H ROESY correlations between H-1 and H-5 . We therefore describe the compound as 3-O-βglucopyranosyl-12β-hydroxy coroglaucigenin (see Supporting Information, Figure S39 for the IUPAC name).

Quantification of Cardenolides
We quantified compounds 1-10 in mg of compound per gram of seeds (dry weight; Figure 4 and Supporting Information Table S2). We used each compound as a standard for their corresponding calibration curve (Supporting Information Figure S2). Compound 5, 4 -O-β-glucopyranosyl frugoside, is the most abundant cardenolide, with 4.5 mg/g of seeds, approximately two times more than 4 -O-β-glucopyranosyl gofruside 6 (2.06 mg/g). The other eight cardenolides are present in amounts below 1 mg/g. Compounds 1 and 4 are the least abundant at 0.01 and 0.004 mg/g, respectively.

Quantification of Cardenolides
We quantified compounds 1-10 in mg of compound per gram of seeds (dry weight; Figure 4 and Supporting Information Table S2). We used each compound as a standard for their corresponding calibration curve (Supporting Information Figure S2). Compound 5, 4′-O-β-glucopyranosyl frugoside, is the most abundant cardenolide, with 4.5 mg/g of seeds, approximately two times more than 4′-O-β-glucopyranosyl gofruside 6 (2.06 mg/g). The other eight cardenolides are present in amounts below 1 mg/g. Compounds 1 and 4 are the least abundant at 0.01 and 0.004 mg/g, respectively. The range of retention times of the isolated compounds was 19 to 30 min (Supporting Information Figure S1, for chromatography conditions see Section 3.1). The minor compounds 1-4 and 10 have a higher polarity than the more abundant compounds ( Figure 4). Recently, López-Goldar et al. [11] reported a predominance of more polar compounds in the seed extracts of A. curassavica. Allomethylose and deoxy-allomethylose are present as sugar moieties in the abundant cardenolides 5-10, whereas allose and glucose were found in the minor compounds 1, 2, and 4. This finding may provide clues regarding the biosynthetic pathways of the rare sugars in Asclepias, where allosyl cardenolides could be intermediates of interest.

Na + /K + ATPase (NKA) Inhibitory Activity
We tested the inhibitory capacity of the new compounds 1 and 2 and the known cardenolides 5-10 against porcine NKA. We used an in vitro assay to determine the IC50 of each compound and used ouabain as a reference. The range of retention times of the isolated compounds was 19 to 30 min (Supporting Information Figure S1, for chromatography conditions see Section 3.1). The minor compounds 1-4 and 10 have a higher polarity than the more abundant compounds ( Figure 4). Recently, López-Goldar et al. [11] reported a predominance of more polar compounds in the seed extracts of A. curassavica. Allomethylose and deoxy-allomethylose are present as sugar moieties in the abundant cardenolides 5-10, whereas allose and glucose were found in the minor compounds 1, 2, and 4. This finding may provide clues regarding the biosynthetic pathways of the rare sugars in Asclepias, where allosyl cardenolides could be intermediates of interest.

Na + /K + ATPase (NKA) Inhibitory Activity
We tested the inhibitory capacity of the new compounds 1 and 2 and the known cardenolides 5-10 against porcine NKA. We used an in vitro assay to determine the IC 50 of each compound and used ouabain as a reference.
There was a significant variation in inhibition of an unadapted NKA from Sus dom ticus (Supporting Information, Tables S3 and S4). The IC50 values ranged from 10 −6 to M; this is consistent with the sensitivity for cardenolide inhibition expected from S. dom ticus NKA [19,39,57]. Gofruside 9 was the most inhibitory compound tested (IC50 = 9.65 10 −8 M). The least inhibitory was 16α-hydroxycalotropin 10, with an IC50 of 3.667 × 10 − ( Figure 5). Compounds 1 and 8 are coroglaucigenin derivatives with allosyl and allomethylo substitution, respectively, but they did not differ in their inhibition properties (Support Information, Tables S3 and S4). Compounds 8 and 9 share the same glycosylation, bu shows a different oxidation state of C-18 (alcohol vs. aldehyde). The higher inhibition 9 compared with 8 suggests that the aldehyde is the main reason for the difference. T can be due to the high reactivity of 9 towards biomolecules. However, 16αdroxycalotropin 10 also contains an aldehyde but has the lowest inhibitory capacity. It been reported that calotropin, the 16-deoxy aglycone of 10, has an IC50 of 2.7 × 10 −7 M (lo −6.56) against porcine NKA [19]. This difference in inhibition may also be attributed to fact that in 10, the 16α-hydroxylation interferes with binding to the biological target. M lecular docking analyses are required for a better understanding.
We found a reduced inhibitory potential when a glucosylation of cardenolides curred. The compounds 8 and 5 (coroglaucigenin-type) and 9 and 6 (corotoxigenin-ty differ only in the glucosylation, the aglycones having the higher inhibition poten However, we also found that compounds 1 and 2, which differ in the presence of gluc as a second sugar unit, have no significant difference in their inhibitory potential w compared with one another. An explanation for this could be the increased bulkines the molecule caused by the higher degree of glycosylation in both cases, which hind the access to the active site of the NKA. In addition, in this case, molecular docking an yses would be needed to unravel the impact of the glycone moieties in the inhibition NKAs. Overall, our Na + /K + ATPase inhibition results by cardenolides from Asclepias se are in line with a previous comparison of structural characteristics and NKA inhibition Petschenka et al. [39], who described a differential response in the inhibition of vertebr NKA by cardenolides with the same aglycons but different glycosylations. Compounds 1 and 8 are coroglaucigenin derivatives with allosyl and allomethylosyl substitution, respectively, but they did not differ in their inhibition properties (Supporting Information, Tables S3 and S4). Compounds 8 and 9 share the same glycosylation, but 9 shows a different oxidation state of C-18 (alcohol vs. aldehyde). The higher inhibition of 9 compared with 8 suggests that the aldehyde is the main reason for the difference. This can be due to the high reactivity of 9 towards biomolecules. However, 16α-hydroxycalotropin 10 also contains an aldehyde but has the lowest inhibitory capacity. It has been reported that calotropin, the 16-deoxy aglycone of 10, has an IC 50 of 2.7 × 10 −7 M (log 10 : −6.56) against porcine NKA [19]. This difference in inhibition may also be attributed to the fact that in 10, the 16α-hydroxylation interferes with binding to the biological target. Molecular docking analyses are required for a better understanding.
We found a reduced inhibitory potential when a glucosylation of cardenolides occurred. The compounds 8 and 5 (coroglaucigenin-type) and 9 and 6 (corotoxigenin-type) differ only in the glucosylation, the aglycones having the higher inhibition potential. However, we also found that compounds 1 and 2, which differ in the presence of glucose as a second sugar unit, have no significant difference in their inhibitory potential when compared with one another. An explanation for this could be the increased bulkiness of the molecule caused by the higher degree of glycosylation in both cases, which hinders the access to the active site of the NKA. In addition, in this case, molecular docking analyses would be needed to unravel the impact of the glycone moieties in the inhibition of NKAs.
Overall, our Na + /K + ATPase inhibition results by cardenolides from Asclepias seeds are in line with a previous comparison of structural characteristics and NKA inhibition by Petschenka et al. [39], who described a differential response in the inhibition of vertebrate NKA by cardenolides with the same aglycons but different glycosylations.
Phytochemical diversity, like that described here, is linked to herbivore community in several systems [58][59][60], but disentangling concentration and inhibitory potency of phytochemicals is challenging and may be related to the costs of producing phytochemicals for the plant and differences in toxicity against herbivores with specific tolerance mechanisms [11]. In our future work we will seek to identify the selective pressures that lead to the differential investment in the compounds, given their different inhibitory effects against porcine NKA, and determine whether these effects vary depending on different cardenolide tolerances.

General Experimental Procedures
Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance III HD spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) equipped with a cryoplatform and a 5 mm TCI CryoProbe, field strengths of 1 H (500.13 MHz)/ 13 C (125.76 MHz): 11.747 T. Spectrometer control and data processing was accomplished using Bruker TopSpin 3.6.1, and standard pulse programs as implemented in Bruker TopSpin 3.6.1. were used. Samples were measured in MeOH-d 3  High resolution mass spectra were recorded on a Bruker Compact OTOF spectrometer (Bruker Daltonics GmbH, Bremen, Germany). Electrospray ionization (ESI) in positive ion mode was used for the analysis in full scan and auto MS/MS modes, scanning masses from m/z 50-1300. Capillary voltage was set at 4500 V, charging electrode at 2000 V, and corona current at 0 nA; nebulizer pressure gas was set at 1.8 bar, drying gas temperature at 220 • C, and drying gas flow at 9.0 L/min. Sodium formate adducts were used for internal calibration with a Quadratic + HPC mode. Bruker Compass ver.1.9 (OTOF Control ver.5.1.107 and HyStar 4.1.31.1) was used for data acquisition and instrument control, and Bruker DataAnalysis ver. 5.1.201 was used for data processing.
Reversed-phase MPLC separations were carried out on a Biotage Isolera One (Biotage SB, Uppsala, Sweden) using a Biotage Sfär C18 D-Duo 100 Å 30 µm 120 g column. A linear gradient, using a mobile phase consisting of MeOH (supplied with 0.1% FA, Carl Roth GmbH) and water (0.1% FA) was used, with a flow rate of 50 mL/min, and UV detection was carried out at 218 nm. For separations on Sephadex LH-20 (VWR GmbH, Dresden, Germany), a column containing 44 g of sorbent was used and eluted with the Isolera MPLC equipment, using water as the mobile phase. MPLC separations on MCI gel CHP20P (Merck KgA, Darmstadt, Germany) were carried out using a linear MeOH-water gradient on the Biotage Isolera equipment. Semi-preparative HPLC separations were carried out on a Shimadzu Prominence HPLC System, consisting of an autosampler SIL-20AC, gradient pump LC-20AT, UV-Vis detector SPD-20A programed for detection at 220 nm and a fraction collector FRC-10A. For separations, isocratic elutions with MeOH-water mixtures were performed at a flow rate of 0.8 mL/min. A C-18 Nucleodur Isis column (4.6 × 250 mm, 5 µm particle size, from Macherey-Nagel, Düren, Germany) was used. The purity of the isolated compounds was calculated by 1 H qNMR experiments, with ouabain as external standard (ERETIC).

Quantification of Cardenolides
We quantified the cardenolides isolated from the A. curassavica seeds using HPLC-HRMS with a linear calibration method. Each compound was diluted in order to obtain ten data points in a concentration range from 0 to 1 mg/mL, and each value was corrected according to the measured purity of the compound. The spectrometric data were obtained using the method described in Section 3.1. We extracted the ion chromatograms on the basis of the m/z of the most abundant peak for each cardenolide ([M+H] + for compounds 1-6 and 8-10 and [M+H-H 2 O] + for compound 7). The area of the corresponding peak was calculated and used as a quantification parameter. The data points that were out of the linear range, especially at high concentration, were excluded. Linear regression for compounds 1-10 was calculated with concentration and peak area as variables (Supporting Information, Figure S2).
Analysis of the content of cardenolides 1-10 in the seeds was performed as follows: 6 g of A. curassavica seeds were collected from plants grown in the greenhouse of the MPI for Chemical Ecology in Jena, Germany (seeds purchased from Jelitto Perennial Seeds, Art. No.: AA974). Seed samples (three technical replicates, each of 2 g) were freeze-dried, ground, and exhaustively extracted with MeOH/H 2 O (1:1) three times and then twice more with MeOH. The extracts were pooled and filtered using grade 50 Whatman paper filter to remove particles. Lipophilic substances were removed by passing the filtrate through a MN HR-X 500 mg cartridge. The samples were then dried using N 2 gas at 36 • C, obtaining 214.9 mg, 230.6 mg, and 218.9 mg, respectively, of raw extract. The HPLC-HRMS measurements were conducted as described above. From each technical replicate a solution of 2.6 mg/mL was prepared, and from each of these solutions, three injections of 5 µL were used for analysis, resulting in nine data points per compound. Data processing was accomplished as for the calibration curve, by extracting the ion chromatogram per compound and measuring the corresponding peak area. The concentration per injection was later expressed in mg of cardenolide per g of seed (dry weight).

Na + /K + ATPase (NKA) Inhibitory Activity Assay
We quantified the NKA inhibitory activity of the isolated cardenolides using purified NKA from S. domesticus cerebral cortex (Sigma-Aldrich-A7510-5UN, Steinheim, Germany), following protocols standardized by Petschenka et al. [15]. Briefly, porcine NKA was diluted in di-water to a final assay concentration of 0.01 U/mL. The ATPase was exposed to exponentially decreasing concentrations from 10 −3 to 10 −8 M of ouabain (Sigma-Aldrich, O3125-1G, Steinheim, Germany) or the cardenolides 1-2 and 5-10. There was an insufficient amount of compounds 3 and 4 to allow testing. We prepared a stock solution of each cardenolide in 100% DMSO, which was used for preparing the concentrations for the assay. The concentration of DMSO in the final solutions did not exceed 2% per well. The reaction mixture with NKA and cardenolides was incubated at 37 • C for 10 min, followed by addition of ATP (Sigma Aldrich, A9062-1G, Steinheim, Germany) and another incubation for 20 min. The NKA activity after cardenolide exposure was determined by quantification of inorganic phosphate released from enzymatically hydrolysed ATP. The ATP levels were measured by photometric determination, reading the absorbance at 700 nm with a microplate reader (BMG Clariostar, BMG Labtech, Germany). The absorbance of each reaction was corrected with the respective background, and the inhibition curves were plotted in R studio [61] with log 10 -transformed cardenolide concentration versus percent of uninhibited control, with top and bottom asymptotes set to 1 and 0, respectively [30,32,62]. For each cardenolide, we carried out three biological replicates, with two technical replicates, resulting in each data point being an average of six measurements. IC 50 values for each cardenolide and ouabain were also calculated from the inhibition curves using R studio. We employed Bonferroni-adjusted significance tests for pairwise comparisons between the IC 50 values for cardenolides 1-2 and 5-10 and ouabain. The isolated cardenolide purity was addressed in the preparation of the stock solutions for compounds 5 and 7-9. For compounds 1, 2, 6, and 10, stock solutions were prepared without this correction, given the low amounts available and/or low purity; instead, the final IC 50 concentrations values were adjusted with purity percent as the factor.

Conclusions
The re-examination of the seeds of A. curassavica allowed us to describe and quantify four new cardenolides (1)(2)(3)(4). We also isolated six known compounds (5-10) and confirmed by spectrometric quantification that 4 -O-β-glucopyranosyl frugoside 5 is the most abundant cardenolide in the seeds. The examination of the isolated compounds against NKA enzymes from a sensitive vertebrate (S. domesticus) revealed that the most inhibitory cardenolide is gofruside 9, and the least inhibitory are 4 -O-β-glucopyranosyl frugoside 5 and 16α-hydroxycalotropin 10. Comparison of the IC 50 values obtained with the structural characteristics of each cardenolide confirmed that glycosylation, when leading to higher polarity, corresponds directly to a decrease in toxic potential of the cardenolides. However, the structure of the glycosyl substituents and the degree of oxidation at position 19 (alcohol vs. aldehyde) in coroglaucigenin-type cardenolides can also influence inhibitory capacity. The chemical defense of A. curassavica seeds varies in quantity and NKA inhibition potential. The cardenolide profile described here may be the result of biosynthetic constraints that result in a high diversity of bioactive compounds, which increases the chances of achieving an effective defense against several herbivore pressures.