A New Alkamide with an Endoperoxide Structure from Acmella ciliata (Asteraceae) and Its in Vitro Antiplasmodial Activity

From the aerial parts of Acmella ciliata (H.B.K.) Cassini (basionym Spilanthes ciliata Kunth; Asteraceae), three alkamides were isolated and identified by mass- and NMR spectroscopic methods as (2E,6E,8E)-N-isobutyl-2,6,8-decatrienamide (spilanthol, (1)), N-(2-phenethyl)-2E-en-6,8-nonadiynamide (2) and (2E,7Z)-6,9-endoperoxy-N-isobutyl-2,7-decadienamide (3). While 1 and 2 are known alkamides, compound 3 has not been described until now. It was found that the unusual cyclic peroxide 3 exists as a racemate of both enantiomers of each alkamide; the 6,9-cis- as well as the 6,9-trans-configured diastereomers, the former represents the major, the latter the minor constituent of the mixture. In vitro tests for activity against the human pathogenic parasites Trypanosoma brucei rhodesiense and Plasmodium falciparum revealed that 1 and 3 possess activity against the NF54 strain of the latter (IC50 values of 4.5 and 5.1 µM, respectively) while 2 was almost inactive. Compound 3 was also tested against multiresistant P. falciparum K1 and was found to be even more active against this parasite strain (IC50 = 2.1 µM) with considerable selectivity (IC50 against L6 rat skeletal myoblasts = 168 µM).


Introduction
Acmella ciliata (H.B.K.) Cassini (basionym: Spilanthes ciliata Kunth; Asteraceae) is a herb native to tropical regions of South America [1]. Known under the common name jambu in Brazil, it is used as a spice and in traditional medicine for the treatment of toothaches and is the active ingredient in the herbal medicinal product Spolera ® used to treat abrasions and distortions [2]. Previous investigations have been carried out with this species that led to the isolation of several alkylamides (alkamides) [3,4], including the well-known bioactive isobutylamide spilanthol responsible for analgesic [5], anti-inflammatory [6] and antimalarial [7] in vitro activities.
The closely related species Acmella oleracea (L.) R. K. Jansen (basionym: Spilanthes oleracea; synonym: Spilanthes acmella var. oleracea) is widely used as spice for food and in folk medicine for stomatitis, cold and toothaches and is one of the plants constituting the traditional African antimalarial drug "Malarial-5" [8].
Malaria is a disease caused by various Plasmodium species and represents a public health problem causing many hundreds of thousands of deaths per year, especially to people living under poor circumstances and in vulnerable communities in tropical countries. According to the latest available data, an estimated 214 million cases occurred and the disease killed about 438,000 people in 2015 [9]. In this way, WHO recommends artemisinin-based combination therapies (ACTs) as the first line treatment of uncomplicated malaria caused by the P. falciparum parasite. However, since 2006, after the first artemisinin-resistant case was noticed in Cambodia, the resistance to artemisinin-derived drugs has become a global concern [10]. Therefore, the search for new antimalarial drugs has received increasing attention, including natural products. In a recent study, several alkamides from Achillea ptarmica (Asteraceae) were found to be active against Plasmodium falciparum and some other unicellular eukaryotic parasites such as Trypanosoma brucei rhodesiense, responsible for East African Human Trypanosomiasis [11]. Hence, an investigation of further alkamide-containing Asteraceae for alkamides with activity against such pathogens was of interest. Therefore, in the present investigation, we concentrated on alkamides of Acmella ciliata with potential activity against the mentioned parasites.
Malaria is a disease caused by various Plasmodium species and represents a public health problem causing many hundreds of thousands of deaths per year, especially to people living under poor circumstances and in vulnerable communities in tropical countries. According to the latest available data, an estimated 214 million cases occurred and the disease killed about 438,000 people in 2015 [9]. In this way, WHO recommends artemisinin-based combination therapies (ACTs) as the first line treatment of uncomplicated malaria caused by the P. falciparum parasite. However, since 2006, after the first artemisinin-resistant case was noticed in Cambodia, the resistance to artemisinin-derived drugs has become a global concern [10]. Therefore, the search for new antimalarial drugs has received increasing attention, including natural products. In a recent study, several alkamides from Achillea ptarmica (Asteraceae) were found to be active against Plasmodium falciparum and some other unicellular eukaryotic parasites such as Trypanosoma brucei rhodesiense, responsible for East African Human Trypanosomiasis [11]. Hence, an investigation of further alkamide-containing Asteraceae for alkamides with activity against such pathogens was of interest. Therefore, in the present investigation, we concentrated on alkamides of Acmella ciliata with potential activity against the mentioned parasites.
Compound 2 was obtained as a yellow oil. It was previously isolated from A. ciliata [4] and its chemical structure was confirmed by comparison of the mass and 1 H-NMR spectroscopic data with those reported in the literature.
Compound 3 was also obtained as a yellow oil.  Compound 1 was obtained as a green oil and identified as spilanthol by comparison of its spectral data with those described in the literature [12].
Compound 2 was obtained as a yellow oil. It was previously isolated from A. ciliata [4] and its chemical structure was confirmed by comparison of the mass and 1 H-NMR spectroscopic data with those reported in the literature.
Compound 3 was also obtained as a yellow oil. The +ESI QqTOF mass spectrum obtained under UHPLC coupling allowed establishment of the elemental formula as C 14  . While the isolate yielded one sharp peak in the UHPLC/MS chromatogram, it was found to show NMR signals of two very similar isomeric forms (ratio of approximately 3:1) in which most signals resonated at very similar or identical shift values with few exceptions. Nevertheless, the chemical shifts in the proton spectrum, even in parts of the spectrum with strong signal overlap, could all be determined with the help of a 2D J-resolved 1 H-NMR experiment and all 1 H/ 1 H-coupling constants became accessible by selective homonuclear decoupling measurements. The NMR spectra (Tables 1 and 2) showed the presence of an isobutylamide moiety with 1 H signals at δ 3.14 (2H, H-1'), 1.80 (1H, H-2') and 0.92 (6H, H-3' and H-4') and 13 C resonances at δ 46.8 (C-1'), 28.5 (C-2') and 20.1 (C-3', C-4') for the amine component as well as the amide carbonyl at 166.0 ppm (C-1), the latter correlated with the proton signal at δ 3.14 ppm in the HMBC spectrum (key HMBC correlations are summarized in Figure 2). 3.14 dd (6.8 and 6.2) 3.14 dd (6.8 and 6.2) 2' 1.80 th (6.8 and 6.7) 1.80 th (6.8 and 6.7)
These NMR data were similar to those described for longipenamide B (N-isobutyl-syn-6,9dihydroxy-2E,7E-decadienamide) known from Heliopsis longipes [14] which, however, has an elemental formula of C23H25NO3, i.e., one degree of unsaturation less than the compound described here. Longipenamide B possesses an E-configured double bond between C-7 and C-8 and free hydroxy groups at C-6 and C-9. The lack of two hydrogen atoms in comparison with longipenamide B, together with the Z-configuration of the double bond indicated the presence of a cyclic structure in this part of the molecule which could then either be a dihydrofuran system, requiring the presence of one free hydroxy group, or a dihydro-1,2-dioxine system, i.e., a cyclic peroxide structure. Since no HMBC correlations were observable either in chloroform-d3 or in acetone-d6 between the protons and carbons at positions 6 and 9, which would be expected in case of a furanoid system, the compound was indeed assigned the depicted structure of (2E,7Z)-6,9-endoperoxy-N-isobutyl-2,7-decadienamide. The observation that especially the NMR signals for the fragment between C-6 and C-10 were shifted between the major and minor isomers pointed towards the presence of two stereoisomers, i.e., that one represents the cis-, the other the trans-configured form with respect to the asymmetric carbons, C-6 and C-9. It might be expected that the cis-oriented protons in the former would show an NOE correlation, as was previously described for a similar endoperoxide from Zanthoxylum [15]. However, NOEs could not be observed in the 2D NOESY spectrum for either isomer so that the relative stereochemistry could not be clarified in this manner. As mentioned above, the NMR signals of the protons around the endoperoxide ring showed significant shift differences in the minor isomer with H-6 being shifted downfield most prominently for Δδ 0.21 ppm, in comparison to the major isomer. Since no directly comparable data exist in literature, molecular modeling using quantum mechanical calculations had to be applied to assign the relative configuration for the major and minor isomers. To this end, the NMR shift values were calculated for geometry optimized models of the energetically most favorable conformers of both diastereomers (6R,9R-trans and 6S,9R-cis) of the de-alkylated carboxamide part of 3 (the isobutyl moiety was omitted to save computation time). The geometry The signals of the protons and carbons especially at this structural fragment (i.e., positions 6-10) appeared to have shifted between the major and minor components (see below).
These NMR data were similar to those described for longipenamide B (N-isobutyl-syn-6,9dihydroxy-2E,7E-decadienamide) known from Heliopsis longipes [14] which, however, has an elemental formula of C 23 H 25 NO 3 , i.e., one degree of unsaturation less than the compound described here. Longipenamide B possesses an E-configured double bond between C-7 and C-8 and free hydroxy groups at C-6 and C-9. The lack of two hydrogen atoms in comparison with longipenamide B, together with the Z-configuration of the double bond indicated the presence of a cyclic structure in this part of the molecule which could then either be a dihydrofuran system, requiring the presence of one free hydroxy group, or a dihydro-1,2-dioxine system, i.e., a cyclic peroxide structure. Since no HMBC correlations were observable either in chloroform-d 3 or in acetone-d 6 between the protons and carbons at positions 6 and 9, which would be expected in case of a furanoid system, the compound was indeed assigned the depicted structure of (2E,7Z)-6,9-endoperoxy-N-isobutyl-2,7-decadienamide. The observation that especially the NMR signals for the fragment between C-6 and C-10 were shifted between the major and minor isomers pointed towards the presence of two stereoisomers, i.e., that one represents the cis-, the other the trans-configured form with respect to the asymmetric carbons, C-6 and C-9. It might be expected that the cis-oriented protons in the former would show an NOE correlation, as was previously described for a similar endoperoxide from Zanthoxylum [15]. However, NOEs could not be observed in the 2D NOESY spectrum for either isomer so that the relative stereochemistry could not be clarified in this manner. As mentioned above, the NMR signals of the protons around the endoperoxide ring showed significant shift differences in the minor isomer with H-6 being shifted downfield most prominently for ∆δ 0.21 ppm, in comparison to the major isomer. Since no directly comparable data exist in literature, molecular modeling using quantum mechanical calculations had to be applied to assign the relative configuration for the major and minor isomers. To this end, the NMR shift values were calculated for geometry optimized models of the energetically most favorable conformers of both diastereomers (6R,9R-trans and 6S,9R-cis) of the de-alkylated carboxamide part of 3 (the isobutyl moiety was omitted to save computation time). The geometry optimized models were energy minimized with the B3LYP density functional and the 6-31++G(2d,p) basis set. The NMR shifts were calculated using the GIAO method at the same level of theory.
As reported in Table 3, the shift differences resulting for the cisand trans-diastereomers were in good qualitative agreement with the major (3a) and minor isomers (3b), respectively, of compound 3. The formation of 3 (see Figure 3) is most likely to result from addition of a molecule of singlet oxygen to the diene moiety of the E,E-isomer of spilanthol (1) in case of the major isomer 3a, and to 1 itself in case of the minor isomer 3b. Light-induced isomerization of the 6,7-Z-8,9-E diene of 1 to the (energetically more favorable) E,E-isomer is in agreement with earlier literature reports on peroxidation of unsaturated fatty acids [15]. In order to evaluate whether compound 3 is a genuine product of the plant's metabolism or formed during the extraction process, the fresh aerial parts (50 g) of A. ciliata were macerated with chloroform (100 mL) for 2 h in a flask protected from light with aluminum foil. The chloroform extract was dried in vacuo and its acetonitrile solution was then subjected to UPLC-ESI-MS analysis. Moreover, fresh flowers and leaves were quickly pounded and the resulting juices were probed by ASAP (Atmospheric solid analysis probe) MS performed according to established protocols as reported previously, e.g., in [16]. Both experiments displayed the pseudo-molecular ion m/z 254.1759 corresponding to 3 so that its formation is likely to occur already in the living plant to some extent. The structure was further supported by MS 2 fragmentation as depicted in Figure 4. Taken together, it is suggested that addition of 1 O 2 proceeds in an uncatalyzed spontaneous manner. This reaction is possible by means of a 1,4-cycloaddition [17], so that the oxygen is added from either face of the molecules with equal probability (see Figure 3). Therefore, racemates should result in both cases. Indeed, the isolated mixture 3 was devoid of optical activity (no CD cotton effects between 450 and 190 nm) so that this result is in agreement with the postulated means of formation. It can thus be concluded that 3 is actually a mixture of all four possible stereoisomers with the 6R,9S-and 6S,9R-cis enantiomers present as major (3a, approximately 40% each) and the 6R,9R-and 6S,9S-enantiomers as the minor constituents (3b, approx. 10% each). Similar alkamides with an endoperoxide bridge have previously been described by Devkota et al. [15] as constituents of a Zanthoxylum species. However, the structure described here is a new natural product and has also not been described as a synthetic product, to the best of our knowledge. For this new compound, we propose the name dioxacmellamide. optimized models were energy minimized with the B3LYP density functional and the 6-31++G(2d,p) basis set. The NMR shifts were calculated using the GIAO method at the same level of theory. As reported in Table 3, the shift differences resulting for the cis-and trans-diastereomers were in good qualitative agreement with the major (3a) and minor isomers (3b), respectively, of compound 3. The formation of 3 (see Figure 3) is most likely to result from addition of a molecule of singlet oxygen to the diene moiety of the E,E-isomer of spilanthol (1) in case of the major isomer 3a, and to 1 itself in case of the minor isomer 3b. Light-induced isomerization of the 6,7-Z-8,9-E diene of 1 to the (energetically more favorable) E,E-isomer is in agreement with earlier literature reports on peroxidation of unsaturated fatty acids [15]. In order to evaluate whether compound 3 is a genuine product of the plant's metabolism or formed during the extraction process, the fresh aerial parts (50 g) of A. ciliata were macerated with chloroform (100 mL) for 2 h in a flask protected from light with aluminum foil. The chloroform extract was dried in vacuo and its acetonitrile solution was then subjected to UPLC-ESI-MS analysis. Moreover, fresh flowers and leaves were quickly pounded and the resulting juices were probed by ASAP (Atmospheric solid analysis probe) MS performed according to established protocols as reported previously, e.g., in [16]. Both experiments displayed the pseudo-molecular ion m/z 254.1759 corresponding to 3 so that its formation is likely to occur already in the living plant to some extent. The structure was further supported by MS 2 fragmentation as depicted in Figure 4. Taken together, it is suggested that addition of 1 O2 proceeds in an uncatalyzed spontaneous manner. This reaction is possible by means of a 1,4-cycloaddition [17], so that the oxygen is added from either face of the molecules with equal probability (see Figure 3). Therefore, racemates should result in both cases. Indeed, the isolated mixture 3 was devoid of optical activity (no CD cotton effects between 450 and 190 nm) so that this result is in agreement with the postulated means of formation. It can thus be concluded that 3 is actually a mixture of all four possible stereoisomers with the 6R,9S-and 6S,9R-cis enantiomers present as major (3a, approximately 40% each) and the 6R,9Rand 6S,9S-enantiomers as the minor constituents (3b, approx. 10% each). Similar alkamides with an endoperoxide bridge have previously been described by Devkota et al. [15] as constituents of a Zanthoxylum species. However, the structure described here is a new natural product and has also not been described as a synthetic product, to the best of our knowledge. For this new compound, we propose the name dioxacmellamide.

In Vitro Antiplasmodial Activity of Isolated Alkamides
As part of our ongoing efforts to find new compounds with activity against "protozoan" parasites causing neglected tropical diseases as well as malaria, the isolated alkamides were tested for activity against Trypanosoma brucei rhodesiense (Tbr) and Plasmodium falciparum (Pf) as well as for cytotoxicity against mammalian cells (L6 cell line) using established protocols [18]. The resulting IC50 values for in vitro growth inhibition are reported in Table 4. While compound 2 displayed only insignificant activity against either parasite, compounds 1 and 3 were somewhat more active against Tbr and, especially Pf (chloroquine sensitive NF54 strain). In case of 1, antiplasmodial activity at a comparable activity level against strains PFB and K1 of Pf had already been reported [7]. Compound 3, due to its new endoperoxide structure that is related to the most important structure element of the potent

In Vitro Antiplasmodial Activity of Isolated Alkamides
As part of our ongoing efforts to find new compounds with activity against "protozoan" parasites causing neglected tropical diseases as well as malaria, the isolated alkamides were tested for activity against Trypanosoma brucei rhodesiense (Tbr) and Plasmodium falciparum (Pf ) as well as for cytotoxicity against mammalian cells (L6 cell line) using established protocols [18]. The resulting IC 50 values for in vitro growth inhibition are reported in Table 4. While compound 2 displayed only insignificant activity against either parasite, compounds 1 and 3 were somewhat more active against Tbr and, especially Pf (chloroquine sensitive NF54 strain). In case of 1, antiplasmodial activity at a comparable activity level against strains PFB and K1 of Pf had already been reported [7]. Compound 3, due to its new endoperoxide structure that is related to the most important structure element of the potent antimalarial artemisinin, was also tested against the multi-resistant K1 strain of Pf where it displayed even higher activity than against the sensitive strain with an IC 50 value of 0.54 µg/mL (2.1 µM) and showed considerable selectivity, i.e., very low cytotoxicity against the L6 cells (S.I. = 78.7). Compound 3 thus appears to be an interesting candidate for in vivo tests against plasmodial infections which are warranted after isolation of larger quantities.

General Procedures
Silica gel 60 and Sephadex ® LH-20 (GE Healthcare, Uppsala, Sweden) were used for column chromatography, while silica gel 60 F 254 aluminum sheets (Silicycle ® , Ville de Québec, QC, Canada) were used for analytical thin layer chromatography (TLC). Compounds were visualized by exposure under UV 254/365 light and by spraying with p-anisaldehyde reagent followed by heating. All solvents for chromatographic procedures were of analytical grade, while those for MS analysis were of spectrometric grade. NMR experiments were acquired at 303 K in CDCl 3 and acetone-d 6 on a Bruker (Billerica, MA, USA) AVANCE III NMR spectrometer and on an Agilent (Santa Clara, CA, USA) DD2 600 spectrometer, operating at 14.1 Tesla, observing 1 H and 13 C at 600 and 150 MHz, respectively. High-resolution MS and MS/MS analysis of the pure compounds were performed on a Bruker MicrOTOF-QII ESI-QTOF mass spectrometer coupled to a Dionex (Sunnyvale, CA, USA) Ultimate 3000 RS UHPLC. LC-MS and MS analyses of CHCl 3 extract and the plant material were carried on a Waters (Milford, MA, USA) Xevo G2-S QTOF mass spectrometer equipped with exchangeable ESI and ASAP sources and coupled to an Acquity UPLC (Waters) chromatograph. A CD spectrum of 3, recorded in MeOH on a Jasco (Groß-Umstadt, Germany) J815 spectropolarimeter, did not show any cotton effects.

Biological Activity Tests
In vitro tests for activity against T. brucei rhodesiense (bloodstream trypomastigotes, strain STIB 900) and P. falciparum (intraerythrosctic forms, strains NF54 and K1) as well as for cytotoxicity against mammalian cells (L6 rat skeletal myoblasts) were performed according to established protocols as reported previously, e.g., in [18].

Molecular Modeling and NMR Shift Calculations for Compound 3
All calculations were performed with molecular models of the dealkylated carboxamide part of 3, i.e., the isobutyl residue was omitted in order to save computation time. A molecular model of each, the 6S,9R-cis-and the 6R,9R-trans-diastereomer was generated with the software package MOE [19] and submitted to a low-mode molecular dynamics conformational search using the MMFF94x force field and standard settings. The energetically lowest conformer of each stereoisomer found in these searches was energy-minimized with Gaussian 03W [20] using the B3LYP density functional and the 6-31G++(2d,p) basis set. Subsequently, the NMR shielding tensors were computed with the same software and at the same level of theory using the GIAO formalism. A molecule of tetramethylsilane was treated in the same manner in order to obtain the reference value (34.9441 ppm). Chemical shift values were obtained by subtracting the TMS reference value from the values calculated for the protons mentioned in Table 1.

Conclusions
The search for alkamides with anti-protozoal activity in the medicinal plant Acmella ciliata has resulted in the discovery of dioxacmellamide (3), a new alkamide with cyclic peroxide structure which shows considerable in vitro activity against P. falciparum with selectivity against the multiresistant K1 strain. It is very likely that this interesting compound is formed by 1,4-cycloaddition of oxygen to the diene moiety of the plant product spilanthol, and it could also be demonstrated that it can easily be obtained from the latter. This accessibility from the main constituent spilanthol will enable us to obtain larger quantities for in vivo tests and detailed studies of the mechanism of action of this new anti-plasmodial hit molecule. Such studies have been initiated.