Selective Cytotoxicity of Damsin Derivatives in Breast Cancer Cells

Cancer is the leading cause of death worldwide, and there is a constant need for new treatment strategies. Sesquiterpene lactones containing a 3-methylenedihydrofuran-2(3H)-one (or -methylene--lactone) moiety, for example damsin (1), are Michael acceptors that affect biological processes such as cell proliferation, death/apoptosis, and cell migration, by interfering with cell signalling pathways. Although the reactivity of the -methylene--lactone moiety is important for these effects, the Michael addition is reversible and it can be assumed that also other parts of the molecules will moderate any given biological activity. In this investigation, the cytotoxicity of 23 -methylene--lactone towards normal breast epithelial MCF-10A cells as well as breast cancer JIMT-1 cells is compared. Most of the investigated compounds are semisynthetic derivatives prepared by the condensation of the natural product damsin (1) with aldehydes. The two cell lines were treated with various concentrations of the compounds in dose response assays, and the 50 % inhibitory concentration (IC50) was determined from dose response curves. The IC50 values were found to depend strongly on the overall structure. The ratio between the IC50 values for MCF-10A and JIMT-1 cells, as a measure for the selectivity of a compound to kill cancer cells, was calculated, and found to vary between just over 1 to more than 10. The most potent derivatives formed from the condensation of 1 with aromatic aldehydes towards JIMT-1 cells are 3a and 3i, both with ratios between the IC50 values for MCF-10A and JIMT-1 cells close to 5. Also some aldol condensation products with acyclic aldehydes, i.e. 3r and 3u, were equally potent, and the latter showed the highest selectivity (ratio > 10). Structure-activity relationships that may explain the observed differences in potency and selectivity are discussed. Maribel Lozano, Wendy Soria, Giovanna R. Almanza, Sophie Manner, Stina Oredsson, Rodrigo Villagomez, Olov Sterner Centre for Analysis and Synthesis, Lund University, P.O.Box 124, 22100 Lund, Sweden Department of Biology, Lund University, Lund, Sweden Instituto de Investigaciones Químicas, Universidad Mayor de San Andrés, La Paz, Bolivia Instituto de Biología Molecular y Biotecnología, Universidad Mayor de San Andrés, La Paz, Bolivia Corresponding author: Olov Sterner, Centre for Analysis and Synthesis, Lund University, P.O.Box 124, 22100 Lund, Sweden, Email: Olov.Sterner@chem.lu.se


Introduction
The incidence of cancer is increasing, and different strategies for controlling the disease are developed. The pool of secondary plant metabolites has always been an important provider of low-molecular anti-cancer drugs, and is expected to be so also in the future. As our understanding about the molecular mechanisms of cancer development and progression, as well as the development of treatment resistance, has increased, our ability to design new anti-cancer drugs has improved. 1,2 One of the cellular molecular pathways that is important for chemoresistance and metastasis in many cancer cells is the NF-κB pathway, where constitutive activation, or over-expression, is part of the tumour aggressiveness. 3 Terpenes containing an α-methylene-γ-lactone moiety are natural products that have been shown to possess promising anti-cancer effects, interfering with biological processes such as cell signalling, proliferation, death/apoptosis, and migration. 4,5 The significant cytotoxic activity of such terpenes is linked to the α-methylene-β-lactone moiety, which via a Michael addition can alkylate nucleophilic residues (primarily free thiols of cysteines) in proteins. A target suggested for such an interaction is the protein p65 in the NF-κB pathway. 6 In general, electrophilic compounds are considered potentially toxic and have been rejected by the pharmaceutical industry in their search for novel drug candidates. 7,8 However, 39 electrophilic drugs, mainly used in oncological therapies, have been approved by the US Food and Drug Administration. 8 These will display more or less off-target toxicity, including immunogenic responses. 8 The efficiency and selectivity of an electrophilic compound will besides the reactivity of the electrophilic moiety depend on the affinity to the target through non-covalent interactions.
Michael acceptors will react with nucleophiles via an addition reaction, not a substitution, in a reaction that is reversible, and will bind tighter to nucleophiles that present a more favourable molecular environment.
However, the identification of specific targets and structural features of non-covalent interactions of Michael acceptors has been elusive and complex to study.
Several studies of structure-activity relationships (SARs) of natural terpenoid Michael acceptors have shown that guaianolide and pseudoguaianolide sesquiterpenes with a α-methylene-γ-lactone moiety possess the most potent anti-cancer and cytotoxic activities. 9,10 Although such sesquiterpenoids are considered to interact with multiple targets in cancer cells, they have been shown to inhibit transcription factors such as NF-κB, STAT3, and AP-1, up-or downregulate the protein kinases MAPK and JNK, and increase the expression of the p53 protein. 11 Besides that, they have been shown to disrupt the redox balance and induce an oxidative stress in cancer cells. 12 In the present study we compare the cytotoxicity of 23 α-methylene-γ-lactones based on the natural product damsin (1) (see Figure 1), in normal breast epithelial MCF-10A cells and breast cancer JIMT-1 cells. 1 is a pseudoguainolide sesquiterpene with a moderate inhibitory activity on NF-κB, 13 available from the plant Ambrosia arborescens together with the 1-hydroxy derivative coronopilin (2). Compound 1 was used as the starting material for the semi-synthetic preparation of 21 novel analogues.

Results
Damsin (1) and coronopilin (2) were isolated from the plant Ambrosia arborescens as previously described. 14 Since 1 can be obtained in good amounts from the plant, we investigated the possibility to modify the structure of 1 while retaining the α-methylene-lactone moiety intact. The latter is important as procedures that require that this functionality is protected/deprotected would waist too much starting material. The Claisen-Schmidt condensation of 1 with aromatic aldehydes, under either acidic or basic conditions, was found to work reasonably well and without affecting the α-methylene--lactone moiety of 1.
The condensation between 1 and benzaldehyde yielded (E)-3-benzylidendamsin (3a), and when the cytotoxicity of 3a towards the two cell lines was compared to that of 1, 3a was found to be more potent in JIMT-1 cells than in MCF-10A cells and thereby more selective for cancer cells than normal cells (see Table 1). Apparently an (E)-3-benzyliden substituent has a favourable effect, either on the reactivity of the α-methylene--lactone moiety, or on the association of 3a with critical cellular components such as proteins. To investigate this further, we decided to expand the investigation and study 3-benzyliden derivatives formed by the Claisen-Schmidt condensation of 1 with the aromatic aldehydes 4a -4n.
The 13 derivatives 3b -3n (see Figure 1) substituted with methyl , triflouro-, methoxy-and hydroxyl groups were prepared, and assayed for cytotoxicity (see Table   1). In general, the condensations proceeded well, and often, but not always, best under acidic conditions (the yields are given in Materials and Methods). However, it was not possible to use the hydoxybenzaldehydes 4l -4n directly (to produce 3l -3n), instead the hydroxyl groups had to be protected as MOMO substituents to form 4x -4z (see Figure 2) prior to the condensation with 1, to form compounds 3x -3z (see Figure 2). These were subsequently deprotected as described in the Experimental section, to yield 3l -3n.
The condensations with aromatic aldehydes were highly stereoselective, only the E-alkenes were obtained. To investigate the influence of the double bond resulting from the condensation, the reduced derivative 3o was prepared. As direct hydrogenation of 3a also reduced the C-11/C-13 double bond in the α-methylene--lactone moiety, it was necessary to first protect the C-11/C-13 double bond of 1 as the 13-phenylthio ether to give compound 5 (see Figure 2). 5 was then condensed with benzaldehyde to give 6 which subsequently was reduced to 7 and deprotected to obtain the desired product 3o (see Figure 2 and Materials and Methods performed with a few alkyl-and alkenyl aldehydes, producing the derivatives 3r -3u for comparison.

Discussion
The structures of all compounds prepared in this The chemical shifts for all proton and carbon signals observed in CDCl 3 are presented in Tables 2 and   3. In all cases, the NMR couplings observed between protons are those that we expected from our previous experience from this type of compounds, and are for reasons of space not reported here. The comparison of the proton and carbon shifts in Tables 2 and 3  is therefore that the differences in cytotoxic activities shown in Table 1 and discussed below, depend on the molecular interaction that the C-3 substituent can provide with a protein at the site where a Michael addition can take place.

Conclusion
The MCF-10A cell line used in this study is a non-tumorigenic breast epithelial cell line 15 while the JIMT-1 cells are breast cancer-derived tumorigenic. 16 Table 1 shows the experimental cytotoxicity IC 50 values in M obtained with the two cell lines, calculated from the dose response curves. In addition, the ratios between the IC 50 MCF-10A and IC 50 JIMT-1 were calculated, and are presented in Table 1  Glutathione has been shown to reduce the toxicity of sesquiterpene lactones 17 and MCF-10A cells have been shown to have a slightly higher glutathione level than JIMT-1 cells. 18 The cytotoxicity of the reduced derivative 3o is similar to that of 3a, indicating that the double bond created at C-3 by the condensation is unimportant.
Replacing the phenyl group of 3a with a cyclohexyl (in 3p) decreases both potency and selectivity, and when comparing 3p with 3q it is obvious that an E configuration of the C-3 double bond is preferable. None of the methylated derivatives 3b -3e is more potent compared to 3a, although the p-substituted derivative 3b has a similar potency, and they are all less selective.
Methyl substituents in the benzene ring are therefore not beneficial. For the trifluoromethyl derivatives 3f -3h there is a strong tendency that p-substitution is better for the potency than m-substitution, and m is better is even worse for both potency and selectivity.
Remarkably, by keeping the same number of carbons in the chain but without branching and with a double bond in the end (as in 3u), the potency is the same as that of 3a while the selectivity is high. With a ratio of IC 50 MCF-10A cells and IC 50 JIMT-1 cells greater than 10, 3u is the most selective derivative prepared in this investigation, and will be a future starting point for the development of this class of compounds.