Dimeric 3,5-Bis(benzylidene)-4-piperidones: Tumor-Selective Cytotoxicity and Structure-Activity Relationships

Background: The objective of this study is to find novel antineoplastic agents that display greater toxicity to malignant cells than to neoplasms. In addition, the mechanisms of action of representative compounds are sought. This report describes the cytotoxicity of a number of dimers of 3,5-bis(benzylidene)-4-piperidones against human malignant cells (promyelocytic leukemia HL-60 and squamous cell carcinoma HSC-2, HSC-3, and HSC-4). Methods: Tumor specificity was evaluated by the selectivity index (SI), that is the ratio of the mean CC50 for human non-malignant oral cells (gingival fibroblasts, pulp cells, periodontal ligament fibroblasts) to that for malignant cells. Results: The compounds were highly toxic to human malignant cells. On the other hand, these molecules were less toxic to human non-malignant cells. In particular, a potent lead molecule, 3b, was identified. A QSAR study revealed that the placement of electron-releasing and hydrophilic substituents into the aryl rings led to increases in cytotoxic potencies. The modes of action of a lead compound discovered in this study designated 3b were the activation of caspases-3 and -7, as well as causing PARP1 cleavage and G2 arrest, followed by sub-G1 accumulation in the cell cycle. This compound also depolarized the mitochondrial membrane and generated reactive oxygen species in human colon carcinoma HCT116 cells. In conclusion, this study has revealed that, in general, the compounds described in this report are tumor-selective cytotoxins.


Introduction
A major interest in our laboratories is the synthesis of novel α,β-unsaturated ketones, which were designed as candidate antineoplastic agents [1].A number of these compounds react with thiols [2][3][4], which is likely an important contributor to cytotoxic effects.In particular, we are interested in 3,5-bis(benzylidene)-4-piperidones, whose general structure is displayed in Figure 1.However, the aim of the present investigation is not only to find compounds with marked antineoplastic properties but those that display greater toxicity to neoplasms than to non-malignant cells.This selective toxicity may arise since after an initial chemical insult, cancer cells may be more susceptible or more sensitive to subsequent toxic effects than non-malignant cells are [5,6].Thus, in Series 2-4, whose structures are displayed in Figure 1, there are four sites at which sequential interactions with cellular constituents could occur.Previous studies with dimeric compounds are as follows [7].The initial study used a diacyl linker and the compounds displayed potent cytotoxicity towards a variety of leukemic cells and lymphomas [7].A subsequent study revealed that these compounds were cytotoxic to colon cancers, as well as leukemias [8].An important observation was the discovery of the multidrug-resistant revertant properties of these compounds [8].Subsequently, Series 2-4 were prepared and most of these compounds demonstrated high potency to various colon cancers, as well as leukemic and lymphoma cells [9].However, limited evidence was provided as to whether greater toxicity for neoplastic cells was being demonstrated or not.Hence, the decision was made to evaluate the compounds in Series 2-4 against the HL-60, HSC-2, HSC-3, and HSC-4 neoplastic cell lines, as well as the HGF, HPC, and HPLF non-malignant cells.Should significant potency and especially selective toxicity be demonstrated, QSAR and mode of action investigations would be envisaged.

Syntheses of Compounds
The synthesis of the compounds has been described previously [7].
Previous studies with dimeric compounds are as follows [7].The initial study used a diacyl linker and the compounds displayed potent cytotoxicity towards a variety of leukemic cells and lymphomas [7].A subsequent study revealed that these compounds were cytotoxic to colon cancers, as well as leukemias [8].An important observation was the discovery of the multidrug-resistant revertant properties of these compounds [8].Subsequently, Series 2-4 were prepared and most of these compounds demonstrated high potency to various colon cancers, as well as leukemic and lymphoma cells [9].However, limited evidence was provided as to whether greater toxicity for neoplastic cells was being demonstrated or not.Hence, the decision was made to evaluate the compounds in Series 2-4 against the HL-60, HSC-2, HSC-3, and HSC-4 neoplastic cell lines, as well as the HGF, HPC, and HPLF non-malignant cells.Should significant potency and especially selective toxicity be demonstrated, QSAR and mode of action investigations would be envisaged.

Syntheses of Compounds
The synthesis of the compounds has been described previously [7].

Activation of Caspases-3 and -7
Both 3b (0.5 µM) and 3c (0.5 µM) were examined for their ability to activate caspases-3 and -7 in HSC cells.This determination was accomplished using the Caspase-Glo 3/7 assay (Promega Corporation, Madison, WI, USA).In brief, this assay uses a tetrapeptide sequence DEVD, which is optimized for caspase and luciferase activities.The cell lysate was incubated with the Caspase-Glo reagent (buffer and substrate) for 1 h and the luminescence was measured.The resultant luminescent signal produced by luciferase was confirmed to be proportional to the amount of caspase-3 and -7 activity.

PARP1 Cleavage and Activation of Bim, Bax, Puma, and Bak
Antibodies against cleaved PARP1, bim, bax, puma, and bak were obtained from Cell Signaling Technology (Beverley, MA, USA) while the anti-actin antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA).
The HSC-2 cells were cultured for 24 h and, then, treated with 0.5 µM of 3b or 3c for 24 h.The cells were scraped and collected in lysis buffer (purchased from Cell Signaling Technology), which contained 1 mM of phenylmethanesulfonyl fluoride and one tablet of protease inhibitor cocktail (Complete, EDTA-free; Roche Diagnostics, Mannheim, Germany).The proteins were subjected to 8% SDS-polyacrylamide gel electrophoresis and, then, transferred to a polyvinylidene difluoride membrane saturated with 5% nonfat dry milk.The membranes were incubated with rabbit monoclonal anti-cleaved PARP (Cell Signaling Technology, Beverly, MA, USA; #9541, 1:1000), or beta-actin antibody (Sigma-Aldrich, St Louis, MO, USA; A5441, 1:10,000) for 1 h at room temperature.The membranes were washed with Tris-buffered saline containing 0.1% Tween 80. And, the secondary antibodies were horseradish peroxidase-conjugated goat antimouse or antirabbit, which reacted with the chemoluminescent substrate system and were exposed to X-ray films.

Cell Cycle Analysis
The effect of 3b (0.5 µM) and 3c (0.5 µM) on the phase distribution of HSC-2 cells after 24 h of incubation was determined using the protocol of a manufacturer (BD Pharmingen, BD BioSciences, San Jose, CA, USA).In brief, treated and untreated cells were suspended at a density of ~1 × 10 6 cells/mL in cooled phosphate-buffered saline (PBS) without calcium and magnesium ions and fixed with 70% ethanol overnight at −20 • C. The cells were washed with cooled PBS and re-suspended in a propidium iodide/RNase staining buffer in the dark at room temperature for 0.25 h.Both acquisition and analysis were performed using EPICS ALTRA (Beckman Coulter, Brea, CA, USA) and EXPO 32 software.

Mitochondrial Membrane Potential Determination
The determination of the IC 50 values of 3b, 3c, and 5-FU on the mitochondrial membrane potential in HCT 116 cells was undertaken using a literature procedure [10].In brief, the cells were incubated at 37 • C in an atmosphere of 5% carbon dioxide for 24 h.The compounds were dissolved in DMSO, added to the cells, and incubated for 48 h.Then, a fluorescent dye, m-chlorophenylhydrazone (CCCP), and 2,4-dinitrophenol (DNP) were added to the cells, which were incubated for 0.5 h.After the addition of tetramethylrhodamine ethyl ester, the fluorescence of the cells was determined using a plate reader.

Reactive Oxygen Species Determination
The determination of the effects of 3b, 3c, and 5-FU on the generation of reactive oxygen species in HCT 116 cells was accomplished using a literature procedure [11].In brief, the cells were incubated at 37 • C in an atmosphere of 5% carbon dioxide for 24 h.Subsequently, solutions of the compounds in DMSO were added to the cells, which were incubated for 48 h.After 2 ′ ,7 ′ -dichlorofluorescein was added to the cells, incubation continued for another 0.5 h and the fluorescence of the cells was measured spectrophotometrically (excitation wavelength 485 nm; emission wavelength, 530 nm).

Quantitative Structure-Activity Relationships
The physicochemical constants for the aryl substituents in 2a-g and 3a-g, which were used in the QSAR determinations, were taken from the literature [12].The linear and semilogarithmic plots were made using a commercial statistical package [13].

Results
The compounds in Series 2 and 3 were prepared by reacting an amount of 3,5-bis(benzylidene)-4-piperidones 1 with oxalyl chloride to produce 2a-i or with malonyl chloride to form 3a-g.In addition, a related compound, Compound 4, was prepared by reacting 1a (1, R 1 = R 2 = R 3 = H) with 1,3-dibromopropane.These unsaturated ketones were evaluated against human promyelocytic leukemia cells (HL-60), as well as the human oral squamous cell carcinoma cell lines HSC-2, HSC-3, and HSC-4.These data are presented in Table 1 and reveal that, in general, the compounds in Series 2-4 are potent cytotoxins, which are much more toxic than the two reference compounds melphalan and curcumin.The compounds displaying the greatest average cytotoxic potencies are 2a, b, g, 3a, b, and g.In addition, these compounds were evaluated against human gingival fibroblasts (HGFs), human pulp cells (HPCs), and human periodontal ligament fibroblasts (HPLFs), which are non-malignant cells.These results are displayed in Table 2. Compounds with the highest CC 50 values, i.e., those that are best tolerated by the non-malignant cells with average CC 50 values in excess of 50 µM, are 2c, h, i, and 3d, as well as the drug melphalan.Selectivity index (SI) figures were generated by dividing the average CC 50 values towards HGF, HPC, and HPLF cells by the CC 50 value towards a neoplastic cell line.These results are displayed in Table 1.The compounds with the highest SI values are 2f and 3b, which are clearly lead molecules.A QSAR study revealed that cytotoxic potencies were correlated positively with the magnitude of the Hammett sigma (σ) values and Hansch pi (π) values of the aryl substituents.In addition, the ADME properties of seven potent cytotoxins were investigated and the result is presented in Table 3.The results reveal that future studies with analogs of Series 2-4 should have lower molecular weights and hydrophobicity than the current analogs.A number of mode of action studies were undertaken with 3b and 3c using HSC-2 cells.The enone 3b induced caspases-3 and -7 (Figure 2), causing the cleavage of PARP (Figure 3) and G2M arrest in the cell cycle (Figure 4).Both compounds were evaluated for their effect on the mitochondrial membrane potential (Figure 5) and generating reactive oxygen species in HCT116 cells (Figure 6). a The abbreviations in some of the headings of each column refer to the molecular weight (MW), logarithm of the partition coefficient (Log P), hydrogen bond acceptor atoms, hydrogen bond donor atoms, and topological polar surface area (TPSA).
A number of mode of action studies were undertaken with 3b and 3c using HSC-2 cells.The enone 3b induced caspases-3 and -7 (Figure 2), causing the cleavage of PARP (Figure 3) and G2M arrest in the cell cycle (Figure 4).Both compounds were evaluated for their effect on the mitochondrial membrane potential (Figure 5) and generating reactive oxygen species in HCT116 cells (Figure 6).    a The abbreviations in some of the headings of each column refer to the molecular weight (MW), logarithm of the partition coefficient (Log P), hydrogen bond acceptor atoms, hydrogen bond donor atoms, and topological polar surface area (TPSA).
A number of mode of action studies were undertaken with 3b and 3c using HSC-2 cells.The enone 3b induced caspases-3 and -7 (Figure 2), causing the cleavage of PARP (Figure 3) and G2M arrest in the cell cycle (Figure 4).Both compounds were evaluated for their effect on the mitochondrial membrane potential (Figure 5) and generating reactive oxygen species in HCT116 cells (Figure 6).

Discussion
The results of evaluating 2a-i, 3a-g, and 4 against the HL60, HSC-2, HSC-3, and HSC-4 neoplastic cells are presented in Table 1.With the exception of 2i and 3d, the compounds are potent cytotoxins.No less than 53% of the CC50 values are submicromolar and, if one This process was repeated using the CC 50 figures of 2a-g towards HSC-2, HSC-3, and HSC-4 cells.Then these determinations were repeated, except the CC 50 values of 3a-g, not 2a-g, were used.All of these determinations were repeated, except semilogarithmic plots instead of linear ones were made.
The results obtained are presented in Tables S1 (data of 2a-g) and S2 (results of 3a-g) in the Supplemental section of this report.The following observations and conclusions were drawn: (1) A statistically significant positive relationship between the CC 50 figures of 2a-g and the sigma (σ) values was obtained in a little over half of the observations made.This result reveals that the electronic properties of the aryl substituents play a significant albeit minor role in determining the magnitude of the CC 50 values.The correlation is positive, revealing that the CC 50 values rise (potency diminishes) as the sigma (σ) values increase in magnitude; (2) The data in Table S1 reveal clearly the positive correlation between the CC 50 values and the π constants.Thus, potency increases as the hydrophilicity of the molecules falls; (3) There is no correlation between the CC 50 values of 2a-g and the MR values of the aryl substituents; (4) The results in Table S2 for 3a-g reveal a similar pattern as displayed by 2a-g.There is a statistically significant positive correlation between the sigma (σ) values and the CC 50 data in half of the determinations.The importance of the π values, but not the MR constants, in controlling potencies is clearly demonstrated.Thus, in planning the expansion in this series of compounds, groups should be placed in the aryl rings, which are either electron-donating, hydrophilic, or have both of these properties.
A further issue to consider is whether the magnitude of the selective toxicity is influenced by the electronic, hydrophobic, and steric properties of the groups in the aryl rings.Consequently, linear and semilogarithmic plots were made between the average SI values of 2a-g and the σ, π, and MR constants in the aryl rings of these compounds.A similar procedure was undertaken with 3a-c, e-g (3d does not have a specific average SI figure).The results of this analysis are presented in Supplementary Table S3.In the case of Series 2, the semilogarithmic plots revealed a negative correlation between the σ and π values.No other correlations were noted (p > 0.05).
Two highly important features of a candidate antineoplastic agent are their potency and SI value.In order to find lead compounds that combine both of these properties, potency-selectivity expression (PSE) figures for the compounds in Series 2-4 were obtained.The values are the products of the reciprocal of the average CC 50 value against neoplastic cells and the average SI figure times 100.These data are presented in Table 2.The following compounds have PSE figures in excess of 100, namely, 2a, b, f, g, 3a, b, g, and 4. The 4-fluoro analogs 2b and 3b have the highest PSE values and are lead molecules.The PSE values of the unsubstituted compounds 3a and 4 are 290 and 120, respectively, indicating the contribution of the oxygen atoms in the spacer group of 3a to its greater PSE value than 4. One may also note that virtually all of the compounds in Series 2-4 have PSE figures far higher than those displayed by melphalan and curcumin.
An effort was made to find out if representative compounds in Series 2 and 3 possess drug-like properties.Accordingly, three compounds in both Series 2 and 3, namely, 2a, b, and g and 3a, b, and g, which have the same aryl substituents, were considered as they have good PSE figures.The rule of five articulated by Lipinski et al. [15] pertains to the size of the molecular weights, LogP values, and the numbers of hydrogen bond acceptor and donor atoms while Veber et al. [16] suggest that the number of rotatable bonds and the total polar surface (TPSA) are important physicochemical parameters.The results are portrayed in Table 3.In general, the compounds have favorable properties in terms of hydrogen bonding, rotatable bonds, and TPSA.However, the size and hydrophobicity of these molecules suggest that smaller molecules should be created with lower hydrophobic properties.The modes of action of these compounds were addressed, initially in general terms and then with specific targets in mind.As mentioned previously, a number of investigations revealed that conjugated enones react readily with thiols [2][3][4].Specifically, various enones react only with thiol groups, even when other functionalities, such as amino and hydroxyl groups, are present [17].
Some of the biochemical mechanisms whereby an important lead compound, 3b, exerts its cytotoxic properties were investigated; also, a question was addressed, namely, why is 3b more potent than related analogs, such as 3c?The following investigations were undertaken using HSC-2 cells and the same concentrations of compounds.Caspases-3 and -7 are activated by both the death receptor and mitochondrial pathways, which lead to apoptosis.After 24 h of incubation, 3b, but not 3c, induced caspases-3 and -7 and this result is presented in Figure 2. Single-stranded DNA breaks can be repaired by poly(ADPribose)polymerase 1 (PARP1) [18] and, thus, compounds that cleave PARP1 reduce the extent of DNA repair and, therefore, may contribute to cytotoxic effects [19].Figure 3 reveals that 3b, in contrast to 3c, leads to the cleavage of PARP1.Cytotoxic agents exert their bioactivity by interfering with the progression of the cell cycle.Cell cycle analysis of 3b and 3c reveals that 3b, but not 3c, causes a G2/M arrest in HSC-2 cells (Figure 4).The effect of 3b and 3c on pro-apoptotic proteins, such as bim, bax, puma, and bak, was examined in HSC-2 cells but no changes in the concentrations of these proteins were observed after 24 h (data not shown).Hence, the way in which 3b displays its cytotoxic properties includes the activation of caspases -3 and -7, cleavage of PARP1, and G 2 /M arrest, leading to producing a sub-G1 population (DNA fragments produced by activated DNases) [20] in the cell cycle.On the other hand, at the same concentration of 0.5 µM, 3c does not exert these effects, which explains, at least in part, the greater potency of 3b than 3c.
These mode of action studies were conducted using HSC-2 cells.The decision was made to investigate possible ways in which cytotoxicity is mediated using another cell line.Both 3b and 3c have been evaluated against human HCT116 colon cancer cells and possess IC 50 values of 0.60 and 0.05 µM, respectively.One way in which cytotoxicity can be mediated is by lowering the mitochondrial membrane potential (MMP).The data presented in Figure 5 were generated using the mitochondriotropic dye tetramethylrhodamine ethyl ester, which is extruded from the mitochondria into the cytosol when the MMP is lowered.The results presented in Figure 5 reveal that 3b has a huge effect on lowering the MMP; whereas, 3c has no effect at the concentrations utilized.Another way in which cytotoxicity can be achieved is by increasing the concentration of reactive oxygen species (ROS) [21].In this case, the magnitude of the fluorescence of the dye 2 ′ ,3 ′ -dichlorofluorescein diacetate revealed that, while 3b and 3c cause an increase in the ROS concentrations, 3b has approximately twice the potency of 3c (Figure 6).One may conclude that the generation of ROS contributes to the cytotoxicity observed.
In summary, this study has led to the identification of three series of enones, 2-4, which, in general, have far greater toxicity to various neoplasms than non-malignant cells.QSAR studies revealed the importance of the electronic and hydrophobic properties of the aryl substituents.The ADME studies revealed that reducing the size and hydrophobicity of the aryl substituents should be undertaken.

Conclusions
The results generated in this study reveal that the majority of the compounds in Series 2 and 3 are potent cytotoxins and display greater toxicity to some neoplasms than to various non-malignant cells.From the data presented in Tables 1 and 2, the compounds 2a, b, g, 3a, b, and g are lead molecules in terms of potency while 2f and 3b have the highest SI values.The PSE values, which take into consideration both potency and SI figures, reveal that the 4-fluoro analogs 2b and 3b have the highest PSE values and are lead

Table 1 .
Evaluation of the compounds in Series 2-4 against human malignant cells.

Table 1 .
Cont.The CC 50 values are the concentrations of the compounds required to kill 50% of the cells.bTheletters SI refer to the selectivity index.The figures are obtained by dividing the average CC 50 values towards non-malignant cells (Table2) by the specific CC 50 figure towards a specific neoplastic cell line. a

Table 2 .
Evaluation of the compounds in Series 2-4 against human non-malignant cells.The CC 50 values are the concentrations of the compounds required to kill 50% of the cells.b The potencyselectivity expression (PSE) figures of the compounds are the products of the reciprocal of the average CC 50 value and the selectivity index (SI) figures times 100. a

, b, f, g, 3a, b, f, g, and 4.
a The abbreviations in some of the headings of each column refer to the molecular weight (MW), logarithm of the partition coefficient (Log P), hydrogen bond acceptor atoms, hydrogen bond donor atoms, and topological polar surface area (TPSA).