Novel A-Ring Chalcone Derivatives of Oleanolic and Ursolic Amides with Anti-Proliferative Effect Mediated through ROS-Triggered Apoptosis

A series of A-ring modified oleanolic and ursolic acid derivatives including C28 amides (3-oxo-C2-nicotinoylidene/furfurylidene, 3β-hydroxy-C2-nicotinoylidene, 3β-nicotinoyloxy-, 2-cyano-3,4-seco-4(23)-ene, indolo-, lactame and azepane) were synthesized and screened for their cytotoxic activity against the NCI-60 cancer cell line panel. The results of the first assay of thirty-two tested compounds showed that eleven derivatives exhibited cytotoxicity against cancer cells, and six of them were selected for complete dose–response studies. A systematic study of local SARs has been carried out by comparative analysis of potency distributions and similarity relationships among the synthesized compounds using network-like similarity graphs. Among the oleanane type triterpenoids, C2-[4-pyridinylidene]-oleanonic C28-morpholinyl amide exhibited sub-micromolar potencies against 15 different tumor cell lines and revealed particular selectivity for non-small cell lung cancer (HOP-92) with a GI50 value of 0.0347 μM. On the other hand, superior results were observed for C2-[3-pyridinylidene]-ursonic N-methyl-piperazinyl amide 29, which exhibited a broad-spectrum inhibition activity with GI50 < 1 μM against 33 tumor cell lines and <2 μM against all 60 cell lines. This compound has been further evaluated for cell cycle analysis to decipher the mechanism of action. The data indicate that compound 29 could exhibit both cytostatic and cytotoxic activity, depending on the cell line evaluated. The cytostatic activity appears to be determined by induction of the cell cycle arrest at the S (MCF-7, SH-SY5Y cells) or G0/G1 phases (A549 cells), whereas cytotoxicity of the compound against normal cells is nonspecific and arises from apoptosis without significant alterations in cell cycle distribution (HEK293 cells). Our results suggest that the antiproliferative effect of compound 29 is mediated through ROS-triggered apoptosis that involves mitochondrial membrane potential depolarization and caspase activation.


Introduction
Plants have always played an important role in human health care [1]. The discovery of novel bioactive compounds from natural plants is one of the most effective trends in natural product research [2]. Among these, naturally occurring triterpenoids have found direct application as drug entities and play an important role as templates for the design, synthesis, and semi-synthesis of novel substances [3,4]. Pentacyclic triterpenoids, such as oleanolic (1) and ursolic (2) acids, contain a biologically active scaffold with a high safety profile in cancer therapy and are suitable to carry out different chemical transformations

Chemistry
We designed a new series of oleanonic (3) and ursonic (4) acids derivatives by the introduction of nitrogen-containing heterocycles to C2, C3, and C28 positions or A-ring skeleton transformation. (3-or 4)-Pyridinylidene and furfurylidene fragments were coupled at the C2-position; the nicotinoyloxy-fragment was introduced at C3-position; N-methylpiperazinyl-, piperazinyl-and morpholinyl-amides were synthesized at the C28-position. Modification of the A-ring included a Fischer indolization reaction to indole-fused derivatives, and Beckmann rearrangement to seven-membered lactame (with the following modification onto azepanes) and 2-nitrilo-3,4-seco-4(23)-en-derivatives were carried out. The synthesis of all of the mentioned derivatives 5-36 is presented on Schemes 1-3.
The esterification of oleanolic acid (1) by nicotinic acid chloride in a pyridine-tributylamine media under reflux led to 3β-nicotinoyloxy-derivative 15, which was further transformed into appropriate amide analogs 16-18.
The next series involving A-ring transformations is shown at Schemes 2 and 3. A-seco-4(23)-enes 19, 33 and lactams 23, 43, 45, 56 were obtained by a Beckmann rearrangement of the corresponding C3-oximes using SOCl2 in dioxane. Reduction of lactams 20 and 34 with lithium aluminum hydride in THF under reflux afforded azepanes 21 and 35 with 56-60% yield. Indoles 22-24 and 36 were obtained from oleanonic 3 and ursonic 4 acids using a Fischer reaction followed by amidation at the C28-position. The structures of the compounds were ascertained by the combined use of spectroscopy and elemental analyses.
Thus, a series of triterpenic acids with modified A-ring (3-oxo-, 3-oxo-C2-nicotinoylidene/furfurylidene, 3β-hydroxy-C2-nicotinoylidene-, 3β-nicotinoyloxy-, 2-cyano-3,4-seco-4(23)-ene, indolo-, lactame and azepane) and their C28 amides was synthesized. Generally, the C2 position of pentacyclic triterpenoids is a preferential site to carry out modification and to prepare analogs with better anticancer activities than the parent acids [5,6]. The principles of oleanolic acid modification, specifically the formation of a 2-cyano-1-en-3-one on the A-ring, the modification of the C-ring by converting the 12(13)ene to 12-oxo-9(11)-en and/or methyl esterification or formation of the imidazolide group of the C28 carboxylic group, has been known since 2000 and led to the development of the effective anticancer agents CDDO, CDDO-Me (methyl bordoxolone), and CDDO-Im, which are under clinical trials [7][8][9][10]. Similar chemical modifications have been conducted to other triterpene cores to improve their potency and overcome the drawbacks. For example, glycyrrhetinic acid has been used for the synthesis of CDDO analogs (soloxolones), which significantly improved cytotoxicity [11][12][13], and recently the effective inhibition by methyl soloxolone TGF-β-driven EMT of tumor cells was shown [14]. The same ursanetype analogs were obtained by an oxidative ozonolysis-mediated C-ring enone formation with a potency of approximately five-fold less than the corresponding oleanolic acid derivatives [15].
The Claisen-Schmidt aldol condensation giving 3-oxo-C2-benzylidene (or chalcone) triterpenoids is an efficient type of chemical modification and the first mentions of the physical and chemical properties of such derivatives date back to 1957 by D. H. R. Barton et al. [16]. Nowdays, these compounds demonstrate high potential as antibacterial and antiinflammatory [17,18], antioxidant [19], antidiabetic [20][21][22][23], and cytotoxic agents [24][25][26][27][28][29]. Most of the C2-benzylidenes among the triterpenoids described are derivatives with the carboxylic group at C28 [24,28,29]. The strategy for modification of both the A-ring with chalcone introduction at C2 and derivatization of 28-COOH seems to be attractive because the obtained hybrid molecules while bearing two different pharmacophores can demonstrate high biological potential. For example, derivatives of the ursane type with p-chlorine-benzylideneor 4-pyridynilidene-at C2 and nitrooxy ethyl substituents at C28 were found to be more cytotoxic than the parent drug (IC 50 ranged between 4.28-12.74 µm) and the lead derivatives could induce cell cycle arrest at the G1 phase and apoptosis in a dose-dependent manner via caspase-8 activation [27,30].
The introduction of heterocyclic fragments, especially piperazine, has been demonstrated to enhance anticancer properties [31]. Recently we have found that modification of triterpenoids to indole derivatives on the A-ring with amidation to C28-amide, as well as the introduction of piperazine or N-methyl-piperazine, have a positive effect on anticancer activity [32,33]. Chalcone derivatives of messagenine and platanic acid [25], polyaminolupanes [34], A-azepano-, and 3-amino-3,4-seco-triterpenoids [35] were also found to be effective antiproliferative agents against different cancer cell lines with submicromolar concentration values of GI 50 < 1 µM.
The esterification of oleanolic acid (1) by nicotinic acid chloride in a pyridine-tributylamine media under reflux led to 3β-nicotinoyloxy-derivative 15, which was further transformed into appropriate amide analogs 16-18.
The next series involving A-ring transformations is shown at Schemes 2 and 3. A-seco-4(23)-enes 19, 33 and lactams 23, 43, 45, 56 were obtained by a Beckmann rearrangement of the corresponding C3-oximes using SOCl 2 in dioxane. Reduction of lactams 20 and 34 with lithium aluminum hydride in THF under reflux afforded azepanes 21 and 35 with 56-60% yield. Indoles 22-24 and 36 were obtained from oleanonic 3 and ursonic 4 acids using a Fischer reaction followed by amidation at the C28-position. The structures of the compounds were ascertained by the combined use of spectroscopy and elemental analyses.

NCI-60 Anticancer Drug Screening
Compounds 5-36 were selected by the National Cancer Institute (NCI) Developmental Therapeutic Program (www.dtp.nci.nih.gov, accessed on 16 October 2019) for the in vitro cell line screening to investigate their anticancer activity. Anticancer assays were performed according to the US NCI protocol, which was described elsewhere [40][41][42][43][44][45]. Compounds 5-36 were evaluated against 58 human tumor cell lines, which were derived from nine different cancer types: leukemia, melanoma, lung, colon, CNS, ovarian, renal, prostate, and breast cancers. At first, compounds were tested at a single high dose concentration (10 µM), and according to the criterion adopted by the NCI, compounds that reduced the growth of any of the cell lines to approximately 32% or less were considered to be active.
Compounds 7, 12, 13, 27, 29, and 32 were selected for complete dose-response studies with five different test concentrations (0.01, 0.1, 1, 10, and 100 µM). The dose-response curves (% growth vs. sample concentration) of these compounds against each cell line in the NCI screening, NMR data as well, can be found in the Supporting Information ( Figures S1-S30). A comparative summary of the single-dose mean growth inhibition (%) for all active compounds, and for those that passed the initial one-dose screening test, the mean (GI 50 , µM) and the most sensitive cell line is provided in Table 1. According to the first stage results which are presented in Table 1, the following cancer cell lines: melanoma (LOX IMVI), leukemia (HL-60(TB), SR), non-small lung cancer (NCI-H460), and colon cancer (HT29) were the most sensitive to oleanolic acid derivatives with a growth percent range from −95.08% to 25.52%. Melanoma (SK-MEL-5), renal cancer (UO-31), and colon cancer (COLO-250) cell lines were the most sensitive to ursolic acid derivatives with a growth percent range from −99.47% to −80.36%.
Compounds 7, 13, 27, and 29 displayed high cytotoxic activity with a mean GI 50 < 5 µM. The log mean values of the parameter for GI 50 , TGI, and LC 50 related to the log values (the maximum sensitivity in excess of the mean) and log range values are given in Table 2. These parameters highlight the selectivity and potency of antitumor agents. Higher values of these deltas and ranges indicate high selectivity against some cancers over others. The lower median log GI 50 value (−6.12) for compound 29 showed it to be the most potent compound for all cell lines. The effective growth inhibition of compound 13 (−5.81) also accounts for its high range log GI 50 and log LC 50 values with 2.3 and 1.23 respectively, among all 60 cell lines. Compounds 13 and 29 displayed the most potent cytotoxic activity with significant inhibition for most of the 60 cell lines, and their mean GI 50 , TGI (concentration of compound that totally inhibits cell growth), and LC 50 (concentration of compound that kills 50% of cells) values across each cell line are shown in Table 3. Thus, compound 13 exhibited a broad spectrum of antiproliferative activity with a GI 50 of <4 µM for 93% and <1 µM for 25% of the tested cell lines. Strong growth inhibition (GI 50 < 1 µM) was observed against all leukemia cell lines with (from 0.365 µM to 0.891 µM), as well as against colon cancer (HCT-116  A comparison of obtained results for leader compounds with respect to the activity reported for the standard drug doxorubicine, used by NCI as control [46], reflects that compound 29 showed activity against prostate cancer PC-3 (GI 50 = 0.275 µM) which is comparable with a standard drug (GI 50 = 0.32 µM), as well as against colon cancer HCT-15 compounds 13 (GI 50 = 0.939 µM) and 29 (GI 50 = 0.654 µM) showed comparable activity, while this value was 0.95 µM for the standard drug doxorubicine. The highest activity was observed for compound 13 against non-small cell lung cancer HOP-92 (GI 50 = 0.0347 µM), that is three-fold times more effective than for doxorubicine (GI 50 = 0.10 µM), as well as against HCT-15 colon cancer cell line compound 13 was 10-fold (GI 50 = 0.654 µM) and 29 (GI 50 = 1.20 µM) was 5-fold more effective then doxorubicine (GI 50 = 6.46 µM).

Mechanisms In Vitro Studies Cell Cycle Analysis
Compounds 13 and 29, found to be the most potent in the NCI cytotoxicity screening, have been further evaluated for cell cycle analysis to clarify the mechanisms of their action. PI (propidium iodide) staining followed by flow cytometry was performed to assess the cell cycle progression in response to compounds 13 and 29 exposure in cancerous (lung adenocarcinoma A549, breast adenocarcinoma MCF-7, neuroblastoma SH-SY5Y) and conditionally-normal (human embryonic kidney HEK293) cells. For cell cycle analysis, compounds 13 and 29 were used at their IC 50 values, which were previously established for the aforementioned cell lines in the additional laboratory cytotoxicity screen (see Table S4

The Cell Apoptosis Assay for Compound 29
In addition, we have examined compound 29-induced apoptosis in HEK293 and MCF-7 cells by Annexin V/SYTOX staining followed by flow cytometry. This approach allows the distinguishing of the early and late apoptotic cells. We have found that HEK293 and MCF-7 cells, treated for 24 h with compound 29, exhibited a moderate increase both of early and late apoptotic cells in (Table 4), and when the treatment with 29 was maintained for 48 h, a pronounced augmentation of late apoptotic cells has been observed. Thus, these data indicate and confirm that compound 29 causes a significant time-dependent increase of apoptosis in HEK293 and MCF-7 cells.

Measurement of Intracellular Reactive Oxygen Species Level for Compound 29
Excessive reactive oxygen species (ROS) production and associated mitochondrial disruption is known to result in oxidative stress and subsequent cell apoptosis [47]. Moreover, ROS have been demonstrated to be highly reactive species that cause DNA damage [48]. To evaluate whether compound 29-induced apoptotic cell death was mediated by ROS generation, levels of intracellular ROS were estimated using 2 ,7 -dichlorofluorescein diacetate (CM-H2DCFDA) as a fluorescent probe. Figure 4 shows a significant time-dependent increase in ROS accumulation in HEK293 cells, treated with compound 29 (~28 folds after 1.5 h and~40 folds after 3 h), while in MCF-7 cells ROS generation was less pronounced (~8 folds after 1.5 h and~9 folds after 3 h). Considering the dissipation of mitochondrial membrane potential (MMP) as the earliest event of the apoptotic cascade and as one of the specific signs of apoptosis [49], we used the JC-1 cationic dye to detect the changes of mitochondrial membrane potential in HEK293 and MCF-7 cells upon compound 29 treatments. As demonstrated in Figure  5A, a progressive time-dependent decrease in the red/green fluorescence intensity ratio in HEK293 cells was observed after the substance treatment (14.7 μM), indicating the mitochondrial membrane depolarization. In MCF-7 cells, compound 29 (11.1 μM) evoked a moderate decline of red/green ratio in a time course-dependent manner ( Figure 5B), suggesting a reduction of mitochondrial membrane potential, although to a lesser degree compared with that in HEK293 cells. Considering the dissipation of mitochondrial membrane potential (MMP) as the earliest event of the apoptotic cascade and as one of the specific signs of apoptosis [49], we used the JC-1 cationic dye to detect the changes of mitochondrial membrane potential in HEK293 and MCF-7 cells upon compound 29 treatments. As demonstrated in Figure 5A, a progressive time-dependent decrease in the red/green fluorescence intensity ratio in HEK293 cells was observed after the substance treatment (14.7 µM), indicating the mitochondrial membrane depolarization. In MCF-7 cells, compound 29 (11.1 µM) evoked a moderate decline of red/green ratio in a time course-dependent manner ( Figure 5B), suggesting a reduction of mitochondrial membrane potential, although to a lesser degree compared with that in HEK293 cells.

Caspase 8, 9 Activity Assay for Compound 29
It is well-established that apoptosis can be triggered through two major pathways: the cell death receptor-mediated extrinsic pathway and the mitochondrial-mediated intrinsic pathway, resulting in activation of caspase-8 and caspase-9, respectively, followed by induction of the downstream executioner caspases-3/7 [50]. The intrinsic apoptosis pathway is initiated by mitochondrial alterations culminating in the release of mitochondrial cytochrome c with a concomitant reduction of the mitochondrial transmembrane potential. To further evaluate whether compound 29 preferentially affects the extrinsic and/or intrinsic apoptotic pathways, the activities of initiator caspases 8 and 9 were assessed. As shown in Figure 6A, compound 29 did not affect caspase-9 activity, whereas caspase-8 activity was increased after 6 h of compound' treatment in HEK293 cells. Interestingly, in MCF-7 cells the substance caused a marked rise of caspase-9 and a less pronounced increase of caspase-8 activities at the 6 h time point ( Figure 6B). Notably, the highest increment in values of caspase activity was observed for caspase-8 in HEK293 cells and caspase-9 in MCF-7 cells, suggesting a role of caspase-8 in mediating apoptosis in HEK293 cells, while caspase-9 may promote apoptotic cell death in MCF-7 cells. Taken together, these data indicate the involvement of caspase-dependent apoptosis and presume that compound 29 dependently on the cell line may evoke apoptosis preferentially by intrinsic or by extrinsic pathway.  point ( Figure 6B). Notably, the highest increment in values of caspase activity was observed for caspase-8 in HEK293 cells and caspase-9 in MCF-7 cells, suggesting a role of caspase-8 in mediating apoptosis in HEK293 cells, while caspase-9 may promote apoptotic cell death in MCF-7 cells. Taken together, these data indicate the involvement of caspase-dependent apoptosis and presume that compound 29 dependently on the cell line may evoke apoptosis preferentially by intrinsic or by extrinsic pathway. In summary, despite the precise targets of compound 29 remaining elusive, overall data clearly demonstrated that the substance raised ROS generation, which in turn, resulted in cell-cycle dependent (MCF-7) or cell-cycle-independent (HEK293 cells) apoptosis. Annexin V/SYTOX staining, evaluation of mitochondrial membrane potential, and initiator caspases activity prove the apoptosis induction and suggested that compound 29 caused the apoptosis in MCF-7 cells mainly via the intrinsic pathway by depolarising MMP and subsequent activation of downstream caspase-9, although the contribution of the cell death receptor-mediated pathway is not excluded as well. Withal, in compound 29-treated HEK293 cells, the accumulation of sub-G1 apoptotic cells occurred without disturbances of the cell cycle and was accompanied by a decrease of MMP and substantial activation of caspase-8, thus, proposing the involvement of the extrinsic apoptosis pathway. However, the extrinsic pathway can converge on the In summary, despite the precise targets of compound 29 remaining elusive, overall data clearly demonstrated that the substance raised ROS generation, which in turn, resulted in cell-cycle dependent (MCF-7) or cell-cycle-independent (HEK293 cells) apoptosis. Annexin V/SYTOX staining, evaluation of mitochondrial membrane potential, and initiator caspases activity prove the apoptosis induction and suggested that compound 29 caused the apoptosis in MCF-7 cells mainly via the intrinsic pathway by depolarising MMP and subsequent activation of downstream caspase-9, although the contribution of the cell death receptor-mediated pathway is not excluded as well. Withal, in compound 29-treated HEK293 cells, the accumulation of sub-G 1 apoptotic cells occurred without disturbances of the cell cycle and was accompanied by a decrease of MMP and substantial activation of caspase-8, thus, proposing the involvement of the extrinsic apoptosis pathway. However, the extrinsic pathway can converge on the intrinsic pathway through the caspase-8-mediated direct cleavage of BID protein, which is responsible for mitochondrial cytochrome c release followed by the subsequent triggering of the mitochondrial-centered control mechanism [51]. According to literature data, the most relevant mechanisms of the anticancer activity of triterpenoids involved cell cycle arrest, apoptosis, and autophagy triggered by the effect of these secondary metabolites on the mitogen-activated different signaling pathways [3][4][5]. Mechanistically, ursolic acid mediates its antitumor potential through inhibition of NF-κB activation induced by carcinogenic agents with targets at cyclooxygenase 2, matrix metallo-proteinase 9, and cyclin D11 [52]. It also inhibits tumor growth through other promising mechanisms involving angiogenesis and metastasis [53]. In a side-by-side comparison, C-2-benzylidene-3-oxo-ursolic acid derivative, contained indole fragments, inhibited glioma cell growth, induced apoptosis, and arrested the cell cycle through metabolic pathway down-regulation [29].

Structure-Activity Analysis with Network-like Similarity Graphs
In order to systematically comprehend the biological data obtained and guide future drug design efforts, we performed a structure-activity relationship analysis with networklike similarity graphs [54] using the Rubberband Forcefield approach implemented in DataWarrior software [55]. In essence, it maps the studied molecules onto 2D-chemical space as graph nodes so that similar structures are located closely. Similarity relationships between them are shown as graph edges. In addition, the so-called Structure-Activity Landscape Index (SALI) is calculated for all pairs of similar molecules. The SALI value is proportional to activity change and inversely proportional to the dissimilarity between molecules. SALI defines the size of nodes and allows easy identification of activity cliffs when an abrupt change in activity is achieved with small structural modification. As an activity measure, mean growth percentages from NCI test panel cell lines were used. Chemical structures were represented with circular fingerprint SkelSpheres for fine-grained chemical similarity and with Flexophore to assess 3D-pharmacophore similarity.
The obtained network-like similarity graphs are shown in Figure 7. According to the SkelSpheres descriptor, which takes into account the mutual arrangement of the atoms and stereochemistry, the compounds were distributed rather uniformly. Two clusters can be recognized. C1 contains C3-benzylidene and indole-fused derivatives. C2 is comprised of C3-oxo and C3-nicotinoyloxy, lactame, and azepane derivatives, with all of them showing low cytotoxicity and flat SAR. The most active compounds belong to C1, but are separated with a large "chemical distance" and inactive analogs (except for 7 and 27, which differ only in C29 methyl position and have similar potency). Compound 13 demonstrates that the C28-morpholinyl fragment is clearly beneficial over piperazinyl and N-methylpiperazinyl (9, 10, 11) or free carboxyl (5, 6), and C2-4-nicotinoylidene is superior to 3-nicotinoylidene (5,9), or furfurylidene (11). At the same time, compounds 7 and 27, comprising C2-furfurylidene and C28-carboxyl, are more active than 13, and the lead compound 29 features C2-3-nicotinoylidene and C28-N-methylpiperazinyl.
Hence, the 2D SkelSpheres fingerprint appears to be unable to correctly perceive SAR in the series. Since it is reasonable to assume that compounds of similar structure share the mechanism of action, i.e., have the same molecular target, we performed a similarity analysis with the Flexophore descriptor. The latter takes into account molecular flexibility and pharmacophoric features responsible for protein-binding behavior. As Figure 7b shows, this approach produced better results. We can see a big area of continuous SAR with inactive compounds on the right side (C1), while hits populate the upper-right corner and are located more closely to each other (C2). Several activity cliffs can be readily recognized here as well. The largest cliff shows that for ursolic derivatives 25 and 26, activity is vastly improved upon the introduction of C2-furfurylidene (compounds 7 and 27). Lead ursolic acid derivative 29 stands out having only two neighbors with close pharmacophore properties, inactive 30 and 31, where C2-3-nicotinoylidene is substituted with C2-4-nicotinoylidene or C2-furfurylidene, respectively. Thus, the introduction of different substituents at the C2 position of oleanonic 3 or ursonic 4 acids showed that only the furfurylidene group led to the higher cytotoxic activity, resulting in the potent analogs 7 and 27, which inhibited a broad spectrum and good antiproliferative activity with a mean growth inhibition percentage between −58.89% and −40.62% in the first stage. As shown in Table 1, compounds 7 and 27 had mean GI 50 (concentration of compound that inhibits cell growth by 50%) values of 3.39 µM and 4.57 µM, respectively, which showed that ursane core is 1.34-fold more active than oleanane. The reduction of 3-oxo-group of the inactive C2-3-pyridinylidene acids 5 and 25 resulted in the more active analogs 12 and 32 with a mean GI 50 of 8.13 µM and of 15.14 µM.
Modification of inactive N-methylpiperazinyl-amides 8 and 28 at C2-positions had a different influence which depends on the triterpene core type. Thus, among C2-3pyridinylidene-N-methylpiperazinyl amides, ursane derivative 29 showed higher activity with a mean growth inhibition percentage of −74.90%, while moderate activity was observed for oleanolic acid analogue 9 with 55.08% of mean growth. Similarly, among the C2-furfurylidene-amides 11 and 31, the activity was observed for ursane type analog 31 with selectivity against renal cancer UO-31 (−97.52%).
type derivative 10 showed inhibitory activity against the leukemia HL-60(TB) cell line in the the one-dose 60-cell assay. The replacement of the N-methyl-piperazinyl fragment (compound 10) by a morpholinyl moiety (compound 13) led to superior results with a mean GI50 of 1.55 μM, and LOX IMVI (melanoma) was the most sensitive cancer cell line. descriptors. Compounds are displayed as nodes and edges indicate molecular similarity relationships. Nodes are colored according to compound cytotoxicity using a continuous color spectrum from green (highest potency in the data set) over red to blue (lowest potency). Nodes are scaled in size according to their contribution to local SAR discontinuity.
We can conclude that compounds presented here are characterized with non-additive SAR, i.e., substituents do not act independently, and the final effect on cytotoxicity could not be ruled out from individual structure modifications. Analysis of clusters represented by active molecules suggests that substituents at C2 and C28 have a strong influence on cytotoxicity, but their direction depends on the core triterpene structure. Therefore, pharmacophore modeling should be used to guide further optimization of lead compounds. The mapping of novel virtual structures onto a network-like similarity graph developed in this work may provide a venue to overcome this issue.

CellMiner and Gene Enrichment Analysis
To govern the mechanism of action studies for lead compounds, we have analyzed their cytotoxic activity spectrum using the CellMiner pattern comparison tool [56]. The Figure 7. Network-like similarity graphs generated with SkelSpheres (a) and Flexophore (b) descriptors. Compounds are displayed as nodes and edges indicate molecular similarity relationships. Nodes are colored according to compound cytotoxicity using a continuous color spectrum from green (highest potency in the data set) over red to blue (lowest potency). Nodes are scaled in size according to their contribution to local SAR discontinuity.
On the other hand, among the C2-4-pyridinylidene amides 10 and 30, only oleanane type derivative 10 showed inhibitory activity against the leukemia HL-60(TB) cell line in the the one-dose 60-cell assay. The replacement of the N-methyl-piperazinyl fragment (compound 10) by a morpholinyl moiety (compound 13) led to superior results with a mean GI 50  We can conclude that compounds presented here are characterized with non-additive SAR, i.e., substituents do not act independently, and the final effect on cytotoxicity could not be ruled out from individual structure modifications. Analysis of clusters represented by active molecules suggests that substituents at C2 and C28 have a strong influence on cytotoxicity, but their direction depends on the core triterpene structure. Therefore, pharmacophore modeling should be used to guide further optimization of lead compounds. The mapping of novel virtual structures onto a network-like similarity graph developed in this work may provide a venue to overcome this issue.

CellMiner and Gene Enrichment Analysis
To govern the mechanism of action studies for lead compounds, we have analyzed their cytotoxic activity spectrum using the CellMiner pattern comparison tool [56]. The premise of this approach is in the assumption that drugs with a similar cytotoxic activity profile share a molecular target or mechanism of action. Hence, pGI 50 values obtained for NCI-60 cell lines for compounds 4-6, 10, 11, and 21 were used as seeds to identify significant (p < 0.05) correlations with compounds that were previously tested at NCI. Results were filtered to exclude weak correlations (Pearson's coefficient r < 0.5) and substances with unknown mechanisms of action (Table 5). We also identified correlations between the 60-cell line gene expression patterns and cancer cell lines sensitivity profiles using CellMiner and Gene Ontology (GO) term enrichment analysis to further elucidate plausible molecular effectors and targets of compounds' action (Table S4). By analyzing the NCI-60 cell lines for a correlation between their transcriptome and their sensitivity to the cytotoxic effects, we found genes that were significantly correlated (p < 0.05) with their in vitro antiproliferative activity. Table 5. Possible mechanism of action for the lead compounds according to CellMiner 1 .

Compound
Pearson's    Alkylating at N 7 position of guanine, tubulin FDA approved 1 The drug activity levels used were expressed as pGI 50 and obtained from the Developmental Therapeutics Program (DTP) at http: //dtp.cancer.gov/index.html. 2 Pearson's correlations between the compound and NCI synthetic library, only correlations with r > 0.5 were considered.
No analogs with known mechanisms of action have been found for compound 7. Gene enrichment analysis revealed several interesting traits. Significant correlations were found for genes involved in interleukin-4 receptor binding, lipid-transporting, and steroltransporting ATPase activity (ABCG1), as well as other genes, mediating immune cell activation (CD2, CD48, CR2, CCR9) and cholesterol metabolism.
The activity distribution of compounds 12, 13, and 29 against NCI-60 cell lines correlates the most with several benzamide type HDAC inhibitors. HDAC inhibitors mostly act via epigenetic regulation and are known to cause cell cycle arrest and apoptosis, reduce angiogenesis, and modulate immune response [57]. Similarly, compound 12 appears to act via CD38, CD52, CCKBR, P2RY1, CXCR4, and RXFP3 genes that are involved in the elevation of cytosolic Ca 2+ concentration (which ultimately leads to apoptosis) and activation of lymphocytes and leukocytes.
Other correlating drugs for compound 13 are alkylating agents and PARP1 inhibitor olaparib, which damage and prevent the reparation of DNA, respectively. Curiously, gene enrichment analysis suggests that 13 regulates adenylate cyclase activity by the G-protein signaling pathway of calcitonin receptor, which probably explains the similarity of cytotoxic specificity between 13 and tyrosine kinase inhibitors vismodegib and LDK-378.
For compound 27, triapine and camptothecin derivatives were found to share a similar activity profile. Triapine (3-aminopyridine-2-carboxaldehyde thiosemicarbazone) is a small molecule inhibitor of ribonucleotide reductase, reducing the availability of deoxyribonucleotides required for DNA synthesis and currently being investigated in clinical trials [58]. Camptothecin derivative NSC 681634 targets topoisomerase I to induce DNA strand breaks [59]. Gene enrichment analysis failed to reveal any additional insights.
The most promising compound 29 appears to act via a wide range of genes, as enrichment analysis shows. Notably, there is a correlation with genes involved in histone deacetylase and chromatin remodeling complexes (TOP2B, RBBP4, HDAC1, HDAC6, PKN1), which corresponds to activity pattern similarity of 29 and HDAC inhibitors. Interestingly, HDAC inhibitor trichostatin A induces G 0 /G 1 phase arrest in hepatoma cells HepG2 and Huh-7, and 29 exhibits a similar action in A549 cells [60]. Another important aspect of compound 29's action is the involvement of genes regulating mitochondrial (NDUFA5, NDUFB11, ATP5A1, NDUFAB1) and ribosomal functions (RPS25, RPS9, RPSA, RPS10, RPS12), which are essential for the survival of cancer cells. These are in agreement with the experimental data on mitochondrial dysfunction, caused by compound 29. This might also serve as an explanation of the cell cycle arrest at the S phase which is energy consuming and critically depends on ribosomal protein synthesis. Additionally, genes participating in pyruvate dehydrogenase activity are also affected (PDHA1, PDHB, DLAT), which suggests that compound 29 might also inhibit anaerobic glycolysis. This multifaceted nature might explain the high cytotoxic activity of 29.
The activity profile of compound 32 significantly correlates with several tyrosine kinase inhibitors and alkylating agents LDK-378 and estramustine. Gene enrichment consistently shows that the cytotoxic activity of 32 is mediated via the interleukin-2 signaling (IL2RA, IL2RB), specifically MAPK/ERK pathway (IL26, NOD1, NOD2, TNF genes). Recombinant IL-2 is approved in the USA and several European countries for the treatment of malignant melanoma and renal cancer. Furthermore, multiple genes involved in cytokine secretion and lymphocyte-mediated immune reactions are enriched, reflecting the greater activity of 32 towards lymphoid cancer cell lines.

Computational ADMET Profiling of Compound 29
Preliminary assessment of the pharmacokinetic and toxicological properties of lead compound 29 was performed with several predicting services utilizing different models. The choice of services was based on applicability criterion since not all of them are trained on compounds of triterpene nature. Consensus results are shown in Table 5. Computational results show that compound 29 is highly lipophilic and, consequently, predicted to have low water solubility. However, no solubility issues were noted in biological experiments. Due to high logP, the compound is likely to have high intestinal absorption and good cellular permeability due to P-glycoprotein inhibition.
The predicted volume of distribution (VDss) is low, which is typical for lipophilic compounds bound to tissue and cellular components (e.g., protein, lipid) and might favor antitumor activity. Compound 29 is predicted to be degraded by CYP3A4, which is known to oxidize steroids and other large molecules. There is also a possibility of CYP3A4 inhibition, which could be circumvented by decreasing the lipophilicity, adding steric hindrance to the heterocycle para to the nitrogen, or adding an electronic substitution (e.g., halogen) that reduces the pKa of the nitrogen [61]. Furthermore, bearing planar amide moiety, compound 29 could be a substrate for CYP1A2. Low predicted clearance values in conjunction with a short half-life (T 1/2 ) might reflect the possibility that the compound is prone to rapid metabolic degradation. Thus, structural modifications might be required to improve its metabolic stability and achieve a more favorable pharmacokinetic profile.
Compound 29 is predicted to be non-mutagenic but might be a hERG inhibitor, which again might be addressed by the introduction of polar substituents [62]. According to predicted moderate oral acute toxicity, the compound can be attributed to Category 3 according to GHS classification.
Overall, computational ADMET profiling renders compound 29 as suitable for future in vivo testing and indicates that possible drawbacks, such as rapid metabolic degradation and cardiotoxicity, might be addressed by a logP decrease (Table 6).

Synthesis of Compounds 7 and 27
Furfural (0.11 mL, 1.3 mmol) and 40% KOH in ethanol (2.5 mL) were added to a solution of compound 3 or compound 4 (0.45 g, 1 mmol) in ethanol (5 mL) under stirring and cooling (from −5 to 10 • C). The mixture was stirred for 24 h at room temperature, pH was adjusted to neutral values with 5% HCl solution, and the mixture was poured into cold water (50 mL). The residue was filtered, washed with water, and dried, then purified by column chromatography on Al 2 O 3 using petroleum ether-CHCl 3 (1:1 to 1:3) as eluent.  To a solution of compound 5-7, 15, 2,3-indolo-olenolic acid [64] or 25-27 (1 mmol) in CH 2 Cl 2 (20 mL) (COCl) 2 (3 mmol; 0.26 mL) was added and stirred at room temperature for 2 h. The mixture was concentrated to dryness under reduced pressure and the resulting acid chloride was dissolved in CH 2 Cl 2 (10 mL), 3 drops of Et 3 N and 1.5 mmol of the corresponding amine were added: (a) N-methylpiperazine (for synthesis of compounds  of compounds 14, 18, and 23). After completion of the reactions (TLC control) the organic layers were treated with 5% HCl (3 × 50 mL) until neutral pH, dried over CaCl 2 , and evaporated under reduced pressure. The residue was purified by column chromatography on Al 2 O 3 using petroleum ether-CHCl 3 (10:1 to 0:10) as eluent.

NCI-60 Cytotoxicity Drug Screen
The NCI-60 cell line panel is organized into nine subpanels with diverse histology representing leukemia, melanoma, non-small-cell lung, colon, kidney, ovarian, breast, prostate, and central nervous system cancers. Details of the NCI-60 cell line screening protocols and reporting procedures have been described previously [69][70][71][72]. Briefly, test compounds were assayed at a single-dose concentration (10 µM) in the full NCI-60 cancer cell line panel. Upon initial indication of activity in the single-dose experiment, compounds were subsequently tested at five doses starting at 100 µM and decreasing by logarithmic dilution to a final concentration of 0.01 µM. Cell viability after 48 h of incubation was visualized using sulforhodamine B. Through the use of a time zero cell control, the total cell growth can be determined for each cell line, thus allowing calculations of GI 50 , TGI, and LC 50 . pyridinylidene]-ursonic N-methyl-piperazinyl amide 29, which exhibited a broad-spectrum inhibition activity with GI 50 < 1 µM against 33 tumor cell lines and < 2 µM against all 60 cell lines. The data for cell cycle analysis indicates that compounds 13 and 29 could exhibit both cytostatic and cytotoxic activity, depending on the cell line evaluated. Our results suggest that the antiproliferative effect of compound 29 is mediated through ROStriggered apoptosis which involves the mitochondrial membrane potential depolarization and caspase activation.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data presented in this study are available on request from the corresponding authors.