Synthesis and Antitumor Potency of 2E,21E-bis-(2-Pyridinylidene)-hollongdione in NCI-60 Panel and Zebrafish Model

TL;DR

A new compound was created and shown to effectively kill many cancer cell types, including melanoma, and work well in a zebrafish cancer model.

Contribution

An efficient site-selective synthesis of potent anticancer dammarane-type chalcones and a promising bis-2-pyridylidene derivative with submicromolar potency.

Findings

Bis-2-pyridylidene derivative 3 showed antitumor activity in 58 of 59 NCI-60 cancer cell lines.

Compound 3 exhibited high selectivity for melanoma, renal, and prostate cancers with an SI of up to 18.82.

In zebrafish xenograft models, compound 3 inhibited tumor growth by 72% without significant toxicity.

Abstract

Michael acceptors, such as chalcones and benzylidenes, are privileged scaffolds for the development of anticancer agents. Taking this into account, we developed a selective Claisen–Schmidt condensation of the dammarane-type triterpenoid hollongdione with pyridine-2-carbaldehyde, enabling controlled synthesis of mono- and bis-substituted triterpenes depending on the reaction conditions. The reaction demonstrated high temperature-dependent regioselectivity, providing C2-mono- 2 or 2,21-bis-substituted 3 triterpenes with yields up to 96% and 95%, respectively. The structures of the newly synthesized triterpene chalcones were elucidated by 1D and 2D NMR spectroscopy and unambiguously confirmed by a single-crystal X-ray diffraction, which established the E configuration of the exocyclic double bond. In biological studies, the bis-2-pyridylidene derivative 3 exhibited a pronounced and broad-spectrum antitumor activity in the NCI-60 panel, inducing cell death in 58 of 59 cancer cell lines. High selectivity toward melanoma, renal, and prostate cancer cell lines was observed, with selectivity indices (SI) of up to 18.82 for melanoma LOX IMVI. In MTT assays, compound 3 displayed a submicromolar cytotoxicity, particularly against the KRAS-mutant PANC-1 cell line (IC50 = 0.22 µM). Anticancer activity was further confirmed in a zebrafish (Danio rerio) xenograft model of human HCT116 colon cancer, where tumor growth inhibition reached 72% without pronounced embryotoxicity (LC50 = 1.4 µM). We have developed an efficient approach for the site-selective modification of hollongdione, providing access to potent anticancer dammarane-type chalcones. The bis-2-pyridylidene derivative 3 emerged as a promising lead compound, demonstrating submicromolar potency, high selectivity towards melanoma, and significant in vivo efficacy in a zebrafish xenograft model.

Figures

Full text

1. Introduction

Michael acceptors, defined as α,β-unsaturated electrophiles capable of undergoing 1,4-conjugate addition with nucleophiles, play a central role in both organic synthesis and medicinal chemistry due to their distinctive chemical reactivity and biological properties [1]. A reactive oxygen species (ROS)-activated prodrug strategy utilizing Michael acceptors has been reported to achieve selective drug release in tumor cells, demonstrating that α,β-unsaturated electrophilic scaffolds can serve as tunable covalent warheads, enabling precise control over reactivity and biological activity in anticancer therapy [2].

Among these, 2-cyano-1,2-enone derivatives and related cyanoenone analogues of CDDO (2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oic acid) have attracted significant attention. The incorporation of a strongly electron-withdrawing cyano-group conjugated with enone functionalities enhances electrophilicity, promoting efficient Michael addition with cysteine residues in the Keap1 protein and leading to potent anti-inflammatory and cytoprotective activities [3,4,5]. The central role of cyanoenone Michael acceptors in modulating redox signaling and anti-inflammatory pathways through targeting cysteine residues in KEAP1 has been clearly demonstrated [6].

Emerging studies have also highlighted the importance of quinone-based electrophilic scaffolds as Michael acceptors with biological activity. In particular, investigations of 5,8-quinolinedione derivatives and their nucleoside conjugates have shown that structural modifications can fine-tune both reactivity and bioactivity, including interactions with thiol-containing enzymes such as DT-diaphorase [7,8]. These findings emphasize the potential of synthetic electrophilic scaffolds as tunable Michael acceptors bridging organic synthesis and medicinal chemistry [9,10].

Concurrently, chalcones (α,β-unsaturated ketones or 1,3-diarylprop-1-en-ones), commonly found in plants, represent a major class of naturally occurring Michael acceptors with a broad pharmacological relevance [11]. Their conjugated enone core enables covalent interactions with biological nucleophiles, underpinning antioxidant, anti-inflammatory, antimicrobial, and anticancer activities. In triterpene chemistry, the introduction of benzylidene fragments at the C2 position has led to compounds with enhanced anti-inflammatory [12], antitumor [13,14,15], and antidiabetic properties [16,17].

Among benzylidene-modified triterpenoids, derivatives of β-boswellic acid bearing 2-pyridinylidene substituents have demonstrated a significant cytotoxicity across cancer cell panels, mediated by apoptosis induction and ROS generation [18]. Chalcone-like derivatives of 20-oxo-lupanes (messagenin analogues) also exhibited potent cytotoxicity in the NCI-60 panel, with E-configured pyridinylmethylidene derivatives showing submicromolar GI_50_ values [19]. Molecular modeling studies further suggest that triterpene-heterocyclic derivatives can effectively target anti-apoptotic Bcl-2 family proteins, providing a mechanistic basis for their cytotoxicity [20].

For the present study, the dammarane-type triterpenoid hollongdione (4,4,8,14-tetramethyl-18-norpregnane-3,20-dione) was selected as the starting scaffold. Hollongdione is a naturally occurring hexanor-triterpenoid with structural similarity to pregnane steroids and has been isolated from several plant sources, where it displays antiviral and moderate cytotoxic activities [21,22,23,24,25,26,27]. Notably, a previously reported bis-3-pyridinylmethylidene derivative of hollongdione exhibited strong antiproliferative and anti-angiogenic effects [28]. These findings prompted us to explore selective synthetic strategies and comprehensive biological evaluation of regioisomeric 2-pyridinylmethylidene hollongdione derivatives.

2. Results

2.1. Chemistry

In this work, hollongdione was obtained via a one-pot synthetic approach from dipterocarpol, the main metabolite of D. alatus resin, and used as the starting scaffold [29]. Hollongdione 1 was reacted with 1 or 2 mmol of pyridine-2-carbaldehyde in ethanol via a classical Claisen–Schmidt aldol condensation in the presence of a strong base (40% KOH). Depending on the reaction temperature, either the mono- or bis-substituted derivatives 2 and 3 were obtained (Scheme 1).

In the case of the interaction of hollongdione, containing two oxo-groups at positions C2 and C21 of rings A and D, a method for the regioselective synthesis of mono- and/or bis-condensed products was developed. Data on the influence of solvents and temperatures on the distribution of by-products and yields of the target 2-[2-pyridinyl]-methylidenohollongdione 2 are summarized in Table 1. The table includes data for four experiments using different solvents (methanol, ethanol) at two temperature conditions (room temperature (r.t.) and 0 °C) (in the Experimental Section, the procedure for the optimized conditions is provided). When methanol was used at room temperature, the yield of product 2 was 65%, while the yield of the bis-product 3 was 27%. At 0 °C, the yield of product 2 increased to 70%, and the formation of bis-product 3 was observed in a yield of 20%. In ethanol, the yield of compound 2 at room temperature was 75%, and the bis-product 3 was obtained in a yield of 18%. Notably, the highest regioselectivity (96%) for the target product 2 was achieved with ethanol at 0 °C, providing the highest yield and completely suppressing the formation of the bis-product 3.

The choice of solvent has a significant impact on the efficiency of the Claisen–Schmidt condensation. As one can see, ethanol provided the best results. Methanol, with its relatively high protic acidity (pKa ~15.5), can hinder enolate formation (the key intermediate of the reaction) and may promote side reactions such as saponification under strong basic conditions (NaOH, KOH). Ethanol offers an optimal balance: it dissolves alkali bases well, facilitates enolate generation, and minimizes side reactions. These features make ethanol the most suitable solvent for the Claisen–Schmidt condensation, as supported by both literature and experimental data [30].

When the reaction mixture was cooled to 0 °С, the only mono-substitution product at the C2 position of the A cycle 2 was selectively obtained in 89% yield after purification by column chromatography. When the reaction was carried out at room temperature for an hour, a precipitate of C2,21-bis substituted product 3 was observed in a yield of 95% after crystallization from methanol.

To evaluate the effect of the 2-pyridinyl-substituent at the C21 position on the antiproliferative activity, we have also synthesized the C-21 mono-substituted derivative 5. At first, 3β-hydroxyhollongdione 4 was obtained by the action of NABH_4_ on hollongdione 1 in methanol at room temperature for 30 min. The desired 3β-hydroxy-17-[3-(pyridin-2-yl)-prop-2-en-1-one]-hollongdione 5 was synthesized from 4 via the aldol condensation with 2-pyridinecarboxaldehyde in a yield of 94%.

The structure of compounds 2, 3, and 5 was established using two-dimensional correlation techniques {^1^H, ^1^H} COSY, {^1^H, ^1^H} NOESY, {^1^H, ^13^C} HSQC, and {^1^H, ^13^C} HMBC (Figure 1 and Supplementary Information Figures S1–S30). DEPT-editing of ^13^C NMR signals for compound 3 made it possible to assign the signals to 9 quaternary, 15 methine, 7 methylene, and 5 methyl groups. The presence of the (pyridin-2-yl)methylene substituent at position C2 for compounds 2 and 3 is confirmed by HMBC cross-peaks between the methylene bridge proton H-31 and the quaternary carbon C3, as well as the methylene carbon C2 of ring A. The basic chalcone substituent contains a 1,3-diphenyl-2-propen-1-one structure, where two aromatic rings are connected by a three-carbon unit that has an α,β-unsaturated carbonyl system. This fragment can be found as either a trans (E) or a cis (Z) isomer, and several studies verified that the E isomer is more stable than the Z isomer [31]. Based on literature data, E-configuration of the methylene double bond in the different classes of compounds, as well as triterpenoids and steroids, was also more preferred [19,32,33,34].

In the ^1^H NMR spectrum of compound 3, the H-31 proton signal is observed as a doublet-doublet with long-range spin-spin interaction constants ^4^JHH = 3.6 and 1.8 Hz with methylene protons at position C1. Analysis of the splitting of cross-peaks in the phase-sensitive variant COSY-DQF allowed us to establish that the long-range interaction with a value of ^4^JHH = 3.6 Hz corresponds to the spin-spin interaction with H_α_-1 (δ_H_ 2.48), and the splitting value with the β-proton at δ_H_ 3.57 is ^4^JHH = 1.6 Hz. A similar pattern of spin-spin interaction of the methylidene proton H-31 with α and β methylene protons at C1 is observed for the acetyl derivative 2, in which ^4^JHH is 3.1 and 1.8 Hz, respectively. For methylene protons at the C1 position of 3 and 2 compounds, the assignment to the α- or β-orientation was established on the basis of cross-peaks in the NOESY spectra: H_α_-1/H-9, H_β_-1/H-19, and H_β_-1/H_eq_-11.

For the compound 3, the observed HMBC cross-peaks of H-21 and H-1′ double bond protons with pyridine atoms and carbonyl carbon C20 (δ_C_ 203.83), as well as HMBC cross-peaks of D cycle protons with carbonyl C20, indicate a 3-(pyridine-2-yl)-prop-2-enoyl substituent in position C17. For the double bond in the 2-(pyridine-2-yl)-prop-2-enoyl substituent, the E-configuration was established based on ^3^JHH = 15.8 Hz, the value of which indicates the trans-arrangement of protons in the double bond. In the NOESY spectrum, cross-peaks of the H-21 double bond proton with protons at positions C16 (δ_H_ 2.03 and 1.88) and C17 (δ_H_ 3.02) are observed. For the compound 2, the acetyl grouping at position C17 is confirmed by the characteristic values of δ_C_ 30.02 and 212.12, the methyl signal at δ_H_ 2.15, and the corresponding cross-peaks in the HMBC spectrum (Figure 1).

The formation of a 2-pyridine-prop-2-en-1-one fragment in compound 5 at position C17 was established based on the presence of characteristic quaternary carbon signals at δ_C_ 203.85 and double bond’s methine signals at δ_C_ 129.86 and 140.68. The signal of the oxo-group at δ_C_ 203.85 correlated with protons H-17 (δ_H_ 2.95) and H_α_-16 (δ_H_ 1.83) from the side of cycle D, while protons H-21 (δ_H_ 7.18) and H-1′ (δ_H_ 7.51) correlated from the side of the double bond. The position of the aromatic substitution was established based on the HMBC correlations of the H-21 proton with quaternary aromatic carbon C2′ at δ_C_ 153.41 and of the H-3′ (δ_H_ 7.45) aromatic proton with the double bond’s methine signals at δ_C_ 140.68. This data correlates with the reported earlier [19,32,33,34]. The NMR spectral characteristics for compounds 2, 3, and 5 are presented in Table 2

X-ray crystallography remains an indispensable tool in pharmaceutical drug discovery, enabling the definitive assignment of absolute configurations and the characterization of non-covalent interactions, such as hydrogen bonding and π-π stacking, which dictate molecular stability and bioactivity [35]. In the present study, single crystals of the 2,21-bis-[2-pyridinyl]-methylidenohollongdione 3, suitable for X-ray diffraction, were obtained via the slow evaporation of ethanol solution. C_24_H_38_O_2_ are monoclinic, space group P2_1_: a = 11.1425(8) Å, b = 10.5767(7) Å, c = 13.5129(10), β = 110.106(2)°, V = 1495.46(18) Å^3^, Z = 2, dcalc = 1.192 g·cm^−3^. 16,127 reflections were collected at SMART APEX II CCD diffractometer (λ(Mo-Kα) = 0.073 Å, graphite monochromator, ω-scans, 2θ < 29°) at 120 K. The structure was solved by the direct methods and refined by the full-matrix least-squares procedure in an anisotropic approximation. 6540 independent reflections [R(int) = 0.0471] were used in the refinement procedure that was converged to wR2 = 0.11110 calculated on F^2^hkl (GOF = 1.006, R1 = 0.0494 calculated on Fhkl using 3436 reflections with I > 2σ(I)). (Figure 2). Obtained crystallographic data and refinement parameters are collected in Table S1 (see Supplementary Information). CCDC 1984684 contains the supplementary crystallographic data for this paper. This data was correlated with the previously study [19], where benzylidene (chalcone) derivatives of messagenin and platanic acid were obtained predominantly as the thermodynamically more stable E isomers.

Thus, the solid-state structure provides direct experimental evidence for the preferential formation of the E isomer under the applied synthetic conditions. In this context, it is important to consider the factors governing the stereochemical outcome of benzylidene-type condensations of triterpenes.

Reaction conditions, including temperature, solvent, and catalyst, have a significant influence on product distribution and E/Z isomerism in benzylidene condensations of triterpenes. Under mild conditions (e.g., room temperature and weakly basic media), the kinetically controlled Z isomer often predominates, whereas elevated temperatures or acidic environments favor formation of the thermodynamically more stable E isomer due to steric and electronic factors associated with the transition state [36]. Substituents on the aromatic aldehyde also play a pivotal role: electron-donating groups (EDGs; e.g., –OMe, –Me) tend to increase electron density and can stabilize intermediates leading to Z-enriched products, whereas electron-withdrawing groups (EWGs; e.g., –NO_2_, –CF_3_, halogens) render the carbonyl moiety more electrophilic and shift selectivity toward the E isomer [37]. Steric effects, particularly those arising from ortho substituents, can further bias the reaction toward Z configurations by increasing steric hindrance in transition states, leading to the E isomer [38]. Overall, the interplay of electronic and steric effects of substituents, together with reaction parameters, allows fine control over the E/Z ratio, which is crucial for optimizing the physicochemical and biological properties of triterpenoid derivatives.

The biological activity of steroidal arylidene derivatives is highly dependent on the geometry of the C=C double bond. Numerous studies have demonstrated that E- and Z-isomers can exhibit markedly different pharmacological profiles, including antiproliferative and enzyme-inhibitory activities. For example, certain 16(E)- and 21(E)-arylidene steroids display enhanced anticancer activity as well as stronger aromatase and 17β-HSD1 inhibition compared with their Z-counterparts, highlighting the critical role of E/Z isomerism in molecular recognition and biological efficacy [39]. Consistently, a messagenin derivative bearing an E-configured arylidene fragment exhibited the highest cytotoxic activity in screening against the NCI-60 human cancer cell line panel, with GI_50_ values ranging from 0.304 to 0.804 μM, indicating the most potent antiproliferative effect within the series [19].

2.2. Biology

2.2.1. NCI-60 Anticancer Drug Screening

Mono-2 and bis-3 derivatives of hollongdione were selected by National Cancer Institute (NCI, Bethesda, MD, USA) Developmental Therapeutic Program (DTP) and tested at one dose assay (10^−5^ M) toward a panel of approximately sixty cancer cell lines representing different cancer types: leukemia, melanoma, lung, colon, CNS, ovarian, renal, prostate, and breast cancers (see Supplementary Material, Figures S34–S36). The results for each compound are reported as the percent growth (GP %) of treated cells compared to untreated control cells (negative numbers indicate cell kill) (Table 2). In accordance with the criterion adopted by the National Cancer Institute (substances are considered active if they inhibit cell growth to 32% of the control or cause cell death). The compound 2 showed no activity, while compound 3 showed the greatest antiproliferative activity against 58 of 59 cell lines, and in the case of 40 cell lines, cancer cell death was observed in the range from −1.54 to −99.62% (Table 3), сomparable to the previously described 2,21-bis-[3-pyridinyl]-methylidenohollongdione, which exhibited a cytostatic effect on 20 and a cytotoxic effect on 54 out of 59 tested cell lines in the NCI panel [28].

The range of percent growth emphasizes the lowest and the highest percent growth found among the different cancer cell lines. As a result of the one-dose assay, compound 3 (2,21-bis-[2-N-pyridinyl]-methylidenohollongdione) was selected for an in-depth five-dose assay against a panel of 59 tumor cell lines, by using 10-fold dilutions of five concentrations (100 µM, 10 µM, 1 µM, 0.1 µM, and 0.01 µM) (see Supplementary Information, Figures S34–S36) [40,41,42,43]. Results showed significant antitumor activity against all cell lines, with optimum cytotoxicity against the MCF7 breast cancer cell line (Table 4).

A raw comparison of the activity of compound 3 with respect to the activity reported for the standard drugs doxorubicin and 5-fluorouracil, used by NCI as control [44], showed that the studied compound were more active against cell lines of leukemia CCRF-CEM, HL(60)-TB, MOLT-4; against cell lines of NSC lung cancer HOP-92, NCI-H226, NCI-H522; against cell lines CNS cancer SNB-19; against cell lines of melanoma SK-MEL-2, UACC-257; against cell lines of ovarian cancer OVCAR-3, OVCAR-4 and OVCAR-5; against cell lines of renal cancer RXF-393; against cell lines of prostate cancer PC-3; against cell lines of breast cancer MDA-MB-31/ATCC, HS-578T, BT-549, T-47D than for 5-fluorouracil.

The selectivity index (SI) calculated by dividing the full panel MG_MID_60_ (μM) of the compound 3 by their individual subpanel MG_MID of the cell line (μM) was considered as a measure of compound selectivity (Table 4). Ratios between 3 and 6 mean moderate selectivity, ratios greater than 6 indicate high selectivity toward the corresponding cell line, while compounds not meeting either of these criteria are rated nonselective [45]. In this context, based on the selectivity ratio SI ranging from 0.41 to 8.57, compound 3 was generally found to exhibit high selectivity in inhibiting the growth of renal and prostate cancer cell lines with SI 8.54 and 8.57 by TGI level (Table 5). Also, this compound demonstrated a certain selectivity profile toward some individual cell lines at the LC_50_ level. Thus, compound 3 was more selective at the LC_50_ levels towards colon, CNS, renal, and melanoma (selectivity indexes 6.07–18.82). Also, compound 3 was high selectivity at IС_50_ level was observed for colon cancer cell lines: HCC-2998 (SI = 6.07), HCT-116 (SI = 11); CNS cancer cell line: U251 (SI = 8.56); melanoma cell lines: LOX IMVI (SI = 18.82), MDA-MB-435-3M (SI = 11.62), SK-MEL-5 (SI = 7.95), UCC-62 (SI = 7.35); renal cancer cell lines: 786-0 (SI = 7.56), A498 (SI = 7.09), RXF (SI = 9.98), U-31 (SI = 9.89). The greatest selectivity was observed for melanoma cells LOX I MVI with SI = 18.82 (Table 6).

A similar bis-substituted hollongdione derivative, 2,21-bis-[3-pyridinyl]-methylidenohollongdione 3, demonstrated a comparable level of activity against the colon cancer cell line HCT-116 (SI = 6.66); CNS cancer cell lines: U-251 (SI = 7.23); melanoma cell lines: LOXIMVI (SI = 6.51), SK-MEL-5 (SI = 7.38); renal cancer cell lines: 786–0 (SI = 6.55), DU (SI = 6.61) ovarian cancer cell line OVCAR-3 (SI = 6.19). However, the antiproliferative effect of 2,21-bis-[3-pyridinyl]-methylidenohollongdione at the IC_50_ level for melanoma cell lines, LOXIMVI (SI = 6.51) [28], was less pronounced than that of compound 3, which contains a 2-pyridinoylmethylidene substituent in its structure.

Thus, compound 3 showed strong and broad-spectrum anticancer activity across the NCI 59-cell-line panel, surpassing both the monosubstituted analogue and the previously reported bis-pyridinyl derivative. It also demonstrated high selectivity, with the strongest effect observed against melanoma LOX IMVI cells. These findings identify compound 3 as the most potent and selective hollongdione derivative to date and a promising lead for further anticancer development.

2.2.2. Cytotoxicity Assay of Compounds 3 and 5 on the Selected Cancer Cells

The effects of compounds 3 and 5 on the growth and proliferation of human cancer cells were examined using the 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT) assay. The IC_50_ values are presented in Table 7. To further characterize its cytotoxic profile, compound 3 was evaluated in a focused panel of human cancer cell lines, and the corresponding IC_50_ values are summarized in Table 6. Consistent with the NCI-60 data, Compound 3 exhibited potent cytotoxicity in several cell lines, with the most pronounced effect observed in PANC-1 cells (IC_50_ = 0.22 ± 0.03 µM), indicating a remarkable sensitivity of this pancreatic cancer line. Strong activity was also detected in DU145 (0.65 ± 0.03 µM), HCT116 (0.99 ± 0.01 µM), and A549 (0.58 ± 0.04 µM) cells, while moderate activity was seen in SKBR3 (3.29 ± 0.07 µM), SK-MEL-28 (3.12 ± 0.09 µM), and SK-OV-3 (6.39 ± 0.07 µM). In contrast, compound 3 displayed weak or negligible cytotoxicity against MCF-7 (>100 µM), MIA Paca-2 (47.99 ± 0.10 µM), HT-29 (11.75 ± 0.04 µM), and U-251 MG (18.21 ± 0.07 µM) cells. Notably, compound 3 exhibited limited activity against MCF-7 breast cancer cells (IC_50_ > 100 µM), indicating a degree of cell line–dependent selectivity.

In contrast, the mono-pyridinylidene derivative 5 showed a distinct cytotoxic profile. While it retained high potency against HCT116 cells (IC_50_ = 0.23 ± 0.02 µM) and displayed moderate activity against DU145 (0.44 ± 0.02 µM) and A549 (1.61 ± 0.05 µM) cell lines, its efficacy against PANC-1 cells was markedly reduced (IC_50_ = 4.94 ± 0.94 µM) compared to compound 3. Conversely, compound 5 demonstrated enhanced activity toward MCF-7 breast cancer cells (IC_50_ = 7.24 ± 0.22 µM), where compound 3 was inactive.

Compound 3 has a high selectivity vs. melanoma LOX IMVI (SI = 18.82), but the MTT assay is weak versus other melanoma lines, e.g., SKMEL-28. LOX IMVI and SK-MEL-28 are both melanoma lines, but have well-documented genetic and phenotypic differences. LOX IMVI is highly metastatic, carries a BRAF V600E mutation, and may have a higher basal ROS load or differential expression of pro-apoptotic proteins. SK-MEL-28, while also being BRAF V600E mutant, has different transcriptional profiles and metabolic states. The divergent sensitivity suggests compound 3’s mechanism exploits a specific vulnerability present in LOX IMVI, but not in SK-MEL-28. This data is valuable as it argues against a general class effect and points toward a biomarker-driven activity.

The same differences show comparison between PANC-1 and MIA PaCa-2 in the MTT assay. The stark difference between PANC-1 and MIA PaCa-2 is a key finding. Both are KRAS mutants, but MIA PaCa-2 has a homozygous G12C mutation, while PANC-1 is heterozygous for G12D; different mutations can lead to distinct signaling outputs. Public databases (CCLE) show significant differences in pathways like oxidative phosphorylation and EMT.

Compound 3 also showed poor activity vs. MCF7 in MTT (IC_50_ > 100µM), but NCI-60 says it is optimally cytotoxic. The primary methodological difference (48 h pre-diluted compound vs. 72 h direct addition in our MTT) may affect outcomes for slow-acting compounds [46].

Overall, the comparison of the cytotoxic profiles of compounds 3 and 5 indicates that the presence of two pyridinylidene moieties in compound 3 significantly enhances antiproliferative potency, particularly against pancreatic, lung, and prostate cancer cell lines. In contrast, mono-substitution at C21 in compound 5 results in a more selective activity pattern, with pronounced efficacy against colorectal and breast cancer cells. These findings underscore the critical role of the substitution pattern in modulating the anticancer activity of hollongdione derivatives.

Based on a comprehensive multi-parameter evaluation that extended beyond in vitro IC_50_ values, compound 3 was chosen for zebrafish xenograft studies. Although it was not uniformly the most potent compound in 2D cell culture, it demonstrated compatibility with fluorescence-based tumor tracking and a favorable preliminary toxicity profile in wild-type zebrafish embryos, which are critical prerequisites for reliable in vivo dosing and efficacy assessment.

2.2.3. Danio Rerio Embryotoxicity

Survival of Danio rerio embryos (n = 5) was evaluated following 24, 72, and 144-h exposure to compound 3 (Figure 3). Embryos incubated with 1.5 μM compound 3 were no different from control embryos during the first day of the experiment. However, spinal deformities were observed starting at 48 h after the start of incubation. By 144 h, three of the five embryos had died, while the remaining two not only survived but also developed despite the spinal deformities. Kaplan–Meier survival analysis (Figure S37) showed that treatment with compound 3 did not induce significant toxicity, as embryo survival remained comparable to that of the control group throughout the observation period.

These results indicate that compound 3 exhibits moderate developmental toxicity at concentrations near the LC_50_, with surviving embryos showing partial resilience to sublethal morphological defects.

This pattern—specific teratogenicity (spinal deformity) at concentrations that do not significantly reduce overall survival—is characteristic of a selective developmental toxicity rather than general systemic lethality. It suggests that compound 3 interferes with specific pathways crucial for somitogenesis or axial development (e.g., ROS-mediated disruption of Wnt/β-catenin or Notch signaling in precursors) without causing catastrophic organ failure at the same dose. The preserved survival indicates a therapeutic window exists, but the morphologic defects highlight a potential toxicity liability that would require careful monitoring in development, particularly for non-cancer indications [47].

A compound with strong cytostatic effects across many lines could have a low (favorable) MG-MID, inflating the SI against a subpanel, even without true cytotoxic selectivity. To address this, we complement the SI analysis by examining the full three-parameter response data (GI_50_, TGI, LC_50_). For compound 3, the high negative Growth Inhibition percentages in sensitive lines correspond primarily to LC_50_ values (full cell kill), confirming true cytotoxic selectivity rather than a cytostatic artifact (false selectivity signal).

2.2.4. Antitumor Effect of Compound 3 on Danio rerio Embryos

Analysis of tumor progression in Danio rerio embryos demonstrated a pronounced inhibitory effect of compound 3. The mean tumor area 24 h after cell implantation was 0.031 ± 0.004 mm^2^ in the control group and 0.026 ± 0.009 mm^2^ in the experimental group, with no statistically significant differences. Following 48 h incubation with compound 3, a significant inhibition of tumor growth was observed, with a tumor growth inhibition (TGI) rate 72%. (Table 8). The mean tumor area was 0.012 ± 0.007 mm^2^ (p = 0.025) in the treatment group and 0.043 ± 0.001 mm^2^ in the control group. Representative photographs of the control and treatment groups are shown in Figure 4. As a result, the tumor growth index decreased to 0.50 ± 0.06 (p = 0.05), indicating a 50% suppression of tumor expansion compared to the untreated controls.

HCT116 is not the most sensitive cell line in vitro, but factors like hypoxia, nutrient starvation, or stromal interactions in the zebrafish model could sensitize HCT116 cells to the compound’s mechanism (e.g., enhanced oxidative stress) [48].

2.3. In Silico Assay

2.3.1. In Silico ADMET Study: Comparison of Physicochemical and Physicochemical Profile

To assess the pharmaceutical relevance of the studied molecules, a comparative analysis was conducted for compounds 2, 3, and 5 across key drug development parameters [49], including physicochemical descriptors, lipophilicity, solubility, pharmacokinetic predictions, and druglikeness criteria, using SWISS ADME (https://www.swissadme.ch/, accessed on 15 December 2025). The calculated data are presented in Table 9 and Figure 5.

The comparative evaluation of the three molecules 2, 3, and 5 highlights notable differences in their physicochemical, pharmacokinetic, and drug-likeness profiles. While all three compounds fall into the “poorly soluble” class based on multiple solubility prediction models (ESOL, Ali, SILICOS-IT), 3 displays the lowest solubility, consistent with its high molecular weight (536.75 g/mol), increased aromaticity, and reduced sp^3^ character (0.56). In contrast, 5 exhibits improved balance, with better solubility estimates and a higher fraction of sp^3^ carbons, indicating more favorable drug-like structural features.

From a pharmacokinetic perspective, 2 and 5 are predicted to have high gastrointestinal (GI) absorption, whereas 3 shows poor predicted absorption, which may limit its oral bioavailability. Notably, 5 stands out for its metabolic stability, as it does not inhibit any major cytochrome P450 isoforms, unlike 2 (CYP2C9 inhibitor) and 3 (CYP1A2 inhibitor), reducing the risk of drug–drug interactions.

In terms of lipophilicity, 3 has the highest consensus LogP (6.52), indicating strong hydrophobicity. While excessive lipophilicity often correlates with poor solubility and potential toxicity, it may enhance membrane permeability and binding affinity to hydrophobic protein targets. Additionally, 3 has the highest molar refractivity (163.20), reflecting increased polarizability and potential for favorable π-π and van der Waals interactions within target binding sites. Its rich aromatic framework (12 aromatic heavy atoms) and moderate flexibility (4 rotatable bonds) suggest high structural complexity, which can be advantageous for selective binding, although such complexity often requires further optimization to improve ADME properties. All three compounds trigger a single Brenk alert due to the presence of a Michael acceptor moiety, but none raise PAINS (pan-assay interference) alerts, supporting their suitability for initial biological evaluation. From a medicinal chemistry perspective, synthetic accessibility scores for all compounds are moderate (5.4–5.8), indicating that further derivatization is feasible with reasonable synthetic effort.

Among the three compounds studied, 5 exhibits the most favorable balance of physicochemical and pharmacokinetic properties, including high predicted oral absorption, metabolic neutrality, and acceptable lipophilicity, making it the most promising candidate for further preclinical development. 2 also shows potential but is limited by CYP2C9 inhibition. Although compound 3 performs poorly in terms of solubility and GI absorption, it offers distinct advantages such as high molecular complexity, aromatic content, and lipophilicity, which could be beneficial for targeting hydrophobic protein pockets. With appropriate structural modifications aimed at improving solubility and absorption, compound 3 may serve as a valuable scaffold for the design of potent and selective drug candidates.

2.3.2. Potential Basis of the Antiproliferative Activity of Compound 3

The observed antiproliferative activity of compound 3 may be attributed to several key structural and physicochemical features that enhance its potential to interact with oncogenic targets. Firstly, 3 displays high lipophilicity (Consensus LogP = 6.52), which likely promotes efficient membrane permeability and accumulation within lipid-rich intracellular compartments such as mitochondria or the endoplasmic reticulum—sites often involved in apoptosis and redox regulation [50]. Secondly, its high degree of aromaticity (12 aromatic heavy atoms) enables π–π stacking interactions with nucleic acids and aromatic residues in proteins, suggesting a potential for DNA intercalation or inhibition of transcription factors and kinases implicated in cancer proliferation [51].

Moreover, 3 contains a Michael acceptor moiety, flagged by a Brenk alert, which may confer electrophilic reactivity towards nucleophilic amino acid residues (e.g., cysteines) in critical cellular proteins. This covalent modification mechanism is common among targeted covalent inhibitors, particularly in cancer drug development [52]. In addition, compound 3 is predicted to inhibit CYP1A2, a key phase-I drug-metabolizing enzyme. This finding is more likely to be relevant for the pharmacokinetic and drug–drug interaction profile of the molecule than for a direct cytotoxic mechanism, although modulation of CYP1A2-dependent metabolic pathways could indirectly influence the formation of pro- or anti-proliferative metabolites in specific contexts [53,54]. A cytochrome P450 isoform known to be overexpressed in certain tumor types and involved in oxidative metabolism and detoxification processes. Disruption of such pathways may sensitize cancer cells to oxidative stress and contribute to cytotoxicity [55].

Finally, the high molar refractivity (163.20) and structural complexity of compound 3 suggest a strong potential for selective engagement with protein binding sites through van der Waals and hydrophobic interactions. Such characteristics are often associated with increased biological selectivity and potency [56]. Taken together, these properties support the hypothesis that the antiproliferative effects of compound 3 may involve multiple intracellular mechanisms, including redox modulation, DNA interaction, covalent protein binding, and enzyme inhibition, warranting further biological investigation and target validation.

2.3.3. PASS Predict for Compound 3

To predict the biological activity of the new compound 3 and guarantee its efficacy, the PASS (v2.0) (Prediction Activity Spectra for Substances) analysis was performed. Table 10 shows the biological properties of compound 3 and its activity score with Pa value based on a cut-off of >0.7 [57]. This threshold was applied to prioritize compounds with significant predicted activity, as commonly used in PASS-based screening studies. The results show that the compound 3 demonstrates strong predicted antineoplastic activity; it may inhibit the growth of malignant cells (e.g., breast, lung, liver, or colon cancer), and has antileukemic potential, including acute and chronic forms. Compound 3 has the ability to modulate transcription factors NF-E2-related factor 2 stimulant (Pa = 0.788) and transcription factor NF-κB stimulant (Pa = 0.781) Activation of Nrf2, a transcription factor regulating antioxidant defense; may be beneficial in oxidative stress, inflammatory and neurodegenerative diseases, cancer prevention, and activation of NF-κB, a key regulator of inflammation and immune response; suggests possible immunoregulatory and anti-inflammatory effects.

The predicted biological activity of compound 3, evaluated using PASS analysis, indicates strong antineoplastic potential, with a high probability of activity (Pa = 0.876) and antileukemic effects (Pa = 0.791). Additionally, compound 3 is predicted to modulate key transcription factors, including NF-E2-related factor 2 (Pa = 0.788) and NF-κB (Pa = 0.781), suggesting possible immunoregulatory and anti-inflammatory properties.

These in silico predictions are in good agreement with the experimental cytotoxicity data obtained in vitro. Compound 3 demonstrated potent cytotoxicity across multiple cancer cell lines in both NCI-60 screening and focused IC_50_ assays, particularly against PANC-1 (IC_50_ = 0.22 µM), DU145, HCT116, and A549 cells, reflecting its strong antineoplastic potential. Moreover, the predicted antileukemic activity aligns with the observed cytotoxicity in leukemia cell lines in the NCI panel. The predicted modulation of transcription factors Nrf2 and NF-κB, although not directly measured in these assays, may underlie the compound’s selective cytotoxicity and potential anti-inflammatory or chemopreventive effects.

Overall, the PASS predictions reinforce the experimental findings, supporting that compound 3 is a promising candidate with broad-spectrum antitumor activity and potential for targeted modulation of transcriptional pathways relevant to cancer and inflammation.

2.3.4. COMPARE Correlations

Standard COMPARE analyses allow the comparison of the selectivity patterns of tested compounds with standard antineoplastic agents of the known mode of action and NCI active synthetic and natural compounds, which are present in publicly available databases [58] (Table 11). Application of this algorithm may provide preliminary information about the mechanism of cell growth inhibition and cell death; however, the resulting correlations should be regarded as hypothesis-generating rather than definitive mechanistic proof. Quantitative evaluation of the obtained results was carried out according to the Chaddock scale using the following interpretation of pair correlation coefficients: insignificant (0.00–0.30), weak (0.30–0.50), moderate (0.50–0.70), high (0.70–0.90), and very high (0.9–1.0) [59].

As follows from Table 11, IC_50_ COMPARE analysis of compound 3 revealed moderate positive correlations with belinostat and oxaliplatin, while the GI_50_ analysis showed a weak positive correlation with belinostat. Taken together, these values indicate only partial phenotypic similarity of the growth-inhibition profiles and do not demonstrate a common primary mechanism of action. The anticarcinogenic effect of the identified drugs, whose mean graphs correlate with that of compound 3, is associated with their ability to induce apoptotic cell death and/or to modulate enzyme systems that enhance the efficacy of anticancer drugs and can ultimately lead to cancer cell death. Belinostat (Beleodaq, PXD101) is a hydroxamic acid-type histone deacetylase (HDAC) inhibitor with antineoplastic activity. Inhibition of histone deacetylases (HDACs) by belinostat indirectly promotes anti-cancer therapeutic effect by provoking acetylated histone accumulation, re-establishing normal gene expressions in cancer cells and stimulating other routes such as the immune response, p27 signaling cascades, caspase 3 activation, nuclear protein poly (ADP-ribose) polymerase-1 (PARP-1) degradation, cyclin A (G2/M phase), cyclin E1 (G1/S phase) [60,61]. The mechanism of action of another antitumor agent, oxiplatine, is mainly related to the action leading to various DNA damages. This drug refers to an alkylating agent, which adds alkyl groups to DNA bases, leading to cross-linking of the DNA strand and, thus, inhibition of cancer cell growth [62]. In this study, the moderate and weak COMPARE correlations with belinostat and oxaliplatin, therefore, suggest that epigenetic modulation and/or interference with DNA integrity could contribute to the overall cellular response to compound 3. A comparison of the structures of the compound 3 and belinostat is shown in Figure 6.

2.4. The Structure Balance of Hollongdione Pyridinylidenes 3 and 5

Compounds containing reactive functional groups such as Michael acceptors fundamentally differ from non-covalent agents by forming stable covalent bonds with target proteins, thereby maximizing binding affinity, prolonging target engagement, and potentially overcoming acquired drug resistance [63]. However, this electrophilic nature also predisposes them to off-target reactivity with cellular nucleophiles (hepatic glutathione conjugation, GSH, albumins, proteasome subunits) [64]. Importantly, Michael acceptor reactivity can be rationally tuned for selective interaction with specific nucleophiles when embedded in conformationally constrained scaffolds [65].

All hollongdione derivatives (2, 3, and 5) trigger a single Brenk alert due to their α,β-unsaturated carbonyl Michael acceptor systems (Table 9). Nevertheless, the observed structure-activity relationship reveals substantial dependence on fragment number and positioning. Mono-substituted analog 2 bearing a single Michael acceptor at C2 shows no activity at 10 μM, while mono-C21 2-pyridinylidene derivative 5 exhibits clear selective cytotoxicity (IC_50_ HCT116 = 0.23 μM, MCF-7 = 7.24 μM). Most pronounced effects were observed for 2,21-bis(2-pyridinylidene) derivative 3, demonstrating a broad anticancer activity (NCI-60 MGMID GI_50_ = 1.16 μM) with exceptional selectivity toward LOX IMVI melanoma (SI LC_50_ = 18.82, Table 6). A primary indicator of non-specific reactivity is uniform, high potency across diverse cell types. In contrast, compound 3 exhibits a highly differential and reproducible pattern of sensitivity and resistance across the NCI-60 cell line panel. For instance, it shows sub-micromolar LC_50_ values in lines like LOX IMVI (melanoma) but negligible activity (>100 µM) in others like MOLT-4 (leukemia). This heterogeneous response profile strongly suggests that cellular context, such as the expression of specific protein targets, detoxification systems (e.g., glutathione levels), or downstream pathway dependencies, modulates its activity, arguing against pure non-specific electrophilic damage. These data can indicate substantial contributions from cell-specific factors, including redox homeostasis, detoxification capacity (GSH/GST, Trx/TrxR systems), and stress-response pathway configuration (Nrf2, NF-κB), which modulate cellular susceptibility to covalent electrophiles [66]. Bis-substitution with pyridine fragments substantially increases the compound’s 3 lipophilicity (logP = 6.52, Table 9), potentially facilitating selective accumulation in lipid-rich melanoma compartments (melanocyte membranes, mitochondria) and enhancing mitochondria-mediated tumor cell damage [28,67].

Literature data on related triterpenoid systems bearing mono- and bis-(pyridinyl)benzylidene substituents confirm activity dependence on electrophilic fragment number and nature. Platanic acid mono-3-pyridinylidene derivative showed moderate activity, while the 2,30-bis-substituted analog induces apoptosis with characteristic morphological changes in tumor cells without toxicity toward normal HaCaT keratinocytes and 1BR3 fibroblasts (GI_50_ > 15 μM) [68]. Among oleananone triterpenoids, C2-[4-pyridinylidene]-oleananone C28-morpholinylamide selectively inhibited HOP-92 (GI_50_ = 0.0347 μM), while C2-[3-pyridinylidene]-N-methylpiperazinylamide ursonic acid exhibited a broad NCI-60 activity (GI_50_ < 2 μM). Cytostatic effects involve cell cycle arrest (S/G0-G1 phases), while cytotoxicity toward normal HEK293 cells proceeds through ROS-induced apoptosis with mitochondrial depolarization and caspase activation [16]. Similarly, 2,21-bis-(3-pyridinylmethylene)hollongdione displays pro-apoptotic and anti-angiogenic activity accompanied by mitochondrial respiration inhibition [28].

Comparing biological profiles, lead compound 2,21-bis-(2-pyridinylidene)hollongdione 3 demonstrates superior melanoma selectivity (LOX IMVI SI = 18.82) versus the bis-3-pyridinylmethylene analog (SI = 6.51) [28]. This enhancement stems from key structural differences in the pyridinylidene substituent, specifically nitrogen positioning. In the 2-pyridinylidene fragment, the ortho-nitrogen exerts stronger electron-withdrawing inductive/resonance effects, enhancing β-carbon electrophilicity within the exocyclic double bond [69]. This enhances its ability to form covalent interactions with biological targets, resulting in potentiated antiproliferative activity.

Although compound 5 exhibits more favorable ADMET characteristics (improved solubility, favorable absorption prediction, Table 9), compound 3 was selected as the primary in vivo candidate due to markedly superior antiproliferative potency (NCI-60 MGMID GI_50_ = 1.16 μM) combined with synthetic accessibility. The combination of electrophilicity, lipophilicity, and structural complexity likely enables both targeted covalent interactions and hydrophobic binding, contributing to potent cellular effects.

Results from NCI-60, targeted MTT, and in vivo zebrafish xenograft experiments demonstrate strong pharmacodynamic profile correlation for compound 3, confirmed by MTT data for PANC-1 (IC_50_ = 0.22 μM) and DU145 (IC_50_ = 0.65 μM) versus weak activity against MCF-7 and MIA PaCa-2 (Table 7). MCF-7 discrepancies (NCI hypersensitivity vs. MTT IC_50_ > 100 μM) reflect analytical endpoint differences (SRB measures total protein content while MTT assesses mitochondria-dependent metabolic activity), incubation time, and free intracellular compound fraction (F_ic_) variability for lipophilic molecules (LogP = 6.52). Nevertheless, relative sensitivity ranking (melanoma/renal/prostate more sensitive than MCF-7/MIA PaCa-2) remains consistent across NCI-60 and MTT, supporting reproducible selective profiling and indicating methodological rather than biological origin of MCF-7 discrepancies. HCT116 xenograft showed TGI = 72% (p < 0.025) despite moderate in vitro sensitivity. In the HCT116 zebrafish xenograft model, compound 3 (1.5 μM) achieved TGI = 72% (p < 0.025), substantially suppressing tumor growth over 48 h. This in vivo effect validates the efficacy of the selective profile identified in vitro.

In summary, the data suggest that the antiproliferative activity of the hollongdione derivatives is likely influenced by the balance between electrophilicity, lipophilicity, and the spatial arrangement of pyridinylidene Michael acceptor groups. Compound 3, with its enhanced electrophilic centers and favorable lipophilicity, shows promising selective cytotoxicity, particularly against melanoma cells. These findings support further investigation of compound 3 as a potential anticancer agent.

3. Materials and Methods

3.1. Chemical Part

The spectra were recorded at the Center for the Collective Use “Chemistry” of the Ufa Institute of Chemistry of the UFRC RAS and RCCU “Agidel” of the UFRC RAS. 1H and 13C NMR spectra were recorded on a “Bruker AM-500” (Bruker BioSpin, Billerica, MA, USA, 500 and 125.5 MHz respectively, δ, ppm, SSCC, Hz) in CDCl_3_, internal standard—tetramethylsilane. Melt-ing points were measured on a micro table “Rapido PHMK05” (Nagema, Dresden, Germany). Elemental analysis was performed on a Euro EA-3000 CHNS analyzer (Eurovector, Milan, Italy). The main standard is acetanilide. Optical rotations were measured on a polarimeter “Perkin-Elmer 241 MC” (Perkin-Elmer GmbH, Rodgau, Germany) in a tube of length 1 dm. TLC analysis was performed on Sorbfil plates (Sorbpolimer, Krasnodar, Russia), using the solvent system chloroform—ethyl acetate, 40:1. Substances were detected by a 10% solution of sulfuric acid with subsequent heating at 100–120 °C for 2–3 min. Hollongdione 1 and 3β-hydroxy-hollongdione 2 were synthesized according to prior work [29,70]. All chemicals were of reagent grade (Sigma-Aldrich, St. Louis, MO, USA). XRD data for compound 3 were obtained on a Bruker Kappa Apex II CCD diffractometer using φ,ω-scans frames with Mo Kα radiation (λ = 0.71073 Å) and a graphite monochromator. Absorption corrections were applied by the empirical multiscan method in the SADABS Version 2008-1 software [71]. The structure was solved by direct methods and refined by the full-matrix least-squares method against all F^2^ in an anisotropic approximation using the SHELXT–2014/5 [72] and SHELXL-2018/3 [73] software suites. The positions of hydrogen atoms were calculated using the riding model. The SQUEEZE procedure (PLATON program [74]) was used due to a highly disordered solvent molecule. The obtained crystal structures were analyzed for short contacts between nonbonded atoms in PLATON and MERCURY software packages [75]. CCDC 966944 contains the supplementary crystallographic data for this paper. Detailed crystallographic data, atomic positional parameters, and bond lengths and angles can be obtained free of charge via http://www.ccdc.cam.ac.uk/cgi-bin/catreq.cgi (accessed on 15 December 2025), or from the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: (+44) 1223 336 033; or e-mail: deposit@ccdc.cam.ac.uk.

3.1.1. 2(E)-[2-Pyridinyl]-methylidenohollongdione (2)

A solution of 1 mmol of compound 1 in EtOH (10 mL) was cooled to 0 °C, then 1 mmol or 2 mmol of 2-pyridinecarbaldehyde and 40% KОН in EtОН (2.5 mL) were added. The reaction was stirred for 30 min at room temperature, poured into water (50 mL), and the precipitate was filtered off and washed with water. The crude product was purified by SiO_2_ column chromatography with petroleum ether/ethyl acetate in a ratio of 1:1 as eluent.

Yield 0.43 g (96%), mp 74–79 °C, [α]D^20^ +56° (c 0.01, CHCl_3_). ^1^H NMR (СDCl_3_, δ ppm, J Hz): 0.83 (s, 3H, H-19); 0.92 (s, 3H, H-30); 1.04 (s, 3H, H-18); 1.13 (s, 3H, H-28); 1.19 (s, 3H, H-29); 1.23 (m, 1H, H_α_-15); 1.34 (m, 1H, H_ax_-11); 1.34 (m, 1H, H_ax_-12); 1.37 (m, 1H, H_eq_-7); 1.49 (m, 1H, H_eq_-6); 1.56 (m, 1H, H-5); 1.57 (dd, 1H, ^3^J9-11ax = 11.8, ^3^J9-11eq = 3.3, H-9); 1.59 (m, 1H, H_ax_-6); 1.63 (ddd, 1H, ^2^J = 14.2, ^3^J7ax-6ax = 14.7, ^3^J7ax-6eq = 4.7, H_ax_-7); 1.68 (m, 1H, H_β_-15); 1.72 (m, 1H, H_eq_-11); 1.73 (m, 1H, H_eq_-12); 1.74 (dddd, 1H, ^2^J = 13.4, ^3^J16β-15β = 10.7, ^3^J16β-17 = 6.2, ^3^J16β-15α = 1.8, H_β_-16); 1.96 (ddt, 1H, ^2^J = 13.4, ^3^J16α-17 = 10.8, ^3^J16α-15α = 8.6, ^3^J16α-15β = 8.6, H_α_-16); 2.00 (ddd, 1H, ^3^J13-12ax = 12.4, ^3^J13-17 = 10.8, ^3^J13-12eq = 3.8, H-13); 2.15 (s, 3H, H-21); 2.48 (dd, 1H, ^2^J = 17.9, ^4^J1α-31 = 3.2, H_α_-1); 2.64 (td, 1H, ^3^J17-13 = 10.8, ^3^J17-16α = 10.8, ^3^J17-16β = 6.2, H-17); 3.59 (dd, 1H, ^2^J = 17.9, ^4^J1β-31 = 1.8, H_β_-1); 7.18 (ddd, 1H, ^3^J35-36 = 7.7, ^3^J35-34 = 4.9, ^4^J35-37 = 1.2, H-35); 7.37 (dd, 1H, ^4^J31-1α = 3.2, ^4^J31-1β = 1.8, H-31); 7.39 (dd, 1H, ^3^J37-36 = 7.7, ^4^J37-35 = 1.2, H-37); 7.70 (dt, 1H, ^3^J36-35 = 7.7, ^3^J36-37 = 7.7, ^4^J36-34 = 2.0, H-36); 8.69 (dd, 1H, ^3^J34-35 = 4.9, ^4^J34-36 = 2.0, H-34). ^13^C NMR (СDCl_3_, δ ppm): 14.87 (C18); 15.76 (C30); 15.97 (C19); 20.35 (C6); 21.99 (C11); 22.29 (C29); 25.71 (C12); 25.96 (C16); 29.41 (C28); 30.02 (C21); 31.52 (C15); 34.47 (C7); 36.22 (C10); 40.22 (C8); 45.21 (C1); 45.23 (C13); 45.39 (C4); 48.41 (C9); 50.07 (C14); 53.33 (C5); 54.17 (C17); 122.23 (C35); 126.86 (C37); 134.09 (C31); 136.14 (C36); 138.65 (C2); 149.44 (C34); 155.59 (C32); 208.72 (C3); 212.12 (C20). ^15^N NMR (СDCl_3_, δ ppm): 316.10 (N33). Found, m/z: 448.3164 [M+H]^+^. C_30_H_41_NO_2_. Calculated, m/z: 447.3137. Anal. Calcd. for C_30_H_41_NO_2_: C, 80.49 H, 9.23; N, 3.13. Found: C, 80.24; H, 9.10; N, 3.05.

3.1.2. 2(E),21(E)-bis-[2-Pyridinyl]-methylidenohollongdione (3)

To a solution of compound 1 (0.360 g; 1.0 mmol) in ethanol (20 mL) 2-pyridinylcarboxaldehyde (1.3 mmol, mL) and 40% KOH in ethanol (2.5 mL) were added. The reaction was stirred for 30 min at room temperature, poured into water (50 mL), and the precipitate was filtered off and washed with water. Compound 3 was crystallized from EtOH. Yield 0.51 g (95%), mp 176 °C, [α]^20^D +119° (c 0.01, CHCl_3_).^1^H NMR (СDCl_3_, δ ppm, J Hz): 0.82 (s, 3H, H-19); 0.98 (s, 3H, H-30); 1.06 (s, 3H, H-18); 1.14 (s, 3H, H-28); 1.19 (s, 3H, H-29); 1.27 (m, 1H, H_α_-15); 1.32 (m, 1H, H_ax_-11); 1.34 (m, 1H, H_ax_-12); 1.38 (m, 1H, H_eq_-7); 1.49 (m, 1H, H_eq_-6); 1.56 (m, 1H, H-5); 1.57 (dd, 1H, ^3^J9-11ax = 11.8, ^3^J9-11eq = 3.3, H-9); 1.58 (m, 1H, H_ax_-6); 1.65 (ddd, 1H, ^2^J = 14.2, ^3^J7ax-6ax = 14.7, ^3^J7ax-6eq = 4.7, H_ax_-7); 1.68 (m, 1H, H_eq_-11); 1.72 (m, 1H, H_eq_-12); 1.74 (m, 1H, H_β_-15); 1.88 (dddd, 1H, ^2^J = 13.4, ^3^J16β-15β = 10.7, ^3^J16β-17 = 6.5, ^3^J16β-15α = 1.8, H_β_-16); 2.02 (ddt, 1H, ^2^J = 13.4, ^3^J16α-17 = 10.8, ^3^J16α-15α = 8.6, ^3^J16α-15β = 8.6, H_α_-16); 2.11 (ddd, 1H, ^3^J13-12ax = 12.4, ^3^J13-17 = 10.8, ^3^J13-12eq = 3.8, H-13); 2.48 (dd, 1H, ^2^J = 18.0, ^4^J1α-31 = 3.6, H_α_-1); 3.02 (td, 1H, ^3^J17-13 = 10.8, ^3^J17-16α = 10.8, ^3^J17-16β = 6.5, H-17); 3.57 (dd, 1H, ^2^J = 18.0, ^4^J1β-31 = 1.8, H_β_-1); 7.17 (ddd, 1H, ^3^J35-36 = 7.7, ^3^J35-34 = 4.9, ^4^J35-37 = 1.2, H-35); 7.22 (d, 1H, ^3^J21-38 = 15.8, H-21); 7.28 (ddd, 1H, ^3^J5′-4′ = 7.7, ^3^J5′-6′ = 4.9, ^4^J5′-3′ = 1.2, H-5′); 7.36 (dd, 1H, ^4^J31-1α = 3.6, ^4^J31-1β = 1.8, H-31); 7.38 (dd, 1H, ^3^J37-36 = 7.7, ^4^J37-35 = 1.2, H-37); 7.48 (dd, 1H, ^3^J3′-4′ = 7.7, ^4^J3′-6′ = 1.2, H-3′);7.55 (d, 1H, ^3^J1′-21 = 15.8, H-1′); 7.68 (dt, 1H, ^3^J36-35 = 7.7, ^3^J36-37 = 7.7, ^4^J36-34 = 2.0, H-36); 7.73 (td, 1H, ^3^J4′-3′ = 7.7, ^3^J4′-6′ = 7.7, ^4^J4′-7′ = 2.0, H-4′); 8.67 (dd, 1H, ^3^J6′-5′ = 4.9, ^4^J6′-4′ = 2.0, H-6′); 8.69 (dd, 1H, ^3^J34-35 = 4.9, ^4^J34-36 = 2.0, H-34). ^13^C NMR (СDCl_3_, δ ppm): 14.99 (C18); 15.86 (C30); 16.04 (C19); 20.42 (C6); 22.07 (C11); 22.33 (C29); 25.81 (C12); 26.34 (C16); 29.48 (C28); 31.79 (C15); 34.59 (C7); 36.28 (C10); 40.40 (C8); 45.29 (C1); 45.46 (C4); 46.07 (C13); 48.50 (C9); 50.35 (C14); 52.17 (C17); 53.37 (C5); 122.26 (C35); 124.28 (C6′); 124.57 (C3′); 126.85 (C37); 129.94 (C21); 134.20 (C31); 136.15 (C4′); 136.88 (C43); 138.72 (C2); 140.78 (C1′); 149.52 (C34); 150.18 (C7′); 153.46 (C2′); 155.65 (C32); 203.83 (C20); 208.95 (C3). ^15^N NMR (СDCl_3_, δ ppm): 308.35 (N7′); 309.17 (N33). Found, m/z: 537.3424 [M+H]^+^. C_36_H_44_N_2_O_2_. Calculated, m/z: 536.3403. Anal. Calcd. for C_36_H_42_N_2_O_2_: C, 80.56 H, 8.26; N, 5.22. Found: C, 80.31; H, 8.07; N, 5.02.

Crystallographic data for 3: C_36_H_44_N_2_O_2_, FW = 536.73, Monoclinic, Space group P21, a = 11.1425(8) Å, b = 10.5767(7) Å, c = 13.5129(10), b = 110.106(2)°, V = 1495.46(18) Å^3^, Z = 2, dcalc = 1.192 g × cm^−3^. 16,127 reflections were collected at SMART APEX II CCD diffractometer (l(Mo-Ka) = 0.073 Å, graphite monochromator, w-scans, 2q < 29°) at 120 K. The structure was solved by the direct methods and refined by the full-matrix least-squares procedure in an anisotropic approximation. 6540 independent reflections [R(int) = 0.0471] were used in the refinement procedure that was converged to wR2 = 0.11110 calculated on F2hkl (GOF = 1.006, R1 = 0.0494 calculated on Fhkl using 3436 reflections with I > 2s(I)). CCDC 1984684 contains the supplementary crystallographic data for this paper. These data can be obtained from the Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif (accessed on 15 December 2025).

3.1.3. 3β-Hydroxy-hollongdione (4)

A solution of 1 (0.444 g; 1.0 mmol) in AcOH (30 mL) was refluxed for 2 h; the reaction mixture was cooled to 0 °C, and ozone was passed through. The mixture was then poured into Н_2_O, and filtered and washed with H_2_O. The crude product was dried with air, then treated with 5 % KOH in EtOH (50 mL) for 24 h. The solution was poured into H_2_O, the precipitate was filtered off, and dried with air. Yield 0.35 g (97%), mp 194–195 °C, [α]D^20^ +54° (c 0.01, CHCl_3_). (lit. mp 195–197 °C, [α]D^22^ +54.1° (c 1.33)) [58].

3.1.4. 3β-Hydroxy-21(E)-[2-pyridinyl]-methylidenohollongdione (5)

To a solution of compound 4 (0.360 g; 1.0 mmol) in ethanol (20 mL), 2-pyridinylcarboxaldehyde (1.3 mmol, mL) and 40% KOH in ethanol (2.5 mL) were added. The mixture was stirred for 24 h at room temperature, the pH was adjusted to neutral by adding an aqueous solution of 5% HCl, and the mixture was poured into cold H_2_O. The residue was filtered off, washed with H_2_O, and dried, then purified by column chromatography on Al_2_O_3_ using petroleum ether-EtOAc (40:1 to 1:1), CHCl_3_ as eluent. Yield 0.42 g (94%), mp 123–124 °C, [α]^20^D +54° (c 0.01, CHCl_3_). ^1^H NMR spectrum, δ, ppm (J, Hz): 0.71 (dd, 1H, ^3^J5-6ax = 11.8, ^3^J5-6eq = 2.1, H-5); 0.75 (s, 3H, H_3_-29); 0.82 (s, 3H, H_3_-19); 0.91 (s, 3H, H_3_-30); 0.95 (m, 1H, H_ax_-1); 0.96 (s, 3H, H_3_-28); 0.98 (s, 3H, H_3_-18); 1.18 (m, 1H, H_α_-15); 1.24 (m, 1H, H_ax_-11); 1.24 (m, 1H, H_ax_-12); 1.29 (m, 1H, H-9); 1.29 (m, 1H, H_eq_-7); 1.42 (m, 1H, H_ax_-6); 1.49 (m, 1H, Heq-11); 1.52 (m, 1H, H_eq_-6); 1.55 (m, 1H, H_ax_-2); 1.56 (m, 1H, H_ax_-7); 1.58 (m, 1H, H_eq_-2); 1.64 (m, 1H, H_eq_-12); 1.66 (m, 1H, H_eq_-1); 1.69 (m, 1H, H_β_-15); 1.83 (m, 1H, H_β_-16); 1.98 (dddd, 1H, ^2^J = 13.5, ^3^J16α-17 = 10.8, ^3^J16α-15α = 9.0, ^3^J16α-15β = 8.5, H_α_-16); 2.03 (ddd, 1H, ^3^J13-12ax = 12.0, ^3^J13-17 = 10.8, ^3^J13-12eq = 4.1, H-13); 2.95 (td, 1H, ^3^J17-13 = 10.8, ^3^J17-16β = 10.8, ^3^J17-16α = 6.5, H-17); 3.18 (dd, 1H, ^3^J3-2ax = 11.2, ^3^J3-2eq = 5.0, H-3); 7.18 (d, 1H, ^3^J21-1′ = 15.8, H-21); 7.25 (ddd, 1H, ^3^J5′-4′ = 7.7, ^3^J5′-6′ = 4.9, ^4^J5′-3′ = 1.2, H-5′); 7.45 (dd, 1H, ^3^J3′-4′ = 7.7, ^4^J3′-5′ = 1.2, H-3′); 7.51 (d, 1H, ^3^J1′-21 = 15.8, H-1′); 7.69 (td, 1H, ^3^J4′-5′ = 7.7, ^3^J4′-3′ = 7.7, ^4^J4′-6′ = 2.0, H-4′); 8.63 (dd, 1H, ^3^J6′-5′ = 4.9, ^4^J6′-4′ = 2.0, H-6′). ^13^C NMR spectrum, δ, ppm (J, Hz): 15.40 (C29); 15.63 (C18); 15.96 (C30); 16.24 (C19); 18.26 (C6); 21.24 (C11); 25.63 (C12); 26.32 (C16); 27.41 (C2); 28.03 (C28); 31.78 (C15); 35.61 (C7); 37.19 (C10); 38.98 (C4); 39.11 (C1); 40.60 (C8); 45.89 (C13); 50.25 (C14); 50.74 (C9); 52.20 (C17); 55.88 (C5); 78.84 (C3); 124.20 (C5′); 124.47 (C3′); 129.86 (C21); 136.80 (C4′); 140.68 (C1′); 150.10 (C6′); 153.41 (C2′); 203.85 (C20). Found, Found, m/z: 450.3255 [M+H]^+^. C_30_H_43_NO_2_. Calculated, m/z: 449.3294. Found, %: С 80.10; H 9.69; N 3.13. C_30_H_43_NO_2_. Calculated, %: С 80.13; H 9.64; N 3.11.

3.2. Biological Assay

3.2.1. NCI-60 Data

Primary anticancer assays were performed according to the NCI protocol as described elsewhere (see e.g., http://dtp.nci.nih.gov, accessed on 12 November 2025) [44,45]. The compounds were added at a single concentration, and the cell cultures were incubated for 48 h. The endpoint determinations were conducted with a protein-binding dye, sulforhodamine B (SRB). The percentage of growth was evaluated spectrophotometrically versus controls (samples not treated with test agents) after 48-h exposure and using the SRB protein assay to estimate cell viability or growth. Three antitumor activity dose-response parameters were calculated for each cell line: GI_50_—molar concentration of the compound that inhibits 50% net cell growth; TGI—molar concentration of the compound leading to the total inhibition; and LC_50_—molar concentration of the compound leading to 50% net cell death (presented in negative logarithm). Furthermore, mean graph midpoints (MG_MID) were calculated for each of the parameters, giving an average activity parameter over all cell lines for the tested compound. For the MG_MID calculation, insensitive cell lines were included with the highest concentration tested.

3.2.2. Cell Lines

Human cancer cell lines breast cancer (MCF-7, SKBR-3), prostate cancer (DU-145), glioblastoma (U-251 MG), liver cancer (Huh7, HepG2), pancreatic cancer (MIAPaCa-2, PANC-1), lung cancer (A549), ovarian cancer (SK-OV-3), melanoma (SK-MEL-28) and colon cancer (SW620, HCT116, HT-29) were obtained from the N.N. Blokhin National Medical Research Center Bio Collection. Cells were propagated in RPMI-1640 medium supplemented with 10% fetal calf serum (Gibco) and 1% penicillin/streptomycin in a 5% CO, atmosphere at 37 °C. After 80% confluence, the cells in the exponential growth phase were trypsinized, counted, seeded at a density of 5 × 10^4^ cells/well in 96-well plates, and incubated for 24 h.

3.2.3. Cytotoxicity Assay

Cells were seeded in a 96-well plate at a density of 8 × 10^3^ cells per well. After overnight incubation, the studied substances (compound 3 and compound 5) were added to cells in various concentrations in 100 µL of culture medium. After 72 h of drug exposure, cells were treated with 20 µL of MTT solution (MTT Cell Viability Assay Kit, Servicebio, Wuhan, China). In parallel, MTT was added to the wells with DMEM with NPs without cells that were used further for optical compensation. Incubation with MTT lasted for 4 h, then the medium was aspirated from the wells, and formazan crystals were dissolved in dimethyl sulfoxide (DMSO). The compensation for DMSO absorbance was also performed. Optical density was measured at 570 nm on a Tecan Infinite^®^ M Nano + spectrophotometer (Männedorf, Switzerland). Absorption values were converted to percentages relative to the absorption values of intact cells.

3.2.4. Fish Embryo Acute Toxicity Test (FET)

Studies were conducted on zebrafish (Danio rerio) in accordance with OECD guideline Protocol 236 “Fish Embryo Acute Toxicity (FET) Test” (OECD/OCDE, 2013). Morphological effects were assessed according to [76]. Adult wild-type zebrafish were kept in aquariums with an aeration and recirculation system at a temperature of 28 °C, pH 6.5–7.5, with a photoperiod cycle of 14:10 h (light: dark). The fish were fed twice daily according to conventional recommendations (using zebrafish food). Freshly laid eggs after fertilization (less 1 h post-fertilization (hpf)) were collected and placed in Danio rerio E3 embryo water (5 mM NaCl, 0.33 mM CaCl_2_, 0.33 mM MgSO_4_·7H_2_O, 0.17 mM KCl, and 0.1% methylene blue). Unfertilized eggs and embryos that had significant developmental defects 24 h after fertilization were detected under a Nexcope NSZ-810 microscope (Ningbo, China) and removed from the experiment. Experimental embryos were mechanically dechorionized with tweezers and placed in 24-well plates (2 embryos per well, total 0.5–1.5 mL of solution per well). They contained 1.5, 3, and 6 µM of the compound 3. The toxicity of the agent was assessed after 144 h of incubation. The following endpoints were used to evaluate toxic effects: developmental abnormalities and delays, morphological alterations (including irregular yolk sac shape, tail development impairment), and reduced motor activity. The median lethal concentration (LC_50_) was calculated using GraphPad Prism software, version 9.1.

3.2.5. Antitumor Effect of Compound 3 on Danio rerio Embryos

HCT-116 human colon carcinoma cells were pre-labeled with the fluorescent dye CFSE (Lumiprobe) for subsequent injection. For labeling, the cells were incubated with the dye for 20 min at 37 °C and washed twice with PBS buffer (according to the manufacturer’s protocol). Using a microinjection system (RWD Life Science micro pump R-480 and micromanipulator MP-550), 15 nL of the cell suspension (approximately 200 live cells per embryo) was injected into the yolk sac. Following cell injection (within two hours), the embryos were transferred into a 24-well plate (one embryo per well). The test compound was dissolved in DMSO and diluted to the required concentration (selected based on a prior toxicity test) in embryo water; the final DMSO concentration was kept below 0.1%. The embryos with injected cells were maintained at 32 °C for optimal mammalian cell growth. At 24 h post-injection, embryos with established tumors were selected and randomized into two groups: control and experimental. The compound 3 was added to the water of the experimental group. Tumor cell visualization in the embryonic yolk sac was performed at 24 and 72 h after adding the compound, followed by tumor area quantification using a stereoscopic fluorescence microscope.

Images of the tumor-bearing embryos were processed using ImageJ software Version 1.53: the images were split into green channel and grayscale, after which the tumor regions were identified, outlined based on intensity, and their area was calculated from the dimensions of the selected tumor region.

3.3. Statistical Analysis

Survival data were evaluated by regression analysis using GraphPad Prism v9.5.1 (GraphPad Software, San Diego, CA, USA). Therapeutic activity data were analyzed using descriptive statistics (mean ± SD) calculated in Microsoft Excel 2019 (Microsoft Corp., Redmond, WA, USA). Statistical significance of differences between groups was assessed using SPSS Statistics v26.0 (IBM Corp., Armonk, NY, USA) with application of the non-parametric Mann–Whitney U test. All experiments were performed in three independent replicates (n = 3), and differences were considered statistically significant at p < 0.05.

3.4. Swiss ADME

The physicochemical properties of the compounds 2, 3, and 5 were calculated using online software SwissADME (https://www.swissadme.ch/) (accessed on 7 October 2025).

3.5. PASS Analysis

PASS v2.0 (Prediction of Activity Spectra for Substances) analysis was calculated on https://way2drug.com (accessed on 7 October 2025). The output file represents a list of activities with two probabilities, Pa (probability to be active) and Pi (probability to be inactive). Pa (probability “to be active”) estimates the chance that the studied compound belongs to the sub-class of active compounds (resembles the structures of molecules, which are the most typical in a subset of “actives” in PASS training set). Pi (probability “to be inactive”) estimates the chance that the studied compound belongs to the subclass of inactive compounds (resembles the structures of molecules, which are the most typical in a subset of “inactives” in the PASS training set). The Pa value based on a cut-off of >0.5 [52].

4. Conclusions

An efficient and regioselective approach for the modification of hollongdione via Claisen–Schmidt aldol condensation was developed, enabling controlled formation of mono- or bis-arylidene derivatives by varying the reaction temperature and reactant molar ratios. Ethanol was identified as the optimal solvent, affording high yields and selectivity, while lowering the reaction temperature to 0 °C completely suppressed the formation of bis-condensed byproducts. The structures of compounds 2, 3, and 5 were unambiguously confirmed by comprehensive NMR spectroscopy. The structure of 2(E),21(E)-bis-(2-pyridinyl)-methylidenohollongdione 3 was confirmed by single-crystal X-ray diffraction, which established the E configuration of both exocyclic double bonds in compound 3. In the NCI-60 screening, the 2,21-bis-substituent derivative of hollongdione 3 exhibited broad antiproliferative potency (MG-MID GI_50_ = 1.16 μM) and melanoma selectivity (LOX IMVI SI = 18.82). In MTT assays, compound 3 showed exceptional potency against KRAS-mutant PANC-1 pancreatic cancer cells (IC_50_ = 0.22 µM). In vivo evaluation in an HCT116 xenograft model using Danio rerio embryos confirmed the antitumor efficacy of compound 3, resulting in a tumor growth index of 0.5 without significant toxicity. Collectively, these results identify compound 3 as the most potent and selective hollongdione derivative reported to date and as a promising lead for further preclinical development, particularly in the context of pancreatic cancer therapy.