Multimodal 4-arylchromene derivatives with microtubule-destabilizing, anti-angiogenic, and MYB-inhibitory activities

Aim: Efficient and readily available anticancer drugs are sought as treatment options. For this reason, chromene derivatives were prepared using the one-pot reaction and tested for their anticancer and anti-angiogenic properties. Methods: 2-Amino-3-cyano-4-(aryl)-7-methoxy-4H-chromene compounds (2A-R) were repurposed or newly synthesized via a three-component reaction of 3-methoxyphenol, various aryl aldehydes, and malononitrile. We performed assays to study the inhibition of tumor cell growth [3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyl tetrazolium bromid (MTT) assay], effects on microtubules (immunofluorescence), cell cycle (flow-activated cell sorting analysis), angiogenesis (zebrafish model), and MYB activity (luciferase reporter assay). Fluorescence microscopy was applied for localization studies via copper-catalyzed azide-alkyne click reaction of an alkyne-tagged drug derivative. Results: Compounds 2A-C and 2F exhibited robust antiproliferative activities against several human cancer cell lines (50% inhibitory concentrations in the low nanomolar range) and showed potent MYB inhibition. The alkyne derivative 3 was localized in the cytoplasm after only 10 min of incubation. Substantial microtubule disruption and G2/M cell-cycle arrest were observed, where compound 2F stood out as a promising microtubule-disrupting agent. The study of anti-angiogenic properties showed that 2A was the only candidate with a high potential to inhibit blood vessel formation in vivo. Conclusion: The close interplay of various mechanisms, including cell-cycle arrest, MYB inhibition, and anti-angiogenic activity, led to identifying promising multimodal anticancer drug candidates.

Based on these active anticancer pyran heterocycles, several 2-amino-3-cyano-4-aryl-7-methoxy-4Hchromenes were prepared for testing their antiproliferative and MYB-inhibitory activities. Experiments on cell death induction, microtubule destabilization, drug localization, and anti-angiogenic effects were performed to investigate the underlying mechanisms of action of the most active compounds.
Tube-formation assay EA.hy926 endothelial hybrid cells were maintained for 24 h in EndoPrime low-serum endothelial medium and seeded (3 × 10 5 cells/mL, 50 µL/well) on basement membrane-like matrix Matrigel® on Ibidi µ-Slides. After treatment with test compounds for 4 h until tubular structures had formed in the control wells, results were documented using a Zeiss Axiovert 135 light microscope. Cell vitality was measured using the MTT assay and was above 75% compared to solvent-treated cells. Experiments were performed in triplicate.

Zebrafish angiogenesis assay
Transgenic zebrafish (Tg(fli1:EGFP, casper mutant) were bred at 28 °C. [16] After fertilization, the eggs were transferred to E3-medium (5 mm NaCl, 0.17 mm KCl, 0.33 mm CaCl, 0.33 mm MgSO 4 , 0.01% methylene blue, pH 7.2), followed by incubation for 24 h. The chorion was manually removed, and the larvae were treated with test compounds or solvent in six-well plates (30 fish per concentration, 5 mL/well) for 48 h. The fluorescent vasculature was documented with a Leica MZ10F and Zeiss AxioCam MRrc. To quantify the angiogenesis, the sub-intestinal vein (SIV) area was measured using ImageJ and expressed as means ± SD with the control set to 100%. The significance of SIV reduction through substance treatment was assessed using one-way analysis of variance (ANOVA), *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001, with Dunnett´s multiple comparison test (GraphPad Prism 9).

Luciferase-dependent MYB activity assay
Compounds were tested using the HEK-MYB-Luc reporter cell line, which allows a doxycycline-dependent induction of MYB expression and harbors an MYB-dependent luciferase reporter plasmid, as previously described [17,18] . After 16 h of substance treatment (0.0001-3 µM), luciferase activities were analyzed as described [11] .

RESULTS
The test compounds 2A-R were obtained from 3-methoxyphenol, malononitrile, the corresponding aryl aldehyde, and a cat. amount of Et 3 N in acetonitrile [Scheme 1]. Synthesis and analysis of the known compounds 2B, 2O, and 2R were as described [5,14] . The new compounds 2A, 2C-N, 2P, and 2Q were colorless solids. The structures of the test compounds were confirmed by NMR, IR, and MS analyses. The compounds are racemic mixtures, and no efforts to separate enantiomers were performed. The yields were generally low but acceptable considering the simple one-pot reaction and workup and the commercially available starting compounds [ Table 1].
The new propargyl ether derivative 3 was prepared to conduct localization studies of this compound within cancer cells [Scheme 2]. 3-Propargyloxyphenol was obtained from the reaction of resorcinol with 1.2 equiv. propargyl bromide (80% in toluene) in DMF under basic conditions (K 2 CO 3 ) [19] . Harsher conditions (reflux for 5 h in EtOH) were necessary to synthesize 3 by the described three-component reaction compared with the synthesis conditions of compounds 2A-R. 3 was obtained as an off-white solid in low yields only.
The antiproliferative activity of compounds 2A-R and 3 was evaluated in nine tumor and hybrid cell lines from six entities and compared with previously published data of the known compounds 1A and 1B [ Table 2] [13,20] .
The presentation of the cytotoxicity screening results was simplified by determining the average IC 50 values of all cell lines and sorting them based on their antiproliferative activity [ Table 3]. The halogen atom did not have any major effect on the activity among the compounds with a 3-halo-4, 5-dimethoxyphenyl group (2A, 2B, and 2C). By contrast, in the group of 3, 5-dichloro-to 3, 5-diiodo-4-methoxyphenyl derivatives 2D, 2E, and 2F, a distinct increase in activity was observed from the dichloro compound 2D via the dibromo 2E to the diiodo derivative 2F. Several striking effects were observed in addition to this trend. Comparing the structure of 1C (coumarin-based) and 2F (7-methoxy-4H-chromene), a clear increase in antiproliferative activity based on the methoxychromene structure was observed, while the IC 50 values of the 4H-naphtho(1, 2-b)pyran-3-carbonitriles (1A) and the 7-methoxy-4H-chromenes (2R) were both in the nanomolar range. Some compounds showed certain tumor cell line-specific activities, surpassing positive controls 1A and 1B. 2C was particularly active against HCT116 p53-deficient colon carcinoma cells, verapamil-treated KB-V1 Vbl cervix carcinoma cells and MCF-7 breast cancer cells, whereas 2F and 2O were highly active against multidrug-resistant HT-29 colon carcinoma cells. The 1A-analog 2R showed slightly less overall activity but had a much stronger effect on MCF-7 breast cancer cells than 1A. The vinblastine-resistant KB-V1 Vbl cells were treated with the P-glycoprotein (P-gp) blocker verapamil to identify synergy effects [21] . Compounds 2I, 2B, 2C, and 2E showed 12-, 14-, 20-and 320-fold lower IC 50 values combined with verapamil (1 µM), assuming an inhibition of the efflux pumps increases the substance efficiency by blocked drug removal via the P-gp membrane transporter. A comparison of compounds 2B and 3, prepared for intracellular localization purposes, showed a substantial conformity of activity, indicating a similar mode of action and accumulation behavior in most tumor cell lines. However, compound 2B was more active against verapamil-treated KB-V1 Vbl cells than against untreated cells and HeLa cells (which is the parent cell line of KB-3-1, from which KB-V1 Vbl cells are derived), while 3 was more active against HeLa cells and KB-V1 Vbl cells in the absence of verapamil than against verapamil-treated KB-V1 Vbl cells.
The most active derivatives 2A-C and 2E-F and the positive control 1A were tested for their toxic effects on non-malignant HDFa cells [ Table 3]. The selectivity index (SI) was calculated as a measure of selectivity toward cancer cells compared to non-malignant cells. 2B, 2C, and 2F displayed exceptionally high SI values, highlighting their potential as anticancer drug candidates. Köhler et al. [20] ; [d] Köhler et al. [13] .
For a more detailed investigation of possible drug mechanisms, compounds 2A-C and 2F were tested for their influence on the cell cycle of 518A2 melanoma cells. FACS analysis at doses of 25 nM revealed a significant G2/M cell-cycle arrest for these compounds and the positive control combretastatin A4 (CA4), but to varying extents [ Figure 2]. Among the compounds 2A-C, the 3-chloro-4, 5-dimethoxyphenyl derivative 2A showed more significant cell-cycle arrest than its 3-bromo and 3-iodo congeners 2B and 2C. However, an opposite effect was observed for the 3, 5-dihalo-4-methoxyphenyl derivatives 2D-F, where the 3, 5-diiodo-4-methoxyphenyl derivative 2F triggered by far the most significant arrest. The analogous 3, 5-dichloro (2D) and 3, 5-dibromo (2E) compounds induced only a slight increase in G2/M phase cells (data not shown). Substance 2F arrested about 74% of cells before or during mitosis and exceeded the C-A4 control. C-A4, 1A, and 1B caused G2/M arrest by microtubule depolymerization, thus preventing the formation of a functional spindle apparatus required for cell division [7,11,22] .
Because of the structural similarity of the new cell cycle arresting compounds 2A-C and 2F to the known tubulin-depolymerizing agent C-A4, the effects on the microtubule cytoskeleton were investigated. Timedependent imaging of 518A2 melanoma cells was applied to provide insights into the dynamics of the depolymerization process [ Figure 3]. In solvent-treated cells, the tubulin cytoskeleton consists of fine filaments extending throughout the cytoplasm. Upon treatment with test compounds 2A-C and 2F, these filaments were initially shortened and fragmented, leading to their destruction and distribution within the cytoplasm. However, the timing of this disintegration process was highly dependent on the substance used. For instance, 2A-C showed a breakdown of the cytoskeleton after 6 h, whereas 2F led to a change in the microtubule structure after 1 h. A similar effect was observed for the C-A4 control after 30 min, which can be attributed to a faster uptake or higher target affinity.
The alkynyl derivative 3 was designed to investigate compound uptake and localization within tumor cells. The propargyl group allows orthogonal fluorescence labeling with 3-azido-7-hydroxycoumarin using a copper-catalyzed click reaction. After 10 min, a distinct increase in fluorescence was observed within the treated cells, with most of the fluorescence found in the cytoplasm [ Figure 4]. This finding suggests an accumulation of 3 in the cytoplasm, supporting the hypothesis that cytoplasmic tubulin is the primary target for 3 and its close analogs used in this study. The conformity of the basic structure and the antiproliferative activity of 3 with compound 2 suggests a similar mechanism.
In addition to antiproliferative and cytotoxic effects, microtubule-destabilizing agents possess additional antitumor properties [23] . In the case of C-A4, anti-angiogenic and vascular disruptive effects have been demonstrated. In this context, the tube-formation assay is a suitable method to observe the effects on the two-dimensional (2D) vessel structures by EA.hy926 cells [24] . The inhibition of cell migration and the development of cell-cell junctions (necessary for tube formation) by 2A-C and 2F were investigated [ Figure 5]. At 100 nM, all four test compounds showed anti-angiogenic effects on EA.hy926 cells seeded on Matrigel®. The cells could not form a cross-linked 2D structure within 4 h, as with the negative control or at 25 nM substance concentration. Even if cell-cell junctions were formed in isolated cases, most cells agglomerated due to their spatial proximity. In addition, rounding was observed in many cells, as in C-A4 treated cells, presumably because of microtubule-destabilizing effects.  The development of blood vessels is based on a complex mechanism with various regulators, which are essential targets for the treatment of tumor growth. Angiogenesis can be seen in the embryonal development of zebrafish larvae, where the SIV can be used to measure anti-angiogenic effects [25] . After exposure of 24-h-old zebrafish embryos to substances 2A-C and 2F or positive control axitinib for 48 h, we  Positive controls used axitinib (50, 500 nM). Negative controls used equivalent amounts of DMSO. The SIV area was quantified using ImageJ and expressed as mean ± SD of at least 20 zebrafish. The significance is expressed as n.s. P > 0.05; ***P < 0.001; ****P < 0.0001 against control (one-way ANOVA, with Dunnett´s multiple comparison test). measured the SIV area and compared it to solvent-treated fish [ Figure 6]. Here, 2F showed no significant change in blood vessel growth, and 2F had a toxic effect above 250 nM. 2B and 2C showed no SIV decrease in the tolerated concentration range. However, 2A inhibited angiogenesis to an extent comparable to the known inhibitor axitinib [26] . This finding suggests a different mechanism of action for 2A, a bimodal compound with selective antiproliferative properties against cancer cells and angiogenesis-inhibiting features.
Due to their structural similarity to the potent MYB inhibitor 1B (data to be published elsewhere), compounds 2A-C and 2F were tested for their MYB-inhibitory activity [ Figure 7]. Inhibition of MYB activity by 2A-C and 2F (1.28-2.81 nM) was superior to the inhibition by reference compound 1B (9.07 nM). Because the values are very close, it is challenging to identify a structure-dependent activity trend; however, the activity was in the order 2B > 2C > 2F > 2A.

DISCUSSION
This study's objective was to develop and optimize the lead structures 1A, 1B, and 1C. For this reason, we reduced the size of the benzo[h]chromene backbone to a 7-methoxy-4H-chromene structure, which led to increased activity. In addition, various phenyl substituents were tested to optimize the lead structure. Initial cytotoxicity studies against nine cancer cell lines confirmed that the C-A4-derived 3-iodo-, 3-bromo-, and 3-chloro-4, 5-dimethoxyphenyl motifs (2A-C) and the analogous 3, 5-diiodo-and 3, 5-dibromo-4methoxyphenyl derivatives (2E-F) are excellent pharmacophores with selectivity against malignant cells.
Only the 3, 5-dichloro-4-methoxyphenyl stands out with only micromolar IC 50 values. Interestingly, differences in the specificity for HeLa cells, derived KB-V1 Vbl cells, and KB-V1 Vbl cells treated with the P-gp inhibitor verapamil were observed for the close analogs 2B and 3. 2B is more active against the parent HeLa cells than against the vinblastine-resistant KB-V1 Vbl cells. In contrast, the addition of verapamil strongly sensitized the KB-V1 Vbl cells to treatment with 2B, leading to a higher activity of 2B against verapamiltreated KB-V1 Vbl cells than against HeLa cells. A comparable hypersensitizing effect of verapamil was observed for KB-V1 Vbl cells treated with 2A, 2C, and 2I, surpassing their activity against HeLa cells, which may have explanations beyond mere P-gp inhibition by verapamil. The hypersensitivity effects of resistant P-gp-overexpressing cancer cells upon verapamil treatment were reported (e.g., based on disrupted energy homeostasis upon ATP depletion), which might lead to enhanced anticancer activity combined with other active drug candidates such as 2A-C (but not 2F) compared with their activity against related cell lines without (overexpressed) P-gp transporter [27] . For unknown reasons, compound 3 showed only slightly reduced activity against KB-V1 Vbl cells when combined with verapamil than against HeLa cells.
Because P-gp tends to have relatively hydrophobic substrates with aromatic rings, substituting methoxy groups with halogen atoms may already have an efflux-attenuating effect due to increased hydrophilicity [30] . Treatment with P-gp modulators can provide more significant cytotoxicity by reducing effective concentrations by 100-fold (paclitaxel) or 351-fold (vinblastine) in MDR colorectal cancer SW620/Ad300 cells [31,32] .
The test compounds were obtained and tested as racemic mixtures. An increase in activity might be achieved by the separation of the enantiomers of the active derivatives and evaluating the separated enantiomers via MTT assay to identify any enantiomers which are more active or less active than the mixture. Instead of separating racemic mixtures, chiral synthetic procedures (e.g., using chiral organic bases instead of piperidine or triethylamine) might be applied to generate enantiopure compounds for biological testing.