Two-carbon tethered artemisinin–isatin hybrids: design, synthesis, anti-breast cancer potential, and in silico study

Eleven two-carbon tethered artemisinin–isatin hybrids (4a–k) were designed, synthesized, and evaluated for their antiproliferative activity against MCF-7, MDA-MB-231, and MDA-MB-231/ADR breast cancer cell lines, as well as cytotoxicity toward MCF-10A cells in this paper. Among them, the representative hybrid 4a (IC50: 2.49–12.6 µM) was superior to artemisinin (IC50: 72.4->100 µM), dihydroartemisinin (IC50: 69.6–89.8 µM), and Adriamycin (IC50: 4.46–>100 µM) against the three tested breast cancer cell lines. The structure–activity relationship revealed that the length of the alkyl linker between artemisinin and isatin was critical for the activity, so further structural modification could focus on evaluation of the linker. The in silico studies were used to investigate the mechanism of the most promising hybrid 4a. Target prediction, bioinformatics, molecular docking, and molecular dynamics revealed that the most promising hybrid 4a may exert anti-breast cancer activity by acting on multiple targets such as EGFR, PIK3CA, and MAPK8 and thus participating in multiple tumor-related signaling pathways.


Introduction
Breast cancer consists of a group of biologically and molecularly heterogeneous diseases that originate from the breast and can be divided into four major subtypes: luminal-A, luminal-B, human epidermal growth factor receptor 2 (HER2) positive, and triple-negative breast cancer (Feng et al., 2018;Wang et al., 2020a;Tsang and Tse, 2020).At present, breast cancer is the most prevalent malignancy in women (Akr et al., 2017;Lau et al., 2022).In 2020, breast cancer has overtaken lung cancer as the most common malignancy, and an estimated 2.3 million new cases of breast cancer were diagnosed, accounting for nearly one-fourth of women cancer patients (Jaiswal et al., 2021;Sung et al., 2021).Moreover, 685,000 women died due to breast cancer, leading to one-sixth of cancerrelated deaths in women (Arnold et al., 2022).Unfortunately, the burden of breast cancer is expected to increase further in the coming years (Wang et al., 2020b;Schröder et al., 2022;Sher et al., 2022).
Advancements in the treatment of breast cancer have resulted in an increasing population of patients living with this disease, and chemotherapy occupies an important position in breast cancer therapy.However, the chemotherapy treatment suffers from the generation of drug resistance (Shaikh and Emens, 2022;Tufail et al., 2022).Hence, development of innovative chemotherapeutics is a promising strategy to improve therapeutic outcomes of breast cancer.
Hybrid molecules, generated by combining two or more molecule entities, could affect multiple targets to fight against various diseases including breast cancers (Bérubé, 2016;Waseem and Ahmad, 2022;Wang et al., 2023a).Artemisinin (ART) and its derivatives (dihydroartemisinin/DHA, artesunate, and artemether), from the traditional Chinese medicine drug, possess a unique peroxy bridge structure, and they could exert the anticancer effects through diverse mechanisms, inclusive of cell cycle inhibition, inhibition of tumor angiogenesis, promotion of DNA damage, and promotion of ferroptosis (Zhang S. et al., 2022a;Hu et al., 2022).Hence, ART derivatives may have significant therapeutic effects on cancers (Gao et al., 2020;Mancuso et al., 2021).Isatin is an exceptionally useful template for developing new anticancer scaffolds; on account of this finding, more and more isatin-based molecules are either in clinical use or in trials (Gupta et al., 2019;Ding et al., 2020).Accordingly, a combination of ART with isatin is a promising strategy to discover novel anti-breast cancer candidates.
In recent years, several series of ART-isatin hybrids have been screened for their potential against various cancer cell lines, and some of them demonstrated promising antiproliferative activity against breast cancer cells (Hou et al., 2021;Zhang Z. et al., 2022b;Dong et al., 2022;Wang et al., 2022).The structure-activity relationship (SAR) revealed that the linker between ART and isatin moieties influenced the antiproliferative activity significantly, and as a linker, alkyl was more favorable than 1,2,3-triazole.Hence, a series of two-carbon tethered ART-isatin hybrids were designed, synthesized, and assessed for their antiproliferative activity against both drug-sensitive (MCF-7 and MDA-MB-231) and Adriamycin-resistant (MDA-MB-231/ADR) breast cancer cell lines in this study.Moreover, the cytotoxicity of the synthesized ART-isatin hybrids toward normal MCF-10A breast cells was also tested.Finally, possible mechanisms were investigated by in silico studies.The purpose of this paper was to find the candidates with promising antiproliferative potential against both drug-sensitive and drug-resistant breast cancer cell lines and high selectivity.The ART-isatin hybrids 4a-k were prepared according to our method reported previously, and the detailed synthetic route is shown in Scheme 1 (Wang et al., 2023b).Etherification of dihydroartemisinin 1 with ethylene glycol in the presence of boron trifluoride diethyl etherate (BF 3 •OEt 2 ) generated 2hydroxyethyl dihydroartemisinin intermediate 2. The conversion of the hydroxy group in intermediate 2 to tosylate (OTs) with pyridine as the base yielded intermediate 3. Alkylation of (5substituted)isatins with tosylate 3 provided desired ART-isatin hybrids 4a-c.Finally, ART-isatin hybrids 4a-c reacted with methoxylamine/ethoxylamine/benzyloxyamine hydrochlorides using sodium carbonate (Na 2 CO 3 ) as the base yielded hybrids 4d-k.The structures and yields are listed in Table 1.

Characterization
The desired two-carbon tethered ART-isatin hybrids 4a-k were characterized by high-resolution mass spectrometry (HRMS), proton nuclear magnetic resonance ( 1 H NMR), and carbon-13 nuclear magnetic resonance spectroscopy ( 13 C NMR) (Wang et al., 2023b).The corresponding analytical data and the analytical spectra are included in our previous study and Supplementary Figures S1-S33[27].
Table 2 demonstrates that a significant part of the synthesized hybrids (IC 50 : 2.49-73.3µM) showed considerable activity against MCF-7, MDA-MB-231, and MDA-MB-231/ADR breast cancer cell lines, and the activity was superior to that of ART (IC 50 : 72.4->100 µM) and DHA (IC 50 : 69.6-89.8µM).The SAR illustrated that the introduction of the (methoxy/ethoxy/benzyloxy)imino group into the C-3 position and fluoro or methoxy group into the C-5 position of the isatin moiety decreased the activity.In particular, incorporation of the benzyloxyimino group into the C-3 position of isatin skeleton led to a loss of activity.Moreover, the length of the alkyl linker between ART and isatin seems to have a significant influence on the activity, as evidenced by that the two-carbon tethered ART-isatin hybrids were more potent than the threecarbon analogs reported in reference 17.Frontiers in Molecular Biosciences frontiersin.org All the synthesized two-carbon tethered hybrids (IC 50 : >100 µM) were non-toxic toward normal MCF-10A breast cancer cells as ART (IC 50 : >100 µM) and DHA (IC 50 : >100 µM), and the cytotoxicity was lower than that of Adriamycin (IC 50 : 68.8 µM).
Amongst the synthesized hybrids, the most active hybrid 4a (IC 50 : 2.49-12.6µM) not only was far more potent than the parents ART (IC 50 : 72.4->100 µM) and DHA (IC 50 : 69.6-89.8µM) but also possessed higher activity than Adriamycin (IC 50 : 4.46->100 µM) against all the three breast cancer lines.The RI value of hybrid 4a was 0.65, implying its potential to overcome drug resistance.In addition, hybrid 4a: >100 µM) was non-toxic toward normal MCF-10A breast cells, and the SI value was >7.9.Accordingly, hybrid 4a could serve as a promising candidate for further preclinical evaluations.

In silico studies 2.2.1 Target prediction of 4a
SwissTargetPrediction, a tool for predicting the targets of compounds based on the similarity of two-dimensional and three-dimensional structures of known compounds, has become a common tool for predicting the targets of compounds currently due to its better sensitivity and specificity (Daina et al., 2019).We selected the promising candidate 4a and performed target prediction by SwissTargetPrediction.The results showed that 4a may have 109 potential targets.Supplementary Table S1 displays the full list in detail.

Construction of cross-over genes, protein-protein interaction networks, and enrichment analysis
To better explore the cross-over genes, we selected breast cancerrelated gene (BC-related gene) in GeneCard to intersect with our predicted targets of 4a (potential target) (Safran et al., 2010).Finally, a total of 51 cross-over genes were selected (Figure 1A).These crossover genes are the potential gene targets of 4a in the activities against breast cancer.Supplementary Table S2 displays the full list in detail.
To obtain the protein-protein interaction (PPI) networks, we imported 51 cross-over genes into the STRING database and selected Homo sapiens as the organism.The results showed 462 interactions.Subsequently, the PPI networks were obtained using Cytoscape for visualization and calculation (Figure 1B).The average closeness centrality (CC) was 0.62, the average betweenness centrality (BC) was 31.5, and the average degree value (DV) was 18.5.Among them, SRC, EGFR, CCND1, MTOR, MAPK1, PIK3CA, MMP9, MAPK8, MMP2, and MAPK14 had higher DV, BC, and CC values than the norm.This suggests that they are the central targets of 4a.Supplementary Table S3 displays the specific node properties in detail.
We also performed Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis on these target genes, and the KEGG pathway enrichment analysis showed that the target genes were mainly involved in pathways in cancer (31/51), the PI3K-Akt signaling pathway (20/51), endocrine resistance (15/51), MAPK signaling pathway (14/50), Ras signaling pathway (14/51), and other pathways (Figure 1F).These suggest that 4a may exert anti-breast cancer effects by acting on multiple tumor signaling pathways.

Molecular docking
Molecular docking can be used to assess 4a-central target protein interactions.Generally, lower binding energy suggests Frontiers in Molecular Biosciences frontiersin.orgbetter interactions.For now, there is no single criterion for screening the binding energy.The screening is usually based on a binding energy ≤ −5.0 kcal/mol (Zhang J. et al., 2022c).The results of molecular docking showed that 7 of the 10 selected central targets were below −5.0 kcal/mol.These results suggest that 4a has good interactions with the seven central targets.All the results are shown in Table 4.
In addition, the visualization analysis allowed us to better understand the spatial structure and interactions between 4a and the central targets.The intermolecular force between 4a and the central targets include van der Waals, hydrogen bond (conventional hydrogen bond, etc.), and hydrophobic (pi-alkyl, alkyl, etc.) forces.Notably, 4a also generates interesting intermolecular forces with central targets, which include pi-sulfur with EGFR, pi-anion with MAPK1 and MAPK14, pi-pi T-shaped with MMP9, pi-cation with MAPK8, amide-pi stacked with MAPK8, and pi-pi stacked with MMP2.These interesting interactions may have a specific effect on the affinity of the 4a-central marker and deserve further investigation.All the results are shown in Supplementary Figures S34-S43.

Molecular dynamics
In order to better understand the kinetics of the interaction between 4a and target proteins, we selected three proteins with the best binding energy-EGFR, MAPK8, and PIK3CA-as key target proteins for molecular dynamics simulation studies.
Root-mean-squared deviation (RMSD), which measures the coordinate deviation of a specific atom with respect to a reference structure, is often used to assess whether a ligand-receptor system has reached stability (Sargsyan et al., 2017).A stable RMSD means that the atoms in the corresponding system become stable, whereas a fluctuating RMSD implies fluctuations.This simulation process is 50 ns (Figure 2A).The simulations show that 4a-MAPK8 and 4a-PIK3CA reach equilibrium soon after the start of the simulation process and that the RMSD values are below 0.3 nm throughout the simulation time.This suggests that the protein-ligand system fluctuates less, which indicates that the docking level is reasonable in the dynamics study (Rastelli et al., 2010;Dai et al., 2023).4a-EGFR fluctuated more in the initial stage and reached equilibrium after 32 ns, and the RMSD value was approximately  0.5 nm throughout the simulation time, suggesting poor stability in the dynamics study (Li et al., 2010).
The solvent-accessible surface area (SASA) is calculated by the van der Waals forces interacting with solvent molecules to calculate the solute area (Akoonjee et al., 2022).The lower SASA can be explained by stronger hydrophobic interactions and less intercomplex solvent water, that is, the more compact binding between the 4a-key target protein complexes ( VAN DAN BURG et al., 1994).The 50-ns simulation results (Figure 2B) show that all 4a-key target proteins show an acceptable SASA value of the protein-ligand complex during the complex simulation.In addition, the results indicate that the 4a-PIK3CA complex exhibits the most compact structure, followed by the 4a-MAPK8 complex, while the 4a-EGFR complex is the the loosest structure in the dynamics study (Zhou et al., 2023).
Root-mean-squared fluctuation (RMSF) calculates the increase and decrease of each atom relative to its average position, characterizing the change in structure averaged over time, i.e., giving a characterization of the flexibility of each region of the protein (Işık et al., 2022;Xu et al., 2022).All results are shown in Figures 2C-E.First, the residues with small RMSF fluctuations in the protein complex system are consistent with the active residues, which may be related to the interactions such as hydrogen bonds and hydrophobic interactions generated between 4a-key target proteins and forming stable compounds.The regions with larger fluctuations are located in the inactive regions at the edges of the protein, which may be related to interactions such as water and chloride ions.The RMSF values of the 4a-key target protein showed little fluctuation during the 50-ns simulation, indicating that the protein ligands can bind stably in the dynamics study (Radwan et al., 2022).
To explore the interaction of 4a with key target proteins, first, we performed hydrogen bonding analysis (Figure 2F).The average numbers of hydrogen bonds of 4a-EGFR, 4a-MAPK8, and 4a-PIK3CA were 1.97, 0.58, and 2.63, respectively, suggesting that the binding stability of 4a-PIK3CA may be higher than that of the 4a-EGFR and 4a-MAPK8 systems in the dynamics study.
Molecular mechanics-Poisson-Boltzmann surface area (MM-PBSA) is a method for post-processing molecular dynamics trajectories to estimate binding free energies (Homeyer and Gohlke, 2012).To better explain the interaction energy between the ligand and the receptor, we determined the binding energy of all protein-ligand complexes in the equilibrium phase using the gmx_mmpbsa method (Valdés-Tresanco et al., 2021).In the application of the MM-PBSA method, the total binding energy was decomposed into four independent components (electrostatic interactions, van der Waals interactions, and polar and nonpolar solvation interactions).The results of the binding energy of 4a-key target protein are shown in Supplementary Table S4.In the 4a-key target protein complex system, the total binding energies were all negative, indicating their contribution to the binding of the complex system.Among them, the total binding free energies of 4a-EGFR, 4a-MAPK8, and 4a-PIK3CA were −72.310 kJ/ mol, −109.794kJ/mol, and −110.833kJ/mol, respectively.These results indicate that the total binding free energies of the 4a-key target protein complex system supported the strong binding of 4a-MAPK8 and 4a-PIK3CA complexes, while the 4a-EGFR complex is relatively weaker in the dynamic system (Fu et al., 2017;Kushwaha et al., 2021).

Conclusion
In summary, a series of two-carbon tethered ART-isatin hybrids were synthesized and assessed for their antiproliferative activity against MCF-7, MDA-MB-231, and MDA-MB-231/ADR breast cancer cell lines, as well as cytotoxicity toward MCF-10A cells.SAR (Figure 3) revealed that 1) the length of the alkyl linker between ART and isatin influenced the activity remarkably, and the twocarbon linker was more favorable than the three-carbon linker; 2) the relative contribution order of the functional group at the C-3 position of isatin moiety was carbonyl > methoxy/ethoxyimino > benzyloxyimino; 3) compared with hydrogen, regardless of the electron-donating or electron-withdrawing group at the C-5 position of isatin skeleton, the activity was reduced.
In particular, the representative hybrid 4a (IC 50 : 2.49-12.6µM) not only was superior to Adriamycin (IC 50 : 4.46->100 µM) against all the three breast cancer lines but also exhibited an excellent safety profile and had the potential to overcome drug resistance.Meanwhile, the in silico study provided a preliminary mechanistic study of 4a.The PPI network suggested that 4a may target related proteins expressed by SRC, EGFR, CCND1, MTOR, MAPK1, PIK3CA, MMP9, MAPK8, MMP2, MAPK14, and other genes, which may be central targets for 4a to exert its anti-breast cancer effects.KEGG enrichment analysis suggested that 4a may exert anti-breast cancer effects by participating in the PI3K-Akt signaling pathway, endocrine resistance, MAPK signaling pathway, and other breast cancer-related pathways.Subsequently, molecular docking revealed the binding energy of 4a to the central target and detailed intermolecular force information.Finally, molecular dynamics studies provided dynamic information on the binding pattern of 4a to the three highest interacting key targets (EGFR, MAPK8, and PIK3CA).Therefore, hybrid 4a was a promising anti-breast cancer candidate and merited further preclinical evaluations.
FIGURE 1 (A) Intersection of BC-related gene and potential target (Venn diagram).(B) Protein-protein interaction (PPI) network diagram of candidate targets.(C-E) GO enrichment analysis of target genes: (C) biological processes, (D) cellular components, and (E) molecular functions.(F) KEGG enrichment analysis of target genes.
FIGURE 2 (A) Variation curve of RMSD of the protein with time during the simulation of the 4a-key target protein.(B) Variation curve of SASA of the protein with time during the simulation of the 4a-key target protein.(C) RMSF analysis of 4a-EGFR.(D) RMSF analysis of 4a-MAPK8.(E) RMSF analysis of 4a-PIK3CA.(F) Variation in the number of hydrogen bonds with time during the simulation of the 4a-key target protein.

TABLE 1
Structures and yields of two-carbon tethered ART-isatin hybrids 4a-k.

TABLE 3
Selectivity index and resistance index of ART-isatin hybrids 4a-k.