Synthesis, Biological Evaluation, DNA Binding, and Molecular Docking of Hybrid 4,6-Dihydrazone Pyrimidine Derivatives as Antitumor Agents

In the present paper, on the basis of molecular hybridization, a series of 4,6-dihydrazone pyrimidine derivatives containing the pyridine moiety were synthesized, structurally characterized, and evaluated in vitro for their antitumor activity. According to the results, all the tested compounds demonstrated broad-spectrum antitumor activity against selected tumor cell lines (MCF-7, BGC-823, A549, and BEL-7402) and no obvious toxicity toward normal cells HL-7702. In particular, compounds 10a and 10f were found to be the most promising antitumor agents among the tested compounds against BGC-823 cells (IC50 = 9.00 μM and 7.89 μM) and BEL-7402 cells (IC50 = 6.70 μM and 7.66 μM), respectively. Compounds 10a and 10f exhibited higher potency against BGC-823 and BEL-7402 than the positive control 5-FU (IC50 = 15.18 μM and 15.81 μM). Further mechanism investigations demonstrated that compounds 10a and 10f could significantly increase the level of cellular ROS and induce early apoptosis of BGC-823 cells in a dose-dependent manner. Moreover, the DNA binding results from UV/Vis, CD spectroscopy, and molecular docking studies indicated that 10a and 10f bind with DNA via groove binding and partial intercalation. These results demonstrated that 10a and 10f may serve as novel lead compounds for the discovery of more dihydrazone pyrimidine derivatives with improved antitumor potency and selectivity.


Introduction
Tumor is a serious disease that endangers human life and health. At present, the treatment methods of malignant tumor mainly include surgery, radiotherapy, chemotherapy, and molecular targeted therapy, among which chemotherapy is one of the most effective methods. However, the current chemotherapeutic drugs generally have disadvantages, such as poor selectivity, strong drug resistance, and obvious side-effects; therefore, it remains a major challenge for researchers in developing novel antitumor agents with reduced toxicity and improved selectivity. Since the introduction of the first antimetabolic antitumor drug 5-fluorouracil  [1] in 1957, a series of fluorouracil drugs have been developed successively, and the antitumor activity of pyrimidine derivatives has attracted more and more attention from researchers [2]. The pyrimidine moiety plays an important role in molecular design, synthesis, and bioactivity of new drugs because of its physiological activity [3][4][5][6]. According to literature reports, pyrimidine derivatives have been becoming important drug intermediates for the antitumor, bactericidal and antiviral activities [7][8][9][10].

Synthesis and General Characterization
The target compounds 10a-10f were synthesized via a two-step reaction according to the synthetic routes depicted in Scheme 1. First of all, intermediates 9a and 9b were synthesized by nucleophilic substitution reaction; secondly, 10a-10f were synthesized by condensation reaction of the above intermediates and selected various pyridine aldehydes, respectively. Compounds 10a-10f were all purified by column chromatography using the same eluent, with yields above 50%.   Herein, inspired by the versatility of the pyrimidine skeleton, hydrazone moiety, and pyridine functional group mentioned above, a series of 4,6-dihydrazone pyrimidine derivatives (10a-10f) (Figure 1) based on a merging approach were designed, synthesized, and evaluated for their antitumor activities. Among them, 10e and its Hg(II) and Pb(II) complexes were previously reported in the literature [27]; however, the antitumor activity of compound 10e has not been reported. Substituents at the R 1 position were added to the terminal pyridine ring to investigate the electronic effect. In order to explore the effect of the molecular plane on antitumor activity, phenyl groups were also introduced at the R 2 position of the pyrimidine skeleton for further modification. All target compounds were assayed for their antitumor activity in vitro against selected cell lines MCF-7, BGC-823, A549, BEL-7402, and HL-7702. On the basis of the antitumor results, promising compounds were selected for further mechanism studies. AO/EB double staining, flow cytometry, and ROS detection were carried out to confirm the apoptosis mechanism. The probable interaction mode of promising compounds with DNA was studied by UV/Vis, circular dichroism (CD) spectroscopy, and molecular docking.

Synthesis and General Characterization
The target compounds 10a-10f were synthesized via a two-step reaction according to the synthetic routes depicted in Scheme 1. First of all, intermediates 9a and 9b were synthesized by nucleophilic substitution reaction; secondly, 10a-10f were synthesized by condensation reaction of the above intermediates and selected various pyridine aldehydes, respectively. Compounds 10a-10f were all purified by column chromatography using the same eluent, with yields above 50%. spectra of 10a-10f were similar, all displaying strong absorption peaks near 205 nm because of the existence of C=N functional groups and the big conjugate system in the structure. The absorption peak of C=N functional group is n→π*, which is assigned to the transition of lone pair electrons from unbonded heteroatoms to π* antibonding orbitals; the conjugate system is usually attributed to the π→π* transition overlap and benzene ring vibration. In summary, the absorption peaks near 205 nm of 10a-10f were generated by n/π→π* transition. Therefore, the successful synthesis and purity of 10a-10f were fully characterized by 1 H NMR, 13 C NMR, ESI-HRMS, IR, and UV/Vis spectra. Scheme 1. Synthesis of target compounds (10a-10f).

Analysis of Stability
The stability of a compound is crucial for the biological activity. The time-dependent UV/Vis spectra ( Figure S26-S31) of 10a-10f in Tris-HCl buffer solution (pH 7.4) were recorded to evaluate the stability. As shown, the peak position and intensity of the strong absorption peaks near 205 nm in 10a-10f did not change significantly within 48 h, which indicated that 10a-10f were stable in Tris-HCl buffer solution (pH 7.4) during the observation time. The results suggest that 10a-10f were responsible for the antitumor activity.

In Vitro Cytotoxic Activity
All target compounds 10a-10f were screened for their antitumor activities in vitro against tumor cell lines BGC-823, BEL-7402, MCF-7, and A549 with 5-FU as a positive control. Additionally, HL-7702 was chosen to evaluate the safety of 10a-10f against normal human cell lines. The IC50 values of preliminary biological evaluation were summarized in Table 1 and Figure 2. The results showed that compounds 10a-10f exhibited potent cytotoxic activity against selected tumor cell lines and less cytotoxicity toward normal cells HL-7702, which indicated that they all had broad-spectrum antitumor activity. Compounds 10a and 10f were the most potent compounds against BGC-823 and BEL-7402. Moreover, neither of them had cytotoxicity toward normal cells HL-7702. For 10a, the IC50 values against BGC-823 and BEL-7402 cells were 9.00 µM and Scheme 1. Synthesis of target compounds (10a-10f).
The chemical structures of 10a-10f were fully identified by 1 H NMR, 13 C NMR, ESI-HRMS, IR, and UV/Vis spectra (Supplementary Figures S1-S25). Signals characteristic for the pyridine ring are multiple peaks at 8.61-7.16 ppm in the 1 H NMR spectra. The presence of hydrazone group was confirmed by broad signal of NH protons and CH=N protons at 11.59-11.32 ppm and 8.27-8.15 ppm, respectively. For 10d-10f, the presence of a benzene ring was supported by the chemical shift at 8.04-7.20 ppm in the 1 H NMR spectra. Furthermore, for 10c and 10f, the presence of a methyl group was described by unimodal signal at 2.47 and 2.42 ppm in the 1 H NMR spectra. For the 13 C NMR analysis of 10a-10f, the presence of hydrazone group was confirmed by broad signal of CH=N protons at 140.54-137.26 ppm. For 10c and 10f, the presence of a methyl group was supported by the chemical shift at 24.08 and 24.09 ppm, respectively. Compounds 10a-10f are symmetric, and the results of 1 H NMR and 13 C NMR were consistent with their structures. In the ESI-HRMS analysis of 10a-10f, the molecular ion peaks were observed as [M + H] + for all the compounds.
The IR spectra of 10a-10f showed similar absorption bands. The N-H stretching vibrations were located in the range of 3191-3299 cm −1 [28], and the C=N stretching vibrations were observed in the region of 1525-1598 cm −1 . The C-C stretching vibrations occurred in the region of 1467-1386 cm −1 . The absorption bands at 1450 cm −1 and 1454 cm −1 corresponded to the -CH 3 stretching vibrations of 10c and 10f. In addition, for 10a-10f, the observed absorption bands of 689-985 cm −1 were assigned to stretching vibration of benzene ring. Conventional UV/Vis characterization of 10a-10f is shown in Figure S25. As we can see, different substituents had almost no effect on peak positions. The UV/vis spectra of 10a-10f were similar, all displaying strong absorption peaks near 205 nm because of the existence of C=N functional groups and the big conjugate system in the structure. The absorption peak of C=N functional group is n→π*, which is assigned to the transition of lone pair electrons from unbonded heteroatoms to π* antibonding orbitals; the conjugate system is usually attributed to the π→π* transition overlap and benzene ring vibration. In summary, the absorption peaks near 205 nm of 10a-10f were generated by n/π→π* transition.
Therefore, the successful synthesis and purity of 10a-10f were fully characterized by 1 H NMR, 13 C NMR, ESI-HRMS, IR, and UV/Vis spectra.

Analysis of Stability
The stability of a compound is crucial for the biological activity. The time-dependent UV/Vis spectra ( Figures S26-S31) of 10a-10f in Tris-HCl buffer solution (pH 7.4) were recorded to evaluate the stability. As shown, the peak position and intensity of the strong absorption peaks near 205 nm in 10a-10f did not change significantly within 48 h, which indicated that 10a-10f were stable in Tris-HCl buffer solution (pH 7.4) during the observation time. The results suggest that 10a-10f were responsible for the antitumor activity.

5-FU
15.18 ± 0.05 15.81 ± 0.01 11.32 ± 0.78 11. Note: Data are shown as the mean  standard deviation, >100 stands for . The preliminary structure-activity relationships of 10a-10f were explored. The electronic effect had a remarkable influence on antitumor activity; thus, the effect of substituent (-Br, -H, and -CH 3 ) on pyridine moiety on antitumor activity was discussed. It is evident that the antitumor activity of 10a-10c against BGC-823, BEL-7402, and A549 cell lines all followed the descending order 10a > 10b > 10c. The bromine-substituted compound 10a showed the best antitumor activity, which indicated that the anticancer activity was prone to being sensitive to the electron-pulling substituent in the molecule. It is worth mentioning that the introduction of electron-pulling substituent can really improve the bioactivity of many compounds [32][33][34]. DNA is an important target for many small-molecule agents. Generally speaking, increasing the molecular plane of aromatic ring can enhance the binding ability of small molecule with DNA. Hence, 10d-10f were specifically designed and synthesized to investigate the effect of molecular plane on antitumor activity. It was expected that the introduction of phenyl on pyrimidine skeleton can significantly enhance antitumor activity for the increase of conjugate structure. Regrettably, it was found that the introduction of phenyl did not significantly improve the antitumor activity, compared with the above series without phenyl (10a-10f). Perhaps, the introduction of phenyl also increased the molecules' rigidity and steric hindrance, which is not beneficial to antitumor activity. Interestingly, when phenyl was introduced, the order of antitumor activity was completely reversed compared with 10a-10c and 10d-10f against BGC-823, BEL-7402, and MCF-7 cell lines all followed the descending order 10f > 10e > 10d; the methyl-substituted compound 10f showed the best antitumor activity, which means that the electron-pushing substituent is more beneficial when phenyl is on the pyrimidine skeleton. It should be mentioned that the addition of a phenyl group to the pyrimidine skeleton significantly reduces the toxicity to normal cells HL-7702.  Figure 3. A normal morphology and homogeneous green fluorescence were observed in the control group. However, an increasing number of cells with apoptotic features such as nuclear fragmentation and chromatin condensation were observed in the test group in a dose-dependent manner. The results preliminarily indicate that both 10a and 10f can induce apoptosis of BGC-823 cells in a dose-dependent manner. apoptotic or necrotic cells which have lost their membrane integrity and show oran red fluorescence. Therefore, AO/EB double staining can be used to evaluate the phology of living and dead cells (apoptotic and necrotic). The morphological chang AO/EB dual-stained BGC-823 cells treated with varying concentrations of 10a and μM for control, 0.5 IC50, IC50, and 1.5 IC50) are shown in Figure 3. A normal morph and homogeneous green fluorescence were observed in the control group. Howev increasing number of cells with apoptotic features such as nuclear fragmentation chromatin condensation were observed in the test group in a dose-dependent ma The results preliminarily indicate that both 10a and 10f can induce apoptosis of BG cells in a dose-dependent manner.

Apoptosis Detection
Induction of apoptosis is considered to be one of the pathways for many antit agents. To further confirm whether 10a and 10f can induce apoptosis of BGC-823 ce Annexin V-FITC/PI kit was used to detect the degree of apoptosis by flow cytometr shown in Figure 4, compared with the control group, after BGC-823 cells were incu with 10a or 10f gradient concentrations (0.5IC50, IC50, and 1.5IC50) for 48 h, the propo of apoptotic cells increased in a concentration-dependent manner. As shown in F 4a,c, after BGC-823 cells were treated with 10a (5 μM, 9 μM, and 13 μM) for 48 h percentage of the Q3 zone was significantly increased by 49.7%, 55.1%, and 60.5% spectively. Similarly, when BGC-823 cells were incubated with 10f (4 μM, 8 μM, a μM), the proportion of Q3 zone was also remarkably increased (Figure 4b,c) by 4 54.8%, and 59.1%, respectively. The Q3 zone represents the early apoptotic cells above data revealed that both 10a and 10f could effectively induce early apopto BGC-823 cells in a concentration-dependent manner. The results are consistent with of AO/EB double staining.

Apoptosis Detection
Induction of apoptosis is considered to be one of the pathways for many antitumor agents. To further confirm whether 10a and 10f can induce apoptosis of BGC-823 cells, an Annexin V-FITC/PI kit was used to detect the degree of apoptosis by flow cytometry. As shown in Figure 4, compared with the control group, after BGC-823 cells were incubated with 10a or 10f gradient concentrations (0.5IC 50 , IC 50 , and 1.5IC 50 ) for 48 h, the proportion of apoptotic cells increased in a concentration-dependent manner. As shown in Figure 4a,c, after BGC-823 cells were treated with 10a (5 µM, 9 µM, and 13 µM) for 48 h, the percentage of the Q3 zone was significantly increased by 49.7%, 55.1%, and 60.5%, respectively. Similarly, when BGC-823 cells were incubated with 10f (4 µM, 8 µM, and 12 µM), the proportion of Q3 zone was also remarkably increased (Figure 4b,c) by 44.3%, 54.8%, and 59.1%, respectively. The Q3 zone represents the early apoptotic cells. The above data revealed that both 10a and 10f could effectively induce early apoptosis in BGC-823 cells in a concentration-dependent manner. The results are consistent with that of AO/EB double staining.

ROS Cell Fluorescence Intensity
ROS are key substances in the process of apoptosis. Excess ROS can attack the mitochondrial membrane, leading to the release of cytochrome C and apoptosis-inducing factors [35,36]. 2,7-Dichlorodihydrofluorescein diacetate (DCFH-DA, ROS Assay Kit) was used to evaluate the ROS production ability of 10f against BGC-823 cells. As shown in Figure 5, the fluorescence intensity of the control group was zero, while there was also no significant intensity increase in the Rosup group. Figure 5a shows that the number of cells and intensity of green fluorescence increased with the increase in 10f (4 µM and 8 µM). Meanwhile, Figure 5b revealed that the ROS generation increased when BGC-823 cells were treated with 10f (4 µM and 8 µM); the fluorescence intensities were 27.6 and 32.2, respectively. The results indicated that 10f can enhance the level of the intracellular ROS and facilitate cell apoptosis; this phenomenon has also been widely reported in the literature [37,38].

ROS Cell Fluorescence Intensity
ROS are key substances in the process of apoptosis. Excess ROS can attack the mitochondrial membrane, leading to the release of cytochrome C and apoptosis-inducing factors [35,36]. 2,7-Dichlorodihydrofluorescein diacetate (DCFH-DA, ROS Assay Kit) was used to evaluate the ROS production ability of 10f against BGC-823 cells. As shown in Figure 5, the fluorescence intensity of the control group was zero, while there was also no significant intensity increase in the Rosup group. Figure 5a shows that the number of cells and intensity of green fluorescence increased with the increase in 10f (4 μM and 8 μM). Meanwhile, Figure 5b revealed that the ROS generation increased when BGC-823 cells were treated with 10f (4 μM and 8 μM); the fluorescence intensities were 27.6 and 32.2, respectively. The results indicated that 10f can enhance the level of the intracellular ROS and facilitate cell apoptosis; this phenomenon has also been widely reported in the literature [37,38].

DNA-Binding Studies
Targeted therapy is an important mechanism of antitumor drugs. The interacti small molecules with DNA is the common mechanism of antitumor effect [39]. The three main noncovalent interaction modes of small molecules with DNA [40−44], w are electrostatic interaction, groove binding, and intercalation. The commonly use

DNA-Binding Studies
Targeted therapy is an important mechanism of antitumor drugs. The interaction of small molecules with DNA is the common mechanism of antitumor effect [39]. There are three main noncovalent interaction modes of small molecules with DNA [40][41][42][43][44], which are electrostatic interaction, groove binding, and intercalation. The commonly used detection methods are spectrum, viscometry, and gel electrophoresis [45,46]. Spectral detection is the most widely used method because of speed and convenience. The interaction modes of small molecules with DNA can be determined by the changes of peak intensity and peak position.

UV/Visible Spectral Analysis
Electron absorption spectroscopy is commonly used to examine the binding mode of small molecules with DNA [47]. The absorption spectra of 10a and 10f in the absence and presence of CT-DNA at varying concentrations were measured (Figure 6a,b). As shown, the absorption bands of 10a and 10f were similar and affected significantly. Sharp hyperchromic effects occurred at 205 nm and 260 nm with increasing concentration of CT-DNA; meanwhile, no apparent wavelength shift was observed, which suggested that interactions of 10a and 10f with CT-DNA were nonclassical intercalations, such as groove binding and partial intercalation mode. Hyperchromism has also been observed in other compounds [48][49][50], which is presumed to be due to the damage of the DNA double helix, whereby 10a and 10f are maybe embedded between DNA double helix base pairs, resulting in changes in the base-stacking force [51][52][53]. To investigate the binding affinity of 10a and 10f with CT-DNA, the binding constant (K b ) was calculated according to the equation described in Section 3. The K b values obtained for 10a and 10f were 4.44 × 10 6 M −1 and 4.59 × 10 6 M −1 (Figure 6c,d), which displayed a higher affinity of 10a and 10f with DNA. Although the K b values for 10a and 10f were less than that of classical intercalator EB whose binding constant has been reported to be in the order of 10 7 M −1 [54], they were still within the range of intercalation mode (K b = 1 × 10 5 -1 × 10 11 M −1 ) [54][55][56]. The absorption spectra results revealed that 10a and 10f probably interacted with DNA via groove binding and partial intercalation. The K b values for 10a and 10f also imply that the binding affinity of 10f with DNA was slightly stronger than that of 10a, which might be attributed to the larger planar aromatic structure. In order to further confirm the binding mode of 10a and 10f with DNA, CD spectra and molecular docking were also performed.

Circular Dichroism Spectral Analysis
The CD spectrum is a powerful tool for monitoring structural changes in DNA. The CD spectrum is extremely sensitive to conformational changes of DNA, especially in the range of 180-320 nm [57]. Binding of small molecules with DNA can change the DNA configuration, which will cause changes in the CD spectra. Electrostatic interaction and groove binding of small molecules usually cause no change or perturbation in the CD spectrum of DNA [28], while classical intercalation generally significantly changes the position and intensity of the positive and negative peaks of CD spectrum or leads to the generation of new induced circular dichroism (ICD) signals [58]. Figure 7 shows the CD spectra of CT-DNA in the absence and presence of 10a or 10f. Both the positive (279 nm) and the negative (247 nm) bands decreased in intensity after the addition of 10a or 10f, which suggested that the binding of 10a and 10f with CT-DNA induces certain conformational changes, unwinding the DNA helix [59]. These changes support the groove binding mode [60][61][62], which does not exclude a partial intercalation. It should be pointed out that 10f caused a more obvious decrease in CD peak intensities, suggesting that 10f interacted with DNA more strongly, which is consistent with the calculation results of UV/Vis spectra. A possible reason is that the introduction of phenyl increased the planarity of the molecule, strengthening DNA binding.

Circular Dichroism Spectral Analysis
The CD spectrum is a powerful tool for monitoring structural changes in DNA CD spectrum is extremely sensitive to conformational changes of DNA, especially i range of 180-320 nm [57]. Binding of small molecules with DNA can change the D configuration, which will cause changes in the CD spectra. Electrostatic interaction groove binding of small molecules usually cause no change or perturbation in the spectrum of DNA [28], while classical intercalation generally significantly change position and intensity of the positive and negative peaks of CD spectrum or leads t generation of new induced circular dichroism (ICD) signals [58]. Figure 7 shows th spectra of CT-DNA in the absence and presence of 10a or 10f. Both the positive (279 and the negative (247 nm) bands decreased in intensity after the addition of 10a or which suggested that the binding of 10a and 10f with CT-DNA induces certain co mational changes, unwinding the DNA helix [59]. These changes support the gr binding mode [60−62], which does not exclude a partial intercalation. It shoul pointed out that 10f caused a more obvious decrease in CD peak intensities, sugge that 10f interacted with DNA more strongly, which is consistent with the calculatio sults of UV/Vis spectra. A possible reason is that the introduction of phenyl increase planarity of the molecule, strengthening DNA binding.  To further verify the binding mode, AutoDock Tools-1.5.6 and PyMOL Mo Graphics System were used to investigate the binding of 10a or 10f with DNA ID:1D18) through semiflexible molecular docking. The visualization is shown in F Figure S32, and Figure S33. It can be seen clearly that the interaction mode of 10a with DNA (PDB ID: 1D18) is not the classic parallel but oblique to the base p shown, the pyrimidine groups of 10a and 10f inserted into the base pairs of th while the pyridine ring on 10a and 10f extended toward the grooves in both s DNA. For compound 10f, the phenyl on the pyrimidine skeleton extended tow minor groove and formed π-π stacking (Figure 8b and Figure S33). Figure S32 a ure S33 display that the interaction of 10a and 10f with DNA was groove bind

Molecular Docking Study
To further verify the binding mode, AutoDock Tools-1.5.6 and PyMOL Molecular Graphics System were used to investigate the binding of 10a or 10f with DNA (PDB ID:1D18) through semiflexible molecular docking. The visualization is shown in Figure 8, Figures S32 and S33. It can be seen clearly that the interaction mode of 10a and 10f with DNA (PDB ID: 1D18) is not the classic parallel but oblique to the base pairs. As shown, the pyrimidine groups of 10a and 10f inserted into the base pairs of the DNA, while the pyridine ring on 10a and 10f extended toward the grooves in both sides of DNA. For compound 10f, the phenyl on the pyrimidine skeleton extended toward the minor groove and formed π-π stacking (Figure 8b and Figure S33). Figures S32 and S33 display that the interaction of 10a and 10f with DNA was groove binding and oblique intercalation. The docking poses showed that 10a and 10f stacked among the A, G, and C base pairs of the DNA intercalation site. The N atoms on hydrazone and pyridine ring formed hydrogen bonding with the DG-13 residue. Furthermore, for 10a, the N and H atoms of hydrazone had hydrogen bonds with DG-4 and DG-13 residues; in the case of 10f, the N and H atoms of hydrazone formed hydrogen bonding with DA-14 and DC-5 residues. The oblique docking position of 10a and 10f with DNA loosened the double helix of the DNA and reduced the stacking force of base pairs, which caused a decrease in the positive and negative peaks. The observed results were in agreement with CD spectral data. Apoptosis can be activated by a variety of cellular signals, such as intracellular Ca 2+ concentration, an increase in ROS levels, and DNA damage. Herein, ROS detection results showed that the intracellular ROS concentration increased with the increase in 10f concentration, which can induce apoptosis. The results of UV/Vis, CD spectra, and molecular docking showed that DNA was an important target for the antitumor activity of these compounds, which may also cause cell apoptosis.

Instrumentations
1 H NMR and 13 C NMR were measured using a nuclear magnetic resonance spectrometer (AVANCE III 500 MHz, Bruker, Germany). ESI-HRMS was recorded using high-resolution mass spectrometry (Thermo Scientific Q Exactive, Bruker, Germany). IR absorption spectra were recorded in the range of 4000-400 cm−1 on a WQF-510A FTIR spectrophotometer. UV/Vis spectra were recorded on a UV-3600i plus spectrophotometer (UV/Vis/NIR spectrophotometer, Shimadzu, Japan). CD spectra were recorded by circular dichromatic spectrometer (Chirascan/V100, Applied Photophysics, UK). In the MTT assay, the absorbance of each compound was measured at 570 nm using Multiskan GO (Thermofisher, Thermo Fisher Scientific, Waltham, MA, USA). AO/EB staining was performed under an inverted fluorescence microscope (DMIL LED, Leica, Germany). Cell apoptosis and ROS intensity were detected using the flow cytometry system (Beckman coulter CytoFLEX, Beckman Coulter, Pasadena, CA, USA). In addition, ROS intensity was also detected using an inverted fluorescence microscope (DMIL LED, Leica, Ger- Additionally, the binding energy (∆G b 0 ) for 10a and 10f interacting with DNA is listed in Table S1. The ∆G b 0 values for 10a and 10f were −8.0 kcal/mol and −8.1 kcal/mol, respectively. The negative values of ∆G b 0 indicated that the binding of 10a and 10f with DNA was spontaneous. The ∆G b 0 for 10f was slightly less than that of 10a, indicating stronger binding. The binding constants (K b ) for 10a and 10f calculated according to this equation [63] (∆G o b = −RT ln K b ) were 7.75 × 10 5 M −1 and 9.18 × 10 5 M −1 (Table S1), which are also within the range of intercalation mode (K b = 1 × 10 5 to 1 × 10 11 M −1 ) [54][55][56]. As a result, the binding of 10a and 10f with DNA involved groove binding and partial intercalation, which were consistent with UV/Vis and CD spectra.
Apoptosis can be activated by a variety of cellular signals, such as intracellular Ca 2+ concentration, an increase in ROS levels, and DNA damage. Herein, ROS detection results showed that the intracellular ROS concentration increased with the increase in 10f concentration, which can induce apoptosis. The results of UV/Vis, CD spectra, and molecular docking showed that DNA was an important target for the antitumor activity of these compounds, which may also cause cell apoptosis.

Instrumentations
1 H NMR and 13 C NMR were measured using a nuclear magnetic resonance spectrometer (AVANCE III 500 MHz, Bruker, Germany). ESI-HRMS was recorded using high-resolution mass spectrometry (Thermo Scientific Q Exactive, Bruker, Germany). IR absorption spectra were recorded in the range of 4000-400 cm−1 on a WQF-510A FTIR spectrophotometer. UV/Vis spectra were recorded on a UV-3600i plus spectrophotometer (UV/Vis/NIR spectrophotometer, Shimadzu, Japan). CD spectra were recorded by circular dichromatic spectrometer (Chirascan/V100, Applied Photophysics, UK). In the MTT assay, the absorbance of each compound was measured at 570 nm using Multiskan GO (Thermofisher, Thermo Fisher Scientific, Waltham, MA, USA). AO/EB staining was performed under an inverted fluorescence microscope (DMIL LED, Leica, Germany). Cell apoptosis and ROS intensity were detected using the flow cytometry system (Beckman coulter CytoFLEX, Beckman Coulter, Pasadena, CA, USA). In addition, ROS intensity was also detected using an inverted fluorescence microscope (DMIL LED, Leica, Germany).
The detailed synthesis procedure of 9a was as follows: 8a (15 mmol, 2.235 g) was dispersed in 15 mL hydrazine hydrate and stirred at room temperature for 1 h. Then, the mixture was heated to 60 • C to react continuously, and white solid gradually precipitated during the reaction. The reaction progress was monitored by thin-layer chromatography (TLC). After reaction was completed, the mixture was cooled overnight in a refrigerator at 4 • C to precipitate completely. Then, the precipitate was filtered and washed with mixed solvent of ethanol and water several times to obtain a white solid, which was named 4,6-dihydrazinylpyrimidine (9a), with a yield of ca. 72.0%. 4,6-Dihydrazinyl-2phenylpyrimidine (9b) was also obtained using the same method with a yield of ca. 55.5%.

General Procedure for the Synthesis of Compounds 10a-10f
The synthesis procedures of 10a-10f were similar, as represented in Scheme 1; therefore, only the synthesis of 10a is introduced in detail. 4,6-Bis(2-((E)-(6-bromopyridin-2-yl)methylene)hydrazinyl)pyrimidine (10a). A mixture of 9a (2 mmol, 280 mg) and 6-bromo-2-pyridinecarboxaldehyde (4.2 mmol, 781 mg) in 40 mL ethanol solvent was stirred at room temperature for 2 h, and then refluxed at 78 • C until the reaction was finished. The white solid gradually precipitated during the reaction, and the reaction was monitored by TLC. After the reaction was completed, the resulting solid was collected by filtration, washed with ethanol several times, and purified by column chromatography (petroleum ether: ethyl acetate = 1:1) to obtain target compound 10a, with a yield of ca. 84 The stability of 10a-10f was the basis for the evaluation of antitumor activity, which was tested by UV/Vis. The stock solution of 10a-10f (10 −3 M) was prepared with DMSO. Considering the environment of cytotoxicity assay, the time-dependent UV/Vis absorption spectra of 10a-10f (0.1 µM) at 0 h, 24 h, and 48 h were recorded in Tris-HCl buffer (0.05 M Tris-HCl/0.1 M NaCl, pH 7.4), respectively. The stability of 10a-10f was determined by the changes in the position and intensity of the spectral absorption peaks.

MTT Assay
The MTT method is often used to detect the cytotoxicity of compounds in vitro [64]. The cytotoxicity of 10a-10f was evaluated by MTT assay against BGC-823, BEL-7402, MCF-7, and A549 tumor cells. HL-7702 was selected to evaluate the safety of 10a-10f against human normal cells. 5-FU was also evaluated as a positive control. The well-grown cells were inoculated in 96-well plates at a density of 3000 cells/well and incubated in 37 • C for 24 h. The negative control group, blank control group, and administration group were set up. The cells were incubated with a gradient concentration (1.25, 2.5, 5, 10, and 20 µg/mL) of samples or 5-FU for 48 h. Then, 20 µL MTT solution (5 mg/mL) was added to each well.
After the cells were incubated for 4 h, 150 µL DMSO was added to each well before testing. The absorbance was measured at 570 nm with a microplate reader. The inhibition rate and final IC 50 values were calculated according to the following formula. The values shown are the average of at least three parallel trials.
3.5. Apoptosis Detection 3.5.1. AO/EB Double Staining BGC-823 cells were inoculated in six-well plates and incubated at 37 • C for 24 h, and then incubated continuously for another 24 h with a varying concentration of 10a or 10f (0.5IC 50 , IC 50 , and 1.5IC 50 ). AO (500 µL, 1 mg/mL) and EB (500 µL, 1 mg/mL) were mixed, and then 4 mL PBS was added to the above solution to form AO/EB dye (0.1 mg/mL). AO/EB dye (100 µL) was added to each well. After incubating for 3-5 min and washing twice with cold PBS, BGC-823 cells were observed immediately under an inverted fluorescence microscope.

Flow Cytometry
The Annexin V-FITC/PI Kit was used to distinguish different stages of apoptosis [65,66]. BGC-823 cells were inoculated in six-well plates at 37 • C for 24 h, and then incubated for another 48 h with a varying concentration of 10a or 10f (0.5 IC 50 , IC 50 , and 1.5 IC 50 ). The cells were trypsinized (without EDTA), and no fewer than 10 6 cells were collected. According to the instructions, the Annexin V-FITC/PI kit was used to monitor induced apoptosis.

ROS Detection
BGC-823 cells were inoculated in six-well plates at 37 • C for 24 h, and then incubated for another 24 h with a varying concentration of 10f (0.5IC 50 , IC 50 , and 1.5IC 50 ). According to the instructions of ROS Detection Kit (Solarbio, Beijing, China), in situ probe loading was adopted. ROS intensity changes were determined by inverted fluorescence microscopy and flow cytometry. The whole experiment was carried out under dark conditions. 3.6. DNA-Binding Study 3.6.1. UV/Vis Spectra The binding patterns of 10a and 10f with CT-DNA were measured by UV/Vis spectroscopy in Tris-HCl buffer (0.05 M Tris-HCl/0.1 M NaCl, pH 7.4). The stock solution concentrations of both 10a and 10f in DMSO were 10 −3 M. CT-DNA was dissolved in Tris-HCl buffer and stored at 4 • C for use. The stock solution concentration of CT-DNA was determined by UV/Vis absorbance at 260 nm with the extinction coefficient 6600 M −1 ·cm −1 [67]. Tris-HCl buffer was used as a blank to make preliminary adjustments. Absorbance titration experiments were performed at room temperature with the concentrations of 10a and 10f at 0.1 µM, varying the concentrations of CT-DNA (0.2, 0.4, 0.6, 0.8, and 1 µM). The solution was allowed to equilibrate for 2-3 h before measurement. UV/Vis absorption spectra were recorded in the range of 200-400 nm. The binding constant was calculated according to the following equation [54]: where ε a corresponds to the observed extinction coefficient (A/[DNA]), ε f corresponds to the extinction coefficient of the free compound, and ε b is the extinction coefficient of the compound when fully bound to CT-DNA.
[DNA] is the concentration of CT-DNA added to the compound. The K b values of 10a and 10f were calculated from the slope to intercept ratio, which was obtained by plotting a graph of [DNA]/(ε a − ε f ) versus [DNA].

Circular Dichroism Spectra
The CD spectra of CT-DNA in the absence and presence of 10a or 10f were measured after equilibrating for 1 h in a 1 mm diameter quartz colorimeter at room temperature. The concentration of CT-DNA was fixed at 600 µM in each sample. Both 10a and 10f were added to the above CT-DNA solution at a concentration of 100 µM, respectively. The scanning speed was 200 nm/min. The spectra were scanned on an average of six times in the range of 200-400 nm. All measurements were performed in 0.05 M Tris-HCl buffer (pH 7.4).

Molecular Docking
Molecular docking of 10a and 10f with DNA was carried out. The molecular structures of 10a and 10f were plotted by ChemDraw and converted into MOL2 format by Chem3D. DNA (PDB ID: 1D18) was selected from the Protein database (http://www.rcsb.org./pdb, accessed on 23 November 2022). AutoDock Tools-1.5.6 software was used to process the molecular structures of 10a, 10f, and DNA, which were saved in PDBQT format. The binding energy (∆G o b ) of 10a and 10f interacting with DNA was calculated. The docking posture was visualized using the PyMOL Molecular Graphics System.

Conclusions
In summary, a series of dihydrazone pyrimidine derivatives were synthesized successfully, characterized, and evaluated biologically. MTT results showed that all of them exhibited potential cytotoxic activity against selected tumor cells and less cytotoxicity toward normal cells HL-7702, indicating a good selectivity toward tumor cells. Among them, 10a and 10f displayed more potent cytotoxic activity against BGC-823 and BEL-7402 cells than 5-FU. Mechanism studies showed that 10a and 10f could induce early apoptosis in BGC-823 cells. Compounds 10a and 10f bind with DNA via groove binding and partial intercalation. Our work suggests that dihydrazone pyrimidine is a promising chemical skeleton and could be potentially utilized for designing more excellent antitumor agents.

Data Availability Statement: Not applicable.
Acknowledgments: The authors would like to thank the Shiyanjia lab (www.Shiyanjia.com, accessed on 23 November 2022) for the support of 1 H NMR and ESI-HRMS analysis.

Conflicts of Interest:
The authors declare no conflict of interest.