Synthesis, characterization, and biological activity of sterically hindered fluorine-containing platinum(II) complexes

Abstract Picoplatin is a sterically hindered antitumor compound with unique chemical and DNA binding properties, but its curative effect is not as good as other platinum drugs. Therefore, a series of sterically hindered fluorine-containing platinum(II) complexes Y1–Y5 were synthesized and characterized by introducing fluorine atoms at the 2-position of the pyridine group and different leaving group ligands. In vitro antiproliferative activity assessments of these complexes have been performed against A549, MCF-7, SW480, HCT116, HepG2 human cancer cell lines, and LO2 normal cells. The results showed that Y1 with 2-fluoropyridine coordination and chloride ion as the leaving group exhibited stronger antiproliferative activity and selectivity in the tested cancer cell lines, especially for A549 cancer cells. DNA binding constants were determined by UV absorption and fluorescence titration, and the binding ability of Y1 to DNA is much stronger than that of the others, which shows Y1 has better DNA damage ability. For the determination of apoptosis by the complex, the results were consistent with those of the MTT assay, and the apoptosis rate of Y1 reached 38.9%. Therefore, our work offers an advantageous method for the development of new platinum-based antitumor drugs.


Introduction
Platinum antitumor complexes such as cisplatin, carboplatin, and oxaliplatin have been applied in the clinical treatment of various malignant tumors [1][2][3]. However, the efficacy of these drugs is largely connected with serious adverse reactions such as neurotoxicity, nephrotoxicity, immunosuppression, and drug resistance [4]. Therefore, overcoming the drug resistance of cisplatin and its analogs and reducing the side effects have always been key breakthrough points for the development of new platinum antitumor drugs. Metallothionein and glutathione (GSH) in human cells have high reactivity with platinum drugs, which reduces the probability of drug binding DNA in tumor cells and the effective drug concentration acting on the target [5]. Based on this, pharmacologists and chemists believe that it can enhance the effective drug concentration that reaches the target, by reducing the reactivity of platinumbased drugs with GSH and other biologically active molecules in cells. As a result, many sterically hindered platinum complexes were designed and developed [6][7][8].
Picoplatin (ZD0473) is a sterically hindered antitumor compound that possesses unique chemical and DNA binding properties [9,10]. Due to the sterically hindered "pyridine structure," the reduced affinity of ZD0473 for sulfur ligands shows obvious DNA binding characteristics, and it can bypass the drug resistance mechanism of cisplatin and oxaliplatin-resistant tumor cell lines [11,12]. Also, ZD0473 hydrolyzes with more difficulty than cisplatin in solution, and the process of hydration and dissociation of picoplatin into the cell and the pathway of binding to guanine are shown in Figure 1. It is mainly hydrolyzed into two forms in vivo, including monofunctional water adduct with amino cis or trans. Then the monofunctional water adduct formed is hydrated and dissociated to form a bifunctional water adduct, which is further combined with two guanines to form the final adduct. The rate of hydrolysis of picoplatin is lower than that of cisplatin under the conditions of formation of the same amount of Pt-DNA adducts, due to the spatially-restricted group 2-methylpyridine [13,14]. Therefore, ZD0473 may reduce the drug resistance of cisplatin and its analogues to a certain extent. In clinical studies, it showed hematological toxicity, but no nephrotoxicity and neurotoxicity [13,15]. However, although picoplatin has the advantages of good tolerability and low toxicity and side effects, as well as oral activity, its efficacy is not as good as other clinical platinum drugs, resulting in its failure in phase III clinical trials [10,16]. Oral activity of 200-400 mg of picoplatin is only equivalent to the intravenous activity of 80 mg/m 2 [17].
Introduction of fluorine-containing atoms or groups into bioactive molecules may have great impact on their properties [18,19]. By converting the acidity, lipophilicity, and conformation deviation of the molecule, the properties of the synthesized molecules are changed. In addition, the metabolic mode can be altered by blocking its redox sites [20][21][22]. Since the Food and Drug Administration (FDA) first approved the fludrocortisone steroid in 1955, hundreds of fluoride-containing drugs have been introduced into clinical applications. In recent years, 40% of small molecule drug candidates entering Phase III clinical trials were reported to be fluorochemicals [23,24]. Current research on fluorine-containing platinum compounds showed that the introduction of fluorine atoms or fluorine-containing groups into platinum complexes can significantly enhance their antiproliferative activity, and some compounds even appear to reduce platinum drug resistance with a unique mechanism for exerting their antiproliferative activity [25,26].
Considering the good structural characteristics of picoplatin and the advantages of selective fluorination of bioactive molecules, five sterically hindered fluorine-containing platinum(II) complexes Y1-Y5 have been synthesized and spectrally characterized through 1 H NMR, 13 C NMR, 19 F NMR, IR, ESI-MS, and HPLC. Structures are shown in Figure 2. The change of water solubility of Y1-Y5 was studied by changing the methyl to the fluorine atom in the original 2-position pyridine structure. Biological activities of Y1-Y5 for in vitro antiproliferative activity was evaluated against A549, MCF-7, SW480, HCT116, HepG2, and LO2 cell lines, and the damaging effect on DNA was studied. Subsequently, we observed the apoptosis rate of several complexes by flow cytometry. We hope to improve the anticancer activity of the compounds by introducing fluorine atoms while retaining the superiority of picoplatin. Dissolved cisplatin (300 g, 1 mol, 1 eq) and pure platinum (30 g, 0.15 mol, 0.15 eq) in solvent (1200 mL of HCl) were reacted in the dark at 80 C for 5 h to obtain orange precipitate. Then, cisplatin (120 g, 0.33 mol, 0.33 eq) was added to a mixture of ammonia and water (ammonia:water ¼ 3:5, 400 mL) and reacted at 60 C for 2 h to obtain a clarified solution. The above two reactants were mixed and reacted at room temperature for 2 h. At the end of the reaction, the obtained precipitate was washed with ice water, ethanol, and dried to obtain intermediate e (scheme 1a). K 2 PtCl 6 (100 g, 0.25 mol, 0.25 eq) and the above intermediate e were dissolved in 50 mL of water and reacted at 65 C for 3 h. After subsequent evaporation, a concentrated solution was obtained which was filtered to yield an orange solid. This crude product was dissolved in water (300 mL) and HCl (1.5 mL) to remove impurities. Subsequently, the resulting orange precipitate was washed with cold water, ethanol, and then dried. An orange solid K 2 Pt(NH 3 )Cl 3 (126.70 g) was obtained.

Log P (o/w) determination
The lipid-water partition coefficient plays a necessary role in exploring the mechanism of drugs. We tested the log P (o/w) of the complexes by octanol-water shake flask experiments and a UV spectrophotometer. First, equal volumes of n-octanol and water were placed in a container for 12 h on a shaker at 37 C. Then, Y1-Y5 and picoplatin at known concentrations were added to the above solution and agitated for another 48 h. The final liquid was centrifuged and two-phase of the UV absorption coefficient was measured using a UV spectrophotometer. Then, the log P-value was calculated according to Equation (1): In order to maintain the accuracy of the experiment, each group of data was recorded three times.

Preparation of cell culture
Representative cancer cell lines and normal cell lines were selected: A549 (lung cancer cells), SW480 (colorectal cancer cells), HCT116 (colorectal cancer cells), MCF-7 (breast cancer cells), HepG2 (liver cancer cells), as well as LO 2 (normal hepatocytes), and these cells were inoculated in DMEM medium with 10% heat-inactivated fetal bovine serum and 100 U mL À1 penicillin. They were incubated overnight at 37 C in a damp environment (incubator with 5% CO 2 ).

MTT assay
Antiproliferative activities of Y1-Y5 and positive controls: cisplatin and picoplatin against the cancer cell lines (A549, SW480, HCT116, MCF-7, and HepG2) and normal cell line (LO2) were evaluated by MTT assay. First, tumor cells of the logarithmic growth phase were taken for the experiment and inoculated in 96-well cell culture plates, and the cell density was adjusted so that there were about 2 Â 10 4 cells per well. Then, Y1-Y5 and the positive controls cisplatin and picoplatin were dissolved in DMF and diluted with medium to bring the concentration of DMF below 0.2%. The final volume of each well was 200 lL, and the cells were incubated for 48 h. Subsequently, MTT reagent (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was added to the wells, incubated for 4 h, the supernatant cleared away, and the purple soild was dissolved with addition of 200 lL DMSO. The absorbance value (OD) of each well at 490 nm was measured by a multifunctional enzyme marker, and the inhibition rate was calculated and evaluated for its in vitro antiproliferative activity based on the IC 50 value of the complexes. Each test was repeated three times. The half-maximal inhibitory concentrations (IC 50 ) were obtained by drawing cell viability (%) against complex concentration (mM).

Fluorescence emission quenching
The solution (DNA-EB) of EB (10 lM) and CT-DNA (100 lM) was prepared in Tris-HCl buffer (5 mM Tris-HCl, 50 mM NaCl, pH 7.4). The mixture was incubated with Y1-Y5 and picoplatin with different concentrations (0, 5, 10, 15, 20, 25, and 30 lM) at 25 C for 30 min in the dark. By an excitation at 490 nm, the fluorescence spectra of EB at 550-750 nm were recorded. Then, the Stern-Volmer equation (4) was used to determine the quenching constant (K sv ) of the complexes,

Cell apoptosis assay
The apoptosis assays of A549 with different formulations were assessed via a FITC Annexin V apoptosis detection kit. After culturing (3 Â 10 5 /well) in six-well plates overnight, the cells were divided into six groups, five of which were incubated with Y1-Y5 and one of which was incubated with picoplatin for 24 h. After that the cells were washed with cold water, kept at the binding buffer, and dye with Annexin V-FITC and PI. The samples were analyzed via flow cytometry.

Synthesis and characterization
The synthesis route of Y1-Y5 was designed by referring to the experience of many reports [27][28][29]. The main strategy of synthesizing the target platinum complexes Y1-Y5 is based on retaining the active NH 3 group with 2-fluoropyridine introduced to replace 2-methylpyridine in picoplatin as a new sterically hindered group. Subsequently, different leaving groups were selected to improve the physicochemical properties of the drugs. In the beginning, K 2 Pt(NH 3 )Cl 3 and fluorine-containing intermediate A (A ¼ Pt(NH 3 )(C 5 H 4 NF)(Cl)(I)) were synthesized, which are key to the preparation of fluorine-containing sterically hindered picoplatin analogs. Their synthesis routes are shown in Scheme 1. First, the synthetic route and experimental operation of K 2 Pt(NH 3 )Cl 3 mainly consists of four steps. In the process, the amount of concentrated hydrochloric acid and the cooling time should be strictly controlled to ensure the purity and high yield of the product (73.0%). Then, the prepared K 2 Pt(NH 3 )Cl 3 was used as the starting material to react with potassium iodide (KI), and the product was coordinated with 2-fluoropyridine to prepare the corresponding fluorine-containing intermediate A. The reaction had mild conditions and was easy to control. The purity and yield of the product were high with the yield reaching more than 90%.
Subsequently, two main methods were used for the synthesis of targeted platinum complexes Y1-Y5. The targeted sterically hindered fluorine-containing platinum complexes Y1-Y5 were characterized by 1 H NMR, 13 C NMR, IR, ESI-MS, and HPLC (some data are displayed in the Supplementary Material). From the 1 H NMR data, all complexes had chemical shift values between 4.0-4.5 ppm and 7.0-9.0 ppm. Due to the introduction of fluorine atoms in the pyridine ring, H-F coupling occurs in the aromatic ring part (7.0-9.0 ppm) of Y1-Y5, and this portion is subjected to amplification. The ligand parts of the complexes are consistent, so Y1 will be an example. In 1 H NMR of Y1, the triple peak at 7.51 should be attributed to proton e, which is mainly coupled by ortho fluorine atom and hydrogen atom, with coupling constant of 6.6 Hz. The double peak at 7.59 should be attributed to proton c, whose coupling constant is 8.4 Hz due to the coupling of adjacent hydrogen atoms. The quadruple peak at 8.26 is the most complex, which should be attributed to proton d because of the adjacent coupling of hydrogen atom and the interatomic coupling of fluorine atom, dividing into multiple peaks. The double peak at 8.66 should be attributed to proton a, whose coupling constant is 5.7 Hz due to coupling of adjacent hydrogen atoms. It can be seen from the results that the adjacent coupling of F-H and H-H is similar, but the remote coupling is significantly different. For example, the different splitting of protons c and d show that the H-H remote coupling of proton c is far weaker than that of proton d. The other complexes (Y2-Y5) showed similar results with Y1. For Y1-Y5, the peak of NH 3 ligand is mainly in the chemical shift range 4.0-5.0 ppm. The difference is that for Y3-Y5, there are hydrogen peaks in the leaving groups-dicarboxylate at 1.0-3.0 ppm. The number of protons obtained in the 1 H NMR spectra of Y1-Y5 is consistent with the structural characteristics of the complexes, which proves that Y1-Y5 are synthesized. After analyzing the 13 C NMR data, the peak positions of all carbon atoms correspond to the number and position of carbon atoms in the synthesized complexes, which is consistent with the 1 H NMR data. Similarly, for the 19 F NMR data, the F atom shift of Y1-Y5 occurred between À60 and À65 ppm, thereby proving the introduction of fluorine atoms at the 2-position of fluoropyridine. For the ESI-MS data, Y1-Y5 showed [M þ Na] þ peaks consistent with their exact molecular mass. In the IR spectra, Y1-Y5 showed lower absorption bands of m NH and d NH than those of free ammonia, indicating that the ammonia is coordinated to platinum. The new band at 520-781 cm À1 was assigned to the Pt-N stretch. The new band at 1110-1271 cm À1 was designated as C-F stretching. Therefore, 2-fluoropyridine ligands have been coordinated with platinum(II). Because of the absorption of Pt-Cl at 789 cm À1 , there is a chloride leaving group in Y1. Compared with the free carboxyl group, the absorption of C ¼ O was red-shifted in IR spectra of Y2-Y5, indicating that carboxylate is coordinated with platinum(II) ion. The appearance of the m(Pt-O) peak also further proved that the carboxylate group is coordinated to the Pt center through the oxygen atom. The above results all proved the synthesis of Y1-Y5. The HPLC data reflect the purity of Y1-Y5.

Log P (o/w) determination
A good estimate for platinum complexes to cumulate in cancer cells is their lipophilicity, which is normally indicated through the octanol-water partition modulus, log P. Generally, the log P value trend of picoplatin and Y1-Y5 is determined by the traditional shake flask method [30]. The log P values of Y1-Y5 are shown in Figure 3. The lipophilicity of Y2-Y4 was poor with the experimental log P (o/w) ¼ À0.12 ± 0.06, À0.26 ± 0.03, and À0.36 ± 0.05, respectively, which was close to that of picoplatin (log P (o/w) ¼ À0.36 ± 0.05). This is because oxalate, malonic acid, 1,1-cyclobutanedicarboxylate, and other dicarboxylic acid salts replace chlorides, which can improve the stability of platinum complexes, but their physical and chemical properties will change, thus leading to the improvement of water solubility. Instead, the experimental log P (o/w) values of Y1 and Y5 were 0.07 ± 0.02 and 0.59 ± 0.03. This result proves that the introduction of fluoride ligands can improve the lipophilicity of the complexes to a certain extent, and further improve the lipophilicity due to the formation of the seven-membered chelate ring with succinic dicarboxylate.

In vitro antiproliferative activity
The antiproliferative activity in vitro of these sterically hindered fluorine-containing platinum complexes (Y1-Y5) against five cancer cell lines (A549, MCF-7, SW480, HCT116, and HepG2) and one normal cell line (LO2) were tested by MTT assay. The half-maximal inhibitory concentrations (IC 50 ) of these compounds are shown in Table 1. Compounds Y1-Y5 displayed favorable antiproliferative activity against the tested cancer cell lines and were selective for normal cells. Thus, introduction of the sterically hindered group on the amine carrier ligand has an advantageous influence on the activity of the targeted complexes.
Complex Y1 showed significant antiproliferative activity against four cancer cell lines; its antiproliferative activity in A549 was better than cisplatin and picoplatin. For A549 cells, Y1 (IC 50 ¼ 3.1 ± 0.4 lM) was 2.1 times more potent than cisplatin (IC 50 ¼ 6.4 ± 0.2 lM); Y1 was 3.4 times as potent as picoplatin (IC 50 ¼ 10.6 ± 0.7 lM) against A549 cells. This may be attributed to Y1 having chlorides as leaving groups and the sterically hindered group 2-fluoropyridine ligands, which could increase the drug accumulation of Y1 in cancer cells, thereby exhibiting enhanced antiproliferative activity. The antiproliferative activity of Y5 (IC 50 ¼ 4.5 ± 0.8 lM) against A549 and SW480 cell lines was higher than that of picoplatin , with 1.5 times more activity than cisplatin in A549 cancer cells. This shows that in the structure of picoplatin as a new steric hindrance group, the existence of fluoropyridine and succinate can improve the fat solubility and stability of Y5, making it easier to be absorbed by A549 and SW480 cancer cells. Previous experiments proved that Y5 has good lipophilicity. The activities of Y2, Y3, and Y4 in the tested cancer cell lines were significantly lower than the positive controls cisplatin and picoplatin. When the chelating dicarboxylic acid replaces the chloride as the leaving group, although it is beneficial to improve the water solubility and stability of the compound to a certain extent, it may reduce the rate of hydrolysis of the compound and the rate of cross-linking with DNA, thereby reducing the antiproliferative activity.

UV-vis absorption spectroscopy
Platinum drugs can bind covalently or non-covalently to DNA or cause the fission of the DNA-helix, which eventually leads to cancer cell apoptosis by blocking DNA replication and transcription [31,32]. We investigated the binding capacity of Y1-Y5 to DNA by UV-vis spectroscopy titration in calf thymus-DNA (CT-DNA). Different concentrations of CT-DNA (0-50 mM) were added to constant concentration (15 mM) of Y1-Y5 and picoplatin in equal gradients. To ensure the accuracy of the test, an equal amount of CT-DNA was added to each solution of the complexes. The absorption titration measurements were carried out in a 1 cm path length quartz cuvette at room temperature; the absorption spectra are shown in Figures 4 and S32-S35. The absorbance   Table 2. The binding constants are Y5 >Y1 >Y3 >Y2 >picoplatin>Y4. Especially, Y1 and Y5 bind DNA more strongly than other complexes with binding constant values of 2.90 Â 10 4 and 2.98 Â 10 4 . Complexes Y1-Y5 can bind to DNA in vitro.

Fluorescence emission spectroscopy
Complexes Y1-Y5 are not fluorescent in solution or with adding of CT-DNA. The interaction with DNA cannot be directly determined by fluorescence spectroscopy. Ethidium bromide (EB) is a kind of nucleic acid dye and its fluorescence intensity is increased by strong intercalation interaction with CT-DNA. When the metal complexes were added to DNA solution, the effective binding sites for EB decreased, resulting in a decrease in fluorescence intensity [33]. Hence, an EB replacement test was performed to evaluate the ability of DNA to bind with the compounds. Both ethidium bromide (EB, 10 lM) and CT-DNA (CT-DNA, 100 lM) were mixed into buffer Tris-HCl (5 mM Tris, 50 mM NaCl, pH ¼ 7.2). The mixture was incubated with Y1-Y5 and picoplatin at various concentrations (0, 5, 10, 15, 20, 25, and 30 lM) at 25 C for 30 min in the dark. By excitation at 490 nm, the fluorescence spectra of EB were recorded at 510-750 nm. Figures 5 and S36-S39 show that when the concentration of the complex  in the solution of the DNA-EB system changes, the maximum emission band has a low photochromism and negligible red shift, which proves that Y1-Y5 bind to DNA. The inset Stern-Volmer plots show that the EB-DNA fluorescence emission quenching was consistent with the linear Stern-Volmer equation and the quenching might be due to the replacement of EB from EB-DNA conjugate on the addition of each complex. The K sv constants of Y1-Y5 (Table 2) were computed through the Stern-Volmer Equation (4) and the Stern-Volmer diagram. The lifetime value of the chromophore in the absence of the quencher is s 0 ¼ 23 Â 10 À9 s, and the calculated biomolecular quenching constants (K q ) are shown in Table 2. The K sv values were obtained from the equation for Y1-Y5, and the values are 0.60 Â 10 4 -2.16 Â 10 4 M À1 . Further, the K q values of these complexes were all larger than the limiting diffusion rate constant of biomolecules (2 Â 10 10 M À1 s À1 ), suggesting static quenching. Most importantly, the sequence of K sv and k q values for Y1-Y5 are also observed as Y1>Y5>Y3>picoplatin>Y2>Y4, indicating that Y1 has a stronger binding ability consistent with the above UV titration experiments. Overall, these competitive EB-DNA binding studies showed that the compounds act on CT-DNA due to the hypochromaticity of the fluorescence intensity upon the addition of the compounds.

Apoptosis and necrosis study
To verify the antiproliferative activities of Y1-Y5, we used AnnexinV-FITC/PI co-staining and flow cytometry to investigate the apoptosis of A549 cells treated with different formulations. The dot plot of annexin-V FITC/PI consists of four quadrants, Q1 (upper left, necrosis), Q2 (upper right, late apoptosis), Q3 (lower right, early apoptosis), and Q4 (lower left, viable cells) [34]. The A549 cells were dealt with IC 50 concentrations of Y1-Y5 for 24 h; the degree of apoptosis is presented in Figure 6. A population of Annexin-V positive cell lines (early and late apoptotic) are observed in the right quadrants. Among them, the percentage of early apoptosis of A549 cells incubated with Y1-Y5 was 3.43-6.09. However, the late apoptosis rate reached 23.5-38.9. Especially, Y1 and Y5 were much higher than picoplatin, 38.9 and 32.3, respectively. Overall, the apoptotic rates of Y1-Y5 and picoplatin reflect a similar trend to that of MTT analysis, and the apoptosis is mainly induced by late apoptosis.

Conclusion
We prepared five fluorine sterically hindered platinum(II) complexes, Y1-Y5. The log P (o/w) values show that the lipophilicity of Y1 and Y5 is much higher than that of picoplatin. In addition, the synthesized complexes have selectivity for the antiproliferative activity of the selected cell lines, which proves that the introduction of steric hindrance groups into amine carrier ligands has a significant effect on the potency of the target complexes. Complexes Y1 and Y5 showed more antiproliferative properties against A549 and SW480 cell lines, indicating that improvement of lipophilicity can increase intracellular accumulation and improve antiproliferative activity. As they contain a 2-fluoropyridine ligand with steric hindrance, they can combine with DNA in vitro, resulting in DNA damage. The results of flow cytometry were similar to those of the MTT assay, which proved that Y1-Y5 had the ability to induce apoptosis. According to the experimental data of UV absorption and fluorescence titration, the above argument is confirmed. Therefore, our studies may provide a new strategy to improve the medicinal effect of picoplatin.