Small Heterocyclic Ligands as Anticancer Agents: QSAR with a Model G-Quadruplex

G-quadruplexes (GQs) have become valid targets for anticancer studies in recent decades due to their multifaceted biological function. Herewith, we aim to quantify interactions of potential heterocyclic ligands (Ls) with model GQs. For seven 4-aminoquinazolines and three 2-heteroaryl perimidines, seven of this ten-membered group so far unknown, we use routine quantum chemical modeling. As shown in the literature, a preferred mode of interaction of heterocycles with cellular structures is stacking to exposable faces of G-quadruplexes. To exploit the energy of this interaction as a molecular descriptor and achieve the necessary chemical precision, we use state of the art large-scale density functional theory (DFT) calculations of stacked heterocycles to a GQ. Actually, the GQ has been simplified for the computation by stripping it off all pentose phosphate residues into a naked model of stacked guanine quartets. The described model thus becomes computable. The obtained heterocyclic ligand GQ.L stacking energies, that is, their GQ affinities, are the necessary ligand descriptors. Using the ligand biological inhibitory activities (IC50) on a human malignant melanoma A375 cell line, we obtain a good linear relationship between computed ligand stacking affinities to GQ, and experimental log (IC50) values. Based on the latter relationship, we discuss a putative mechanism of anticancer activity of heterocyclic ligands via stacking interactions with GQs and thereby controlling cell regulatory activity. This mechanism may tentatively be applied to other condensed five- and six-membered small heterocycles as well.


Introduction
Alkaloids and their chemical analogs have long been among the most popular and sought organic natural, laboratory, and industrial products for a leading reason-their beneficial physiological activity on human health [1]. Recently, their activity and applications have increasingly been related to their capability to interact with a particular category of nucleic acids (NAs)-the four-stranded G-quadruplexes [1]. While not directly involved in preserving and transferring genetic information, G-quadruplexes have been disclosed as decisive participants in a plethora of cellular processes such as NA biosynthesis, replication, transcription, oncogenesis, etc. Telomeres are known sites accumulating G-quadruplexes, which are essential to their functioning in cell reproduction, aging, genetic stability, and cancer. A G-quadruplex may inhibit telomerase activity, directly affecting cancer cells and primary tumors [2]. A G-quadruplex may dissociate telomere-binding proteins, thus leading to dysfunction and, finally, to apoptosis or senescence [3]. A G-quadruplex interferes with telomeric replication by impairing replication fork progression [4]. Thus, knowledge of ligand structures stabilizing G-quadruplexes would allow for the specific design of heterocyclic systems targeting cancer cell function [1,5].

Synthesis
The studied ligands involve seven novel and three known compounds representing two groups of heterocycles, 4-aminoquinazolines and 2-heteroaryl perimidines, summarized in Table 1. Aminoquinazolines are synthesized from the parent 4-quinazolinone via a two-step protocol. The intermediate chlorides are obtained according to a literature procedure [22] and are used in the second step without purification to avoid decomposition. The conditions are optimized, and pure compounds are isolated in moderate to excellent yields depending on the amine reactivity. Perimidines are prepared from 1,8-diaminonaphthalene and a heteroaromatic aldehyde via a two-step, one-pot protocol [23]. The conditions are optimized, and pure compounds are isolated in moderate to excellent yields depending on the amine reactivity. Perimidines are prepared from 1,8-diaminonaphthalene and a heteroaromatic aldehyde via a two-step, one-pot protocol [24]. yields depending on the amine reactivity. Perimidines are prepared from 1,8-diaminonaphthalene and a heteroaromatic aldehyde via a two-step, one-pot protocol [24]. yields depending on the amine reactivity. Perimidines are prepared from 1,8-diaminonaphthalene and a heteroaromatic aldehyde via a two-step, one-pot protocol [24].

Biological Evaluation of the Effect of the Studied Heterocyclic Ligands
The cytotoxic activity of the studied heterocyclic ligands has been evaluated in vitro by assessing their cytotoxicity on melanoma cancer cells (A375 cell line). Cell viability has been estimated after 24 h of incubation with ligands. The cytotoxicity potential has been examined using several concentrations in the range 5-200 µg/mL. The obtained results for concentrations of 10 and 50 µg/mL. (Figure 1) have shown that the treatment of A375 cells with five of the ligands, 5,7,8,9, and 10, with the lower concentration of 10 µg/mL slightly decreased cellular viability (between 3% and 7%). In comparison, the higher concentration of 50 µg/mL has significantly suppressed cell viability-up to 84%. At the low concentration, two other ligands, 4 and 6, have reduced cell viability by 21 and 29%, whereas at the high concentration, the reduction has been 71 and 83%, respectively. Ligands 2 and 3 have induced 46% and 38% inhibition of cell viability at the low concentration, and 81% and 92% at the high concentration. The most potent cytotoxicity effect on A375 melanoma cells has been demonstrated by ligand 1 with an 81% reduction at the low concentration and 91% at the higher concentration ( Figure 1). The data obtained on cytotoxicity of the small heterocyclic ligands have been used for calculations of half-maximal inhibitory concentrations (IC 50 ) summarized in Table 1. The results in Table 1 show that 2-quinolinyl-quinazolines 1-3 are the most active ligands within the series studied. A comparison between quinazolines with identical 4-amino substituents, tetrahydroisoquinolinyl entry 1 vs. entry 5) and morpholinyl (entry 2 vs. entry 8), indicates that replacement of the 2-methyl substituent with 2-quinolinyl leads to a significant reduction in IC 50 , by approximately an order of magnitude. Therefore, it can be suggested that the aromatic substituent at the 2-position is essential for the activity of the particular ligands, possibly completing the preferred crescent ligand shape [8,10] and in line with the recent suggestion that quinoline residues would amplify G4 affinity to the corresponding ligand [14].

Computational Modeling
We have chosen the simplest G-quadruplex model, consisting of two guanine quartet sandwich layers and a single stabilizing potassium ion in between [8,10]. With these definitions, the ligand affinity has the simple form of where E QL , E Q , and E L are the computed total energies in vacuum for the quadruplexligand complex, free quadruplex, and free ligand, each completely optimized at the chosen theoretical level. An excerpt of the results is given in Table 1, and all data are summarized in Table S1. A plot of experimental IC 50 values against computed ligand affinities is shown in Figure 2.
where EQL, EQ, and EL are the computed total energies in vac and complex, free quadruplex, and free ligand, each comple theoretical level. An excerpt of the results is given in Table 1,  in Table S1. A plot of experimental IC50 values against comput in Figure 2.

Discussion
The clear trend revealed between experimental IC50 of st ands indicates, at first sight, the good likeliness of the sugge their interactions with G-quadruplexes. Thus, the stacking of molecules is probably a valid interaction mechanism, apart f action with somewhat larger anticancer ligands targeting tel tion does not eliminate different methods of attachment of sm ruplexes, let alone branched and macrocyclic ligands [14]. Th of finding the minima of potential energy surfaces for these solution, even from a purely mathematical viewpoint. Som may be found in the earlier observation that molecular dynam are relatively flat with deep global minima for bound ligands

Discussion
The clear trend revealed between experimental IC 50 of studied small heterocyclic ligands indicates, at first sight, the good likeliness of the suggested stacking mechanism of their interactions with G-quadruplexes. Thus, the stacking of relatively small heterocyclic molecules is probably a valid interaction mechanism, apart from known modes of interaction with somewhat larger anticancer ligands targeting telomeres [16,24]. This suggestion does not eliminate different methods of attachment of small heterocycles to G4 quadruplexes, let alone branched and macrocyclic ligands [14]. The multidimensional problem of finding the minima of potential energy surfaces for these interactions has no unique solution, even from a purely mathematical viewpoint. Some optimism in this direction may be found in the earlier observation that molecular dynamics potential energy surfaces are relatively flat with deep global minima for bound ligands [25,26]. We may then focus on the structural properties of small ligands and the variations of their quadruplex interaction energies elicited by ligand characteristics. A case of deviations of interaction energy may arise from internal structural variations of a given ligand-the possibility for tautomeric forms and rotational isomerism. Examples of this point are given by tautomers of 4-amino quinazoline, with heterocyclic substituents at N 4 , 2-pyridyl, and 8-quinolyl, as shown in Scheme 1.
The above intra-ligand processes may change the ligand affinities within a rang 0.5 to several kcal.mol −1 . More serious changes are possible in cases where the ligan non-planar and can attach with either its "concave" or "convex" side. At the same t the overall energy changes of the G4-L complex energy remain within a coupl kcal.mol −1 , these attachments may also induce changes in the shape of the G4-stack in complex, see Figure 3. These ligand's stacking and affinity variations may cause chan in the overall trend of proportionality of ligand affinities against quadruplex function expressed in the above affinity against IC50 relationship. In the specific case of N pyridyl and N4-8′-quinolyl substituents (entries 9 and 10 in Table 1), the compounds in fact, outliers to the generally linear relationship of affinities AL to activities, log (IC Figure 3. Complexes of model two-layered quadruplex Q2 with hetero-ligands, left: Q2 + 4; ce Q2 + 9, N3H tautomer; right: Q2 + 1. In addition, see Figure S1 for opposite conformation of 1. Scheme 1. Tautomerism and rotational isomerization of N4-heterosubstituted quinazolines. The above intra-ligand processes may change the ligand affinities within a range of 0.5 to several kcal·mol −1 . More serious changes are possible in cases where the ligand is non-planar and can attach with either its "concave" or "convex" side. At the same time, the overall energy changes of the G4-L complex energy remain within a couple of kcal·mol −1 , these attachments may also induce changes in the shape of the G4-stack in the complex, see Figure 3. These ligand's stacking and affinity variations may cause changes in the overall trend of proportionality of ligand affinities against quadruplex functioning, expressed in the above affinity against IC 50 relationship. In the specific case of N4-2 -pyridyl and N4-8 -quinolyl substituents (entries 9 and 10 in Table 1), the compounds are, in fact, outliers to the generally linear relationship of affinities A L to activities, log (IC 50 ). The above intra-ligand processes may change the ligand affinities within a range of 0.5 to several kcal.mol −1 . More serious changes are possible in cases where the ligand is non-planar and can attach with either its "concave" or "convex" side. At the same time, the overall energy changes of the G4-L complex energy remain within a couple of kcal.mol −1 , these attachments may also induce changes in the shape of the G4-stack in the complex, see Figure 3. These ligand's stacking and affinity variations may cause changes in the overall trend of proportionality of ligand affinities against quadruplex functioning, expressed in the above affinity against IC50 relationship. In the specific case of N4-2′pyridyl and N4-8′-quinolyl substituents (entries 9 and 10 in Table 1), the compounds are, in fact, outliers to the generally linear relationship of affinities AL to activities, log (IC50). The demonstrated variability of ligand-G-quadruplex model interactions certainly takes place in their interactions, in reality, thus bringing some scattering into computed The demonstrated variability of ligand-G-quadruplex model interactions certainly takes place in their interactions, in reality, thus bringing some scattering into computed ligand affinities, also expressed in deteriorating the correlation coefficient of the relationship. The latter deterioration of the correlation is even more pronounced with calculated RI-MP2 ligand affinities, only with R = 0.56, where the geometries of ligand complexes are not optimized at the used level of theory. Thus, the apparent suggestion from Figure 3 is that higher ligand affinity to G4 is associated with planar, crescent-like structures 1 to 4 as frequently noted in earlier work [8,10]; see also Figures 4 and S2. Here, we manage to quantify the intuitive trend into a structure-activity relationship at the wB97XD/6-31G(d,p) level of DFT theory [19,20] and remain convinced of the possibility of a more extensive selection of example heterocyclic molecules to yield better correlations of computed structural data, G4 stabilization affinities, against the experimental anticancer activity. ligand affinities, also expressed in deteriorating the correlation coefficient of the relationship. The latter deterioration of the correlation is even more pronounced with calculated RI-MP2 ligand affinities, only with R = 0.56, where the geometries of ligand complexes are not optimized at the used level of theory. Thus, the apparent suggestion from Figure 3 is that higher ligand affinity to G4 is associated with planar, crescent-like structures 1 to 4 as frequently noted in earlier work [8,10]; see also Figures 4 and S2. Here, we manage to quantify the intuitive trend into a structure-activity relationship at the wB97XD/6-31G(d,p) level of DFT theory [19,20] and remain convinced of the possibility of a more extensive selection of example heterocyclic molecules to yield better correlations of computed structural data, G4 stabilization affinities, against the experimental anticancer activity.

Synthesis
General: All reagents were purchased from Aldrich, Merck, and Fluka and used without further purification. The deuterated solvents were purchased from DeuteroGmbH. Fluka silica gel (TLC-cards 60778 with fluorescent indicator 254 nm) were used for TLC and Rf-values determination. Merck Silica gel 60 (0.040-0.063 mm) (Darmstadt, Germany) was used for flash chromatography purification of the products. The melting points were determined in capillary tubes on SRS MPA100 OptiMelt (Sunnyvale, CA, USA) automated melting point system with a heating rate of 1 °C per min. The NMR spectra were recorded on Bruker Avance II+ 600 or NEO 400 spectrometers (Rheinstetten, Germany) in an appropriate solvent; the chemical shifts were quoted in ppm in δ-values against tetramethylsilane (TMS) as an internal standard, and the coupling constants were calculated in Hz. The assignment of the signals is confirmed by applying two-dimensional COSY, NO-ESY, HSQC, and HMBC techniques. The spectra were processed with the Topspin 3.6 program. The mass spectra were recorded in positive mode on Q Exactive Plus Hybrid Quadrupole-Orbitrap Mass Spectrometer Thermo Scientific (ESI HR-MS). The spectra are processed with Xcalibur Free Style version 4.5 (Thermo Fisher Scientific Inc., Waltham, MA, United States.) program.

Synthesis
General: All reagents were purchased from Aldrich, Merck, and Fluka and used without further purification. The deuterated solvents were purchased from DeuteroGmbH. Fluka silica gel (TLC-cards 60778 with fluorescent indicator 254 nm) were used for TLC and R f -values determination. Merck Silica gel 60 (0.040-0.063 mm) (Darmstadt, Germany) was used for flash chromatography purification of the products. The melting points were determined in capillary tubes on SRS MPA100 OptiMelt (Sunnyvale, CA, USA) automated melting point system with a heating rate of 1 • C per min. The NMR spectra were recorded on Bruker Avance II+ 600 or NEO 400 spectrometers (Rheinstetten, Germany) in an appropriate solvent; the chemical shifts were quoted in ppm in δ-values against tetramethylsilane (TMS) as an internal standard, and the coupling constants were calculated in Hz. The assignment of the signals is confirmed by applying two-dimensional COSY, NOESY, HSQC, and HMBC techniques. The spectra were processed with the Topspin 3.6 program. The mass spectra were recorded in positive mode on Q Exactive Plus Hybrid Quadrupole-Orbitrap Mass Spectrometer Thermo Scientific (ESI HR-MS). The spectra are processed with Xcalibur Free Style version 4.5 (Thermo Fisher Scientific Inc., Waltham, MA, USA) program.
The studied ligands (Table 1) include seven novel (2-7 and 10) and three known (1, 8 and 9) compounds and can be divided into two series; aminoquinazolines and perimidines. Aminoquinazolines are synthesized from the corresponding quinazolinone via a two-step protocol. The intermediate chlorides are obtained according to a literature procedure [22,27,28] and are used in the second step without purification to avoid decomposition. The conditions are optimized, and pure compounds are isolated in moderate to excellent yields depending on the amine reactivity. Perimidines are prepared from 1,8-diaminonaphthalene and aromatic aldehyde via a two-step, one-pot protocol [23,29].

Synthesis of 4-Chloroquinazolines
4-chloro-2-methylquinazoline was prepared according to a known procedure [22] from commercially available 2-methyl quinazoline-4(3H)-one. To a solution of 2-methyl quinazoline-4-one (3 mmol) and Et 3 N (5 mmol) in benzene (15 mL), POCl 3 (4.5 mmol) was added, and the mixture was refluxed with stirring for 2.5 h. After cooling to room temperature, the reaction mixture was poured into ice water and was consecutively washed with aq. NaHCO 3 , brine, citric acid, brine, NaHCO 3 , and brine. The organic layer was dried over MgSO 4 and evaporated to dryness to give the crude product, which was further used without purification.
4-chloro-2-(2-quinolinyl)quinazoline was prepared via a two-step protocol: Step 1. A solution of anthranilamide (3 mmol), quinoline-2-carbaldehyde (3.3 mmol), and p-TsOH (0.15 mmol) in THF (25 mL) was stirred at room temperature for 2 h. Iodine (4.5 mmol) was then added, and the mixture was stirred at room temperature for 4 h. The products were partitioned between EtOAc and aq. Step 2. To a solution of 2-(2-quinolynyl)quinazoline-4-one (3 mmol) and Et 3 N (5 mmol) in benzene (15 mL) POCl 3 (4.5 mmol) was added, and the mixture was refluxed with stirring for 2.5 h. After cooling to room temperature, the reaction mixture was poured into ice water and was consecutively washed with aq. NaHCO 3 , brine, citric acid, brine, NaHCO 3 , and brine. The organic layer was dried over MgSO 4 and evaporated to dryness to give the crude product, which was further used without purification.

WST-1 Cell Proliferation Assay
WST-1 assay (Sigma-Aldrich Co., Darmstadt, Germany) was performed to assess the cytotoxicity of the heterocyclic ligands as previously described [30,31]. Briefly, the cells were seeded into 96-well plates at a density of 2 × 10 4 cells per well and incubated for 24 h at 37 • C and 5% CO 2 . On the following day, the culture medium was replaced with fresh medium, and the cells were exposed to increasing concentrations of the tested heterocyclic ligands for another 24 h. At the end of incubation, the cell medium was aspirated, and a new medium was provided. After that, the WST-1 reagent was added directly to the cells in a ratio of 1:10 according to the manufacturer's instructions. After 2 h incubation at 37 • C in the dark, the amount of the produced formazan by the cells was measured by absorbance at 450 nm using a standard microplate reader (Thermo Scientific Multiskan Spectrum, Waltham, MA, USA). The cell proliferation data were normalized to the percentage of the untreated control. The corresponding half-maximal inhibitory concentration (IC 50 ) values were calculated using GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA).

Computational Details
To obtain the necessary DFT [19,20] affinities of heterocyclic ligands to a model Gquadruplex, we use the hybrid long range and dispersion corrected wB97XD functional [32,33] at the 6-31G(d,p) basis set level, as implemented in the Gaussian 16 program system [34]. Default optimization criteria have been applied in Gaussian. Explicit electron correlated energies are calculated at the RI-MP2/6-31G(d,p) level using GAMESS-US [35], with an SVP auxiliary basis set [36] as single-point calculations at the optimized DFT geometries.

Conclusions
We have developed a model of the ligand with G-quadruplex interactions whereby the approximately planar heterocyclic ligand system is stacked to a plane of the quadruplex [16]. Quantum chemical DFT calculations indicate that computed ligand affinities to the G-quadruplex should correlate with ligand activities as anticancer agents. To verify this hypothesis, we have synthesized some 4-aminoquinazolines and 2-hetarylperimidines and have determined their anticancer activity quantitatively in the form of IC 50 . We have found a good linear relationship of theoretically computed DFT ligand affinities, A L , and log (IC 50 ). This quantitative structure-activity relationship, QSAR, provides a means for the design of novel small heterocyclic G4-ligands to be tested as anticancer agents under the described putative stacking mechanism of novel drug-like heterocycles [37,38] to G-quadruplexes.