Synthesis, Biological Evaluation and Stability Studies of Some Novel Aza-Acridine Aminoderivatives

Several new amino-substituted aza-acridine derivatives bearing a basic side chain have been designed and synthesized. The antiproliferative activity of the target compounds has been evaluated against three cancer cell lines—namely HCT-116 (colorectal), the uterine sarcoma MES-SA, and its doxorubicin-resistant variant MES-SA/Dx5. A limited number of the new acridines showed marginal cytotoxicity against the tested cell lines; nevertheless, these analogues possessed a similar substitution pattern. The moderate biological activity of these derivatives was attributed to their instability in aqueous media, which has been studied by mass spectrometry and computational chemistry experiments at the density functional level of theory (DFT).


Introduction
Acridine derivatives have been extensively studied due to their wide range of biological activities and pharmaceutical properties. Regarding their anticancer activities, substantial efforts have been directed at designing and synthesizing novel compounds with improved pharmacological and toxicological profiles. Acridines are characterized by a unique, semi-planar, and hydrophobic scaffold that interacts with diverse biomolecular targets. These properties are mainly attributed to the effectiveness of the π conjugated structure to intercalate between double-stranded DNA, thereby inhibiting DNA replication in the rapidly growing cancer cells [1][2][3][4], However, intercalation alone cannot fully support the cytotoxicity shown by acridine derivatives; thus, the inhibition of various DNA-related enzymes such as topoisomerases [1,5], telomerases [6], and poly(ADP-ribose)polymerases (PARPs) [7], is implicated in their mode of action.
It has been reported that the effectiveness of acridin-9-amines is related to their ability to exist in two tautomeric forms; the heterocyclic nitrogen adopts an acceptor or donor conformation, which has a radical effect on the binding properties of the molecule [8,9]. On the other hand, structure-activity relationship (SAR) research indicates that the substitution with a 7-methoxy group, and/or an electron-deficient nitro group are crucial for their activity and selectivity. Furthermore, activity relationship (SAR) research indicates that the substitution with a 7-methoxy group, and/or an electron-deficient nitro group are crucial for their activity and selectivity. Furthermore, the incorporation of a flexible dialkylaminoalkylamino-substituted side chain results in a noticeable increase in potency. The distance between the two amino groups is essential and the optimal length is equal to two or three methylene units. Interestingly, if the ability of the amino group to participate in hydrogen bonding is hindered, the antitumor activity is significantly reduced or demolished [10][11][12].
In order to further improve the pharmacodynamic and pharmacokinetic properties of acridines and develop more efficient analogues, several structural modifications have been reported. Many of these derivatives exhibited noticeable cytotoxic activity and many have been proved to be clinically effective. The most characteristic examples include pyrazoloacridines [10,[13][14][15], thiadiazinoacridines [12], triazoloacridones [16], and imidazoloacridones [17]; among them, the pyrazoloacridine PZA [13,14], an amino-substituted 5-nitropyrazolo [3,4,5-kl] acridine derivative, the imidazoloacridine C-1311 [11,18], the acridine carboxamide DACA [4], and the natural product acronycine [19,20] are of great interest (Figure 1). Although considerable progress has been made in this field, more work is needed to further improve the potency and selectivity of these compounds. In the course of our involvement in this area, we had previously studied several amino-substituted acridine derivatives with interesting anticancer properties [15,[21][22][23]. These compounds share common structural features with the compounds described above, bearing a basic side chain and a nitro group, and exhibit substantial cytotoxic activities against a panel of cancer cell lines, presumably due to DNA binding and intercalation.
In an effort to explore the optimal structural requirements for acridines, in order to express biological activity and since there are only limited examples of acridines with an additional nitrogen participating in the ring junction, we designed a number of new amino-substituted aza-acridines and present herein the study of their synthesis, biological evaluation and stability. The rationale for this modification was prompted by the correlation of these compounds with potent acridines. In the new derivatives, the orientation of the amino chain substitution is strictly defined due to the strong intramolecular hydrogen bond between the acridone carbonyl and the adjacent amino group. On the other hand, due to the ring-junction nitrogen, the resulting scaffold is locked in one of the tautomeric forms, in which the acridine nitrogen adopts only the acceptor conformation. This alteration could induce a radical effect on the binding properties of the molecule.

Chemistry
Commercially available methyl anthranilate (1) or methyl 5-methoxyanthranilate (2) were used as starting materials (Scheme 1). Treatment of esters 1 and 2 with 2,6-dichloro-3-nitropyridine (3) afforded only the desired methyl esters 4 and 5, respectively. The structure of these derivatives was elucidated using 1 H and 13 C NMR spectral data, using both direct and long-range homonuclear and heteronuclear correlation experiments (HMBC, HMQC, and NOESY sequences). Each one of the esters 4 and 5 was then reacted with the appropriately substituted N,N-dialkylethylenediamines or N,N-dialkylpropylenediamines providing the target amines 6-13. Mild saponification of the esters 6- Although considerable progress has been made in this field, more work is needed to further improve the potency and selectivity of these compounds. In the course of our involvement in this area, we had previously studied several amino-substituted acridine derivatives with interesting anticancer properties [15,[21][22][23]. These compounds share common structural features with the compounds described above, bearing a basic side chain and a nitro group, and exhibit substantial cytotoxic activities against a panel of cancer cell lines, presumably due to DNA binding and intercalation.
In an effort to explore the optimal structural requirements for acridines, in order to express biological activity and since there are only limited examples of acridines with an additional nitrogen participating in the ring junction, we designed a number of new amino-substituted aza-acridines and present herein the study of their synthesis, biological evaluation and stability. The rationale for this modification was prompted by the correlation of these compounds with potent acridines. In the new derivatives, the orientation of the amino chain substitution is strictly defined due to the strong intramolecular hydrogen bond between the acridone carbonyl and the adjacent amino group. On the other hand, due to the ring-junction nitrogen, the resulting scaffold is locked in one of the tautomeric forms, in which the acridine nitrogen adopts only the acceptor conformation. This alteration could induce a radical effect on the binding properties of the molecule.

Chemistry
Commercially available methyl anthranilate (1) or methyl 5-methoxyanthranilate (2) were used as starting materials (Scheme 1). Treatment of esters 1 and 2 with 2,6-dichloro-3-nitropyridine (3) afforded only the desired methyl esters 4 and 5, respectively. The structure of these derivatives was elucidated using 1 H and 13 C NMR spectral data, using both direct and long-range homonuclear and heteronuclear correlation experiments (HMBC, HMQC, and NOESY sequences). Each one of the esters 4 and 5 was then reacted with the appropriately substituted N,N-dialkylethylenediamines or N,N-dialkylpropylenediamines providing the target amines 6-13. Mild saponification of the esters 6-13 resulted in the carboxylic acids 14-21, respectively, which, without further purification, were Each compound's structure was unambiguously established by 1 H and 13 C NMR spectroscopy, using both direct and long-range heteronuclear correlation experiments. Structural discrimination resulted from the observation that H-8 of the aza-acridines 22-29 exhibits 4 J coupling with the carbonyl group, while in the case of esters 6-13 such a correlation is not possible. In general, the 1 H NMR spectra of the esters 6-13 as well as the aza-acridines 22-29 showed slight differences, especially Each compound's structure was unambiguously established by 1 H and 13 C NMR spectroscopy, using both direct and long-range heteronuclear correlation experiments. Structural discrimination resulted from the observation that H-8 of the aza-acridines 22-29 exhibits 4 J coupling with the carbonyl group, while in the case of esters 6-13 such a correlation is not possible. In general, the 1 H NMR spectra of the esters 6-13 as well as the aza-acridines 22-29 showed slight differences, especially in the

Biological Evaluation
The in vitro antiproliferative activities of the new aza-acridine aminoderivatives were evaluated against three human tumor cell lines-namely, HCT-116 (colorectal), the uterine sarcoma MES-SA and its variant MES-SA/Dx5, reported to be 100 times more resistant to doxorubicin (Dx) [24]. The results of the MTT dye reduction assay, expressed as 50% inhibitory concentrations (IC 50 ) in µM, are summarized in Table 1. For comparison, Dx and mitoxantrone (Mitox) were used as positive controls. Most of the new compounds showed low cytotoxic activity and only three of them, namely 26, 27 and 29, were detected to possess a certain degree of cytotoxic activity which is enhanced against the two uterine sarcoma cell lines (MES-SA and MES-SA/Dx5). Even so, it is noticeable that all three compounds possess the methoxy substitution, providing evidence that the presence of the methoxy group is favorable for the cytotoxicity of this class of compounds. This finding is consistent with previous observations [2,11,12] concerning the improved cytotoxic activity of methoxy-substituted analogues.
From a direct comparison of the cytotoxic activity against the Dx-sensitive (MES-SA) and Dx-resistant (MES-SA/Dx5) cell lines, it is observed that derivatives 22-25 exhibit equal cytotoxicity against both cell lines, as indicated by the relevant resistant factor (RF) values (~1.0). On the contrary, the methoxy-substituted compounds 26-29, proved to be more active against the Dx-resistant (MES-SA/Dx5) cell line and thus retain full antiproliferative activity against P-glycoprotein-overexpressing cells, with a relevant resistant factor of~0.5. It must be pointed that even if the overall activity of the derivatives is moderate, the potency of compounds 26-29 is enhanced against the two uterine sarcoma cell lines (MES-SA and MES-SA/Dx5). This probably indicates that the methoxy-substituted derivatives could overcome Multi Drug Resistance (MDR).
Overall, the activity of the new compounds was remarkably reduced when compared to the structurally related lead compounds ( Figure 1). This is most likely due to the insertion of the additional nitrogen atom, which blocks the tautomerism of the acridine scaffold and could also interfere with the strength of the intramolecular hydrogen bond. On the other hand, it has been reported that aza-acridines are amenable to ring-opening [25] and we considered that the diminished activity of the new aza-acridines could potentially be due to hydrolytic instability. A hydrolysis mechanism should involve the nucleophilic attack of H 2 O to the carbonyl group, with concomitant ring-opening resulting in their precursor acids (14-21, Scheme 2). Therefore, the observed antiproliferative activity could be attributed to a mixture of aza-acridines and ring-opened compounds. In order to verify our hypothesis, we performed a series of computational chemistry-and mass spectrometry-based hydrolysis experiments. Two compounds of the synthesized set of derivatives were selected as representative examples; i.e., 25 and its methoxy-substituted analogue 29, aiming to evaluate the effect of the methoxy group on the stability and therefore on the activity.

Mass Spectrometry
In order to reveal the stability of compounds 25 and 29, an initial estimation of the corresponding hydrolysis rates was performed. Both derivatives were found to be very unstable in an aqueous environment, albeit with different rates of hydrolysis, with the methoxy derivative 29 being far more stable. The results are presented in Figure 2. In order to assimilate the conditions used for the biological testing of the substances, the substances were analyzed at the 20 μg/mL level (~60 μΜ). It should be pointed out that the use of mass spectrometry was chosen since the low concentration could make detection through NMR spectroscopy difficult. Furthermore, such a concentration is expected not to induce any matrix effect. Due to the fact that the hydrolysis products (i.e., 17 and 21) afforded much cleaner spectra, the negative ion mode was employed. The kinetic equation used has been transformed to fit the data linearly. Thus, the equation In order to verify our hypothesis, we performed a series of computational chemistry-and mass spectrometry-based hydrolysis experiments. Two compounds of the synthesized set of derivatives were selected as representative examples; i.e., 25 and its methoxy-substituted analogue 29, aiming to evaluate the effect of the methoxy group on the stability and therefore on the activity.

Mass Spectrometry
In order to reveal the stability of compounds 25 and 29, an initial estimation of the corresponding hydrolysis rates was performed. Both derivatives were found to be very unstable in an aqueous environment, albeit with different rates of hydrolysis, with the methoxy derivative 29 being far more stable. The results are presented in Figure 2. antiproliferative activity could be attributed to a mixture of aza-acridines and ring-opened compounds.

Scheme 2. Hydrolysis of compounds 25 and 29.
In order to verify our hypothesis, we performed a series of computational chemistry-and mass spectrometry-based hydrolysis experiments. Two compounds of the synthesized set of derivatives were selected as representative examples; i.e., 25 and its methoxy-substituted analogue 29, aiming to evaluate the effect of the methoxy group on the stability and therefore on the activity.

Mass Spectrometry
In order to reveal the stability of compounds 25 and 29, an initial estimation of the corresponding hydrolysis rates was performed. Both derivatives were found to be very unstable in an aqueous environment, albeit with different rates of hydrolysis, with the methoxy derivative 29 being far more stable. The results are presented in Figure 2. In order to assimilate the conditions used for the biological testing of the substances, the substances were analyzed at the 20 μg/mL level (~60 μΜ). It should be pointed out that the use of mass spectrometry was chosen since the low concentration could make detection through NMR spectroscopy difficult. Furthermore, such a concentration is expected not to induce any matrix effect. Due to the fact that the hydrolysis products (i.e., 17 and 21) afforded much cleaner spectra, the negative ion mode was employed. The kinetic equation used has been transformed to fit the data linearly. Thus, the equation In order to assimilate the conditions used for the biological testing of the substances, the substances were analyzed at the 20 µg/mL level (~60 µM). It should be pointed out that the use of mass spectrometry was chosen since the low concentration could make detection through NMR spectroscopy difficult. Furthermore, such a concentration is expected not to induce any matrix effect. Due to the fact that the hydrolysis products (i.e., 17 and 21) afforded much cleaner spectra, the negative ion mode was employed. The kinetic equation used has been transformed to fit the data linearly. Thus, the equation used was where A 0 is the initial concentration of each molecule, B the concentration of the respective product after hydrolysis and k is the hydrolysis constant. The two hydrolysis reactions were evaluated as linear models, based on the equation shown above. The statistical evaluation is presented in Table 2. The results are consistent with a first-order degradation reaction, as the fitting employing this equation order is linear. Rapid degradation was observed for the non-methoxy derivative under the conditions examined, whereas substantial degradation was observed for the methoxy analogue. Comparing the two slopes, which are the degradation coefficients, it was found that the degradation of 25 is 26 times faster than the corresponding hydrolysis of the methoxy derivative 29. Assuming that the reactions are pseudo first-order the corresponding t1/2 values are 10.6 min −1 for 25 and 277 min −1 for 29.

Computational Chemistry
The two derivatives exhibit roughly the same geometry features. As the region of interest is located at the internal amide bond, the geometry features of the involved bonds are presented in Table 3. Both structures are planar, denoting a high aromaticity degree, whereas the bond lengths are essentially the same. Therefore, the difference in reactivity should be revealed by the electronic properties of the two molecules. Calculation of the electrostatic potential (ESP) (see Supplementary Materials, Figure S1) shows that the carbonyl C is depleted of electronic charge and therefore it is susceptible to an electrophile attack, which is consistent with the first step of hydrolysis (nucleophilic attack of water). Nevertheless, no appreciable difference has been observed between the two ESP surfaces, denoting that the differences calculated by the ESP methodology are too subtle to be detected. where A0 is the initial concentration of each molecule, B the concentration of the respective product after hydrolysis and k is the hydrolysis constant. The two hydrolysis reactions were evaluated as linear models, based on the equation shown above. The statistical evaluation is presented in Table 2.
The results are consistent with a first-order degradation reaction, as the fitting employing this equation order is linear. Rapid degradation was observed for the non-methoxy derivative under the conditions examined, whereas substantial degradation was observed for the methoxy analogue.
Comparing the two slopes, which are the degradation coefficients, it was found that the degradation of 25 is 26 times faster than the corresponding hydrolysis of the methoxy derivative 29. Assuming that the reactions are pseudo first-order the corresponding t1/2 values are 10.6 min −1 for 25 and 277 min −1 for 29.

Computational Chemistry
The two derivatives exhibit roughly the same geometry features. As the region of interest is located at the internal amide bond, the geometry features of the involved bonds are presented in Table 3. Both structures are planar, denoting a high aromaticity degree, whereas the bond lengths are essentially the same. Therefore, the difference in reactivity should be revealed by the electronic properties of the two molecules. Calculation of the electrostatic potential (ESP) (see Supplementary Materials, Figure S1) shows that the carbonyl C is depleted of electronic charge and therefore it is susceptible to an electrophile attack, which is consistent with the first step of hydrolysis (nucleophilic attack of water). Nevertheless, no appreciable difference has been observed between the two ESP surfaces, denoting that the differences calculated by the ESP methodology are too subtle to be detected. where A0 is the initial concentration of each molecule, B the concentration of the respective product after hydrolysis and k is the hydrolysis constant. The two hydrolysis reactions were evaluated as linear models, based on the equation shown above. The statistical evaluation is presented in Table 2.
The results are consistent with a first-order degradation reaction, as the fitting employing this equation order is linear. Rapid degradation was observed for the non-methoxy derivative under the conditions examined, whereas substantial degradation was observed for the methoxy analogue.
Comparing the two slopes, which are the degradation coefficients, it was found that the degradation of 25 is 26 times faster than the corresponding hydrolysis of the methoxy derivative 29. Assuming that the reactions are pseudo first-order the corresponding t1/2 values are 10.6 min −1 for 25 and 277 min −1 for 29.

Computational Chemistry
The two derivatives exhibit roughly the same geometry features. As the region of interest is located at the internal amide bond, the geometry features of the involved bonds are presented in Table 3. Both structures are planar, denoting a high aromaticity degree, whereas the bond lengths are essentially the same. Therefore, the difference in reactivity should be revealed by the electronic properties of the two molecules. Calculation of the electrostatic potential (ESP) (see Supplementary Materials, Figure S1) shows that the carbonyl C is depleted of electronic charge and therefore it is susceptible to an electrophile attack, which is consistent with the first step of hydrolysis (nucleophilic attack of water). Nevertheless, no appreciable difference has been observed between the two ESP surfaces, denoting that the differences calculated by the ESP methodology are too subtle to be detected.  The electronic charge on carbonyl C as well as that of the connected N belonging to the internal amide bond also act as a measure of reactivity of the two derivatives. The Merz-Kollmann (MK) ESP-fitting atomic charges have been employed to estimate the electron partitioning scheme and showed that carbonyl C of the derivatives is the second most electron deficient carbon of the molecule with a charge of 0.761, slightly less than the pyridine carbon that is connected to the amine group, which bears a charge of 0.763. The attachment of the methoxy group renders carbonyl C the most electrophilic site of the molecule with a charge of 0.828 with the aforementioned pyridine carbon having an atomic charge of 0.797. Furthermore, the bond order of the C-N bond could also help explain the reactivity difference. The Mayer bond order analysis scheme that has been employed showed that the bond order increased by nearly 0.07 to the methoxy-bearing molecule, rendering the derivative more stable and thus resistant to hydrolysis. The delocalization of the HOMO orbital over the molecular skeleton provides visual assessment of the bonding across the atoms ( Figure 3A,B). The spread of electronic density across the C-N bond of the internal amide in the case of the methoxy derivative indicates that the bond exhibits a higher overlap between the atomic orbitals (AOs) and therefore should be less prone to hydrolysis. The ELF function has been calculated for both molecules and the results can be visualized in Supplementary Figures S2A and S2B. In order to evaluate the concentration of electronic density the ELF value at the CP of the C-N bond has been assessed and was found to be 0.898 in both cases. Values larger than 0.7 indicate substantial π-character, but the ELF theory was not able to differentiate between the two molecules. Furthermore, the condensed Fukui function analysis was employed in order to pinpoint the electrophilic reactivity difference of the two carbonyl carbon atoms. The corresponding values were found to be 0.0207 and 0.0211 for the non-methoxy and methoxy derivative, respectively. The slight difference observed does not reflect their hydrolytic stability. Furthermore, as expected, their electrophilicity indices (e*eV) are also substantially similar (0.066 and 0.065, respectively). Interestingly, the condensed local softness (Hartree*e) for electrophilic attack is substantially different between the species. Thus, the corresponding values are 0.0874 for the non-methoxy derivative and 0.0933 for its methoxy counterpart. It is noteworthy that the computed local electrophilicity indexes (based on the Frozen Molecular Orbitals (FMOs) approximation) are also substantially different with values of 0.9608 and 0.7619 for the two molecules, but their order is opposite to the hydrolysis rate showing that this approach is not consistent with the experimental results in this case. Figure S3a,b shows the distribution of the Fukui function between the two molecules. Finally, Bader's QTAIM (quantum theory of atoms-in-molecules) theory has been employed in order to study the electronic density partitioning between atoms. The AIM charges were calculated as well as the electron density Rho at the bond critical point connecting the carbonyl C to the pyridine N. The values obtained are for the carbon and nitrogen-+0.991 and −0.911, respectively-whereas for the BCP (bond critical point) a value of +0.253 was obtained for the non-methoxylated molecule. The corresponding values were +0.997, −0.911 and +0.255 for the methoxy derivative. The partitioning scheme does not indicate appreciable differences between the two molecular species. Finally, in order to estimate the energy difference between each molecule and its hydrolyzed counterpart, a full geometry optimization with the same level of theory has been performed. The corresponding values are −568.95 kcal/mol for the non-methoxylated derivative and -616.91 kcal/mol for its hydrolyzed derivative leading to an energy difference of 47.96 kcal/mol. The corresponding values for the methoxy molecule are −640.81 and -688.76 kcal/mol, leading to an energy difference of 47.95 kcal/mol. These results show that the thermodynamic result of the reaction is negligible and thus could not be the driving force of the more extensive degradation of the methoxy-bearing molecule. Overall it seems that the differentiation of the carbonyl carbon electronic charge as also depicted by the shape of the HOMO orbital could explain the difference of the hydrolytic stability, whereas the aromatic character as calculated by the topological analysis is not affected to a large extent.

General Information
All commercially available chemicals and solvents were used as received without any further purification. Melting points were determined on a Büchi apparatus and are uncorrected. 1 H NMR spectra and 2D spectra were recorded on a Bruker Avance DRX 400 instrument (Bruker BioSpin GmbH, Rheinstetten, Germany), whereas 13 C NMR spectra were recorded on a Bruker AC 200 spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) in deuterated solvents and were referenced to TMS (δ scale). The signals of 1 H and 13 C spectra were unambiguously assigned by using 2D NMR techniques: 1 (4): NaH (460 mg, 11.5 mmol, 60% in mineral oil) was added to a solution of methyl-2-aminobenzoate (1.63 g, 10.78 mmol, 1) in dry THF (100 mL) at 0 °C under argon and the resulting suspension was stirred at room temperature for 30 min, followed by dropwise addition of a 2,6-dichloro-3-nitropyridine (2.26 g, 10.78 mmol, 3) solution in dry THF (10 mL). The reaction mixture was then stirred at 60 °C for 2 h; an excess of NaH was quenched with methanol and the solvents were vacuum evaporated. The residue was dissolved in CH2Cl2, washed with a 10% Na2CO3 solution and water, dried over Na2SO4, and the solvent was evaporated to dryness. Flash chromatography on silica gel, using a mixture of cyclohexane-CH2Cl2 3:1 as the eluent, afforded the title compound (2.22 g, 67 %) as an oil. 1

General Information
All commercially available chemicals and solvents were used as received without any further purification. Melting points were determined on a Büchi apparatus and are uncorrected. 1 H NMR spectra and 2D spectra were recorded on a Bruker Avance DRX 400 instrument (Bruker BioSpin GmbH, Rheinstetten, Germany), whereas 13 C NMR spectra were recorded on a Bruker AC 200 spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) in deuterated solvents and were referenced to TMS (δ scale). The signals of 1 H and 13 C spectra were unambiguously assigned by using 2D NMR techniques: 1