The Application of CA and PCA to the Evaluation of Lipophilicity and Physicochemical Properties of Tetracyclic Diazaphenothiazine Derivatives

The subject of the study was 11 new synthetized tetracyclic diazaphenothiazine derivatives. Using thin-layer chromatography in a reverse phase system (RP-TLC), their RM0 lipophilicity parameter was determined. The mobile phase was composed of 0.2 M Tris buffer (pH = 7.4) and acetone (POCH S.A., Gliwice, Poland) in different concentrations. Using computer programs, based on different computational algorithms, theoretical values of lipophilicity (AClogP, ALOGP, ALOGPs, miLogP, MLOGP, XLOGP2, and XLOGP3) as well as molecular descriptors (molecular weight, volume of a molecule, dipole moment, polar surface, and energy of HOMO orbitals and LUMO orbitals) and parameters of biological activity: human intestinal absorption (HIA), plasma protein binding (PPB), and blood-brain barrier (BBB), were determined. The correlations between the experimental values of lipophilicity and theoretically calculated lipophilic values and also between experimental values of lipophilicity and values of physicochemical or biological properties were assessed. A certain relationship between structure and lipophilicity was found. On the other hand, the relationships between RM0 and physicochemical or biological properties were not statistically significant and therefore unusable. For all analysed values, an analysis of similarities and principal component analyses were also made. The obtained dendrograms for the analysis of lipophilicity and physicochemical and biological properties indicate the relationship between experimental values of lipophilicity and structure in the case of theoretical lipophilicity values only. PCA, on the other hand, showed that ALOGP, MLOGP, miLogP, and BBB and molar volume have the largest share in the description of the entire system. Distribution of compounds on the area of factors also indicates the connections between them related to their structure.


Introduction
Designing new compounds which could find the application as drugs (or their precursors) is an expensive and timeconsuming process. ere is ongoing research concerning a new solution that could limit the number of necessary syntheses and at the same time obtain the products with optimal physicochemical properties correlated with biological activity [1]. QSAR (quantitative structure-activity relationship) is the leading method used in obtaining new drugs. It describes the quantitative relationship between the necessary biological activity of the compound and its chemical structure [2,3]. To define the differences between a series of similar substances, the QSAR descriptors are used.
ey are physicochemical parameters that are an interpretation of the biological properties of the compounds investigated [4]. One of the most often used parameters in QSAR analysis is lipophilicity. It can be connected with all drug interaction phases in the body, i.e., pharmaceutical, pharmacokinetic, and pharmacodynamic phases [5]. In the pharmaceutical phase, it affects the form of the drug, the way it is administered, and the release in the body. Also, the processes of absorption and distribution of biologically active substances depend significantly on lipophilicity, which is the main factor determining the bioavailability of the drug, and therefore its solubility in bodily fluids and the ease with which it is transported through biological membranes. Moreover, the lipophilic properties of the drug substance affect the way it interacts with the target receptor, and thus the pharmacological effect [6]. Lipophilicity determines the affinity of the molecule to the organic phase. It is expressed as the partition of substances in a two-phase system: liquid-liquid or liquid-solid; most often, lipophilicity is described by partition processes between polar (organic) and nonpolar (water) phases [7]. e traditional method of its determination is extraction in the octanol-water system [8,9]. Currently, the chromatographic methods in which the test substance undergoes partition in a dynamic system composed of the mobile and stationary phases are of great importance in the determination of lipophilicity parameters [10,11]. is method precisely and simultaneously determines many test substances, whose distribution coefficient can be in a wide range. In thin-layer chromatography and high-performance liquid chromatography, octadecylsilanized silica gel (RP-18) is used as the stationary phase, which to some extent imitates the structure of long-chain fatty acids in biological membranes. e mobile phase is usually the water-organic solution [11][12][13]. One of the groups of the drugs that are important from the point of view of medicinal chemistry is phenothiazine neuroleptics, known since the 1950s. Initially, these compounds were used in psychiatry as antipsychotics [14,15]. e basic structural unit of neuroleptic phenothiazine is a tricyclic system in which two benzene rings are connected by a nitrogen and sulfur atom to form a 1,4-thiazine ring, with an attached alkyl chain in the N-10 position [16]. It is a highly lipophilic system, which is related to the strong affinity of the phenothiazine molecule to the lipid bilayer of cell membranes of neurons and other lipid-rich tissues.
is property of phenothiazines allows them to penetrate the blood-brain barrier, which determines the mechanism of neuroleptic action of these compounds [17,18]. Modification of the basic tricyclic phenothiazine structure can be accomplished by introducing new pharmacophore substituents into the thiazine ring or benzene rings and by replacing the benzene rings with heterocyclic systems, e.g., pyridine, quinoline, or pyrimidine. ese types of structure transformations can lead to new compounds with a different direction and impact force [19,20]. Literature reports indicate a number of promising types of biological activity of classical phenothiazines and their new derivatives, including antibacterial [21][22][23][24], antifungal [25], antiprional [26], antiprotozoal [27,28], and antiviral [29,30]. In addition, the antitumor properties of phenothiazines and their ability to modify multidrug resistance of certain tumor cell lines have been described [31,32]. A new group of phenothiazine derivatives is tetracyclic quinobenzothiazine derivatives, which were obtained by replacing one of the benzene rings with a quinoline ring [33,34]. For a number of derivatives of this type, the lipophilicity parameters R M0 and log P TLC (theoretical and experimental) were determined [35]. Moreover, their interesting anticancer properties have been described in relation to human tumor cells of the MDA-MB-231 line (breast cancer), SNB-19 (brain glioma), and C-32 (cutaneous melanoma) [36]. Using the conversion of the benzene ring in the quinobenzothiazine system to the pyridine ring, a number of new compounds have been synthetized containing different substituents at various positions of the pyridine ring. e aim of this work is to determine the lipophilicity parameters (R M0 , log P TLC , and log P calc ) of 15 new pyridoquinothiazine salts by thin-layer chromatography in the reverse phase of RP-TLC and using computational methods and correlating them with each other. In addition, the R M0 lipophilicity parameter will be correlated with other molecular descriptors, and with theoretical values of biological activity.

Chemicals.
e series of pyridoquinothiazine derivatives, described by symbols 1-11, were investigated. e structures of compounds are shown in Figure 1.
Derivatives 1-11 were obtained through synthesis of thiochininantrenodiic chloride with a series of substituted aminopyridine derivatives. e methodology was described in our previous work [37]. e basic structure of the unsubstituted pyridoquinothiazine salt containing the nitrogen atom at position 8 of the tetracyclic system was modified by introducing various types of electron-withdrawing substituents and electron donors (Br, Cl, F, I, CH 3 , and OCH 3 ) in positions 9, 10, and 11 of the pyridine ring. e previously unsubstituted 11 derivative has a nitrogen atom at position 10 of the pyridoquinothiazine system. e compounds have an additional methyl group on the pyridine or imine nitrogen atom. e structure of all compounds was confirmed by ESI-HRMS spectrometry and 1 H, 13 C NMR spectroscopy, using two-dimensional techniques HSQC, HMBC, NOESY, and COSY.

in-Layer Chromatography.
e lipophilicity parameters were experimentally determined by reverse phase thinlayer chromatography (RP-TLC). Chromatograms were prepared on RP-18F 254s plates (1.05559.0001, Merck, Germany) precoated with nonpolar silicone oil. Plates were developed in glass chromatography chambers (Chromdes, Lublin, Poland) previously saturated with the vapour of the mobile phase. Solutions of pyridoquinothiazine salts 1-11 were prepared by dissolving 1 mg of particular compound in 2 ml of ethanol (POCH S.A., Gliwice, Poland). e solution was spotted into chromatographic plates in the amount of 2 μL by use of microcapillary. e mobile phase was composed of 0.2 M Tris buffer (pH � 7.4) and acetone (POCH S.A., Gliwice, Poland) in different concentrations, i.e., 50%, 60%, 70%, 80%, and 90%. e chromatograms were visualised in UV light (λ � 254 nm). Determination of the R F coefficient was carried twice for all compounds and in all acetone concentrations used. e final value of R F is the mean of the two measurements. e obtained R F coefficient was used to calculate the value of the R M parameter, according to the following equation: By extrapolating R M values to zero acetone concentration, a relative R M0 lipophilicity parameter was obtained, according to the following equation: where b is the slope and C is the concentration of acetone in mobile phase.

Computational Programs.
For pyridoquinothiazine derivatives 1-11, the theoretical values of the log P calc parameter were determined using computer modules whose calculations are based on atoms, molecular fragments, and molecular properties (AClogP, ALOGP, ALOGPs, miLogP, MLOGP, XLOGP2, and XLOGP3) [38,39]. e calculation methods used for obtaining the log P values were described earlier in our work concerning another group of new synthetized compounds [40]. Molecular descriptors (molecular weight, volume of a molecule, dipole moment, polar surface, and energy of HOMO orbitals and LUMO orbitals) were calculated using the DFT (density functional theory) method. A hybrid B3LYP function was used. Parameters of biological activity: human intestinal absorption (HIA), plasma protein binding (PPB), and blood-brain barrier (BBB), were determined using virtual computational bases [41].

Correlation, Cluster Analysis, and Principal Component
Analysis. Based on the values of experimental lipophilicity (R M0 , log P TLC ) and theoretical parameters (log P calc ), as well as the determined values of molecular descriptors and parameters of the predicted biological activity, correlation analysis, cluster analysis (CA), and principal component analysis (PCA) were performed. All data used for CA and PCA were standardised. e cluster analysis was based on the Euclidean distance, a single linkage method. e PCA analysis was based on the correlation matrix, using the Kaiser criterion and scree plot. e entire analysis was carried out using the Statistica 13.1 software.

Results and Discussion
e values of log P TLC , for compounds investigated, were calculated on the basis of known values of R M0 , which were obtained by chromatographic analysis. In order to obtain the values of the log P TLC parameter for the tested derivatives 1-11, first, the analysis of standard substances with the wellknown lipophilicity value (log P lit ) was performed. e analysis was carried out under the same chromatographic conditions as for the tested substances. e reference compounds analysed were acetanilide I, p-cresol II, p-bromoacetophenone III, benzophenone IV, and anthracene V. Values of lipophilicity parameters of reference substances: literature (log P lit ), experimental (R M0 ), and log P TLC , are presented in Table 1. Differences between the value of log P lit , and log P TLC for the standards did not exceed ±0.190 and were below ±0.150 for three compounds.
Using the log P lit relation from the experimentally determined R M0 parameters for the standards, a calibration curve was made. e linear function describing the calibration curve led to the formulation of the equation according to which log P TLC was determined for the tested derivatives: e R M0 parameter values and log P TLC values for derivatives 1-11 are shown in Table 2. e log P TLC parameters for the tested derivatives were within the range of 2.92-5.78. e collected results indicate the dependence of lipophilicity on the structure of the molecule as well as the presence and type of substituents at various positions of the pyridoquinothiazine system. Compounds with only an additional alkyl group: 11-CH3 (9, log P TLC : 4.19) and 9-OCH 3 (10, log P TLC : 4.22), were characterised by lower values of lipophilicity. e introduction of the halogen atom in the system containing the − 11 methyl group resulted in a significant increase in the log P TLC parameter in relation to the initial compound, as observed for the derivatives 7 (9-Cl, log P TLC : 5.72) and 8 (9-F, log P TCL : 5.78). A significant reduction in lipophilicity was noted for the unsubstituted derivative 1 (log P TLC : 3.53) and for the isomeric derivative 11 (log P TLC : 2.92) in which the nitrogen atom of the pyridine ring is located in position 10 of the diazaphenothiazine system.
One of the stages of the research was the determination of the log P parameter using computer methods. Depending on the mathematical module of the program used, the obtained values of the log P calc parameter occurred in a very wide range from − 1.16 to 4.70 (Table 3).
It was not possible to obtain a difference fewer than 0.5 units between the parameters calculated log P calc and experimental log P TLC for all tested compounds in any of the programs used. Differing values of results stem from the method of counting lipophilicity. e computer programs calculating the lipophilicity parameters are based on the structure of neutral molecules. It does not take into account the influence of conformation, tautomerisation, ionisation, changes in electron density, or the formation of hydrogen bondsor ion pairs through compounds investigated. Moreover, literature reports indicate a significant influence of changes in the thiazine ring conformation and changes in electron density at individual nitrogen atoms on the lipophilicity of quinobenzothiazine tetradenic derivatives [45].
In order to determine the relationship between experimental and theoretically calculated parameters, a correlation analysis was performed for the R M0 and log P calc parameters. In the studied group of derivatives, high and very high values of correlation coefficients were obtained for all computer modules used (r � 0.7555-0.9604, p < 0.05). Also, the log P values analysed correlated well with each other. With respect to these results, it can be stated that there is a relationship between the structure and the lipophilicity of the compounds tested. e correlation equations obtained are presented in Table 4.
As part of the work, attempts were made to correlate the lipophilicity parameters of derivatives 1-11 with the values of physicochemical properties such as the molar mass, volume of the molecule, dipole moment, polar surface, and HOMO-LUMO gap (Table 5).
In the group of analysed compounds, the best correlation of the R M0 parameter was obtained for the relation with the volume of the molecule (r � 0.8225). Statistically significant correlations were also obtained for relations with dipole moment (r � 0.6626) and with gap energy (r � 0.6496). Correlations with other physicochemical properties were statistically insignificant.
As mentioned earlier, lipophilicity has a significant impact on the behaviour of biologically active compounds in the human body, including their absorption when taken orally, binding to proteins in the bloodstream and penetration of the blood-brain barrier or blood-placenta barrier. erefore, it seemed appropriate to carry out a correlation study of the relative lipophilicity parameter R M0 with the calculated HIA parameter (human intestinal absorption), the PPB parameter (plasma protein binding), and the BBB parameter (blood-brain barrier-blood-brain penetration factor). e values of biological activity coefficients are presented in Table 6.
e results suggest very good oral bioavailability of all test compounds 1-11 (HIA > 96%) ( Table 6). e bloodbrain barrier penetration ratio in both groups is also high for newly obtained salts and is in the range of 89.51-97.28%, which classifies them as potential neuroleptic drugs. e highest BBB parameter value was calculated for derivative 10 containing a 9-methoxy group (BBB = 97.28%). e degree of protein binding is more diverse and depends on the structure of the compound being analysed. For salts 1-11, it ranged from 20.32 to 88.16%. e theoretical value of this coefficient affects the concentration of the free fraction of the compound in the bloodstream and thus its biological activity: the higher the PPB, the lower the biological activity. e lowest values of this parameter were noted for the isomeric derivative 11 (PPB = 20.32%). e correlation of the R M0 parameter with the calculated parameters of biological activity resulted in different regression coefficients. In the group of analyzed compounds, only for the relationship with the PPB parameter, a high positive statistically significant correlation was obtained (r � 0.7284). e remaining results indicate that there is no significant relationship between the analysed features; hence, correlation equations describing such relationships cannot be used to determine the R M0 parameter. Taking into account the above results, Table 7 presents correlation equations describing the relationships between lipophilicity values and physicochemical or biological parameters, but only those which are statistically significant.
For an additional description of the analysed compounds, the similarities were analysed, taking into account the lipophilicity values separately (experimental and theoretically calculated), values of physicochemical properties, and HIA, PPB, and BBB values. e obtained results are presented in the form of dendrograms. e first dendrogram (Figure 2) presents the similarity between analysed compounds based on their values of lipophilicity.
As a result of the analysis of similarities of lipophilicity values for the analysed compounds, several clusters are observed. Compounds 7 and 8 contain a hydrophobic methyl group at position 11. Moreover, they contain also the highly ellectronically accepting atoms at position 9, F in the case of compound 8 and Cl in the case of compound 7.  Compounds 5, 9, and 10 contain electronegative F atom at position 9 and a methyl or methoxy group. e compounds 2, 3, and 4 contain in their structure, in the 9 or 10 position, a Cl or Br atom. e last cluster is formed by compounds 1 and 11. ey do not contain any additional substituents in the pyridine ring. In this case, it can be assumed that the change in the position of the N atom from position 8 to 9 has no significant effect on lipophilicity. e second similarity analysis was done for values of physicochemical properties for all compounds investigated (Figure 3).

Journal of Analytical Methods in Chemistry
Based on the data analysis of the physicochemical properties, there is one large cluster containing all of the compounds tested, except for 6, 10, and 11. is may be due to the fact that compound 6 has an iodine atom with a large volume in its structure, whereas compound 10-methoxy group-and compound 11 have a nitrogen atom at position 10 of the pyridine ring. e remaining compounds show great similarity taking into consideration the physicochemical properties. e next similarity analysis was done for values of biological properties (HIA, PPB, and BBB) for all compounds investigated (Figure 4).
Several clusters were obtained as a result of the analysis. Compounds 2, 3, 4, and 7 contain a Cl or Br atom in positions 9 and 10. In addition, it turned out that compounds 10 and 11 are similar in biological properties, while compounds 1, 5, 8, and 9 are the most similar and form a separate cluster. It should also be emphasised that the compounds investigated show the largest similarity when only the lipophilicity (theoretical and experimental) values were taken into account. For them, the Euclidean distance was the smallest.
For a better description of the obtained experimental and theoretical data for the tested compounds, the principal components analysis (PCA) was also used. Eigenvalues were extracted based on all data. ree main components were selected using the Kaiser criterion, the value of which exceeds 1. ese selected components in 94.13% describe the variability of the system. However, when analysing the scree plot presented in Figure 5, it can be concluded that the main components should be 4. en, the variability of the system would be described in almost 97%.       analysed compounds, which was confirmed by previous analyses.
rough PCA analysis, the distribution of cases (compounds) on the area of factors can be presented (Figure 7).
It is clearly visible that the analysed compounds practically form one group, with the exception of compound 10.
It can most certainly be related to the structure of the compound, as compound 10 is the only one with a methoxy group in position 9 in its structure. e location of compounds 1 and 11 in one place may also indicate a connection with their structure because only they, from all tested compounds, have no substituents.

Conclusion
e subject of the study was new synthetized tetracyclic diazaphenothiazine derivatives. Using thin-layer chromatography in an inverse phase system (RP-TLC), the R M0 lipophilicity parameter for these was determined. Using computer programs, based on different computational algorithms, theoretical, mainly based on compound structure, lipophilicity values, as well as physicochemical and biological properties were determined. It can be concluded, through analysis of the obtained correlations between the experimental values of lipophilicity and the theoretically calculated lipophilic values, that there is a certain relationship between structure and lipophilicity. e relationships between R M0 and ALOGP and AClogP values are characterised by high values of correlation coefficients, 0.9604 and 0.9382, respectively. On the other hand, relationships between R M0 and physicochemical or biological properties were not statistically significant and therefore unusable. For all analysed values, an analysis of similarities and principal component analyses were also made. e obtained dendrograms for the analysis of lipophilicity and physicochemical and biological properties indicate the relationship between experimental values of lipophilicity and structure, but only in the case of theoretical lipophilicity values. PCA, on the other hand, showed that ALOGP, MLOGP, miLogP, and BBB and molar volume have the largest share in the description of the entire system. e distribution of compounds on the area of factors also indicates the connections between them related to their structure.

Data Availability
All chromatographic data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.