Molecular electrostatic potential and pattern recognition models to design potentially active pentamidine derivatives against Trypanosoma brucei rhodesiense

Molecular electrostatic potential (MEP) and pattern recognition (PR) were used to draw potentially active pentamidine derivatives against Trypanosome brucei rhodesiense ( T . b. rhodesiense ). PR models: Principal Component Analysis, PCA model; Hierarchical Cluster Analysis, Analogy, SIMCA model; and Stepwise Discriminant Analysis, SDA model, were built by reducing the dimensionality of a data matrix to twenty-eight pentamidine derivatives and allowed the compounds to be classified into two classes: more active and less active, according to their degrees of activity against T. b. rhodesiense. The study outlined that the properties HOMO (highest occupied molecular orbital) energy, VOL (molecular volume), and ASA_P (water accessible surface area of all polar (  qi  0. 2) atoms) are the most relevant for the construction of the models. The key structural features required for biological activity investigated through MEP were used as guidelines in the design of thirteen new compounds, which were evaluated by PR models as more active or less active against T. b. rhodesiense. The application of PR models indicated nine promising compounds (29, 30, 31, 32, 33, 36, 37, 39, and 40) for synthesis and


Introduction
Human African Trypanosomiasis (HAT) is endemic in 36 sub-Saharan countries: 24 of them experience Trypanosoma brucei gambienese (T. b. gambiense) transmission, leading to West African sleeping sickness, while in 13 countries Trypanosoma brucei rhodesiense (T. b. rhodesiense) is present, causing the most acute East African sleeping sickness syndrome, Uganda being the only country reporting both types (WHO, 2019;Selby, et al., 2019). HAT is maintained by an animal reservoir (Holmes, 2015;Franco, et al., 2020) and, currently, its treatment requires a long period of parietal administration of drugs of high toxicity and poor efficacy (Doleželovȧ, et al., 2018). Thus, the development of new drugs to treat sleeping sickness is necessary (Doleželovȧ, et al., 2018;WHO, 2019). Bakunova, et al., (2009) evaluated the diamidina (pentamidine) and some its derivatives in the standard screens used to identify its antiprotozoal drug candidates. Analysis of the structure-activity relationships showed high antitrypanosomal activity in (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 vitro.
The MEP (Bernardinelli, et al., 1994;Jefford, et al., 2000) and PR (Gangwal, et al., 2016) approaches have been used as independent strategies in the study of active compounds and in proposing new molecules for synthesis and biological testing (Bernardinelli, et al., 1994;Gangwal, et al., 2016). The combination of these two powerful tools to unravel the structure-activity relationship of bioactive compounds and, therefore, propose new molecules was shown by our research group on the designing of potentially active compounds against human hepatocellular carcinoma HepG2, Plasmodium falciparum, and Trypanosoma cruzi, respectively (Barbosa, et al., 2011;Cristino, et al., 2012;Santos, et al., 2020).
In this article, we present a study of the application of MEP and PR models for the investigation of pentamidine and derivatives with activity against T. b. rhodesiense reported in literature (Bakunova, et al., 2009), and the designing of new potentially active derivatives. In the first step of our study, MEP maps (Mehmood, et al., 2018;Aray, 2019;Rzesikowska, et al., 2019) were used in an attempt to identify key structural features of the compounds that are necessary for their activities, to investigate their probable interactions with a receptor through recognition in a biological process, and to propose the new derivatives. PR approaches (Varmuza, 1980;Varmuza, 2018;Beebe, et al., 1998) are used to build the models that will be applied later to the forecast set obtained with the MEP studies.

Molecular Electrostatic Potential in the Ligand-Receptor Biological Process
MEP is a very useful topological index in the study of hydrogen bond interactions as well as in the understanding of electrophilic and nucleophilic attack sites in a molecule obtained from electron density (Scrocco & Tomasi, 1979;. This property is also well suited for analyzing processes based on the "recognition" of one molecule by another, as in drug-receptor, and enzyme-substrate interactions (Scrocco & Tomasi, 1978;Politzer, et al., 1985). Because it is a physical property, MEP can be determined experimentally by diffraction or by computational studies (Politzer, et al., 2019). To investigate the reactive sites for pentamidine and its derivatives, the MEP was evaluated by using the HF/6-31G* approach (See Results and Discussion section). The MEP at a given point (x, y, z) in the vicinity of a molecule is defined in terms of the interaction energy between the electrical charge generated from the electrons and nuclei of the molecule and a positive charge test (a proton) located at ⃗. For the studied molecules, the MEP values were computed using Eq. 1 .
where N is the number of nuclei with charges Zi, located at position Ri and  ( ⃗) is the electronic charge density. The first term on the right side of Eq. (1) represents the contribution of the nuclei, which is positive; the second term brings in the effect of the electrons, which is negative.

PR approaches
PR approaches were introduced in the Chemistry in the 1970s to analyze various types of spectroscopic data (Kowalski & Brender, 1972). Since then, PR has been an excellent tool to aid in the interpretation of chemical data to obtain relevant information in different application sectors of chemical science (Varmuza, 1980;Brown, 2017;Varmuza, 2018).
In this section, we will make a brief presentation of the PR approaches used to build the models in this work. A deeper and detailed description of these matters can be found elsewhere (Varmuza, 1980;Varmuza, 2018;Beebe, Pell & Seasholtz, 1998). Research, Society andDevelopment, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 2.2.1 PCA, HCA, KNN, SIMCA, andSDA approaches PCA approach reduces the dimensionality of a data matrix consisting of a large number of interrelated variables, while retaining, as much as possible, the variation present in the data set. The final result of using the technique provides a small number of variables (molecular properties) that are believed to be best related to the dependent variable (Beebe, et al., 1998;Vidal, et al., 2016), in this case, biological activity against T.b. rhodesiense.
A two-dimensional dendrogram illustrating the fusions or divisions made at each successive step of the analysis represents the HCA approach. The purpose of using it is to display the data in such a way as to emphasize its natural clusters and patterns in this two-dimensional space. The similarity among the clusters is given by the length of their branches so that compounds presenting low similarity have long branches, whereas compounds of high similarity have short branches (Beebe, et al., 1998;Ghosal, et al., 2020).
The KNN approach (Beebe, et al., 1998) is based on the class dependency of an unknown sample. Classification by this method is related to the comparison of distance between samples after calculating multivariate Euclidean distances between each sample of the training set. The model is constructed and a set of tests to classify the sample into the predicted class, considering the multivariate distance from this sample to the K training set samples, is performed.
A model is constructed by means of the SIMCA approach for each class of the training set. In this method, a multidimensional box is built for each class, which means that shape and position of the samples in the classes are considered. The classification of a set sample is achieved by determining which space the sample occupies. It can be in the number of one, more than one or none of the classes (Beebe, et al., 1998).
The SDA approach (Johnson & Wichem, 1992;Zhang, et al., 2017;He, et al., 2020) allows you to separate samples from distinct populations and allocate new samples to previously defined populations. In this approach, at each step, the most powerful variable is inserted into the discriminant function. The approach is based on the F-test for the significance of the variables, that is, at each step, a variable is selected based on its significance and, after several steps; the most significant variables are extracted from the whole set investigated.

Pentamidine and studied derivatives
In Figures 1 and 2  Research, Society andDevelopment, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 The atomic numbering of the compounds is the same adopted in pentamidine. The studied derivatives, from the point of view of the relationship experimental structure-activity, are well reported and documented on paper from literature (Bakunova et al., 2009). Here is a summary of the conclusions reached by the researchers for T. b. rhodesiense: unsubstituted amides were more active in vitro compared to the N-substituted derivatives; the introduction of 2,2´-dichloro substituents further improved antitrypanosomal properties; and the replacement of oxygen atoms in the alkyl chain with secondary amino groups improved activities of the resulting congeners against T. b. rhodesiense. In this work, the derivatives molecules were defined as more active against T. b rhodesiense, when IC50 (M)  0.075, and as less active when IC50 (M)  0.075.

Molecular calculations
In the implementation of the calculations with pentamidine ( Figure 1a) and derivatives ( Figure 2) several theoretical approaches were used. For the quantum-chemical study, the methods available in the Gaussian program package (Frisch & Frisch, 1998) were used, namely: Austin model 1 (AM1) method (Dewar, et al., 1985), Parametric Method Number 3 (PM3) (Wu, et al., 2013), Hartree-Fock (HF) method (Roothaan, 1951), Becke's three-parameter hybrid method (Becke, 1993), the Lee-Yang-Parr (LYP) (Lee, et al., 1988), and standard basis sets (Hehre, et al., 1986) also available in the same program package. In pentamidine, geometry optimization was performed by AM1, PM3 approaches (Dewar, et al., 1985;Wu, et al., 2013) HF/3-21G*, HF/6-31G, HF/6-31G*, B3LYP/3-21G*, B3LYP/6-31G, and B3LYP/6-31G* approaches (Roothaan, 1951;Becke, 1993;Lee, et al., 1988) and basis sets (Hehre, et al., 1986). The calculations were carried out to find the approach and basis set that would present the best compromise between computational time and accuracy of the information relative to experimental data. The experimental structure of pentamidine was retrieved from the Cambridge Structural Database -CSD (Srikrishnan, et al., 2004). PCA and HCA approaches available in the Pirouette program (Infometrix, Inc, 2002) were used to compare the computed structures with different approaches of quantum chemistry with the experimental structure of pentamidine to identify the appropriate approach and the basis set for further calculations. The analyses were performed on an auto-scaled data matrix with dimension 57x9, where each row was related to 56 computed geometries and to 1 experimental geometry, and each column represented one of the 9 geometrical parameters of the pentamidine (bond lengths and bond angles). In order to compute all structures and to perform calculations to obtain the Research, Society and Development, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 molecular properties, the HF/6-31G* approach was selected (See Results and Discussion section), and the initial geometries of the derivatives (1-28) were built with the optimized geometry of the pentamidine selected by PCA and HCA approaches and subjected to the conformational search through optimization.

Molecular descriptors
In this study, all properties were obtained in the most stable conformation of each molecule and the descriptors were computed to give information about the influence of electronic, steric, hydrophilic, and hydrophobic features on the antitrypanosomal activity of the studied derivatives. Quantum-chemical descriptors such as total energy of molecules (TE), highest occupied molecular orbital (HOMO) energy, one-level below to highest occupied molecular orbital (HOMO-1) energy, lowest occupied molecular orbital (LUMO) energy, one-level about lowest occupied molecular (LUMO+1) energy, total dipole moment (), electronegativity of Mulliken (), atomic charges on the Nth atom (QN), molecular hardness (HD) and molecular softness (MS) were computed with the Gaussian 98 program (Frisch & Frisch, 1998). The QN descriptor was derived from the electrostatic potential as implemented in the Gaussian program. The electrostatic potential is obtained through the calculation of a set of punctual atomic charges so that it represents the possible best quantum molecular electrostatic potential for a set of points defined around the molecule (Williams & Yan, 1998;Chirlian & Francl, 1987). The charges derived from electrostatic potential present the advantage of being, in general, physically more satisfactory than the Mülliken charges (Singh & Kollman, 1984), especially regarding biological activity. Physicochemical descriptors such as total surface area (TSA), VOL, molecular refractivity (MR), molecule hydration energy (MHE), and coefficient of molecular partition octanol-water (log P) were obtained with the program ChemPlus (Hyperchem, Inc, 2008). In order to represent different sources of chemical information in terms of size, symmetry, and distribution of atoms in the molecule we have also included molecular holistic (MH) descriptors and these descriptors were obtained with the Dragon program (Todeschini & Consonni, 2009;Grisoni, et al., 2018).

Quantum chemical approach and basis set selection for the description of the geometry of pentamidine and derivatives
There is an advantage to using PCA and HCA at this stage of the study: all structural information is considered simultaneously and the correlations between them are considered. Table 1 shows the theoretical and experimental structural information (bond lengths and bond angles) of the geometry of the pentamidine. It was used with the aim to select using of PCA approach and HCA approach, which quantum-chemical approach and basis set give results closet to the experimental data (Srikrishnan, et al., 2004). (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207  (Srikrishnan, et al., 2004). Source: Authors (2020).
The first two principal components retain 96.47% of the true information as follows: PC1 = 64.79 and PC2 = 28.68%. The Research, Society and Development, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 PC1 versus PC2 scores plot is shown in Figure 3a, from which it can be seen that the methods are discriminated into two classes, according to PC1. The semi-empirical approaches (AM1 and PM3) are on the right side, while the other theoretical (HF and B3LYP) approaches and experimental data are on the left side. Moreover, it is noticed that the HF/3-21G* and HF/6-31G* approaches/basis sets are the closest to the experimental data, indicating that either of them could be used in the development of this study. Also, so as to investigate the most appropriate approach and basis set for further calculations, we used HCA approach. Figure 3b shows the dendrogram obtained with complete linkage method, in this figure one can conclude that the theoretical approaches are distributed in a similar way as in PCA, i.e., HCA confirmed the PCA results. Moreover, we can observe that the HF/6-31G* approach/basis set is closer to the experimental data being, therefore, the most suitable to carry out this study. Figure 1a shows the 2D molecular structure of pentamidine, a bisbenzamidine with two aromatic nuclei and one unbranched saturated five-carbon chain inserted between two phenolic ether oxygens. Electronically, the molecule exhibits nonbinding electron pairs in amidine groups, leading to the formation of a resonance effect between carbon and nitrogen atoms,  electrons in aromatic nuclei, and non-binding electron pairs in the atoms of oxygen from phenol ethers. We can still identify in the pentamidine structure two hydrophilic regions and a hydrophobic region. As one can see, these structural and electronic features presented by the molecule seem to be relevant in any biological recognition processes. Figure 1b shows the MEP map of the Research, Society and Development, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 pentamidine. In this figure, we notice that the molecule has contour surface, in the benzene rings,  electrons (yellow color), in the region between them, associated with the oxygen lone pairs (yellow and red colors), and at the ends of the molecular structure, associated with the lone pairs in amidine groups (yellow and red colors), corresponding to the regions indicated in Figure 1a, related to negative MEP, on which the lowest value for the electronic density is -51.88 kcal.mol -1 (red color), indicating an electronic density concentration due to the lone pairs of electrons in the oxygen (O1 and O2) and nitrogen (N1 and N3), also the difference of electronegativity between N1 and N3, relative to carbon, provides the displacement of the bonding electrons to the proximity of these atoms. From MEP map, it would be predictable that an electrophile attack can occur in these negative regions of ( ⃗)), and that they may be important in a possible biological recognition process, confirming the structural and electronic features presented by the molecular in Figure 1a. Importantly, the key to planning new active derivatives: the inclusion of substituents at positions C2, C16, O1, and O2 must maintain the structural and electronic features presented by the molecular structure in Figure 1a and evidenced by values of MEP in Figure 1b. The molecule also shows surface contour characterized by positive MEP (blue color), whose highest value corresponds the +39.01 kcal.mol -1 (blue color). It is expected that the positive regions de ( ⃗) will be preferred for nucleophilic processes.

MEP of the derivatives
In Figure 4, the MEP maps for four studied derivatives (3 and 28 more active and 11 and 26 less active) are shown. In this figure, we can notice that the derivatives have similar MEP to pentamidine (Figure 1b) in regions of rings, between the rings and at the end of the molecular structures. It may also be noted that the inclusion of groups with electron-withdrawing function in the ring or the replacement of the oxygen atoms by sulfur atoms in the chain provides greater distribution of electron density in the molecule and increase the anti-T.
b. rhodesiense activity of the derivatives. The contour surfaces corresponding to the lower MEP (red color), ( ⃗) = -56.16 kcal.mol -1 , and higher electron density, yet are located around the oxygen or sulfur and nitrogen atoms. The lower values for MEP also stem from the existence of isolated electron pairs in atoms. Still, from MEP maps, it would be predictable that an electrophilic attack can occur in these regions (red color), and that they may be important in a possible biological recognition process. On the other hand, Figure 4 shows a positive region with ( ⃗) =+43.42 kcal.mol -1 (blue color) indicating that this site is probably susceptible to nucleophilic attacks.

PCA, HCA, KNN, SIMCA, and SDA models
To carry out the PR models, all variables were auto scaled as preprocessing so that they could be standardized, and they Research, Society and Development, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 could also have the same importance regarding the scale. The number of multivariate variables was reduced by using the Pearson coefficient (r). When two variables presented values of r  0.8, one of the two was excluded from the matrix a, randomly, since they, theoretically, describe the same property (Ferreira, et al., 2002), that is, they also have a high correlation with anti-T. b.
rhodesiense activity and only one of them is enough to be used as independent variable in the predictive models. Thus, properties that showed either the same values for most samples or a small correlation with the activity (r 3) were eliminated too.
Three molecular descriptors were selected for PCA model. The molecular descriptors (HOMO energy, VOL, and ASA_P), the experimental data of activity IC50 (µM) for anti-T. b. rhodesiense derivatives and correlations including all data for 28 compounds can be seen in Table 2.  Research, Society andDevelopment, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 The correlation between descriptors is less than 0.65. The first three principal components describe 100% of the true information for the 28 molecules. The PC1-PC2 (a) scores and (b) loadings plots and (c) dendrogram obtained with Hierarchical Cluster Analysis (HCA) for the separation of the derivatives into two classes: (+) more active and (-) less active against T. b. rhodesiense are shown in Figure 5. From Figure 5a, one can see that the derivatives are distributed into two distinct regions in PC1.
The more active derivatives are on the left side (3, 6, 8, 9, 12, 14, 17, 20, 22, 25, 27, and 28) and the less active are on the right side   (1, 2, 4, 5, 7, 10, 11, 13, 15, 16, 18, 19, 21, and 26). According to Figure 5b, the ASA_P descriptor is responsible for displaying more active derivatives on the left side, while the HOMO energy and VOL descriptors displayed less active analogs for the right side of this figure. Source: Authors (2020).  The results of the HCA model are displayed in the dendrogram in Figure 5c and are similar to those of PCA model. The pentamidine derivatives are fairly well grouped according to their activity. From this figure, one can see the two clusters: more active (3, 6, 8, 9, 14, 17, 20, 22-25, 27, 28) and less active (1, 2, 4, 5, 7, 10, 11, 13, 15, 16, 18, 19, 21, 26). Table 4 shows the results obtained with one (1NN) to four (4NN) nearest neighbors. To all results the percentage of correct information was 100%. We used the model 4NN because the larger the number of the nearest neighbors, the better reliability of the KNN method (Ferreira, 2015), and this was used for validation of the training set from Figure 2. Source: Authors (2020).
In this figure, one can see that SIMCA models have a good discriminating power, i.e., high interclass distances and high between classes residuals.
Research, Society and Development, v. 10, n. 12, e261101220207, 2021 (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 With the discriminant functions, Eqs 3 and 4, and the value of each descriptor for the compounds studied, the classification matrix for the training set in Table 6 is shown. According to this table, the derivatives were classified as more active and less active, with 100% certainty. The reliability of the SDA model was assessed by cross-validation testing using the leave-one-out method. In this approach, one compound is omitted of the data set and the classification functions are built based on the remaining compounds. Afterwards, the omitted compound is classified according to the classification functions generated. In the next step, the omitted compound is included, and a new compound is removed, and the procedure goes on until the last compound is removed. Table 7 shows a summary of the results obtained with the cross-validation model. Correct information (%) 100 100 Source: Authors (2020).
As HOMO energy, VOL, and ASA_P properties were selected in the PCA, HCA, KNN, SIMCA, and SDA models as most important variables to describe the antitrypanosomal activity, some considerations about them may be relevant to the understanding of the behavior of more active pentamidine derivatives. The frontier molecular orbital (FMO) theory states that the transition state formation in a chemical reaction is due to the interaction between the frontier orbitals (HOMO e LUMO) of reactive species.
According to the literature, these orbitals play a major role in governing many chemical and pharmacological processes and are also responsible for the formation of many charge-transfer complexes (Karelson, et al., 1996;Fukui, 1997;Bulat, et al., 2021).
From Table 2, one can see that the HOMO energy for more active derivatives must present lower values than the less active ones, similar to pentamidine. This is an indication that biological processes occur allowing the electronic interaction between the most active derivatives and an eventual biological target, producing an increase of the antitrypanosomal activity. For the hypothesis of ligand/receptor complex formation due to nucleophilic attack of biological target in the ligand, we can verify a reaction with electrons transfer from the HOMO of the ligand molecule to the LUMO of the biological target.
The VOL property represents steric interactions between compounds and the biological receptor. It is determined the transport characteristics of the molecules, such as intestinal absorption or blood-brain barrier penetration (Cruciani, et al., 2000). In Table 2, it can be seen that, in general, the more active derivatives present lower VOL than less active derivatives, similarly to pentamidine. This is an indication that biological processes occur allowing the steric interactions between the most active pentamidine derivatives and an eventual biological target, producing an increase of the antitrypanosomal activity.
The ASA is a hydrophilic property, by increasing the water-accessible surface of a molecule; one increases the drug's permeability to cells. The ASA_P property refers to the water accessible surface (in Å 2 ) area of all polar atoms (qi 0.2) using a probe radius of 1.4 Å for molecule of water. Let qi denote the partial charge of atom i. The ASA_P property depends on the storage of partial charges of the molecule and their conformation (Stanton & Jurs, 1990). The most active derivatives have hydrophilic regions constituted by the substituents -CNH2NH at the ends of their chains. These regions accessible to the solvent may interact with the biological receptor through hydrogen-bridged interactions. In Table 2 derivatives with larger surface areas of all polar (CCBY4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i12.20207 atoms accessible to water are more active than those with these smaller areas, similar to pentamidine. This is an indication that biological processes occur allowing the hydrophilic interactions between the most active derivatives and an eventual biological target, producing an increase of antitrypanosomal activity.
Knowing the performance of the five PR models (PCA, HCA, KNN, SIMCA and SDA) for 28 studied derivatives, we decided to apply them to a series of thirteen compounds (Figure 7) designed to maintain the key structural features that are necessary for their activities evidenced by the MEP maps of the derivatives of the training set. The thirteen molecules proposed for the study of prediction of activity were drawn with the help of one of the collaborators of this work, belonging to the research group in organic chemistry of the Federal University of Pará, Brazil, and for the most promising syntheses in progress. In the future, antitrypanosomal tests with the most promising derivatives can be used to validate our PR models.  Tables 8 and 9, respectively.  Source: Authors (2020).

MEP maps for the most promising compounds of the prediction set
In Figure 8, the MEP maps for four of the most promising compounds (30, 33, 37 and 40) of the prediction set are shown.

Concluding Remarks
MEP allowed us to identify the key structural features in pentamidine and derivatives that are necessary for their activities against T. b. rhodesiense. PR methods (PCA and HCA, KNN, SIMCA, and SDA) allowed the selection of variables (HOMO energy, VOL, and ASA_P) as the most important properties to describe the antitrypanosomal activity, to indicate the types of interaction between pentamidine and derivatives with a possible target in a biological process and it is interesting to notice that these properties represent three distinct classes of interactions between the molecules and a biological receptor: HOMO energy (electronic), VOL (steric), and ASA_P (hydrophilic). The key structural features of the compounds responsible for their biological activities, evidenced by the study with the MEP, were used to design 13 new pentamidine derivatives that, when submitted to the PR models, allowed us to infer that 9 of the proposed compounds (29, 30, 31, 32, 33, 36,37, 39, 40) present promising potential for syntheses and biological evaluation, what in the future can be used to validate our PR models.