Synthesis, Molecular Docking, and Antimycotic Evaluation of Some 3-Acyl Imidazo[1,2-a]pyrimidines

A series of 3-benzoyl imidazo[1,2-a]pyrimidines, obtained from N-heteroarylformamidines in good yields, was tested in silico and in vitro for binding and inhibition of seven Candida species (Candida albicans (ATCC 10231), Candida dubliniensis (CD36), Candida glabrata (CBS138), Candida guilliermondii (ATCC 6260), Candida kefyr, Candida krusei (ATCC 6358) and Candida tropicalis (MYA-3404)). To predict binding mode and energy, each compound was docked in the active site of the lanosterol 14α-demethylase enzyme (CYP51), essential for fungal growth of Candida species. Antimycotic activity was evaluated as the 50% minimum inhibitory concentration (MIC50) for the test compounds and two reference drugs, ketoconazole and fluconazole. All test compounds had a better binding energy (range: −6.11 to −9.43 kcal/mol) than that found for the reference drugs (range: 48.93 to −6.16 kcal/mol). In general, the test compounds showed greater inhibitory activity of yeast growth than the reference drugs. Compounds 4j and 4f were the most active, indicating an important role in biological activity for the benzene ring with electron-withdrawing substituents. These compounds show the best MIC50 against C. guilliermondii and C. glabrata, respectively. The current findings suggest that the 3-benzoyl imidazo[1,2-a]pyrimidine derivatives, herein synthesized by an accessible methodology, are potential antifungal drugs.

Fungal infections have become a growing problem for patient health, implying great economic losses in hospitals [13,14]. Among the main fungal infections is invasive candidiasis, caused by species

Synthesis
Imidazo[1,2-a]pyrimidines were presently synthesized from N,N-dimethyl-N-pyrimidilformamidine, which is obtained in quantitative yield by condensation of 2-amino pyrimidine 1 and an excess of N,N-dimethylformamide dimethyl acetal under reflux conditions. The synthetic route herein employed (Scheme 1) is a variation on our previous work [29]. The treatment of amidine 2 with different phenacyl bromides under inert atmosphere at room temperature for three hours gave the 3-benzoyl derivatives in synthetically useful yields (62-98%). Thus, imidazo[1,2-a]pyrimidine derivatives with a benzoyl group having electron-donating groups (4b-e) or electron-withdrawing groups (4f-j) were prepared and fully characterized (Table 1). The 1 H NMR spectroscopic data are summarized in Table 2. Some compounds were analyzed in CDCl 3 and others in deuterated trifluoroacetic acid.  Figure 2. Binding mode on CYP51 Ca of the 3-benzoyl imidazo [1,2-a]pyrimidines and reference compounds. The five selected compounds and the reference compounds were overlaid on this receptor. CYP51 Ca is shown as a flat ribbon, while the compounds are illustrated with a stick model. Accordingly, the heme group is depicted in red, fluconazole in cyan blue, the ketoconazole in orange, 4a in dark blue, 4d in pink, 4f in yellow, 4i in green, and 4j in purple.
Once the binding mode of imidazo[1,2-a]pyrimidine 3-benzoyl derivatives to the CYP51 of Candida spp. had been established, the affinity of each of these compounds to the distinct CYP51 models was analyzed. For this purpose, 100 different conformations were obtained for each compound, selecting the one with the lowest binding energy. Each of the 3-benzoyl imidazo[1,2-a]pyrimidine derivatives proved to have a better docking energy (−6.11 to −9.43 kcal/mol) in all CYP51 of Candida spp. than that found for fluconazole (−3.16 to −5.68 kcal/mol) or ketoconazole (48.93 to −6.16 kcal/mol) ( Table 3). The current results are comparable to those found in other studies for fluconazole [32,33]. Of the five imidazo[1,2-a]pyrimidine derivatives with the best binding energies in CYP51 Ck , 4f had the lowest value (−9.43 kcal/mol). The interactions of ligands 4a, 4d, 4f, 4i and 4j with each of the CYP51 from Candida spp. (Figures S61-S103 of Supplementary Materials) were analyzed to determine the residues involved. Within the active site of each of the CYP51 enzymes, functional groups or rings in imidazo[1,2-a]pyrimidine derivatives and reference compounds (fluconazole and ketoconazole) exhibited binding with similar polar and non-polar amino acid residues, such as Tyr108, Gly283, Leu352, Thr287, Thr98, Met493, Tyr94 and Phe102.
Once the interactions between the compounds and each of the CYP51 of the Candida species had been examined, it was found that the CYP51 of Candida krusei displayed the best binding energies. Hence, CYP51 Kru was utilized as the receptor model to illustrate the interactions of the five docked compounds (Figure 3a-g).  An extension the 2D model is provided to depict the interactions between the imidazo[1,2-a] pyrimidine derivative 4f and CYP51 Ck (Figure 4). Interestingly, polar and non-polar amino acid residues are shared between the reference compounds and imidazo[1,2-a]pyrimidine derivatives. The most frequently shared residues were non-polar, including Tyr94, Leu97, Thr98, Phe102, Tyr108, Phe199, Gly273, Val274, Gly277, Gly278, and Leu344. This suggests an important role of the amino acid residues of a hydrophobic character in the binding mechanism of the compounds to CYP51 Ck [34].

Susceptibility of Candida spp. to 3-Benzoylimidazo[1,2-a]pyrimidines
We evaluated the susceptibility of the Candida strains to the imidazo[1,2-a]pyrimidine derivatives by the microdilution method [35]. The MIC50 was calculated for each of the compounds according to materials and methods ( Table 4), finding that the five test compounds have a value much lower than that determined for either fluconazole or ketoconazole. The imidazo[1,2-a]pyrimidine derivative 4f showed the lowest MIC50 values for C. dubliniensis, C. glabrata and C. krusei, while 4j had the lowest values for C. albicans and C. guilliermondii. For C. kefyr, only compound 4d presented a MIC50 slightly higher than the remaining 3-benzoyl imidazo[1,2-a]pyrimidine derivatives.

Discussion
The mechanistic pathway for the synthesis of imidazo[1,2-a]pyrimidines is herein reported. It involves the formation of the pyridinium salt followed by the intramolecular cyclization of the carbanion (adjacent to the quaternary nitrogen) on the amidine carbon via a 5-exo-trig process. Finally, aromatization took place with the loss of dimethylamine. IR and NMR ( 1 H and 13 C) spectroscopy as well as mass spectrometry analysis confirmed the structures of the synthesized compounds. With the 1 H NMR spectrum of compounds 4a-j, an abnormal chemical shift to downfield of proton 5 (~9.88-10.10 ppm) was found. This effect, also observed in 3-acyl imidazo[1,2-a]pyridines [36], is due to the interaction of C-H····O with the oxygen of the carbonyl group.
The identification of compound 4b was unequivocally established. Its molecular weight was found to be m/z = 283 (as determined by HREIMS). In the 13 C NMR spectrum, the conjugated carbonyl group is evidenced by the signal at 183.7 ppm. The HMBC experiment showed two correlations between C-8a and H-5, as well as another interaction of C-8a with H-2. Moreover, an important correlation existed between C-3 and H-2, an interaction that confirms the presence of a quaternary carbon. On the other hand, the carbonyl group exhibited two correlations, with H-12 and H-16. The proton resonances of 4e-i were shifted due to the solvent used for the dilution (Table 2). In all cases, the compounds examined in CF 3 COOD were downfield compared to those analyzed in CDCl 3 , due to the greater polarity of CF 3 COOD versus CDCl 3 . The most significant changes (over 1 ppm) were detected for the equivalent protons (H-13 and H-15) in 4c with respect to 4g, where displacement strongly depended both on the solvent [37] and the electron-withdrawing and electron-donor nature of the nitro and methoxy substituents.
To evaluate whether the compounds synthesized herein are linked to the CYP51 enzyme of yeasts, models of CYP51 of Candida spp., were generated. The CYP51 of various species of the Candida genus proved to have a high structural similarity, which indicates that this structure is conserved among species [38], and emphasizes the importance of CYP51 as a therapeutic target of new drugs with antifungal activity [34].
Ramachandran diagrams were constructed to allow for visualization of the energetically favorable regions that are established by the dihedral angles of a protein [39]. For each CYP51 analyzed, 90% of the residues fall within the favorable regions. Hence, the models obtained are of good quality and comparable to those employed elsewhere [40]. The docking studies revealed that the 3-benzoyl imidazo[1,2-a]pyrimidine derivatives act on the same target as the azoles [40]. Consequently, these derivatives likely exhibit a broad spectrum of antifungal action on various species of the Candida genus, similar to that found previously for other azoles [32].
According to the docking results (Table 3), 4f and 4d have the best binding energies in relation to CYP51 Ca and CYP51 Cd , 4i with respect to CYP51 Cg (and to a lesser extent in CYP51 Cgui and CYP51 Ck ), as well as 4a and 4j in relation to CYP51 Cke . Thus, 3-benzoyl imidazo[1,2-a]pyrimidines bound with distinct affinity in each of the CYP51 presently analyzed. Moreover, the docking studies on the different CYP51 enzymes demonstrated that similar amino acid residues interact with groups in the structures of the five 3-benzoyl imidazo[1,2-a]pyrimidine derivatives and the reference compounds. These findings indicate that the test compounds likely have a mechanism similar to that reported for azoles [33,41]. Additionally, hydrophobic, hydrophilic, and electrostatic interactions were evidenced between different substituents in the test ligands, such as methyl and halogen (Cl or F). Therefore, these connections are favored in the para position of the benzene ring.
According to the results, the best binding energies with CYP51 Ck were shown by the five imidazo[1,2-a] pyrimidine derivatives selected in this study. Compound 4f displayed the greatest affinity for the active site of the enzyme, which is due to the C-H ·····O interactions of the imidazo[1,2-a]pyrimidine ring with the polar side chain of Tyr108 and Gly273 C-H. The current affinity data imply that imidazo[1,2-a]pyrimidine derivatives inhibit CYP51 Ck more effectively than other CYP51. We propose that the cyano substituent in the structure of 4f may play an important role in the binding mode of this compound to the CYP51 Ck .
During the docking study, the heme group of the protein exhibited hydrophobic interactions with the heterocyclic rings (imidazole and pyrimidine) of the 3-benzoyl imidazo[1,2-a]pyrimidine compounds, as well as with the triazole and imidazole rings of fluconazole and ketoconazole, respectively. For imidazo[1,2-a]pyrimidine derivatives 4a and 4i, hydrogen bonds were detected between amino acid Tyr108 of CYP51 Ck and H-5 of the pyrimidine ring. Moreover, halogen bonds (Cl) were observed between compound 4i and the Tyr94, Leu97, Phe199, and Phe102 residues. Also found was another hydrogen bond between the oxygen of the carbonyl group in compound 4j and the amino acid residue Tyr281. Hydrophobic interactions could be appreciated between Gly277 and the heterocyclic moiety in compound 4a and 4i, and between this residue and the benzene ring of compound 4d. For fluconazole, hydrogen bonds were identified between Gly277 and the triazole ring, and hydrophobic interactions between Tyr94 and the other triazole moiety. For ketoconazole, hydrogen bonds were formed with the side chain of Tyr108, and type π-alkyl hydrophobic interactions occurred between Leu344 and the benzene ring, as well as between Leu344 and the Cl substituent in the benzene ring. Another important interaction was between the chlorine in the benzene ring and Ser346 [17,34,42].
Regarding the susceptibility of the Candida species to inhibition, the test compounds had lower MIC50 values than fluconazole and ketoconazole. However, if these results are compared with those reported in the literature for another class of azoles, such as voriconazole, we observe that the MIC50 for this antifungal in C. krusei is 0.25 [35], similar to that found in compound 4f, for C. glabrata and C. albicans MIC50 of 0.125 [43] and 1 [44] were found, respectively, observing that in C. glabrata, four compounds showed an MIC50 similar or better than voriconazole, while in C. albicans, the five derivatives showed better results. Regarding C. kefyr, four of the compounds showed MIC50 equivalent to that reported in voriconazole (0.125) [45,46], in C. guilliermondii [45,46], three of the compounds exhibited similar MIC50 (0.0625), while derivative 4j obtained a better value than the reference compound, and finally, for C. tropicalis [45,46], compound 4i showed a similar value (0.0312). The comparison of the MIC50 of the imidazo[1,2-a]pyrimidine derivatives indicate that different classes of azoles exhibit different inhibitory effects on the seven species of Candida, although it is observed that some are better in some species than in others, and it is necessary to search for new drugs or to improve those already proposed, that can be used in combinatorial therapy with already known compounds with the aim of improving the MIC50 reported. Given the above, these observations suggest that 3-benzoyl imidazo[1,2-a]pyrimidines may offer an alternative to azoles [33,41,47,48] in the treatment of infections caused by Candida spp.

Chemicals and Instruments
All glassware was thoroughly oven-dried. Chemicals and solvents were purchased from commercial suppliers. Melting points were determined on a Melt Temp II apparatus and are reported without correction. By using chloroform-d and CF 3 COOD, the 1 H and 13 C NMR spectra were recorded on a Varian NMR system (Palo Alto, CA, USA) at 500 MHz ( 1 H NMR) and 125 MHz ( 13 C NMR), as well as on a Bruker Advance III (Bruker Biospin, Ettlingen, Germany) at 300 MHz ( 1 H NMR) and 75 MHz ( 13 C NMR). Chemical shifts are given in parts per million with reference to internal TMS (Sigma-Aldrich, San Luis, MO, USA). EI-MS spectra were recorded on a JEOL JMS-AX505 (Akishima, Tokyo, Japan) and a JEOL GCmate spectrometer (Akishima). IR spectra were obtained with a Perkin Elmer FT-IR SPECTRUM 2000 spectrophotometer (Waltham, MA, USA).

Synthesis of Imidazo[1,2-a]pyrimidin-3-yl(phenyl)methanones (3a-j)
To a solution of N,N-dimethyl-N -(pyrimidin-2-il)formamidine (2, 1 mmol) in anhydrous N,N-dimethylformamide (7 mL), at room temperature and under N 2 atmosphere, was added the appropriate phenacyl bromide (3a-j, 1 mmol) in anhydrous N,N-dimethylformamide (7 mL). The reaction was stirred for three hours at room temperature. Then the mixture was added to a flask containing H 2 O (30 mL) and extracted with EtOAc (3 × 20 mL). The organic extracts were combined and dried (anhydrous Na 2 SO 4 ), and the solvent was removed under reduced pressure to give the title compounds 4a-d and 4j, which were further purified by column chromatography on silica gel with a mixture of hexane and EtOAc as eluent. In the case of the phenacyl bromides 3e-i, once the reaction was completed the mixture was added to a flask containing ice (30 g), then the product was precipitated, filtered under vacuum, and recrystallized from ethanol-water to deliver the title compounds 4e-i.
The quality of the models was evaluated with the DOPE (discrete optimized protein energy) method [31]. The model having the lowest DOPE score was selected for ligand-protein interaction studies. In addition, Ramachandran plots [39] were calculated by using the PDBsum database [52] for validation of the 3D structure.

Molecular Docking Studies
A ligand-protein interaction study was previously validated with molecular docking software Autodock version 4.0 (The Scripps Research Institute, La Jolla, CA, USA) [53]. The 2D structure of each ligand was sketched in editor chemical MedChem Designer 3.0 (http://www.simulationsplus.com/software/medchem-designer) and converted to 3D, mol2 format in the Open Babel GUI program [54]. Hydrogens were added to the models generated with the MolProbity program [55] and prepared with Visual Molecular Dynamics (VMD 1.9.1) [56]. All ions were added by utilizing the optimization Nanoscale Molecular Dynamics (NAMD) software program (Illinois University, Urbana and Champaign, IL. USA) [57]. The resulting structures were used for docking.
The selected test compounds and the reference compounds (fluconazole and ketoconazole) were docked in the active site of CYP51. For the preparation of docking, the following parameters were estimated in AutoDock Tools (ADT) [53]. The grid dimensions were 48 × 42 × 40 Å 3 and the points were separated by 0.375 Å. The following grid centers were calculated for CYP51 from C. albicans (

Antifungal Activity Tests
The minimum inhibitory concentration (MIC50) was determined according to CLSI guidelines in the document M27-A3 for yeasts [35]. The preparation of the dilutions of the reference compounds and five 3-benzoyl imidazo[1,2-a]pyrimidines selected was carried out by the method of serial double additive dilutions. For the water-soluble compound (fluconazole), the concentrations tested were 64-0.125 µg/mL, using RPMI 1640 as diluent with glutamine and without sodium bicarbonate, buffered with morpholino propane sulfonic acid (MOPS) at 0.164 M, adjusted to pH 7 ± 0.1, and with 0.2% glucose. For the water insoluble antifungals (ketoconazole and the test compounds), the concentrations ranged from 16 to 0.0312 µg/mL, using DMSO as diluent.
For the preparation of the inoculum of Candida spp., the optical density was adjusted in a spectrophotometer (530 nm) to 0.5 McFarland. Subsequently, a 1:1000 dilution was made with RPMI medium (at concentrations of 1 × 10 3 -5 × 10 3 ). The antifungal assay was performed with the latter dilution. The 96-well plates were inoculated with 100 µL of yeast suspension. RPMI was utilized as the sterility control and DMSO without antifungal as the growth control. The plates were incubated for 24 h at 37 • C. The optical density was determined in a Multiskan™ GO microplate spectrophotometer by agitation of the plates to obtain a homogeneous suspension, followed by a spectrophotometric reading at 530 nm. The MIC50 is the antifungal concentration whose optical density equals 50% of the growth in the control well. The value reported herein represents the average of three different experiments.

Conclusions
A series of 3-benzoyl imidazo[1,2-a]pyrimidines was synthesized and tested in silico and in vitro. By docking the test compounds in the active site of fungal CYP51, the binding mode and binding energy could be predicted in each case. The MIC50 was determined for each compound as well as for two reference compounds (fluconazole and ketoconazole), in order to evaluate the respective capacity for growth inhibition of distinct Candida species. The docking results show that for each species of Candida spp., the binding mode of each test compound shares at least three amino acid residues with the reference drugs. We analyzed and described the interactions of the electron-donor and electron-withdrawing substituents in the aromatic ring with key amino acid side chains in the active site of CYP450. Although the antifungal activity of imidazo[1,2-a]pyrimidines has been studied, the present findings should certainly be instrumental in the design and development of new antifungal drugs derived from 3-benzoyl imidazo[1,2-a]pyrimidines. Further research is needed for this purpose, and to test new derivatives in combination with conventional drugs in animal models, and possibly later, in clinical trials as promising candidates for antifungal therapy.
Supplementary Materials: The supplementary materials are available online. Figures S1-S103.